AutoCAD interviews should reveal whether a candidate can turn design intent into accurate, coordinated, readable drawings—not just recall command names. Recruiters can assess precision through units and coordinates, production judgment through layers and references, and delivery quality through dimensions, layouts, and plotted output. A strong candidate also explains how they verify changes without damaging shared project information.

This guide to 100+ AutoCAD Interview Questions and Answers contains 120 questions across beginner, intermediate, and experienced levels. It covers everyday drafting, reusable content, external references, annotation, plotting, collaboration, 3D modeling, troubleshooting, and automation. The examples distinguish full AutoCAD desktop capabilities from features that vary in AutoCAD LT, AutoCAD for Mac, AutoCAD web, and specialized toolsets.

How can recruiters detect potential AI cheating, deepfakes, impersonation, and unauthorized assistance during AutoCAD interviews?

An AutoCAD assessment might ask a candidate to diagnose a wrongly scaled reference, repair a dimension workflow, or produce two correctly scaled views on one sheet. Unauthorized assistance can hide gaps in the practical reasoning needed to preserve geometry, coordinate drawings, and issue dependable documentation.

Sherlock AI helps recruiters detect potential AI-assisted cheating, deepfakes, candidate impersonation, and unauthorized real-time assistance during AutoCAD interviews.

Combine integrity checks with a short drawing exercise and ask the candidate to explain their units, snaps, reference strategy, and final plot checks. Evaluate suspicious signals in context rather than treating a signal alone as proof of misconduct.

AutoCAD Interview Questions

Choose questions that fit the actual drafting role and installed AutoCAD edition. State drawing units, deliverable standards, operating system, and project constraints before judging a workflow; a command available in full AutoCAD for Windows may not be available in another edition.

AutoCAD Interview Questions for Beginners

Assess precise drawing habits, basic editing, layers, annotation, reusable geometry, and the relationship between model space, layouts, and plotted output.

1. What is AutoCAD, and what work is it designed to support?

2. How do AutoCAD, AutoCAD LT, AutoCAD for Mac, and AutoCAD on the web differ?

3. What are DWG, DXF, and DWT files used for?

4. What should you check before starting a new AutoCAD drawing?

5. Does changing UNITS resize existing geometry?

6. Why should model geometry normally be drawn at full size?

7. What is the difference between absolute and relative coordinates?

8. How would you enter a line using relative polar coordinates?

9. What is the difference between WCS and UCS?

10. What are object snaps, and why are they more reliable than clicking by eye?

11. How do grid snap, Ortho mode, and polar tracking differ?

12. What is object snap tracking used for?

13. How do you use the command line effectively?

14. What is Dynamic Input, and how can it cause coordinate-entry mistakes?

15. What is the difference between a window selection and a crossing selection?

16. How do LINE and PLINE differ?

17. How would you draw a precise 120-by-80 rectangular outline?

18. How do radius and diameter inputs differ when creating a circle?

19. What is the difference between MOVE and COPY?

20. When would you use OFFSET instead of COPY?

21. How do TRIM and EXTEND work?

22. How do FILLET and CHAMFER differ?

23. How would you rotate or mirror geometry accurately?

24. How is STRETCH different from MOVE or SCALE?

25. What are layers, and how should you organize a drawing with them?

26. What does ByLayer mean, and when would you use a property override?

27. What is the difference between turning a layer off, freezing it, and locking it?

28. How do linetype, lineweight, and polyline width differ?

29. What are grips, and when is the Properties palette useful?

30. How would you measure distance, perimeter, and area accurately?

31. What is a hatch, and what should you check when it fails to fill an area?

32. How do TEXT and MTEXT differ?

33. What is a dimension style, and why should dimensions remain associated with geometry?

34. When would you use linear, aligned, angular, radius, or diameter dimensions?

35. What is a block, and why use one instead of copying loose objects?

36. What are block attributes?

37. What is an external reference, and how is it different from an inserted block?

38. What is the difference between model space and paper space?

39. How do you set and protect a layout viewport scale?

40. What checks would you make before saving and issuing a drawing or PDF?

AutoCAD Interview Questions for Intermediate

Explore reliable production workflows for blocks, attributes, references, annotations, viewports, plot styles, and coordinated drawing delivery.

41. How would you redefine a reusable block without losing control of existing drawings?

42. How do parameters and actions work together in a dynamic block?

43. How would you design a dynamic block that changes length without stretching every component?

44. When would you use dynamic-block visibility states instead of separate block definitions?

45. How do block attributes differ from ordinary text?

46. What precautions would you take before using ATTSYNC?

47. How would you produce a reliable block-attribute schedule?

48. How would you prevent inserted blocks from arriving at the wrong size or orientation?

49. What is the practical difference between attaching and overlaying an xref?

50. How would you repair a missing xref without accidentally referencing the wrong file?

51. What changes when you bind an xref, and how do Bind and Insert differ?

52. How would you control xref layer appearance without changing the consultant’s source drawing?

53. When would you use XCLIP rather than trim or explode an xref?

54. How would you coordinate an xref update while other people are working on the project?

55. How would you diagnose a consultant background that is offset, rotated, or scaled incorrectly?

56. How would you set up annotative text for two different viewport scales?

57. Why might annotative text appear in one viewport but disappear in another?

58. How do associative dimensions differ from ordinary dimension objects?

59. How would you distinguish dimension scale from a unit-conversion factor?

60. How would you use fields to reduce stale drawing annotations?

61. How would you calculate a metric viewport scale rather than guessing it?

62. How would you set a viewport to an imperial architectural scale?

63. Why should you lock a layout viewport, and what does locking not do?

64. How would you show different layer visibility in two viewports on the same sheet?

65. Would you dimension in model space or paper space?

66. What is the difference between CTB and STB plotting?

67. How would you convert between CTB and STB without changing the drawing’s appearance unexpectedly?

68. What would you include in a reusable page setup?

69. How would you investigate a PDF that does not match the drawing on screen?

70. How do LTSCALE, object linetype scale, and PSLTSCALE interact?

71. Why can a dashed polyline look different from a dashed line with the same linetype?

72. How would you troubleshoot a hatch that leaks or fails to find its boundary?

73. When should a hatch be associative, and how would you verify it?

74. How would you clean up imported 2D linework without damaging its intended geometry?

75. How can ALIGN combine moving, rotating, and scaling geometry?

76. How would you use geometric and dimensional constraints without overconstraining a sketch?

77. When is an associative array better than repeated COPY operations?

78. What would you check before sending a DWG package to another team?

79. How would you use DWG Compare to review a revision?

80. A drawing looks correct in Model space but is wrong on one sheet. How would you isolate the cause?

AutoCAD Interview Questions for Experienced

Use these scenarios to assess CAD standards, recovery decisions, performance diagnosis, automation, 3D modeling, interoperability, and release responsibility.

81. How would you establish CAD standards for a multidisciplinary team?

82. What is the difference between a DWT template and a DWS standards file?

83. How would you use CHECKSTANDARDS without damaging valid project exceptions?

84. How would you migrate an old project to a new company template?

85. How would you organize drawing ownership when several people work on one project?

86. How do you prevent template and support-file changes from breaking existing drawings?

87. How would you verify a CAD deliverable beyond checking that it opens?

88. How would you diagnose a large drawing that has become slow?

89. When would you change hardware acceleration settings?

90. Why can geometry far from the origin cause problems, and how would you respond?

91. How would you investigate unexpectedly huge drawing extents?

92. What is the difference between PURGE and deleting drawing objects?

93. How would you remove duplicate geometry safely?

94. How do AUDIT, RECOVER, and RECOVERALL differ?

95. How would you recover work after AutoCAD crashes?

96. What would you do if a repaired drawing opens but important objects are missing?

97. How would you distinguish a network problem from a drawing-file problem?

98. When would you choose an AutoCAD script instead of AutoLISP?

99. How would you make command automation less dependent on workstation settings?

100. What error-handling responsibilities should an AutoLISP routine have?

101. Which AutoCAD automation interfaces are portable across products and platforms?

102. How would you safely load a third-party AutoLISP routine or plug-in?

103. How would you validate a batch routine before using it on hundreds of drawings?

104. How do you approach an AutoCAD plug-in upgrade between releases?

105. How would you keep automated drawing changes auditable?

106. How do you choose between solid, surface, and mesh modeling?

107. How would you establish a reliable UCS for modeling on an inclined face?

108. Why might EXTRUDE create a surface instead of a solid?

109. When would you choose REVOLVE instead of EXTRUDE?

110. How would you troubleshoot a failed SWEEP operation?

111. How do UNION, SUBTRACT, and INTERSECT differ in solid modeling?

112. How would you verify the volume of a modeled part?

113. What is the difference between a live section and permanently cutting a model?

114. How would you create dependable 2D documentation from a 3D model?

115. Why is a visually correct 3D model not necessarily a valid deliverable?

116. How would you handle proxy objects received from another Autodesk product?

117. How would you prepare a drawing for a supplier using an older or different CAD product?

118. How would you investigate a supplier drawing that arrives at the wrong size and location?

119. What would you do if the issued PDF and the current DWG disagree?

120. How would you assess whether an experienced AutoCAD candidate can deliver reliable work?

AutoCAD Interview Questions and Answers

These answers follow the question bank in the same order and provide direct responses candidates can adapt in interviews. Command sequences and calculations are illustrative. Check command prompts, settings, feature availability, and results in the installed release using a disposable drawing before applying a workflow to project files.

AutoCAD Interview Questions and Answers for Beginners

The beginner answers connect core commands to accuracy and verification. They distinguish model geometry from its display, dimensions, and plotted representation.

1. What is AutoCAD, and what work is it designed to support?

AutoCAD is a computer-aided design application for creating and documenting precise geometry. I would use it for tasks such as a dimensioned component drawing, a floor-plan detail, or a coordinated set of construction sheets. The deliverable is not just a picture: objects have coordinates and properties that can be measured, edited, reused, and plotted at a controlled scale.

A useful distinction is between drafting competence and engineering approval. Drawing a detail accurately does not establish that its loads, materials, clearances, or regulatory requirements are correct. Those decisions need the appropriate design process and review. In an interview, I would describe the actual drawing types I can produce, the standards I follow, and how I check the resulting DWG and plotted sheet rather than claiming that knowing commands is equivalent to being qualified in every discipline.

2. How do AutoCAD, AutoCAD LT, AutoCAD for Mac, and AutoCAD on the web differ?

Full desktop AutoCAD provides 2D drafting and a broader set of 3D modeling and customization capabilities. AutoCAD LT is primarily a 2D drafting and documentation product; I would not promise full AutoCAD solid-modeling functionality in LT. Mac and web editions have their own interfaces and supported feature sets, so a Windows workflow is not automatically portable.

For a role, I would confirm the exact product, release, operating system, and required commands before choosing a workflow. Specialized industry toolsets are distinct from plain AutoCAD and have platform requirements of their own. Similarly, it is outdated to say that every LT release lacks AutoLISP: Windows LT added support in the 2024 generation, but that does not imply identical API support across LT, Mac, and web. I would check the applicable product documentation rather than treating an AutoCAD subscription as proof of feature parity.

3. What are DWG, DXF, and DWT files used for?

DWG is the native drawing format I normally edit and deliver. DXF is a drawing interchange format used when another application or downstream process requests it. DWT is a drawing template used to start new drawings with an approved initial setup rather than rebuilding that setup each time.

For example, I might create a component in a project DWG, export a required DXF version for a fabrication workflow, and start the next component from a DWT containing the project's layers and styles. Exporting does not guarantee that every object, font, or application-specific feature will transfer perfectly. I would reopen the exchange file and verify units, geometry, text, and the recipient's supported version. A template is also not a live link: changing the DWT later does not automatically update drawings already created from it.

4. What should you check before starting a new AutoCAD drawing?

I would first establish the intended units, coordinate basis, deliverable, and project template. Then I would confirm layers, text and dimension styles, expected sheet sizes, and the approved location for drawing files and references. This prevents avoidable rework such as discovering at plotting time that a drawing was modeled in feet but interpreted as inches.

For a small metric bracket, I might agree that one drawing unit represents one millimeter, use a suitable metric template, draw the actual part size, and reserve sheet scale for the layout. I would also save a named working DWG early instead of leaving the file as an unnamed session. The template is a starting point, not evidence that everything is correct: a known-length test and a quick inspection of the intended styles are inexpensive checks before substantial geometry is created.

5. Does changing UNITS resize existing geometry?

Changing the display format or precision in UNITS does not resize existing objects. A line whose geometric length is 100 remains 100 drawing units even if I change the displayed decimal places or switch the displayed length format. I distinguish that display setting from the physical meaning the project assigns to a drawing unit.

If an existing inch-based model must become millimeter-based, I first confirm the source units and make a controlled copy. The required geometric conversion is 25.4 millimeters per inch, so a 10-unit inch length should become 254 millimeter units. I would use an appropriate conversion workflow, check dimensions and references, and avoid applying the factor twice. Insertion-unit settings can affect how other content is scaled when brought into the drawing; they are not a retroactive repair for incorrectly sized geometry.

6. Why should model geometry normally be drawn at full size?

Full-size modeling keeps geometry and measurements consistent with the chosen drawing unit. If a wall is 6,000 millimeters long, I draw it 6,000 units long in a millimeter-based model, not 60 units because the sheet will be printed at 1:100. The viewport or plotting setup controls the representation on paper.

For that example, the 6,000-unit wall occupies 60 millimeters on a correctly configured 1:100 view when model and paper units are millimeters. This separation allows the same model to appear at 1:100 on an overall sheet and 1:20 in a detail without maintaining differently scaled copies of the design. I would verify the printed output as well as the model: a correct model can still be printed incorrectly if the PDF viewer or printer applies an additional fit-to-page scale.

7. What is the difference between absolute and relative coordinates?

Absolute coordinates locate a point relative to the current coordinate-system origin. Relative coordinates locate it relative to the last point supplied to the command. At the command line, with Dynamic Input off, 30,20 means an absolute XY position, while @30,20 means an offset of 30 in X and 20 in Y from the previous point.

For example, starting at 10,10 and entering @30,20 ends at 40,30; it does not end at 30,20. I use absolute coordinates for known survey or layout positions and relative coordinates for local dimensions between features. Before entering a sequence, I check the UCS and input mode. Dynamic Input can change the interpretation of an unprefixed entry, so I do not assume that the same bare numbers mean the same thing in every input context.

8. How would you enter a line using relative polar coordinates?

I would specify the start point, then enter the next point as a distance and angle relative to that start. For example, with the usual degree-based angle settings, positive angles counterclockwise, and the UCS aligned as intended, @100<30 creates a 100-unit segment at 30 degrees from the positive UCS X direction.

Its XY displacement is approximately 86.603 units in X and 50 units in Y. That gives me a useful independent check if I inspect the endpoints. I would not confuse the length of the sloped segment with its horizontal projection. This entry also depends on the configured angle direction and base angle, so I state those assumptions in a reproducible example. For a one-off known bearing, explicit polar entry is usually clearer than trying to point the cursor close to the desired direction.

9. What is the difference between WCS and UCS?

The World Coordinate System is the fixed reference system for the drawing. A User Coordinate System is a movable and rotatable working coordinate system that makes a particular task easier. Changing the UCS changes how I enter and interpret working coordinates; it is not the same operation as moving the design geometry.

For example, on a plan with a wing rotated 25 degrees, I can align a UCS with that wing so local horizontal and vertical construction is convenient. The objects still have positions in the drawing's world system. Before exchanging coordinates or reporting a point, I identify which coordinate system is intended. To return to the world working orientation, I can use UCS and its World option. A separate view change may be needed if I also want the screen view aligned with that UCS.

10. What are object snaps, and why are they more reliable than clicking by eye?

Object snaps locate defined geometric points such as an endpoint, midpoint, center, or intersection. They let me attach new geometry to an actual feature rather than a screen pixel that only looks close. For example, a hole center should use the intended center or constructed intersection, not a visual estimate made at the current zoom.

I keep a small useful set of running snaps and apply a temporary override when needed. If Endpoint and Nearest compete in a dense detail, I explicitly request Endpoint and confirm the marker before accepting the point. Snapping accurately to the wrong object is still an error, so I zoom in, inspect overlapping geometry, and verify critical distances. Object snapping helps preserve design relationships; it does not decide which relationship the design actually requires.

11. How do grid snap, Ortho mode, and polar tracking differ?

Grid snap restricts point movement to configured increments. Ortho restricts cursor-based movement to horizontal or vertical directions relative to the current UCS. Polar tracking provides alignment guidance at configured angles, such as 30 or 45 degrees, while allowing me to specify a distance along the chosen direction.

I use grid snap for a genuinely modular grid, Ortho for orthogonal construction, and polar tracking for repeated directional work. None is a substitute for checking a connection to existing geometry with an appropriate object snap. For example, a component endpoint may fall between the grid increments, so forcing grid snap can move the intended point. I also remember that Ortho and polar tracking are alternative modes rather than two independent constraints that can simply be combined for every task; I check the active mode before entering a distance.

12. What is object snap tracking used for?

Object snap tracking helps locate a new point by aligning it with existing snap points without drawing permanent construction lines. I acquire the relevant points by hovering over their snap markers, then use the temporary alignment paths to position the new point.

For instance, a circle center may need to share the X position of one endpoint and the Y position of another. Tracking can locate that intersection of alignments while leaving the drawing free of temporary lines. I make sure the desired points have actually been acquired and that the tracking directions match the intended UCS. In a crowded drawing, an unintended acquired point can produce a plausible but wrong result. I would confirm the new center with coordinates or measurements rather than treating the appearance of a tracking path as a complete quality check.

13. How do you use the command line effectively?

I read the current prompt before entering a value or selecting an object. A command is a sequence of states: a number might be a radius at one prompt, a scale factor at another, or an option value at a third. Reading the prompt prevents a remembered keystroke sequence from silently doing the wrong thing.

I use full command names when documenting a procedure because aliases can be customized. Bracketed options identify alternatives to the main action, and the displayed default can often be accepted with Enter. If I lose track of the command state, I cancel with Esc and restart deliberately rather than keep typing. I also inspect command history for measurements and warnings. Ribbon buttons and command-line entry are different ways of starting many of the same operations, not competing definitions of drafting skill.

14. What is Dynamic Input, and how can it cause coordinate-entry mistakes?

Dynamic Input displays prompts and coordinate or dimension entry fields near the cursor. It can make drafting faster, but I need to know whether the field expects relative coordinates, absolute coordinates, or a direct distance. An unprefixed entry can be interpreted differently from command-line coordinate entry with Dynamic Input disabled.

For a controlled exercise, I can disable Dynamic Input and use explicit command-line coordinates. If I use pointer input, an absolute point can be identified with the # prefix, while @ explicitly identifies a relative coordinate entry. For example, confusing an absolute location with an offset is enough to place a hole far from its intended position even though both input values were typed correctly. I check the result using Properties or a measurement and restore the team's usual input settings after a temporary change.

15. What is the difference between a window selection and a crossing selection?

A window selection includes only objects completely enclosed by its boundary. A crossing selection includes objects inside the boundary and objects that it touches. In the usual rectangular selection behavior, dragging or specifying the rectangle left to right gives a window; right to left gives a crossing selection.

I choose the method according to the edit. A window is useful when I want a contained group without catching adjacent long lines. A crossing selection is useful when I need to select a row of objects quickly, and it is particularly important for vertex-based STRETCH behavior. Before accepting a destructive operation, I inspect the highlighted set and remove unintended objects. Selection settings and lasso behavior can vary, so for an unambiguous command-line workflow I can request the Window or Crossing option explicitly rather than relying only on gesture direction.

16. How do LINE and PLINE differ?

LINE creates separate line objects, even when I draw several connected segments in one command. PLINE creates a polyline object that can contain connected straight and arc segments. I choose a polyline when a boundary or path should behave as one object, and separate lines when independent segments are more appropriate.

For example, a closed room boundary is easier to select and inspect as one polyline than as four unrelated lines. I still check that it is actually closed: endpoints that look coincident do not necessarily mean the polyline's Closed property is set. A polyline can also have width, which is a geometric property rather than merely a plotted lineweight. Exploding it changes its structure, so I would not explode a boundary just to make a routine edit that PEDIT or grips can perform while preserving the intended object.

17. How would you draw a precise 120-by-80 rectangular outline?

I would first confirm that the current UCS is the intended drafting plane and that the drawing unit has been agreed. With Dynamic Input off for unambiguous command-line entry, I can start RECTANG, enter 0,0 as the first corner, and enter @120,80 as the opposite corner. That creates the intended dimensions without depending on mouse accuracy.

I would check that retained RECTANG settings such as fillet radius, chamfer, elevation, and width are appropriate before drawing; previous command settings can affect a new rectangle. For a simple sharp-cornered outline, those extra effects should not be active. Finally, I would inspect the geometry: the side lengths should be 120 and 80, the perimeter 400, and the area 9,600 square drawing units. The area represents square millimeters only if a drawing unit represents a millimeter.

18. How do radius and diameter inputs differ when creating a circle?

A radius is the distance from the circle's center to its circumference; a diameter is twice that distance. In the normal center-based CIRCLE workflow, I specify the center and then either give the radius or select the Diameter option and give the diameter. I read the active prompt so that a diameter is not accidentally entered as a radius.

For a 20-millimeter-diameter hole in a millimeter-based drawing, the radius is 10 units. Entering 20 at a radius prompt would create a 40-millimeter-diameter circle. After creation, Properties or a diameter dimension gives a quick check. I would also place the center using coordinates or a precise snap. Correct size does not compensate for a misplaced center, and visually matching a nearby circle is not reliable enough for a fabrication drawing.

19. What is the difference between MOVE and COPY?

MOVE changes the position of the selected objects. COPY creates additional objects while retaining the originals. Both can use a base point and a destination point to define the same displacement, so I choose the command based on whether the original geometry should remain.

For example, to shift a symbol 250 units in X and 100 units in Y, I select it, finish the selection, specify a base point, and use @250,100 as the second point in a command-line coordinate workflow. With MOVE the original location becomes empty; with COPY it remains occupied. The base point does not have to lie on the selected object, but using a meaningful snap point makes placement easier to explain and verify. After copying repeated parts, I also check that I have not left unintended overlapping duplicates at the original location.

20. When would you use OFFSET instead of COPY?

OFFSET creates a parallel or concentric result at a specified perpendicular distance from supported source geometry. COPY translates geometry by a displacement. Those operations can look similar for a straight horizontal line but are not equivalent for a bent boundary or a circle.

For example, to draw the second face of a 150-unit wall along a polyline, I would use OFFSET with a distance of 150 and select the intended side. Copying the polyline 150 units in X would not maintain a 150-unit perpendicular separation around every change in direction. I would inspect corners and short segments after offsetting because tight geometry can produce an unwanted or impossible result. The layer and erase options also matter: I confirm whether the new object should remain on the source layer and whether the source should be retained.

21. How do TRIM and EXTEND work?

TRIM removes portions of objects at cutting boundaries; EXTEND lengthens supported objects to boundary geometry. I use them to clean intersections after creating the main construction, rather than trying to click every endpoint perfectly during the initial sketch.

For example, after offsetting two wall faces, I can trim the portions crossing a doorway and extend a short line to the intended wall edge. I select near the part to remove or the end to lengthen so that the intended side is clear. Modern AutoCAD provides Quick and Standard operating modes, which differ in how boundaries are selected. Therefore, I inspect the command's current mode rather than assuming an older sequence of Enter presses applies. If the result is unexpected, I check the boundary, elevation, and projection settings before repeatedly editing until the drawing merely looks right.

22. How do FILLET and CHAMFER differ?

FILLET joins suitable objects with a tangent arc of a specified radius. CHAMFER creates a straight beveled transition, usually controlled by two distances or a distance and an angle. I choose the operation according to the actual design requirement rather than because both can remove a sharp corner.

For a plate requiring a 5-unit corner radius, I set the FILLET radius to 5 and select the intended edges near the corner. For a 5-by-5 bevel on perpendicular edges, I use the corresponding chamfer distances. These do not produce the same shape or remove the same material. I also check trim settings, selection positions, and available edge length. A zero-radius fillet can be useful for making a sharp intersection, but it is not a substitute for specifying a real radius when the drawing requires a rounded corner.

23. How would you rotate or mirror geometry accurately?

For ROTATE, I select the objects, choose a meaningful base point, and enter the required angle or use the Reference option to align a known direction. For MIRROR, I define the mirror axis with two precise points and decide explicitly whether to retain the source objects.

For example, a symmetric bracket can be drafted as one half and mirrored about its centerline. I use endpoint or midpoint snaps to define that axis so a small cursor error does not create a nearly symmetric part. I would inspect text and annotations separately: mirrored geometry may be correct while a label's orientation is unsuitable, and text behavior depends on settings and object context. After either operation, I verify key locations and orientation rather than assuming a visually plausible result establishes the intended handedness or angle.

24. How is STRETCH different from MOVE or SCALE?

STRETCH changes selected portions of supported geometry while leaving other portions in place. MOVE translates the selection without changing its shape, while SCALE changes size about a base point. For STRETCH, the crossing selection is central: included vertices move, while vertices outside the crossing boundary can remain fixed; objects wholly included may move as complete objects.

To lengthen a rectangular outline from 120 to 150 units without changing its 80-unit height, I can cross-select only the right-hand vertices and stretch them 30 units horizontally. Scaling the whole rectangle by 1.25 would also increase its height to 100, which is a different result. I inspect the highlighted selection and verify dimensions afterward. Not every object type stretches in the same way, so I would not expect a circle to turn into an ellipse through this operation.

25. What are layers, and how should you organize a drawing with them?

Layers organize objects by purpose and provide shared control of visibility, plotting, and graphical properties. I would normally separate meaningful categories such as outlines, hidden features, centerlines, dimensions, and reference information, using the project's naming convention.

The goal is controllable information, not one layer per object or arbitrary colors without a plan. For example, a reviewer should be able to hide dimensions temporarily without removing the model geometry. I set the intended current layer before drafting and check new objects afterward when object-type default layers are configured. I also avoid solving a visibility problem by moving unrelated objects onto a random visible layer. That may make today's screen look correct while breaking later sheet visibility, plotting, or team coordination. A layer convention works best when object properties consistently inherit the approved settings.

26. What does ByLayer mean, and when would you use a property override?

ByLayer means an object's property is inherited from its layer rather than assigned a separate explicit value. I normally use it for color, linetype, and lineweight so a controlled layer change can update related objects consistently. An explicit override is a deliberate exception, not my default drafting habit.

For example, if all hidden edges should use the same dashed linetype, I assign that linetype to their layer and leave the objects ByLayer. If one edge is explicitly set to Continuous, changing the layer later will not repair that exception. I would inspect Properties and remove accidental overrides instead of repeatedly changing the layer. Exceptions can be appropriate for a documented presentation requirement, but I make them intentional and easy to find. ByLayer is also distinct from ByBlock, which is relevant to property inheritance in inserted block content.

27. What is the difference between turning a layer off, freezing it, and locking it?

Turning a layer off hides its objects. Freezing also suppresses display but excludes the frozen content from operations such as drawing-extents calculation and can reduce regeneration work. Locking protects objects from normal editing while allowing them to remain visible and usable as reference geometry.

I might lock a background layer while adding dimensions, turn a temporary category off to declutter a view, and freeze large unused reference categories when appropriate. These are different intentions: locking is not a visibility control, and hiding is not deletion. Before a deliverable check, I inspect the relevant states so a missing object is not mistaken for missing design information. In layouts, viewport-specific freezing offers another level of control, so I also distinguish a layer hidden in one viewport from a layer hidden throughout the drawing.

28. How do linetype, lineweight, and polyline width differ?

Linetype controls the pattern along an object, such as continuous, dashed, or centerline. Lineweight controls its plotted stroke weight. Polyline width is a geometric width assigned to the polyline itself. They can all influence appearance, but they represent different things and should not be substituted casually.

For example, a centerline should use the approved center linetype and a suitable lineweight. Making it a wide polyline would not create the required dash pattern. A 0.35-millimeter plotted lineweight is also not the same as a 0.35-unit-wide modeled strip. I inspect both model and layout display, and check a PDF at the intended sheet scale. If a dashed line looks continuous, I investigate linetype loading and scale rather than breaking the line into hand-drawn dashes that are harder to maintain.

29. What are grips, and when is the Properties palette useful?

Grips are editing handles displayed on selected objects. They provide quick access to changes such as moving a vertex, adjusting a circle radius, or repositioning an object, depending on object type and the active grip operation. The Properties palette exposes actual object values and settings, which is useful when a change needs a precise numerical value or a diagnosis rather than an approximate drag.

For example, if a circle should have radius 12.5, I can inspect and edit that value in Properties instead of dragging until it appears correct. If several objects are selected, I check which properties are shared before making a batch change. I also distinguish selected-object properties from current defaults when nothing is selected: changing a default can affect newly created objects without repairing existing ones. After a grip edit, I still verify the resulting dimensions.

30. How would you measure distance, perimeter, and area accurately?

I use measurement tools and object properties rather than infer dimensions from screen appearance. DIST measures between two specified points, while a closed polyline's properties or the AREA workflow can provide boundary information. I choose precise snaps and confirm that the intended boundary is planar and complete.

For a 120-by-80 rectangle, the expected perimeter is 2 × (120 + 80) = 400 and the area is 120 × 80 = 9,600 square drawing units. That simple independent calculation is useful for checking whether I selected the intended object. For an irregular room, I inspect gaps, overlaps, and internal exclusions instead of trusting a number merely because AutoCAD reports it. Measurement units follow the model's unit convention, and displayed rounding can conceal small differences. I increase displayed precision when necessary without confusing it with a change to geometry.

31. What is a hatch, and what should you check when it fails to fill an area?

A hatch is an object representing a pattern or fill within a boundary. I use it to communicate material or section information, not as a replacement for accurate boundary geometry. An associative hatch can update when its associated boundary changes, provided that relationship remains valid.

If a fill fails or floods an unintended area, I first inspect the boundary for gaps, overlapping segments, unexpected elevations, and internal islands. I then check the selected pattern, scale, and origin. A very large or very small pattern scale can make a valid hatch look empty or almost solid. For a simple plate section, I would confirm the outline is closed and coplanar, create the hatch, and inspect its perimeter and holes. I would not conceal an open boundary by casually increasing gap tolerance without understanding whether that changes the intended region.

32. How do TEXT and MTEXT differ?

TEXT is suitable for separate single-line text objects, while MTEXT is suited to paragraphs and notes that need wrapping and richer formatting. I would use MTEXT for a multi-sentence fabrication note and simple single-line text where independent labels are the intended structure.

The choice affects editing as well as appearance. A paragraph built from many unrelated text lines is harder to reflow when wording changes, while an MTEXT box can wrap within its defined width. I would use approved text styles and check font availability rather than manually formatting every note. I also inspect the plotted height, justification, and drawing scale. Text that is easy to read while zoomed in can be unreadable on the issued sheet, so the final PDF—not only the editing view—is part of the check.

33. What is a dimension style, and why should dimensions remain associated with geometry?

A dimension style groups settings such as text appearance, arrows, units, precision, and spacing so dimensions communicate consistently. Association links a dimension to the geometry it measures, allowing the measurement to update when the associated feature changes.

I would use the approved style and attach dimensions to actual features with precise object snaps. For example, if a bracket length changes from 120 to 150, an appropriately associated dimension should reflect the revised geometry. I still verify it after editing because associations can be lost or attached to the wrong feature. I avoid replacing a measured value with arbitrary text to make the sheet appear correct; that hides a disagreement rather than fixing it. A dimension is documentation, not generally a command that forces the underlying geometry to the displayed size, unless a separate constraint workflow has been deliberately used.

34. When would you use linear, aligned, angular, radius, or diameter dimensions?

I choose the dimension type that expresses the intended requirement. A linear dimension reports a horizontal or vertical projected distance. An aligned dimension follows the distance between its definition points. Angular dimensions report an angle, while radius and diameter dimensions describe circular features in their respective terms.

For example, a sloped segment with endpoint displacement of 30 units in X and 40 in Y has a 50-unit aligned length, but its horizontal linear dimension is 30. Those are not conflicting measurements; they describe different properties. A drilled hole is commonly called out by diameter, while a corner round may be specified by radius according to the drawing convention. I also choose an appropriate datum and avoid redundant chains that can create ambiguity. The objective is an unambiguous requirement, not simply placing every available dimension type on the sheet.

35. What is a block, and why use one instead of copying loose objects?

A block definition groups reusable geometry under a name, and inserted block references place instances of that definition in the drawing. I use blocks for repeated symbols, components, or title blocks because they preserve a consistent structure and can be updated through the shared definition.

For example, twenty instances of one valve symbol should not become twenty independently edited collections of lines. If the approved symbol changes, updating the definition can update its references in that drawing. Each reference can still have its own insertion location, rotation, and allowed scale. I choose a useful base point, verify units, and inspect properties before adding it to a library. Exploding every insertion discards much of the benefit and can damage attributes or other behavior, so I edit the definition or the reference properties when that is the appropriate change.

36. What are block attributes?

Attributes are named data fields associated with block references. They can carry information such as an equipment tag, part number, or sheet identifier, allowing repeated geometry to have different instance-specific values. The attribute definition establishes properties such as its tag and prompt; the inserted instance carries the corresponding value.

For example, the same pump symbol may appear as P-101 and P-102 without creating a separate block definition for each tag. I use meaningful, consistent attribute tags and verify the values rather than relying solely on visible text. Attributes can be invisible or formatted differently, and a displayed label is not automatically a reliable unique identifier. If a block's attribute definition changes later, I check how existing references are synchronized instead of assuming every previously inserted value and position will automatically match the new setup.

37. What is an external reference, and how is it different from an inserted block?

A drawing external reference, or xref, links a host drawing to another DWG. The referenced design remains in its source file and can be reloaded when that source changes. Inserting a drawing as a block instead brings its content into the host's block structure rather than maintaining the same external-file relationship.

I would xref an architectural background into a services drawing so each discipline can maintain its own source. I would use a block for a reusable symbol that belongs within the drawing. An xref is not a guarantee of coordination: the path, source version, units, insertion point, scale, and rotation must all be correct. Before handoff, I check that the recipient can resolve the required files. Sending only the host DWG can leave a drawing apparently empty of essential background information even though the host file itself opens successfully.

38. What is the difference between model space and paper space?

Model space contains the design at its real modeled size. Paper space is used to compose a sheet with items such as a title block, sheet notes, and layout viewports that display the model. A layout viewport is a view of model space, not an independent duplicate of its geometry.

For example, one sheet can show an overall plan at 1:100 and a detail at 1:20 while both reference the same model. I keep the model's physical dimensions unchanged and set each view's scale deliberately. I also pay attention to which space is active before editing: double-clicking inside a layout viewport can put me into model editing through that viewport. If I intend to move a sheet note or viewport boundary, I return to paper space first rather than accidentally moving design geometry.

39. How do you set and protect a layout viewport scale?

I select the layout viewport and assign the intended standard or custom scale, then position the view and lock its display. Locking helps prevent an accidental zoom inside the viewport from changing the established sheet scale. I keep its boundary on an appropriate non-plotting layer when the border should not appear in the output.

For a millimeter-based model and millimeter paper space, a 1:50 viewport displays 5,000 model units as 100 millimeters on the sheet. The layout itself is normally plotted at 1:1; I do not apply 1:50 again to the entire layout. That would double-scale the output. I verify the actual paper size and plotted measurement as well as the scale label, since a manually written '1:50' note does not control the viewport or stop a printer from fitting the PDF to a different sheet.

40. What checks would you make before saving and issuing a drawing or PDF?

I would check geometry, dimensions, notes, layer visibility, references, and the intended revision first. Then I would verify the page setup: output device, paper size, plot area, orientation, plot scale, plot-style behavior, and lineweights. Plot preview is useful, but I also open the generated PDF and inspect the actual pages.

For a scale-sensitive issue, I check a known length and make sure the recipient prints at the intended size rather than fit-to-page. I confirm fonts, symbols, clipping, and referenced images, and save the DWG to the required version and project location. If the editable drawing is being delivered, I include the authorized dependencies rather than assume the PDF is the complete package. Finally, I keep the required revision record; a successful save or plot does not itself mean the design has been reviewed and approved for construction or manufacture.

AutoCAD Interview Questions and Answers for Intermediate

The intermediate answers explain both the workflow and its failure checks, including scale calculations, reference behavior, and the difference between a visual fix and a correct underlying drawing.

41. How would you redefine a reusable block without losing control of existing drawings?

I would treat the block definition as shared content inside the current drawing, not as an independent object for each insertion. Editing the definition with BEDIT changes references of that definition when the edit is saved. Editing one reference's position or rotation does not rewrite its definition.

Before redefining a production symbol, I would save a revision, inspect its base point, units, nested content, layers, and attributes, and test the change on representative insertions. Moving the definition's geometry relative to its base point can shift its apparent position everywhere. A replacement with a different unit convention can introduce an equally serious scale error.

I would verify rotated and scaled instances as well as ordinary ones. Attribute-definition changes require a separate synchronization decision; changing geometry alone is not an instruction to reset attribute formatting. Updating a library DWG also does not automatically revise every drawing that previously inserted it. Those drawing definitions need a controlled redefinition process.

42. How do parameters and actions work together in a dynamic block?

A parameter exposes a property such as a distance, angle, or position; an associated action specifies what selected geometry does when that property changes. A linear parameter without an appropriate action does not automatically stretch the block into a usable product family.

For a rectangular desk symbol, I would define the width direction, associate a stretch action with the intended parameter endpoint, and select the geometry affected by that action. I would then test the grip, property values, and several permitted widths in BTESTBLOCK. Explicit value sets can restrict a symbol to supported sizes rather than allowing an impossible manufactured dimension.

The action's selection set and base point matter as much as the visible grip. I would avoid building multiple actions that unintentionally move the same geometry twice. Parameters and actions are not the same as geometric and dimensional constraints; authoring support differs across product editions, and AutoCAD LT does not provide full constraint authoring.

43. How would you design a dynamic block that changes length without stretching every component?

I would separate components that must stretch, components that must move rigidly, and components that must stay fixed. A stretch action uses a stretch frame and an action selection set; the frame determines which relevant endpoints are stretched, while selected objects wholly inside it can move together.

For a cabinet plan, the left end can remain fixed while the right end and its handle move, with the long horizontal edges extending between them. I would not simply scale the entire symbol because that could distort door thicknesses, circles, and hardware. A separate move action may be clearer for geometry that must translate without changing size.

I would test minimum, nominal, and maximum lengths and inspect nested blocks and attributes. The goal is a consistent symbol, not just a grip that moves. If an attribute must move with the geometry, I would verify its position-lock and action setup in the actual block rather than assuming text follows every stretch.

44. When would you use dynamic-block visibility states instead of separate block definitions?

I would use visibility states when several configurations genuinely belong to one symbol and share enough geometry and metadata to be maintained together. A valve symbol might have open and closed representations, or a fixture might have a small set of presentation options. A visibility parameter selects which authored objects appear; it is not an arbitrary per-layer switch for the whole drawing.

I would name states clearly, decide which objects are common to all states, and test every state before distributing the block. Hidden geometry still belongs to the definition, so a large collection of unrelated alternatives can become difficult to maintain.

Separate definitions are preferable when configurations have different identities, extensive unrelated geometry, or distinct scheduling requirements. I would also check how the intended extraction or counting workflow handles dynamic properties. A visually correct state does not guarantee the schedule records the intended product code.

45. How do block attributes differ from ordinary text?

An attribute is structured information associated with a block reference, while ordinary text is primarily an independent annotation object. An attribute definition provides a tag, prompt, default, and behavior; inserted references can carry different values for the same tag. A door block might contain DOOR_ID, TYPE, and FIRE_RATING attributes.

I would use stable, meaningful tags and separate the identifier from the displayed description. This makes editing and extraction more reliable than trying to interpret free-form text near the symbol. Invisible attributes can hold non-graphical information, but I would document them so users know the data exists.

I would also distinguish constant attributes from values intended to vary by insertion. Attributes do not make the drawing a validated database: duplicate IDs, inconsistent spelling, and missing values remain possible. A schedule workflow therefore needs explicit checks for required values and uniqueness, not just attractive attribute formatting.

46. What precautions would you take before using ATTSYNC?

I would use ATTSYNC to apply attribute-definition changes to existing references, but only after checking the consequences on a saved copy. It is not a general block-cleaning command. Existing assigned attribute values are retained, while instance-level formatting or property adjustments can be reset to the definition.

For example, after adding a required equipment-ID attribute to a standard symbol, I would synchronize a sample drawing and check old values, attribute positions, text styles, rotation, visibility, and dynamic configurations. I would then inspect a schedule to confirm the tags still mean what downstream tools expect.

The command can also remove extended data associated with the block and affect dynamic or third-party content. That makes indiscriminate synchronization risky in discipline-specific drawings. I would obtain the application's recommended update method where custom objects or integrations are involved. A successful command completion is not evidence that business data and presentation survived unchanged.

47. How would you produce a reliable block-attribute schedule?

I would first define what qualifies for inclusion, which attribute tags are required, and whether repeated instances should be individual rows or combined counts. In full AutoCAD for Windows, DATAEXTRACTION can select drawings or objects, filter relevant block types, choose properties, and produce a table or external output. I would save the extraction settings for repeatable updates.

For an equipment schedule, I would check that each required asset has an ID and that the same ID is not accidentally reused. I would inspect how nested blocks, xrefs, dynamic blocks, and hidden alternatives affect the selected population rather than assuming the displayed total equals the purchase quantity.

Finally, I would reconcile a small manual sample against the result and document the source drawing revision. The full Data Extraction wizard is unavailable in AutoCAD LT and AutoCAD for Mac. I would choose and validate an appropriate alternative for those products instead of promising Windows feature parity or silently replacing missing data with guessed values.

48. How would you prevent inserted blocks from arriving at the wrong size or orientation?

I would verify the source geometry's real unit convention, the block or source drawing's declared units, the target insertion units, and the reference's scale and rotation. Changing the displayed unit format is not the same as converting geometry. Automatic insertion scaling can help when the declarations are correct, but it cannot recognize a drawing that was drafted in millimeters and mislabeled as inches.

I would measure a known feature before accepting the insertion. A 1-inch feature should become 25.4 units in a millimeter model, not 1 or 645.16 units. A second conversion applied manually after correct automatic scaling would multiply the error.

I would also inspect the block base point and the current UCS when positioning it. For repeated library use, I would correct the source definition and its metadata rather than compensating with unexplained scale factors on every insertion. Existing references should be checked separately after a library correction.

49. What is the practical difference between attaching and overlaying an xref?

Attachment allows a reference to travel as a nested reference when its host is referenced into another drawing; overlay does not propagate in that way. Both can display in their immediate host. Overlay therefore helps prevent background information from appearing repeatedly through a coordination hierarchy.

If a mechanical drawing references an architectural background as an overlay, a composite drawing that references the mechanical file does not automatically inherit that architectural overlay through it. The composite can attach the architectural model directly once. If the architectural reference was attached instead, it could arrive through the mechanical file as a nested dependency.

I choose the type from the intended reference graph, not from a rule that one is always better. Some nested dependencies are deliberate. I would inspect the External References tree, avoid circular relationships, and test the final host drawing. Unloading a reference is a temporary display/loading choice; it is not equivalent to changing attachment into overlay or detaching the reference.

50. How would you repair a missing xref without accidentally referencing the wrong file?

I would identify the expected source revision and inspect both the saved path and the path AutoCAD actually resolves. A file with the same name in another folder is not necessarily the correct reference. I would compare project location, revision information, and a known feature before repathing it.

Relative paths are useful when the host and references move together within a stable project folder tree. Full paths can be appropriate for a controlled shared location, but are less portable. A no-path reference depends on the application's search behavior and can be ambiguous when filenames are reused.

After selecting the correct source in the External References palette, I would reload it, verify location and scale, save the host, and reopen the package from its intended location. I would inspect nested dependencies too. Fixing the top-level DWG path does not resolve missing images, fonts, or references used by that DWG.

51. What changes when you bind an xref, and how do Bind and Insert differ?

Binding converts an external DWG reference into local block content, so future changes to the external file no longer update that bound instance. It is a deliberate handover or archival operation, not a routine way to refresh a background.

The Bind option retains distinct dependent names by adding a prefix such as Reference$0$Layer. The Insert option merges named definitions without that prefix; existing host definitions with matching names can therefore influence the resulting appearance. I would choose based on whether preserving distinct layers and styles or consolidating names is the required outcome.

Before either option, I would save an unbound master, resolve the full reference tree, and compare the bound copy's layers, linework, annotation, and plots against the source. I would not promise that binding includes every external dependency or converts specialized toolset objects into ordinary geometry. Those require separate compatibility and packaging checks.

52. How would you control xref layer appearance without changing the consultant’s source drawing?

I would apply host-side xref layer overrides, such as a subdued color or a viewport-specific lineweight, rather than editing the consultant's source just to improve my sheet. Xref-dependent layers remain identifiable by their reference-qualified names, which helps separate presentation changes from model ownership.

I would review VISRETAIN and VISRETAINMODE as part of that policy. They influence whether host settings are retained and which source layer properties synchronize on reload. I would not toggle them blindly: resetting properties can remove intentional overrides across multiple sheets.

For a background that is too dark only on one layout, a viewport override may be better than a drawing-wide change. I would test a reload and reopen before considering the setup stable, and check object-level colors if layer overrides seem ineffective. An explicit object property and a ByLayer property are different cases, so changing a layer is not guaranteed to change every visible entity.

53. When would you use XCLIP rather than trim or explode an xref?

I would use XCLIP when I need to display only a selected portion of a referenced drawing while retaining the source relationship. The clipping boundary limits the visible reference; it does not delete the underlying source geometry. That is useful for showing one floor wing or a detail area without modifying the consultant's model.

I would establish a clear boundary, check the inside/outside clipping direction, and inspect the result in the intended layout. If some nested or specialized content behaves unexpectedly, I would investigate its object type and reference structure rather than immediately exploding it.

Trimming or exploding a bound copy changes local geometry and can destroy useful reference organization. It is only appropriate for an explicitly approved conversion. I would also distinguish the clipping boundary's display/plot settings from the actual clipped content: hiding a boundary does not remove the clip, and displaying it for editing does not necessarily mean it should appear in the issued PDF.

54. How would you coordinate an xref update while other people are working on the project?

I would separate source ownership, revision approval, and host reload. The source author should save the agreed revision, and downstream users should know what changed before accepting it into an issued sheet. An available reload is not automatically an approved design change.

I would compare the new reference with the previous accepted version, inspect changed interfaces such as grid locations and openings, and check dependent dimensions and annotations. For a source edit, opening the source DWG explicitly often makes ownership clearer than editing in place. If REFEDIT is used, I would confirm which reference is being changed and whether saving the reference will affect other hosts.

I would preserve a recoverable prior revision and avoid simultaneous uncontrolled writes to a shared DWG. Cloud storage does not by itself turn arbitrary desktop file editing into safe real-time coauthoring. The team's file-locking, versioning, and approval process must match the storage service and AutoCAD workflow actually in use.

55. How would you diagnose a consultant background that is offset, rotated, or scaled incorrectly?

I would compare the reference transform with the agreed project coordinate and unit convention before moving any geometry. The important checks are insertion point, scale factors, rotation, source base point, and a few known control points. A visually plausible alignment can still be wrong by a small rotation or by a feet-to-meters conversion.

I would measure at least two separated control distances and inspect elevations where relevant. If the source is in meters and the host in millimeters, the unit conversion is 1000, but I would first determine whether insertion scaling has already applied it. Arbitrary dragging should not replace that diagnosis.

Once the mismatch is understood, I would correct the source metadata or the documented reference transform and record the decision. I would avoid moving a georeferenced source to the origin merely to make a sheet look convenient. A viewport can present distant model coordinates without redefining the project's coordinate agreement.

56. How would you set up annotative text for two different viewport scales?

I would specify the intended paper text height and give the annotative object the scale representations needed by the viewports. For example, a 2.5 mm paper-height note can appear at the same printed height in both 1:50 and 1:100 views, even though its model-space representations have different effective sizes.

With model and paper measurements both expressed in millimeters, the corresponding model heights are 125 mm and 250 mm. I would not create two unrelated notes and manually scale one unless there is a deliberate reason to maintain separate content. The object's supported annotation scales and each viewport's annotation scale must agree.

I would test the actual layout and PDF, not just the Model tab. Annotation visibility settings can temporarily show representations that would otherwise be hidden, masking a missing scale assignment. Adding an annotation scale is also different from changing the model geometry's real size.

57. Why might annotative text appear in one viewport but disappear in another?

The first thing I would check is whether the object includes the second viewport's annotation scale. Annotative visibility depends on the current scale and the object's available representations. A note can therefore be valid and visible at 1:50 but absent at 1:100 without being deleted.

I would then check the object's layer, viewport freeze settings, viewport clipping, and the current annotation-visibility setting. I would inspect the viewport's annotation scale separately from its geometric view scale, because free zooming can leave those settings inconsistent. Enabling the display of all annotation scales can help diagnose the issue, but is not a substitute for assigning the intended representations.

After correcting the scale assignment, I would review placement in both views. A note that appears at the correct height can still overlap different geometry at another scale. Finally, I would verify a plot preview or PDF so a temporary diagnostic setting does not hide the real sheet configuration.

58. How do associative dimensions differ from ordinary dimension objects?

An associative dimension is linked to the geometry it measures, so supported geometry edits can update its measurement and placement. A nonassociative dimension can still be one editable dimension object, but it does not have the same geometric relationship. An exploded dimension is separate text and linework, which is weaker still.

I would normally create new dimensions with DIMASSOC set to 2 and use precise object snaps to select the intended geometry. Setting the variable does not repair every existing dimension. For a broken association, I would inspect the attachment points and use DIMREASSOCIATE where appropriate, then make a controlled test edit.

I would never replace a wrong measurement by typing the desired number as a text override. That conceals the problem and can become dangerous after revisions. I would establish whether the geometry, association, units, or dimension style is wrong, correct the cause, and check the displayed value against a direct measurement.

59. How would you distinguish dimension scale from a unit-conversion factor?

I would separate annotation appearance from the numeric measurement being reported. Overall dimension scaling affects features such as text and arrows in a nonannotative workflow; it is not a reason to resize the design. A measurement scale factor such as DIMLFAC changes reported linear values and must have a documented purpose.

For a millimeter model that must also report inches, a conversion factor of 1/25.4 may be appropriate in a deliberately configured dimension or alternate-unit style. A 254 mm length then reports 10 inches. That is different from using a 1:50 viewport, which reduces the image on paper but should not make a 254 mm part read as 5.08 mm.

I would inspect alternate units, precision, rounding, and text overrides before diagnosing a wrong dimension. For ordinary model-space dimensions in a consistent unit system, I would not use an unexplained measurement factor to compensate for an incorrect viewport or incorrectly scaled geometry.

60. How would you use fields to reduce stale drawing annotations?

I would use a field when annotation should derive from a controlled property instead of being retyped. Examples include a layout name, drawing property, or a selected closed polyline's area. A field stores a relationship or expression whose displayed result can be updated; it is not merely text that happens to resemble a value.

I would choose the correct property, format its units and precision, and test the update behavior with a controlled change. For an area field, the selected object's area must represent the intended boundary. Deleting and redrawing that object can break the relationship even if the replacement looks identical.

I would check FIELDEVAL and use UPDATEFIELD when a deliberate refresh is needed, then inspect the issued output. A field can be stale or invalid, and its presence does not prove the source property is correct. For revision dates, I would choose a controlled issue value rather than allowing an automatic current-date field to silently rewrite historical information.

61. How would you calculate a metric viewport scale rather than guessing it?

I would compare paper units with model units explicitly. If the model is drawn in millimeters and the layout uses millimeters, a 1:50 view has a custom viewport scale of 1/50, or 0.02. A 5000 mm wall then occupies 100 mm on the sheet.

If the model is drawn in meters but the layout still uses millimeters, the same physical 1:50 view needs 1000/50, or 20 paper millimeters per model unit. A 5-unit wall then occupies the same 100 mm. Applying 0.02 to that meter model would be a unit error, not a different drafting convention.

I would select the viewport in paper space, set the correct custom or named scale, position the view, and lock it. I would also set the intended annotation scale and check a known length in the PDF. The layout itself is normally plotted at 1:1; scaling the sheet again would compound the viewport scale.

62. How would you set a viewport to an imperial architectural scale?

I would convert the printed scale into a ratio using the model's actual unit convention. At 1/4 inch equals 1 foot, a model drawn in inches uses a viewport ratio of 0.25/12, or 1/48. A 20-foot wall modeled as 240 inches should occupy 5 inches on the sheet.

If that model instead uses feet as its numerical unit and paper space uses inches, the custom viewport scale is 0.25, not 1/48. The same wall is 20 model units long and still prints at 5 inches. A named architectural scale is only safe when its definition matches the model/paper unit relationship.

I would set the scale deliberately, check annotation behavior, and lock the viewport after positioning it. I would inspect the final PDF at its declared page size and print without an additional fit-to-page transformation. The scale label in a title block is a statement to verify, not proof of the viewport's configuration.

63. Why should you lock a layout viewport, and what does locking not do?

I would lock a viewport after establishing its intended scale and view position so routine navigation does not accidentally change the view's zoom scale. This protects sheet presentation while I continue inspecting the layout. It does not make the model geometry read-only or protect the viewport object from every paper-space edit.

My workflow is to create the viewport, set its scale, pan to the required composition, return to paper space, and set Display Locked to Yes in its properties. I would then confirm that another navigation step has not changed the custom scale.

If the sheet border needs a larger opening, I can edit the viewport boundary in paper space without treating that as an instruction to rescale the underlying model. I would still recheck the composition because clipping changes what is visible. Layer locking, viewport locking, and file permissions solve different problems; confusing them can leave either the model or the issued view insufficiently protected.

64. How would you show different layer visibility in two viewports on the same sheet?

I would use viewport-specific layer controls rather than globally turning off the layer. With the intended layout viewport active, VP Freeze can hide a layer in that viewport while allowing it in another. Global Off or Freeze affects a broader drawing context and can unintentionally remove information from other views.

For example, a general plan can show furniture while a reflected ceiling plan suppresses it. I would keep the same model and use viewport-specific visibility instead of duplicating geometry into separate drawings solely for presentation. Viewport color, linetype, and lineweight overrides can also support different sheet purposes when permitted by the CAD standard.

I would verify which viewport is active before changing its overrides and check the resulting layout rather than only Model space. I would also inspect the viewport object's own layer: placing its boundary on a nonplot layer is useful, but freezing or turning off that layer can have different consequences from merely disabling the border's plotting.

65. Would you dimension in model space or paper space?

I would follow a consistent project method rather than claim one location is always correct. Model-space dimensions work well for reusable design annotation, especially with annotative styles. Paper-space dimensions can serve sheet-specific presentation and can associate through a viewport to model geometry when created correctly.

For model-space annotation, I would manage the supported scales and avoid drawing duplicate dimensions solely to change printed size. For paper-space annotation, I would lock the viewport and verify associations and reported values after model or view changes. Snapping to a viewport boundary or to unrelated paper-space geometry is not the same as associating with the intended model feature.

The deciding factors are reuse, detail ownership, coordination, and the team's checking process. I would demonstrate the chosen workflow with a controlled model edit and a revised viewport scale. If the resulting dimension becomes misleading, the workflow needs correction; manually overriding the text is not an acceptable substitute for a robust association.

66. What is the difference between CTB and STB plotting?

CTB uses color-dependent plot styles, whereas STB uses named plot styles assigned to objects or layers. With a CTB workflow, an indexed color commonly maps to a prescribed printed lineweight and color. With STB, the displayed color and the named plotting role can be independent.

I would choose the system required by the project standard and ensure the drawing and selected plot-style file use the same type. A missing CTB in a drawing configured for named styles is not necessarily a search-path problem. It can be a type mismatch.

I would also inspect ByLayer properties, object overrides, and True Color objects rather than assuming every entity follows an indexed-color mapping. Neither system automatically makes a drawing readable. I would check a representative output with thin detail lines, heavy outlines, screening, and color-dependent information before accepting the configuration for an entire set.

67. How would you convert between CTB and STB without changing the drawing’s appearance unexpectedly?

I would convert a copy and compare output before changing the production standard. Moving from CTB to STB typically involves CONVERTCTB to create a named-style table from the color-dependent table, followed by CONVERTPSTYLES to change the drawing's plot-style mode and apply the appropriate mapping.

The reverse change also requires attention: a drawing that relied on different named styles for objects of the same color may not retain those distinctions automatically in a color-dependent workflow. I would inspect layer and object assignments rather than treating conversion as a file-extension rename.

I would plot the same representative layouts before and after conversion and compare lineweights, screening, colors, and any special linetype behavior. Referenced content must be included in that check. I would then distribute the approved style files and document the chosen mode in templates. A technically successful conversion command is not sufficient evidence that the visual standard was preserved.

68. What would you include in a reusable page setup?

I would include the approved output device or PDF configuration, paper size, orientation, plot area, plot scale, plot-style table, and relevant lineweight or transparency settings. A named page setup makes these choices repeatable and reduces the chance that an operator issues one sheet with a different paper size or style.

For a conventional paper-space layout, I would plot the layout at 1:1 and let the viewports control model scales. I would not use Fit to Paper to compensate for a border that does not fit the selected printable area. That can silently change the sheet's scale and border margins.

Before applying a setup widely, I would preview a representative sheet and confirm that the printer/PDF configuration and plot-style file are available on the machines doing the work. Device names, printable margins, and platform-specific configuration files can differ. A setup imported from another drawing still needs a local compatibility check.

69. How would you investigate a PDF that does not match the drawing on screen?

I would compare the plot configuration, not just the current viewport display. The common suspects are missing or wrong plot styles, object and layer lineweights, transparency settings, nonplot layers, viewport overrides, and font substitution. A screen color is not necessarily the intended printed color.

I would reproduce the problem with one layout, inspect Plot Preview, and create a PDF with the approved page setup. If the preview is correct but the PDF or physical print is not, I would investigate the output driver, PDF viewer, page size, and print scaling. If the preview is already wrong, I would work backward through drawing and page-setup settings.

For missing text, I would identify the actual font dependency and licensing or embedding constraints rather than replacing it arbitrarily. I would compare known geometry and a scale bar at the final output size. The aim is to isolate one reproducible cause instead of changing multiple plotting variables and losing the baseline.

70. How do LTSCALE, object linetype scale, and PSLTSCALE interact?

I would distinguish the drawing-wide pattern multiplier from an individual object's multiplier and from viewport normalization. LTSCALE changes the global linetype scale; an object's Linetype Scale further affects that object's pattern. PSLTSCALE controls how linetype display is handled through layout viewports.

For a sheet that needs consistent printed dash lengths across differently scaled views, PSLTSCALE=1 is commonly appropriate. I would regenerate the layouts with REGENALL after changing relevant settings, then inspect both short and long objects. MSLTSCALE separately affects how model-tab linetypes respond to annotation scale.

I would also verify whether the drawing uses metric or imperial linetype definitions, because similarly named patterns can have very different numeric lengths. A dashed line can appear continuous because its segments are too short or too long for the displayed object. Randomly increasing every scale value is not a reliable fix; I would change one factor at a time and inspect the plotted result.

71. Why can a dashed polyline look different from a dashed line with the same linetype?

A polyline can generate the pattern separately for its segments or continuously along the polyline, depending on its linetype-generation setting. Short segments can therefore restart the pattern so often that the result looks solid or irregular even when a long single line displays correctly.

I would compare the objects' actual linetype, object scale, layer assignment, and polyline linetype-generation property before changing the global drawing scale. I would also inspect whether the polyline has width, unusual segment lengths, or a mixture of straight and arc segments that changes its visual rhythm.

After making a targeted adjustment, I would regenerate and inspect the relevant viewport and PDF. A screen-only improvement at one zoom level may not solve the sheet output. If the project uses a special linetype with text or shapes, I would confirm its dependencies as well. Missing supporting fonts or shapes is a different problem from an unsuitable dash scale.

72. How would you troubleshoot a hatch that leaks or fails to find its boundary?

I would first inspect the boundary rather than raising the gap tolerance until the hatch appears. Tiny gaps, duplicate edges, self-intersections, noncoplanar geometry, and unintended objects can prevent a clean region. A visual meeting of endpoints is not proof that their coordinates coincide.

I would isolate the intended boundary, check its elevations, and use precise endpoint edits or a closed polyline where appropriate. A small hatch gap tolerance can help a deliberate imperfect boundary, but it does not physically repair the geometry. If the area is used for quantities, I would require a valid measured boundary independently of the hatch.

I would then check island detection, associativity, pattern scale, and draw order. A hatch that exists but is too dense, behind another object, or on a frozen layer can be mistaken for a failed creation. Finally, I would make a controlled boundary edit to verify that the hatch updates as intended before relying on it in a production detail.

73. When should a hatch be associative, and how would you verify it?

I would use an associative hatch when its boundary is expected to change and the hatch should follow those changes. That reduces the risk of a material indication remaining at an obsolete extent. A nonassociative hatch can be intentional for a fixed graphic, but it needs separate maintenance.

I would select the hatch, inspect its associativity and boundary relationship, then make a reversible test change to a known boundary edge. If the hatch does not respond, I would investigate whether the boundary was deleted and recreated, whether the relationship was lost, or whether the edited object was never part of that boundary.

I would not assume that merely turning on the setting for new hatches repairs every existing hatch. Where needed, I would recreate or reassociate the boundary using the available hatch tools and then verify the result. The displayed hatch area should also not be treated as a checked quantity unless the geometry and unit conversion have been independently reviewed.

74. How would you clean up imported 2D linework without damaging its intended geometry?

I would retain the original and work on a controlled copy. I would first inspect units, elevations, duplicate objects, fragmented polylines, and short gaps. Commands such as OVERKILL, JOIN, and PEDIT can help, but their tolerances and selection sets determine what they change.

For example, a duplicate-removal tolerance suitable for a rough site background may erase intentionally separate detail lines in a mechanical drawing. Likewise, joining nearly touching segments can conceal an actual design gap. I would set a tolerance justified by the drawing's units and required accuracy, not by a convenient default.

After cleanup, I would compare object counts, key lengths, closed areas, and a visual overlay with the original. I would keep changes localized so a failure is easy to reverse. Flattening elevations is only appropriate when the content is genuinely intended to be planar; it is not a safe universal repair for survey or 3D data.

75. How can ALIGN combine moving, rotating, and scaling geometry?

With two source/destination point pairs, ALIGN can move and rotate selected objects and optionally scale them to match the destination distance. I would use precise object snaps for the point pairs and decide explicitly whether scaling is intended.

For example, if a scanned or imported detail has a known span of 98 units that should be 100 units, the correction factor is 100/98, approximately 1.020408. Two-point alignment can establish both its orientation and that uniform scale. If the object is already dimensionally correct and only misplaced, I would answer No to scaling.

I would measure a second independent feature afterward. One matched span does not prove that an image or imported drawing is free from distortion, and uniform scaling cannot correct different horizontal and vertical errors. For 3D alignment, the number and arrangement of point pairs matter further; I would not apply a two-point planar shortcut to arbitrary noncoplanar geometry.

76. How would you use geometric and dimensional constraints without overconstraining a sketch?

I would constrain design intent in a deliberate order. Geometric constraints express relationships such as parallel, perpendicular, coincident, or tangent; dimensional constraints control sizes or relationships through values and expressions. I would anchor the necessary reference geometry, apply the essential relationships, and add dimensions that leave the intended degrees of freedom.

For a symmetric bracket outline, I might establish a centerline, symmetry, and a controlled width rather than fixing every vertex independently. If the solver reports an overconstraint, I would look for redundant or conflicting requirements instead of repeatedly deleting constraints until the warning disappears.

I would test a realistic parameter change and confirm that the sketch moves in the intended way. Constraint authoring is not available in AutoCAD LT, although LT can work with constrained drawings within its supported limitations. A dynamic block using actions is also not automatically a fully constrained parametric model. The required product and editing workflow should be stated before evaluating the candidate's solution.

77. When is an associative array better than repeated COPY operations?

I would use an associative array when repeated objects have a pattern whose count, spacing, or path is likely to change. It preserves that pattern as an editable relationship, making a row of seats or regularly spaced fasteners easier to revise than many independent copies.

I would select rectangular, polar, or path arrangement based on the design intent, then verify the base item, spacing, alignment, and any item overrides. For six objects at 200 mm center-to-center spacing, the first-to-last center distance is five intervals, or 1000 mm, not 1200 mm. That simple check catches a frequent count-versus-interval mistake.

Independent copies can be better when the placements are intentionally unrelated or downstream editing requires separate objects. Exploding an associative array removes useful control, so I would do it only for a clear compatibility or handover reason and keep the editable original. Pattern convenience does not replace checking end clearances and clashes.

78. What would you check before sending a DWG package to another team?

I would check the intended DWG version, all references, images, plot-style files, necessary permitted fonts, and any specialized object dependencies. I would also confirm the recipient's product and platform, because a file that opens is not necessarily fully editable or visually identical there.

ETRANSMIT can assemble a package and report dependencies, but I would review its file list and settings rather than assume every required asset is included. I would avoid binding references, converting formats, or stripping information unless the handover specification calls for those changes. Font redistribution must respect licensing.

The strongest practical check is to extract a copy into a separate test folder and reopen it without relying on my original project paths. I would verify reference resolution, representative layouts, and a supplied PDF baseline. I would include revision identification and a concise note about known limitations. Sending a ZIP alone is not evidence of a complete, reproducible handover.

79. How would you use DWG Compare to review a revision?

I would compare the proposed revision against the correct previously approved drawing and use the highlighted differences to guide a structured review. COMPARE helps identify added, removed, and unchanged content, but it does not explain the engineering reason for a change or certify that it is correct.

I would inspect the comparison settings and visibility filters so relevant text, hatches, or layers are not unintentionally excluded. A coordinate shift or widespread style change can produce a large visual difference that needs a different investigation from one revised opening. I would relate important changes to the issue log and review dependent annotation and plots.

I would not import changed objects into the current drawing simply because the comparison tool offers that operation. Acceptance is a separate decision. After approved edits, I would save a clearly identified revision and verify the final sheet set. The comparison result is useful evidence for review, not a substitute for design approval or dependency checks.

80. A drawing looks correct in Model space but is wrong on one sheet. How would you isolate the cause?

I would begin with the affected layout rather than modifying the model globally. I would inspect its viewport scale, annotation scale, view direction, clipping, lock state, and viewport layer overrides. Then I would compare those settings with a working layout that shows the same content.

If the issue is missing notes, I would check annotative scale assignments and VP Freeze. If it is line appearance, I would inspect per-viewport overrides and linetype scaling. If the view looks correct but the PDF is wrong, I would move to the layout's page setup, plot styles, printable area, and output configuration. This sequence separates model content, viewport presentation, and plotting.

I would make one reversible change at a time and keep a known-length or known-text-height check. Rescaling model geometry to fix one sheet can damage every other sheet and external coordination. Once the local cause is corrected, I would reopen the drawing and verify the output, not merely stop when the screen looks better.

AutoCAD Interview Questions and Answers for Experienced

The experienced answers emphasize controlled changes, measurable validation, reversible troubleshooting, and platform-aware automation rather than indiscriminate cleanup or untested scripts.

81. How would you establish CAD standards for a multidisciplinary team?

I would start with the deliverable requirements and agree on a small, enforceable standard before distributing a template. The standard should define drawing units, coordinate ownership, layer naming, annotation, plotting, file naming, reference paths, and review responsibilities. Discipline-specific requirements can extend that baseline without silently changing shared conventions.

For example, I would test one architectural base, one structural reference, and one sheet together before rolling the standard out. A drawing that looks correct alone can still insert at the wrong scale or produce the wrong lineweights in another discipline's host.

I would assign an owner and revision to each template and support file, publish a migration note, and retain a representative regression set. Compliance checks help, but they do not establish engineering correctness. A project lead still needs to verify dimensions, coordination, and the intended issue status.

82. What is the difference between a DWT template and a DWS standards file?

A DWT provides the starting contents and settings for a new drawing; a DWS provides named-object standards against which supported drawings can be checked. I would use the template to prevent avoidable setup work and the standards file to detect selected forms of drift afterward.

My template might contain approved layers, styles, layouts, title blocks, units, and page setups. A DWS can define standards for supported named objects such as layers and dimension styles. It is not a complete validator for every object, coordinate, annotation, or contractual requirement.

Changing a DWT does not automatically update drawings already created from it. I would issue a controlled migration for those files rather than assume that replacing the template fixes the project. CAD Standards commands and checking interfaces also vary by product and platform, so I would verify availability on the actual workstation.

83. How would you use CHECKSTANDARDS without damaging valid project exceptions?

I would associate the approved standards file, run CHECKSTANDARDS in a supported AutoCAD installation, and review each discrepancy before accepting a correction. A nonstandard name may be a mistake, but it can also represent a documented client requirement or a valid discipline-specific exception.

For example, replacing a layer's color could alter plotted lineweight in a color-dependent plotting workflow. I would confirm the intended mapping and compare a test PDF before approving a batch of changes. The checker is evaluating configured standards, not understanding the design intent.

I would retain a before-and-after copy, record accepted exceptions, and rerun the check after targeted corrections. An empty violation list is evidence about the checked properties only; it does not prove that geometry is accurate, references are current, or the drawing is ready for construction.

84. How would you migrate an old project to a new company template?

I would treat migration as a controlled conversion, not paste the entire project into a new file and hope the styles resolve correctly. First I would inventory units, coordinates, layouts, references, styles, title-block attributes, and custom objects. I would then map old conventions to the approved replacements on a representative copy.

For example, two dimension styles can have similar names but different precision or measurement factors. Replacing one without checking its dimensions can change the issued meaning even if the geometry remains unchanged. Likewise, importing a title block can introduce conflicting block definitions.

I would compare key dimensions, object counts where meaningful, reference positions, and plotted sheets before and after migration. Files would be released in a controlled batch with a rollback copy and a record of exceptions. Existing approved drawings should not be silently reissued just because an internal standard changed.

85. How would you organize drawing ownership when several people work on one project?

I would partition ownership by a clear deliverable or discipline and use references to assemble the coordinated result. Each source drawing needs an accountable owner, a documented coordinate basis, and a defined update process. I would avoid two people independently editing competing copies of the same authoritative file.

For a building project, the architectural base might be maintained by one team while structural sheets reference its approved issue. An update should have an identifiable revision and a coordination check rather than merely a newer timestamp.

I would distinguish working files from issued files, define how changes are requested, and verify the storage platform's locking and synchronization behavior. A shared folder alone is not a revision-control policy. When disconnected copies are unavoidable, I would reconcile them explicitly and preserve the source copies until the merged deliverable has been reviewed.

86. How do you prevent template and support-file changes from breaking existing drawings?

I would version the support package and test it against a small set of representative drawings before deployment. That package may include templates, fonts, linetypes, plot-style tables, printer configurations, blocks, and approved automation. I would record dependencies rather than replace files with the same names without notice.

For example, changing a CTB can alter output from many old drawings even when none of their geometry has changed. I would compare plotted PDFs, including thin lines, screened backgrounds, and annotation, before making the new package the default.

The rollout should identify which projects adopt the change and which remain on their existing package. I would keep a rollback version and a change log. User interface customization is also distinct from drawing content: a working toolbar on one machine does not demonstrate that another machine has the same fonts, plot styles, or command dependencies.

87. How would you verify a CAD deliverable beyond checking that it opens?

I would verify the deliverable at three levels: file integrity, drawing meaning, and issued output. Opening without an error addresses only part of the first level. I would check references, required support files, units, coordinates, intended objects, annotation, and the actual plotted sheets.

For a floor plan, I might measure a known grid spacing, confirm a reference's insertion location, inspect critical dimensions, and verify the issue number in the title block. I would then review the PDF for missing fonts, clipped notes, incorrect lineweights, and an unintended sheet scale.

Finally, I would test the package in a clean review location rather than rely on hidden paths available only on my workstation. Acceptance evidence should identify the file revision and reviewer. Neither a successful AUDIT nor a visually attractive PDF independently proves that the model and contract deliverables agree.

88. How would you diagnose a large drawing that has become slow?

I would first isolate whether the delay follows the drawing, workstation, storage location, or a particular operation. I would record a repeatable action, such as opening the file or selecting a known group, and compare a safe local copy with the original working environment.

If only one drawing is slow, I would inspect unresolved references, excessive objects, dense hatches, imported geometry, and database errors. If every drawing is affected, I would look at graphics configuration, drivers, plug-ins, and the workstation environment. I would change one factor at a time and retain the baseline timings.

I would not start by deleting unfamiliar content or disabling security controls. A useful diagnosis identifies the cause and the improvement with the same test. Cleanup is justified only after confirming what can be removed and validating the resulting geometry and plotted output.

89. When would you change hardware acceleration settings?

I would change hardware acceleration as a controlled diagnostic when the symptoms point to graphics, such as display artifacts, orbit lag, or selection highlighting problems. It is not a universal speed switch: the result depends on the GPU, driver, remote-session environment, and drawing workload.

On a supported Windows installation, I would inspect GRAPHICSCONFIG, record the current settings and driver, and repeat a representative navigation test with a targeted setting changed. If the problem improves, I would still investigate the supported driver and graphics configuration instead of assuming that the temporary setting is the best permanent solution.

I would also distinguish a display defect from corrupted geometry by checking object properties and output. A line that looks broken on screen may plot correctly, while incorrect coordinates remain incorrect regardless of the graphics mode. I would avoid promising that a more expensive GPU fixes database or network delays.

90. Why can geometry far from the origin cause problems, and how would you respond?

Very large coordinate magnitudes can make display and precision-sensitive operations harder, especially when tiny details coexist with a geographically large coordinate system. I would inspect the actual coordinates and extents before concluding that the file is corrupt or moving everything to zero.

A survey drawing may legitimately use an established coordinate system. Arbitrarily translating it could break every reference and invalidate coordination. I would first distinguish intentional survey coordinates from a stray object that has expanded the drawing extents.

For a local detail model, a documented local origin and a controlled transform back to the project coordinates may be appropriate. I would preserve the authoritative coordinate relationship, test reference alignment, and document the translation. There is no single coordinate threshold that guarantees failure for every drawing; object scale, operation, and product behavior matter.

91. How would you investigate unexpectedly huge drawing extents?

I would determine whether the extents come from legitimate content, a misplaced object, or a reference before erasing anything. After saving a copy, I would use an overall view, inspect object locations and layers, and isolate references or suspicious object groups in stages.

For example, a small text object accidentally placed millions of units away can make the intended building appear as a dot after Zoom Extents. However, a distant survey benchmark may be intentional and must not be discarded solely because it is inconvenient.

I would identify the owner and purpose of the distant content, correct only confirmed mistakes, and compare the resulting model and sheets. If the object is inside an xref, the correction belongs in the appropriate source drawing, not a misleading host-level workaround. A cleaner view is not enough; the coordinated geometry must remain intact.

92. What is the difference between PURGE and deleting drawing objects?

PURGE removes eligible unused definitions and other supported unused data; deleting an object removes an instance from the drawing. A block definition can remain after its last insertion is erased, and that unused definition may then be purgeable. Conversely, a definition still referenced by other content cannot simply be purged as unused.

I would inspect the purge candidates and retain a backup rather than treat every unused item as unwanted. A project may deliberately carry approved styles or blocks for later use. Purging them can reduce convenience or violate the required template baseline even if existing geometry looks unchanged.

For abnormal file growth, I would determine the cause before cleanup and compare the saved result. PURGE does not repair every database error, remove all duplicate geometry, or establish that remaining objects are correct. Those are different checks with different risks.

93. How would you remove duplicate geometry safely?

I would first confirm that the overlap is accidental rather than two meaningful objects on different layers or in different disciplines. Where available, OVERKILL can remove or combine eligible duplicate and overlapping geometry, but its tolerance and property settings determine what it treats as equivalent.

I would run it on a selected area of a copy, use a tolerance appropriate to the drawing units, and avoid ignoring properties without a reason. For example, coincident lines representing demolition and new construction must not be merged just because their coordinates match.

Afterward I would inspect junctions, closed boundaries, lengths, and any downstream hatch or quantity calculation. I would compare the output with the original and expand the cleanup only after the sample passes. Reducing the object count is a diagnostic result, not proof that the design has been preserved.

94. How do AUDIT, RECOVER, and RECOVERALL differ?

AUDIT checks the current drawing database and can correct certain errors. RECOVER attempts to repair a selected drawing as it opens. RECOVERALL extends recovery to the selected file and its attached references, including nested references, and is therefore much broader than a harmless read-only check.

I would choose the narrowest appropriate operation and preserve the complete dependency set first. RECOVERALL can repair and resave referenced drawings, so running it against shared project paths can change files owned by other people. Its recovery log deserves review, and repaired files must be checked for missing or changed content.

For example, if only a detached supplier copy is damaged, I would not point a broad recovery operation at live team references. I would work on an isolated package, review the result, and coordinate any replacement with the file owners. Recovery success does not guarantee complete preservation of design intent.

95. How would you recover work after AutoCAD crashes?

I would preserve the available files first, then compare the original DWG, backup, autosave, and any recovery candidates. Drawing Recovery Manager can help identify candidates, but the newest timestamp is not automatically the most complete or trustworthy version.

I would copy candidates to a safe location before opening or renaming them. A BAK can be copied and renamed with a DWG extension; an available SV$ autosave can similarly be copied and renamed for recovery. I would avoid overwriting the original while investigating.

After opening a candidate, I would verify the latest known changes, critical geometry, references, layouts, and a representative plot. If no candidate contains all work, I would reconcile recoverable changes into a reviewed copy. Autosave is a recovery aid, not a substitute for intentional saves and a tested backup process, and it cannot guarantee recovery of the last unsaved operation.

96. What would you do if a repaired drawing opens but important objects are missing?

I would stop treating the repaired drawing as authoritative and compare it with untouched backups and recovery candidates. A repair can make a database readable without reconstructing every lost object or custom relationship. I would preserve the recovery log and identify what changed.

I would check whether apparently missing content is actually hidden by layers, clipping, annotation visibility, unresolved references, or missing custom-object support. If the objects are genuinely absent, I would recover them from a known-good source and reconcile subsequent edits rather than invent replacements from a plotted image without disclosure.

The decision to return the file to production should include the responsible designer, not just the person who ran recovery. I would document the recovered revision, remaining uncertainty, and any sheets that require rechecking. A file that opens successfully can still be unsafe to issue.

97. How would you distinguish a network problem from a drawing-file problem?

I would compare the same authorized drawing package in the normal location and an isolated local test location, keeping references and dependencies intact. If local operations are consistently fast while network operations are slow, storage latency, connectivity, synchronization, or path resolution becomes a stronger hypothesis.

If the same delay follows the local copy, I would investigate drawing content, corruption, graphics, and plug-ins instead. I would record the operation and elapsed time rather than rely on a general impression that AutoCAD feels slow.

I would not let the test copy become an untracked alternate production file. Nor would I disable endpoint protection or bypass company storage rules to obtain a faster result. I would hand reproducible evidence to IT and coordinate any infrastructure change, then retest under the actual project conditions.

98. When would you choose an AutoCAD script instead of AutoLISP?

I would use a script for a predictable sequence of command-line actions and AutoLISP when I need decisions, calculations, selection logic, or stronger handling of varying drawing conditions. A script is simple to inspect, but it depends closely on the expected command prompts and state.

For example, applying a known sequence to a controlled sample may suit a SCR file. Renaming layers based on their existing properties and reporting exceptions is better handled by a program with explicit validation. I would not force a fragile script to simulate conditional logic through blind keystrokes.

Before batch use, I would test the exact product and release, quote paths containing spaces, use appropriate command-line variants instead of dialogs, and ensure the script has its required final blank line. I would keep backups and verify outputs rather than assume a command sequence succeeded because it reached the end.

99. How would you make command automation less dependent on workstation settings?

I would make the automation's assumptions explicit and avoid aliases or implicit defaults wherever practical. For built-in commands, the underscore prefix requests the global command or option name, and the period prefix requests the original built-in command if a command has been redefined. Command-line variants reduce reliance on dialog interaction.

I would also control relevant state, such as the current layer, coordinate system, object snaps, file dialogs, and selection, while preserving the user's previous values. A routine that works only when a particular UCS or snap mode happens to be active is not reliable automation.

Testing should include different initial states and a canceled operation. I would avoid claiming complete portability merely because commands use English names: prompts, supported commands, paths, and APIs can still differ across releases, platforms, and AutoCAD versus LT.

100. What error-handling responsibilities should an AutoLISP routine have?

A reliable routine should validate inputs, handle cancellation, restore changed settings, and leave the drawing in a predictable state. I would capture any system variables that the routine changes and restore them both after success and on a handled error, rather than hard-code what I think their defaults should be.

For example, a routine that disables object snaps to construct geometry must restore the previous OSMODE value even if the user presses Escape. It should also avoid saving over the production file automatically just because some entities were created successfully.

I would group changes into an appropriate undoable operation where supported and report partial completion clearly. Cleanup logic itself needs testing: missing entities, locked layers, read-only files, and unexpected selection types can all interrupt execution. A broad error handler that suppresses every message may hide a failed drawing modification rather than make the routine robust.

101. Which AutoCAD automation interfaces are portable across products and platforms?

I would select the interface only after confirming the target product, platform, and release. Full AutoCAD for Windows supports a broader set of interfaces, including managed .NET and ObjectARX. Full AutoCAD for Mac supports AutoLISP and its supported ObjectARX environment, but Windows ActiveX and managed .NET code are not portable to it as ordinary AutoCAD plug-ins.

AutoCAD LT for Windows gained AutoLISP support in the 2024 release, with limitations; that does not make it equivalent to full AutoCAD's APIs. AutoCAD LT for Mac does not support AutoLISP. AutoCAD web has its own supported AutoLISP scope and should not be assumed to run desktop-dependent routines unchanged.

I would inventory functions, external files, COM dependencies, and user-interface assumptions before promising compatibility. A routine consisting of portable core AutoLISP operations is a better candidate for reuse than one built around Windows-only automation.

102. How would you safely load a third-party AutoLISP routine or plug-in?

I would treat executable customization as software, not as harmless drawing content. Before loading it, I would verify its source, intended function, supported release, and deployment approval. For a readable routine, I would inspect file access, external process calls, startup hooks, and drawing modifications; unfamiliar compiled code needs an appropriate trust and review process.

I would use a controlled trusted location rather than make a supplier's whole project folder trusted. SECURELOAD and TRUSTEDPATHS help control executable loading, but a trusted location is not proof that every file placed there is safe.

I would test on disposable drawings with least-necessary permissions and record the installed version. I would not recommend setting SECURELOAD to zero just to make an unknown routine load. A successful load also does not establish that the routine is compatible or that its changes are correct.

103. How would you validate a batch routine before using it on hundreds of drawings?

I would build a representative test set and define measurable acceptance checks before running the batch. The set should include ordinary drawings plus exceptions such as locked layers, missing references, unusual units, custom objects, and read-only destinations.

The routine should log the input file, attempted operation, result, warnings, and output location. I would preserve originals and write reviewed outputs separately until the process is proven. A batch that reports success after silently skipping half the files is not acceptable.

For a layer-renaming routine, I would verify mappings, collisions, object properties, and plotted appearance, then run it again on a copy to check that it does not create additional unwanted changes. I would expand gradually only after the sample passes. Local syntax checks or arithmetic tests cannot replace execution in the intended AutoCAD product and inspection of the resulting DWGs.

104. How do you approach an AutoCAD plug-in upgrade between releases?

I would check the official SDK and compatibility requirements before deploying the old binary into the new release. Source compatibility, binary compatibility, runtime requirements, and drawing behavior are separate concerns. A plug-in that loads can still behave incorrectly with changed APIs or dependencies.

For AutoCAD 2027, Autodesk's SDK guidance calls out a binary-compatibility break from 2025 and 2026, so I would follow the required rebuild and dependency updates rather than assume an unchanged DLL is supported. I would use the precise guidance for the plug-in technology, not apply an ObjectARX build recipe indiscriminately to every interface.

My regression set would cover loading, commands, cancellation, custom-object persistence, save/reopen, and representative customer drawings. I would retain the previous supported environment and a rollback plan until the upgraded plug-in and affected files are accepted.

105. How would you keep automated drawing changes auditable?

I would make each batch traceable to a specific routine version, input set, settings, and output set. The log should distinguish changed, unchanged, skipped, and failed files and include actionable reasons rather than just a total count.

For example, a title-block update should record which attribute tag was changed, the intended value, and which drawings lacked the expected block or contained multiple ambiguous candidates. I would avoid silently choosing the first matching object when the business rule requires exactly one.

I would preserve the pre-change files and compare selected outputs at both data and plotted-sheet levels. The audit record should not expose unnecessary confidential drawing content. Accountability comes from reproducible evidence and explicit exception handling, not merely from having a macro or an undo command available after a large batch has run.

106. How do you choose between solid, surface, and mesh modeling?

I would choose a solid when the object represents a closed volume and operations such as Boolean subtraction or volume measurement matter. I would use surfaces to develop or control shapes without assuming a closed interior, and meshes when polygonal form and local vertex or face editing suit the task.

For a machined bracket, a solid is usually a useful representation because holes, intersections, and volume can be checked directly. A conceptual freeform skin may begin as surfaces or a mesh, but conversion is not automatically lossless or suitable for manufacturing.

I would inspect the resulting object type rather than infer it from shaded appearance. A closed-looking set of surfaces is not necessarily a valid solid. These are full AutoCAD 3D workflows; LT and web should not be assumed to provide the same modeling commands or editing capabilities.

107. How would you establish a reliable UCS for modeling on an inclined face?

I would align the UCS deliberately to the intended working plane, confirm the origin and axis directions, and then construct the profile in that plane. A convenient camera view does not by itself redefine the coordinate system used by commands.

For example, before placing a hole pattern on an inclined plate, I would use the appropriate face or three-point UCS workflow, verify a known edge direction, and check the first point numerically. I would save a useful named UCS if the operation will be repeated and return to World when project-coordinate work resumes.

I would also check the objects' actual elevations and normals. A profile that appears to lie on the face in one view may be offset or rotated in 3D. I would verify with a second view or object properties before extruding or cutting the part.

108. Why might EXTRUDE create a surface instead of a solid?

The input may be an open profile, or the command's mode may be set to create a surface. To create the intended solid, I would verify that the profile encloses a valid planar area without self-intersections and that the selected mode supports the desired result.

A rectangle that only looks closed can contain a tiny gap, separate unjoined segments, or noncoplanar vertices. I would inspect its properties and geometry, repair the intended boundary, and use a suitable closed polyline or region rather than rely on its appearance. I would also check whether source geometry is retained or removed under the current DELOBJ behavior.

For a simple 100-by-50 profile extruded 20 units, the expected volume is 100,000 cubic drawing units. That arithmetic provides a useful independent expectation, but I would still inspect the actual solid and units in AutoCAD before accepting it.

109. When would you choose REVOLVE instead of EXTRUDE?

I would choose REVOLVE for a shape generated by rotating a profile about an axis, such as a shaft, turned sleeve, or rotational housing. EXTRUDE is more natural when a profile extends along a direction or path without that rotational construction.

For a stepped shaft, I would draw a clean longitudinal half-profile in a known plane, identify the centerline, and revolve through the required angle. I would check that the profile does not cross the axis in a way that creates invalid or unintended geometry. A partial revolution can be useful, but it is a different result from a full 360-degree part.

I would verify object type, dimensions, and any internal openings afterward. Selecting a convenient line as the axis without confirming its location can produce a plausible-looking but incorrectly sized object. The modeling method should make the design intent easier to inspect, not just produce a shaded shape.

110. How would you troubleshoot a failed SWEEP operation?

I would inspect the profile and path separately before increasing tolerances or rebuilding the entire model. The profile must be appropriate for the desired solid or surface, and the path must permit the section to travel without producing invalid self-intersections or excessively tight geometry.

For example, a large circular section swept around a very tight bend can intersect itself. I would test a simpler path, a smaller section, or one path segment to identify the failing region, then decide whether the design or modeling approach needs to change.

I would also check alignment, base point, scale, and twist choices. A sweep can succeed but orient the section incorrectly. I would inspect cross-sections and end conditions against the intended geometry rather than accept success as proof of correctness. Source geometry should be preserved in a controlled copy while investigating.

111. How do UNION, SUBTRACT, and INTERSECT differ in solid modeling?

UNION combines selected compatible objects, SUBTRACT removes the volume of one selection from another, and INTERSECT retains the common volume. For subtraction, the order matters: I select the objects to keep first and the cutting objects second.

To drill a simple hole through a plate, I would create a cylindrical cutter that passes completely through it, run SUBTRACT on the plate first and the cylinder second, and inspect the resulting hole from more than one view. Reversing the selections does not mean the same operation.

I would keep a recoverable copy of important construction geometry and check that the operands are the intended object types and positions. Two objects can appear to overlap in a projected view while being separated in depth. A Boolean operation is a geometric modification, not a substitute for checking tolerances, wall thickness, or manufacturing requirements.

112. How would you verify the volume of a modeled part?

I would use a suitable mass-properties measurement on a valid solid and compare it with an independent geometric expectation where possible. I would state the drawing units explicitly: a reported volume is in cubic drawing units, not automatically cubic millimeters or cubic inches.

For a 100-by-50-by-20 millimeter plate with a through-hole of radius 5 millimeters, the ideal volume is 100 × 50 × 20 − π × 5² × 20, or about 98,429.20 cubic millimeters. A large discrepancy suggests a unit, dimension, cutter, or object-selection problem.

I would not call that value the physical mass without a material density and consistent unit conversion. Nor would a matching volume prove that every feature is correct; two different shapes can have the same volume. I would combine the numerical check with critical dimensions and visual inspection.

113. What is the difference between a live section and permanently cutting a model?

A live section is a viewing and documentation aid that exposes interior geometry without necessarily removing it from the source model. A permanent modeling operation, such as a solid cut, changes the geometry itself. I would choose based on whether the deliverable needs a view or a modified part.

For a coordination review, I would position a section plane to reveal a connection and preserve the full model. If the design actually requires material removal, I would perform the appropriate solid operation and verify the resulting geometry separately.

I would make the distinction clear to reviewers: seeing an interior through a section does not mean the hidden material has been deleted. Generated section geometry can also require updating when the source changes, depending on the workflow. I would confirm the relevant command support in the actual product and platform before promising identical documentation behavior.

114. How would you create dependable 2D documentation from a 3D model?

I would establish the required views, scales, hidden-line conventions, and update process before generating the drawings. The goal is traceable documentation, not a collection of screenshots that happens to resemble the model.

For a bracket, I would choose orthographic views that expose the important features, add a section where internal geometry is ambiguous, and dimension manufacturing-critical sizes. I would distinguish associative model-documentation workflows from exported or flattened geometry that will not necessarily update when the model changes.

I would then make a controlled model revision and verify what updates automatically and what requires regeneration. A correct initial sheet is insufficient if later model changes leave stale dimensions or section views. Model-documentation commands differ by platform and product, so I would test the chosen full AutoCAD workflow rather than assume LT, Mac, and web offer identical tools.

115. Why is a visually correct 3D model not necessarily a valid deliverable?

A shaded view can hide gaps, overlapping bodies, reversed design assumptions, wrong units, and missing internal features. I would assess the model against its intended downstream use rather than judge it only by appearance.

For fabrication, I would check valid solids, critical dimensions, hole locations, wall thickness, and the exchange format required by the recipient. For coordination, I would also verify placement, orientation, and which revision the model represents. A mesh suitable for presentation may not satisfy a request for editable solid geometry.

I would reopen the exported result in the agreed review workflow and compare it with the source. If the recipient needs parametric feature history or discipline-specific intelligence, a generic DWG solid may not preserve that requirement. The acceptance criteria should state what information must survive, not merely the file extension.

116. How would you handle proxy objects received from another Autodesk product?

I would identify the creating application and the recipient's required use before trying to explode or replace the objects. A proxy is a substitute for a custom object when its defining application is unavailable, and the permitted viewing or editing behavior depends on that object.

A compatible object enabler may provide more appropriate access, but it is not the same as having the full authoring toolset. I would verify the product, release, and platform support before installation and preserve the original file.

If the recipient only needs conventional geometry, I would ask the source owner for a controlled export and document the loss of intelligence. A wall that becomes lines or solids may no longer carry the relationships and data needed by the original application. Suppressing a proxy warning does not resolve compatibility or prove that all objects are displayed correctly.

117. How would you prepare a drawing for a supplier using an older or different CAD product?

I would agree on the required file format, object types, units, coordinates, references, and editable information before exporting. Saving to an older DWG version addresses file-format compatibility, but it does not guarantee preservation of every custom object or newer feature.

I would create a delivery copy, package the required dependencies, and test the result in the agreed recipient workflow when available. For a critical exchange, I would include a reference PDF and a few known dimensions so the supplier can verify appearance and scale independently.

I would record any intentional conversion, such as replacing intelligent objects with conventional geometry, and retain the authoritative source separately. If recipient-side testing is unavailable, I would say so and request confirmation rather than claim compatibility based solely on a successful SAVEAS operation.

118. How would you investigate a supplier drawing that arrives at the wrong size and location?

I would separate scale, translation, rotation, and coordinate-system issues instead of repeatedly moving and scaling the file by eye. First I would measure a known feature in both drawings and inspect insertion units, insertion scale, base points, and the agreed coordinate basis.

If a 10-foot feature is represented as 120 drawing units in an inches-based file, inserting it into a feet-based model requires a scale factor of 1/12 when unit conversion is not otherwise correctly applied. I would verify the actual settings so I do not apply that conversion twice.

After resolving units, I would use agreed control points to check location and orientation. One matching point proves only translation; a second known direction or additional control points help reveal rotation and scaling errors. I would document the agreed transformation and request a corrected source if the supplier's coordinate assumptions remain uncertain.

119. What would you do if the issued PDF and the current DWG disagree?

I would first establish which revisions were actually issued and preserve both artifacts. The current DWG may include legitimate later work, or the PDF may have been plotted from stale references, a different page setup, or an earlier file. I would not overwrite one to make them appear consistent before understanding the discrepancy.

I would compare issue identifiers, timestamps, reference revisions, and the changed geometry or annotation. Then I would reproduce the plot from the identified source package and determine whether the difference is design content or output configuration.

The responsible project lead should decide whether a corrected issue is necessary. I would retain an audit trail, update the revision documentation, and verify the replacement package before release. An informal resend can create two conflicting authoritative documents, so the correction needs the same control as an ordinary issue.

120. How would you assess whether an experienced AutoCAD candidate can deliver reliable work?

I would use a small realistic drawing problem and ask the candidate to explain decisions while making verifiable changes. The task should combine precision, organization, references, annotation, and output rather than reward memorized command aliases alone.

For example, I might provide a host drawing with one unit mismatch, one unresolved reference, a questionable dimension override, and a layout that plots incorrectly. I would ask the candidate to diagnose the causes, preserve the original, fix only justified issues, and produce a checked delivery copy.

I would assess the evidence they use: known dimensions, object properties, coordinate checks, dependency inspection, and plot review. A strong candidate explains uncertainty and avoids destructive shortcuts. I would also match the task to the advertised product and platform; a specialist in LT should not be evaluated as though full Windows AutoCAD 3D APIs are part of the same role.

AutoCAD Documentation and Further Reading

The baseline is full AutoCAD desktop, using AutoCAD 2027 documentation where applicable. AutoCAD LT, Mac, web, and specialized toolsets have different command and API availability; verify the installed edition and release. Examples have not been executed in AutoCAD.

These curated Autodesk resources support further study and product-specific verification. They are not an exhaustive research log. Documentation review and local arithmetic checks do not validate a drawing, plot, script, or recovery operation in AutoCAD.

• Autodesk: AutoCAD 2027 release notes

• Autodesk: Coordinate entry

• Autodesk: Precision drafting

• Autodesk: Layer visibility and locking

• Autodesk: Layouts and viewport scales

• Autodesk: Attaching and detaching drawing references

• Autodesk: CAD standards

• Autodesk: Recovering drawings and referenced files

• Autodesk: Secure executable loading

• Autodesk: Extruding solids and surfaces

• Autodesk: Block Editor and dynamic block authoring

• Autodesk: Annotative objects and styles

• Autodesk: Plot style tables

• Autodesk: LTSCALE and PSLTSCALE

• Autodesk: Supported programming interfaces by platform