Latest / Mechanical Engineering Made Simple / (#32) GD&T Decoded: ASME Y14.5 Updates, ISO Fit Codes and the 57% Bonus of True Position

Transcript

(#32) GD&T Decoded: ASME Y14.5 Updates, ISO Fit Codes and the 57% Bonus of True Position · Mechanical Engineering Made Simple

• 0:00
  Welcome back, design jackies to mechanical engineering made
• 0:02
  simple. Good to be back.
• 0:03
  Today we're doing a deep dive into, well, really the language
• 0:06
  that defines manufacturing quality, geometric dimensioning
• 0:10
  and tolerancing, GD and T, And specifically, we're hitting the
• 0:14
  absolute bedrock of interchangeable design limits
• 0:18
  and fits for holes in shafts. Yeah, this is fundamental stuff.
• 0:22
  We are simplifying a huge stack of documentation today.
• 0:26
  Sounds ambitious. Well, our mission is to take the
• 0:28
  latest ASME Why 14.52018 changes, sort of pair them up
• 0:34
  with the global ISO 286 fit system and give you clear
• 0:38
  guidance on selecting the exact tolerance code you need.
• 0:41
  Right, so parts actually fit together.
• 0:43
  Exactly to ensure your parts mate perfectly cost effectively
• 0:47
  every single time without, you know, just guessing.
• 0:50
  We've synthesized material covering these latest standard
• 0:52
  revisions, the international ISO fit codes, and, importantly, the
• 0:56
  harsh practical realities of tolerance analysis and
• 0:59
  metrology. Because you're drawing, it
• 1:01
  really is a contract, isn't it? It absolutely is.
• 1:03
  And if that language isn't precise, the contract fails.
• 1:06
  That opens the door to, well, massive scrap and rework costs.
• 1:10
  Definitely want to avoid that. So let's start with the language
• 1:12
  itself, GD and T. How has it been evolving?
• 1:15
  Well, GD and T is essentially that concise, universally
• 1:19
  understood symbol based language.
• 1:22
  It defines geometry and crucially, controls part
• 1:25
  variation. Making clear design definition
• 1:28
  possible. Exactly, it makes it possible.
• 1:30
  And you mentioned the ASMEY 14.5 standard from 2018.
• 1:34
  It seems like it cleaned up quite a bit of legacy ambiguity.
• 1:37
  It really did. What are the biggest updates
• 1:39
  engineers really need to know right now?
• 1:42
  I'd say the most foundational shift is probably the dimension
• 1:44
  target rule rule 4.1 Q. It clarifies that the nominal
• 1:49
  dimension value, let's say 32.0 on the drawing.
• 1:52
  That value is not the manufacturing target.
• 1:55
  Wait hang on, if I put say 32.0 ± .02 on a drawing, the target
• 2:00
  isn't 32 point O? Nope.
• 2:02
  But isn't that what we've like? Always intuitively assumed the
• 2:05
  supplier should aim for. That assumption is precisely
• 2:07
  what's gone now. The standard says the target
• 2:09
  must be the middle of the tolerance zone.
• 2:12
  So if your tolerance zone is asymmetric, like 32 point O + 0
• 2:19
  point O 2 - 0.00. Right, a common scenario.
• 2:22
  Then the suppliers production goal, their target must be
• 2:26
  32.01. Yeah, right in the middle.
• 2:28
  OK, that eliminates potential arguments about, you know, where
• 2:32
  they should center their production runs.
• 2:33
  Exactly. No more ambiguity there.
• 2:35
  That's a pretty fundamental change in thinking it is.
• 2:37
  And Speaking of cleanup, what about those those confusing
• 2:40
  legacy symbols? Engineers have kind of avoided
• 2:42
  them for years, haven't they? Did Why 14.5 Finally get rid of
• 2:46
  concentricity and symmetry? They did.
• 2:48
  Thankfully, the standard has officially eliminated
• 2:51
  concentricity. That's the circle symbol, and
• 2:53
  symmetry the three horizontal line symbol.
• 2:55
  The term, the symbol, everything.
• 2:57
  Term symbol and the whole concept gone, mostly due to
• 3:00
  decades of confusion and frankly, misapplication.
• 3:03
  So what's the recommended practice now for controlling
• 3:06
  coaxial features? If you need to control the
• 3:08
  location of coaxial features, the modern and frankly superior
• 3:12
  alternatives are position and run out.
• 3:14
  They do the job much more clearly.
• 3:16
  OK, that simplifies things a lot by just removing tools that were
• 3:19
  often misused or misunderstood. Definitely.
• 3:21
  What about defining the actual object itself?
• 3:24
  Any changes there? Yeah, they clarify the
• 3:25
  definition of a surface too. That's 4.1 S It says elements of
• 3:30
  a surface must now explicitly include surface texture and
• 3:34
  flaws. Flaws like burrs and scratches.
• 3:36
  Exactly. Burrs, scratches, waviness, all
• 3:39
  those elements have to stay within the tolerant zone
• 3:42
  boundaries unless you specifically state otherwise.
• 3:44
  OK. So that's the ASME world trying
• 3:47
  to be clearer. But manufacturing, it's global,
• 3:50
  right? And that means we run smack into
• 3:53
  the fundamental challenge, the philosophical differences
• 3:56
  between ASME Y 14.5 and the ISO standards like ISO eleven, O 1
• 4:01
  and 286. Right.
• 4:02
  And this is a big one. They agree maybe what, 65 to 95%
• 4:06
  of the time depending on the specific aspect.
• 4:08
  Yeah, something like that, but. The biggest risk, the place you
• 4:10
  can really get tripped up, lies and how size is defined and
• 4:13
  controlled. How so?
• 4:14
  Well, ASME generally treats a size tolerance as controlling
• 4:18
  both the local size and the actual mating size or the
• 4:20
  envelope. Which seems intuitive, right?
• 4:23
  If my whole measure is within its plus minus tolerance, I
• 4:26
  assume it's round enough and straight enough to mate
• 4:28
  properly. And that is exactly the trap
• 4:31
  when you step into the ISO world.
• 4:33
  Under ISO 11 O1, the basic size tolerance only controls A2 point
• 4:37
  size measurement like caliper jaws.
• 4:40
  Yeah, if you rely on that ASME intuition, you completely lose
• 4:44
  control over the form, the roundness, the straightness,
• 4:48
  unless you explicitly add the envelope symbol.
• 4:51
  That's the circle with an E inside to control that essential
• 4:54
  mating size. Wow, so just relying on old
• 4:57
  habits when switching standards could mean a part is technically
• 5:02
  in stack based on a 2 point check, but it's functionally
• 5:05
  useless because it won't assemble.
• 5:06
  Precisely. That's a huge hidden risk,
• 5:08
  especially for multinational companies working across
• 5:10
  different regions or with different suppliers.
• 5:12
  That is a critical distinction. And just to sort of closeout
• 5:15
  this ambiguity section, right, the 2018 standard also cleared
• 5:18
  up that old headache around radius or R tolerance.
• 5:21
  Oh yes, the R versus diameter confusion.
• 5:23
  Right, the clarification is if the center of the radius is
• 5:27
  located using dimensions, then the resulting arcs must be
• 5:30
  concentric. This gets rid of that old
• 5:32
  ambiguity where radius didn't necessarily equal half the
• 5:36
  diameter in terms of form control.
• 5:38
  The good practice is still only use plus minus tolerances on
• 5:43
  features you can easily check with calipers like widths or
• 5:46
  simple diameters. Use GD and T for anything more
• 5:49
  complex. That focus on clear measurement,
• 5:52
  especially for features that aren't just simple widths,
• 5:54
  brings us right back to mating parts.
• 5:57
  If we can't just use a caliper for everything, we need that
• 6:00
  more sophisticated language of fits, which is primarily defined
• 6:03
  by ISO 286. This is how we ensure parts
• 6:06
  actually interact the way we design them to.
• 6:08
  Exactly. The fit describes that required
• 6:10
  relationship between the mating parts, typically a hole in a
• 6:13
  shaft. ISO 286 sets up the whole code
• 6:15
  system for tolerances on these linear sizes.
• 6:17
  And these fits fall into basically 3 categories, right?
• 6:20
  Based on how they're tolerance zones interact.
• 6:22
  That's right, It's all about whether you guarantee clearance,
• 6:25
  guarantee interference, or allow for either possibility.
• 6:28
  OK, let's start with the two extremes.
• 6:29
  What's the first one? So the first is the clearance
• 6:32
  fit. This one always provides a gap,
• 6:34
  a clearance between the hole and the shaft after assembly.
• 6:38
  Meaning the smallest possible hole, it's lower limit or LLS is
• 6:43
  still greater than or equal to the largest possible shaft, it's
• 6:46
  upper limit or ULS. And you'd use this when.
• 6:49
  You use this when parts absolutely need to slide or
• 6:53
  rotate easily. Think loose bushings, sliding
• 6:56
  guides. No tool should be needed to
• 6:59
  assemble them. OK, easy movement.
• 7:01
  What about the opposite end? The real tight high stress
• 7:04
  joint. That would be the interference
• 7:05
  fit. This one always provides an
• 7:07
  interference, basically a negative difference in size
• 7:10
  before assembly. So the shaft is always bigger
• 7:12
  than the hole essentially. Correct.
• 7:14
  The upper limit of the hole size is smaller than or equal to the
• 7:17
  lower limit of the shaft size. And this requires.
• 7:20
  This always requires force or thermal methods for assembly.
• 7:23
  Think press fits for bearings or dowel pins, shrink fits for gear
• 7:27
  hubs. It creates a permanent high
• 7:29
  torque or high load carrying joint.
• 7:32
  Right, so clear for movement interference for fixed joints?
• 7:35
  What about the middle ground where it's not quite guaranteed
• 7:38
  either way? That's the transition fit here.
• 7:41
  The tolerance intervals, the zones for the hole and the shaft
• 7:45
  actually overlap. Meaning the final assembly.
• 7:48
  It may result in a small clearance, or it may result in a
• 7:51
  small interference. It depends on where the actual
• 7:53
  manufactured sizes fall within their allowed ranges.
• 7:57
  So why use that? Seems uncertain.
• 7:59
  It's ideal for parts that need a really secure, reliable location
• 8:03
  and connection, but perhaps might need to be disassembled
• 8:05
  later without, you know, destroying everything.
• 8:08
  Accurate location fits like for inner bearing rings under
• 8:11
  moderate loads. OK, makes sense.
• 8:13
  Now, when designers are applying these fits, they often hear
• 8:16
  about the whole basis system, usually denoted with the capital
• 8:19
  H versus the shaft basis system with a lowercase H.
• 8:24
  If I'm setting up, say, a new production line or just
• 8:27
  standardizing practices, which one should I generally lean
• 8:30
  towards and why? Does that choice really impact
• 8:32
  manufacturing cost? Great question.
• 8:35
  You almost universally choose the whole basis system
• 8:38
  symbolized by that capital H for the holes fundamental deviation.
• 8:42
  Why the whole? The reason is purely
• 8:45
  manufacturing practicality and cost.
• 8:48
  Think about how holes are typically made.
• 8:50
  Drills, reamers. These are often fixed,
• 8:53
  standardized, sometimes expensive tools.
• 8:56
  Right? By fixing the hole size,
• 8:58
  specifically setting its lower limit deviation EI to 0, which
• 9:02
  is what the H signifies. You mean the smallest allowed
• 9:05
  hole is exactly the nominal size?
• 9:07
  Precisely by doing that you minimize the need for lots of
• 9:10
  different specialized oil making tools.
• 9:13
  You achieve the desired fit clearance transition
• 9:15
  interference by varying the tolerance on the the shaft,
• 9:18
  which is usually easier and cheaper to machine to a specific
• 9:21
  side. Got it.
• 9:22
  So you standardize the expensive, often less flexible
• 9:25
  manufacturing step making the hole and manage the actual fit
• 9:28
  by adjusting the component. That's typically easier and
• 9:31
  cheaper to customize, which is the shaft.
• 9:33
  You've got it. The shaft basis system using the
• 9:35
  small H where the shafts upper limit deviation is 0 is much
• 9:39
  less common. When might you use it?
• 9:41
  It's typically reserved for more specialized applications.
• 9:43
  Maybe you're using expensive pre finished precision ground
• 9:47
  shafting as your raw material and you don't want to machine it
• 9:50
  further. Or perhaps you have a single
• 9:52
  long shaft where you need to mount several different
• 9:55
  components, each requiring a different type of fit along its
• 9:58
  length. Then it might make sense to keep
• 10:00
  the shaft consistent and vary the holes.
• 10:02
  But generally whole basis is the way to go.
• 10:06
  Let's make this really concrete. Then for the listener, give us,
• 10:09
  say, the three most essential ISO fit codes using that
• 10:13
  preferred whole basis system that every mechanical designer
• 10:17
  should probably have in their mental toolbox.
• 10:19
  OK, three key examples using whole bases H holes.
• 10:22
  Remember holes are capital letters, shafts are lowercase.
• 10:25
  Got it. First, for a normal running fit,
• 10:27
  you need parts to move relatively freely and maybe like
• 10:29
  sleeve bearings running at decent speeds.
• 10:31
  You'd specify H7F7. That F shaft designation
• 10:35
  guarantees a defined clearance with the H7 hole.
• 10:38
  H fiven F7 go running fit? What about a press fit like for
• 10:41
  a gear? Right.
• 10:42
  If you need a gear wheel or maybe a bearing Bush pressed
• 10:45
  into a housing. Something that has to withstand
• 10:48
  high torque and basically be permanent.
• 10:50
  That requires an interference fit.
• 10:52
  A common choice would be a medium press fit like H7P6H7P6.
• 10:57
  That P shaft is designed to be consistently slightly oversized
• 11:01
  relative to the H7 hole tolerance zone.
• 11:03
  You'll need significant force. OK.
• 11:05
  And and the middle one, the transition fit like for locating
• 11:08
  something accurately but maybe needing to remove it later.
• 11:11
  Exactly for that kind of transition.
• 11:13
  Fit Secure, reliable location, but potentially removable, like
• 11:17
  positioning the inner ring of a standard ball bearing under
• 11:20
  average floating conditions. A good choice is the push fit,
• 11:23
  often specified as H7K6. H10K6.
• 11:27
  Here that K shaft tolerance zone actually overlaps the H7 whole
• 11:31
  zone. So you might get slight
• 11:32
  interference, might get slight clearance, but it provides very
• 11:35
  good centering and location without extreme.
• 11:37
  Force H77 for running, H7P6 for press fit, H7K6 for transition
• 11:41
  location. That clarity is exactly what
• 11:44
  designers. Need those 3 cover a huge range
• 11:46
  of common applications. But you know, even with perfect
• 11:49
  GD and T on the drum and the right ISO fit, specified
• 11:53
  measurement itself can still throw a wrench in the works.
• 11:56
  Oh, absolutely. Let's talk about the inspectors
• 11:58
  dilemma, especially this issue of measurement divergence.
• 12:01
  Yeah, this is where theory meets reality, often painfully.
• 12:05
  Even when the drawing is crystal clear, how the part is actually
• 12:08
  measured is surprisingly complex.
• 12:10
  How so? Different CMMS coordinate
• 12:12
  measuring machines use different software algorithms to interpret
• 12:15
  the cloud of points they collect off a surface, and different yet
• 12:19
  equally valid algorithms like say, least squares, minimum zone
• 12:23
  or maxim inscribed can measure the very same feature on the
• 12:27
  very same part and return significantly different results
• 12:30
  for things like size or position.
• 12:32
  And this isn't just theoretical, right?
• 12:33
  This has caused real problems. I seem to recall reading about
• 12:36
  AGI Develop alert like an industry warning back in 1988.
• 12:40
  That's right, GIDEP or F3AA88-O1 and.
• 12:44
  It basically shut down CMM use for verifying GD and T and some
• 12:47
  high reliability industries for a while because engineers just
• 12:50
  couldn't trust the numbers coming off different machines or
• 12:52
  different software. Exactly.
• 12:54
  The results weren't repeatable or correlatable enough between
• 12:58
  systems, leading to good parts being rejected and bad parts
• 13:01
  potentially being accepted. Why does the most common
• 13:04
  algorithm, least squares, cause so much controversy?
• 13:08
  It sounds reasonable. Well, least squares is
• 13:11
  mathematically straightforward, easy to implement in software,
• 13:14
  and it gives very repeatable results on the same machine.
• 13:17
  That's its appeal. But it works by averaging all
• 13:20
  the data points to find the best fit geometry.
• 13:22
  If there's an anomaly on the surface, like piece of dirt, a
• 13:26
  small Burr, maybe some waviness, it averages that error into the
• 13:30
  definition of the feature. So it might smooth out a
• 13:33
  critical local deviation that could actually affect function.
• 13:36
  Precisely. It might make a slightly out of
• 13:38
  spec feature look acceptable on paper because the average fits.
• 13:42
  This is why some quality professionals kind of cynically
• 13:44
  joke that relying solely on least squares means sometimes
• 13:47
  the statistically averaged wrong is best, simply because it gives
• 13:50
  you the same wrong number every time.
• 13:52
  It doesn't always reflect true functional reality.
• 13:55
  That adds a whole layer of fundamental uncertainty to every
• 14:00
  measurement. Doesn't it?
• 14:00
  It really does. Measurement is not as absolute
• 14:03
  as we sometimes like to think. OK, now if we multiply the
• 14:06
  single part measurement uncertainty across an entire
• 14:09
  assembly of multiple parts, we run head first into the massive
• 14:14
  headache of tolerance stacking or tolerance analysis.
• 14:16
  Yeah, managing the dimensional chains in an assembly is
• 14:19
  critical. You basically have two core
• 14:21
  methods for predicting the final assembly variation.
• 14:25
  What's? The first way.
• 14:26
  The first is arithmetic tolerancing, sometimes called
• 14:28
  worst case tolerancing. This is the most conservative
• 14:31
  approach. How's it work?
• 14:33
  It assumes the absolute extreme of every single tolerance in the
• 14:36
  chain combined simultaneously in the worst possible way to
• 14:39
  maximize the final assembly variation.
• 14:42
  Smallest possible dimension here, largest possible there.
• 14:45
  That sounds incredibly safe, guaranteeing 100% assembly
• 14:48
  success, but probably results in cripplingly tight and expensive
• 14:53
  tolerances on individual parts if the chain has more than a few
• 14:56
  components. It absolutely does.
• 14:59
  The required assembly tolerance grows linearly with the number
• 15:02
  of parts. It guarantees certainty, but
• 15:05
  often at an unfeasible manufacturing cost.
• 15:07
  So what's the more practical alternative?
• 15:09
  The far more common, pragmatic and cost effective approach is
• 15:13
  statistical tolerancing, often using the root sum of squares or
• 15:18
  RSS method. Right where?
• 15:19
  You're basically making a statistical bet, aren't you?
• 15:22
  You are. You assume that the variations
• 15:24
  on individual parts are independent and random, often
• 15:28
  following something like a normal distribution centered in
• 15:30
  their tolerance zones. And the bet is that the
• 15:33
  deviations will usually tend to cancel each other out, some
• 15:36
  parts slightly big, some slightly.
• 15:37
  Small. Exactly.
• 15:39
  It's statistically very unlikely that all parts in an assembly
• 15:42
  will be at their absolute worst case tolerance limit
• 15:45
  simultaneously. And the benefit is?
• 15:48
  The huge benefit is that the predicted assembly variation
• 15:51
  grows much more slowly, only proportional to the square root
• 15:54
  of the number of parts. Statistical tolerancing allows
• 15:57
  designers to assign wider, cheaper manufacturing tolerances
• 16:01
  to the individual components while still achieving a very
• 16:04
  high predictable level of certainty, say 99.73% for a + -,
• 16:09
  3 Sigma process in the final assembly outcome.
• 16:12
  OK, so the practical device for designers seems pretty clear
• 16:15
  then Don't just slapped tight tolerances everywhere.
• 16:17
  Please don't use the general tolerances in your title block
• 16:21
  like ISO 2768 medium or fine or equivalent ASME general notes
• 16:26
  For features that aren't critical to function or
• 16:28
  assembly, let those float within reasonable shot practice and
• 16:32
  then and only call out specific GD and T controls like true
• 16:35
  position, maybe line profile or surface profile for complex
• 16:38
  shapes for those features that are absolutely critical to
• 16:40
  function interchangeability or the key assembly interfaces.
• 16:43
  If you over tolerance everything.
• 16:45
  If you put tight GD and T on non critical features, you are
• 16:49
  forcing specialized manufacturing processes, complex
• 16:52
  inspection setups, maybe mandatory CMM checks
• 16:55
  unnecessarily. You're just multiplying your
• 16:57
  cost, often significantly, without adding any real
• 17:01
  functional value to the product. Makes perfect sense.
• 17:04
  Use precision where it matters, not everywhere.
• 17:06
  That's the essence of cost effective design.
• 17:08
  So wrapping up, we've covered quite a bit the evolution in GD
• 17:12
  and T standards like ASME Y 14.5, getting rid of ambiguity,
• 17:17
  the fundamental principles of the ISO fit system and how
• 17:20
  picking the right notation like H7F7 or H7P6 is really the key
• 17:24
  to predictable cost effective interchangeability.
• 17:27
  We have understanding that language is crucial.
• 17:29
  But if there's one single #1 statistic that really justifies
• 17:34
  mastering all this complexity, especially the GD and T side, it
• 17:37
  seems to come back to the power of true position tolerance.
• 17:40
  Oh. Yes, the bonus tolerance.
• 17:42
  That is probably the greatest hidden efficiency bonus unlocked
• 17:45
  by GD and T. Explain that again, how does
• 17:47
  true position give you bonus tolerance?
• 17:49
  OK, think about locating a hole using traditional coordinate
• 17:52
  tolerancing, like X + -, .1 and y + -, .1.
• 17:57
  That defines a square tolerance zone for the center of the hole
• 18:01
  0.2 by .2. Right, a square.
• 18:03
  But true position using the position symbol defines A
• 18:06
  cylindrical or circular tolerance zone around the
• 18:09
  theoretically exact location. OK, a circle instead of a
• 18:12
  square. So that circular zone has a
• 18:14
  diameter equal to the diagonal of the old square zone.
• 18:18
  It fully utilizes the corners that were previously unusable
• 18:21
  with coordinate tolerancing. Which means the total
• 18:24
  permissible area where the holes axis can actually fall is
• 18:28
  bigger. Much bigger.
• 18:29
  That shift from a square zone to an equivalent circular zone,
• 18:32
  which GD and T enables, results in up to a 57% increase in the
• 18:37
  usable tolerance area for that holes location compared to the
• 18:40
  coordinate method with the same deviation limits along the axis.
• 18:42
  57%, that's huge. It's massive, this hidden
• 18:47
  volume, this extra geometric space you game translates
• 18:50
  directly into greater manufacturing flexibility, fewer
• 18:53
  parts being scrapped for positional errors, and
• 18:55
  ultimately significant cost savings, especially in high
• 18:58
  volume production. That increase in usable
• 19:01
  tolerance area is genuinely shocking, especially when
• 19:04
  engineers are often fighting for just a few extra microns or
• 19:08
  tenths of a thou in other dimensions.
• 19:10
  It really proves that mastering these standards isn't just
• 19:12
  academic, it's essential for maximizing efficiency.
• 19:16
  Absolutely. It's money left on the table if
• 19:18
  you're not using it correctly. So for you, our listener, the
• 19:21
  immediate call to action seems crystal clear after this
• 19:24
  discussion. It really is confirmed today if
• 19:26
  you don't already know for sure whether your organization, your
• 19:28
  department, your key suppliers operate primarily under the
• 19:31
  ASMEY 14.5 standard or the ISO 286-1101 system.
• 19:37
  Because knowing that difference. Understanding that fundamental
• 19:40
  difference, especially regarding how size is controlled and
• 19:43
  whether you need that envelope symbol with ISO, is probably the
• 19:46
  single most critical first step. It impacts your design choices,
• 19:49
  your tolerance analysis, and how you interpret inspection
• 19:52
  reports. Get that wrong and you're
• 19:54
  building risk right into your design process.