When “Within Tolerance” Isn’t Trustworthy: Measurement Errors, Calibration & Gage R&R

WEBVTT
00:00:03.714 --> 00:00:08.085
You're listening to china manufacturing decoded from the Sofeast group
00:00:13.664 --> 00:00:19.847
hello it's episode 350 today and we're looking at a quality issue that's easy
00:00:19.847 --> 00:00:25.797
to overlook sometimes those are the measurements that your factory gives to
00:00:25.797 --> 00:00:28.597
you can you actually trust them so
00:00:29.517 --> 00:00:33.867
even if they tell you that a part that you've ordered is within tolerance different
00:00:33.867 --> 00:00:37.417
equipment the operators working on the line
00:00:37.997 --> 00:00:43.197
measurement methods these could all produce different results and ultimately
00:00:43.197 --> 00:00:47.197
that might lead you to bad quality decisions so
00:00:47.797 --> 00:00:53.907
i need someone to help to explain this in a bit more detail and paul adams has
00:00:53.907 --> 00:00:58.413
joined the fray once more to explain where these sort of measurement errors
00:00:58.860 --> 00:01:04.247
might come from and how maybe better to control them so paul hello
00:01:05.122 --> 00:01:10.317
Hello adrian good to be back 350 i can remember i'm way back in the day so i
00:01:10.317 --> 00:01:12.137
can't remember what hit 300 i'll
00:01:12.137 --> 00:01:17.157
probably say that at the same 250 yes 350 that's a milestone it is it's a.
00:01:17.157 --> 00:01:19.597
Milestone yeah it's a and to
00:01:19.597 --> 00:01:20.881
You for hosting probably all of them.
00:01:21.224 --> 00:01:24.987
Six years almost all almost all yeah Yeah, Renault sometimes as well.
00:01:24.987 --> 00:01:31.317
But like, it's been about six years of free manufacturing knowledge for the world.
00:01:31.317 --> 00:01:35.377
And we don't use ads, we don't do anything annoying like that.
00:01:35.377 --> 00:01:40.217
So yeah, I mean, I'm grateful for everybody that's kept tuning in from the start
00:01:40.217 --> 00:01:41.719
and for our new listeners as well.
00:01:42.790 --> 00:01:47.904
Indeed, indeed. Yeah, yeah. It's just saying, it's all knowledge that you're sharing.
00:01:48.404 --> 00:01:52.159
Yeah, and hopefully you can learn from this and avoid some mistakes.
00:01:52.804 --> 00:01:57.036
Yeah, absolutely. That is the name of the game here, Paul. You know,
00:01:57.036 --> 00:02:01.626
we've got people with lots of experience in our team, like yourself,
00:02:01.626 --> 00:02:06.801
and that's what we're trying to tap into and share it. So, yeah, absolutely.
00:02:07.341 --> 00:02:12.605
So you mentioned trying to help people avoid mistakes. That's kind of what today's topic's about.
00:02:13.427 --> 00:02:17.514
It is. Yeah, that's a nice sort of segue straight into it. It's really.
00:02:18.043 --> 00:02:22.226
So yeah, let's keep on. There's a few things that we should be covering today.
00:02:22.226 --> 00:02:27.706
So I'll break it down into a number of different sections, let's say,
00:02:27.706 --> 00:02:29.593
or points that we can discuss.
00:02:29.895 --> 00:02:34.226
This is important to understand because when someone says the dimension is okay,
00:02:34.226 --> 00:02:37.592
he says, yeah, it's okay, and it says on the report, it's okay,
00:02:38.466 --> 00:02:43.246
but then you start to dig into it and say, okay, then the report last batch
00:02:43.246 --> 00:02:45.166
was okay, but it's okay, it's okay.
00:02:45.506 --> 00:02:48.275
Is this a copy and paste or is it actually real?
00:02:48.643 --> 00:02:53.096
But you really need to start to understand some questions with respect to what
00:02:53.096 --> 00:02:57.486
the tolerances are on your part, which we've talked about immensely in various
00:02:57.486 --> 00:03:00.907
other podcasts, but also how,
00:03:02.126 --> 00:03:08.384
the dimension's being measured and who is measuring it, and what's the accuracy
00:03:08.680 --> 00:03:10.746
of the measurement gauge.
00:03:10.746 --> 00:03:14.664
For example, if you've got something with a really tight tolerance.
00:03:15.952 --> 00:03:18.316
Just going around the houses of the environment, I'll get into the meat in a
00:03:18.316 --> 00:03:24.197
minute, but if you've got a tight tolerance for 0.001 as someone gets out their
00:03:24.423 --> 00:03:27.186
steel rule i'd be a little bit concerned,
00:03:27.993 --> 00:03:35.049
So there's certain pieces of equipment required for certain tolerances and types
00:03:35.049 --> 00:03:38.870
of measurement, again, which we'll get into this itself.
00:03:39.195 --> 00:03:43.195
So let's crack on. So measurements itself can vary.
00:03:43.508 --> 00:03:49.162
And what I mean by this is that a reading is not automatically the truth.
00:03:49.830 --> 00:03:53.447
Let's call it the standard 12.5. Okay.
00:03:54.130 --> 00:03:57.294
So what does 12.5 mean? is at 12.5.
00:03:58.037 --> 00:04:04.863
And if you look at the dimensional tolerance box on your 2D drawings,
00:04:05.275 --> 00:04:10.354
you'll see that if it's 12.5, in other words, one digital point,
00:04:11.975 --> 00:04:15.377
it's going to be a certain tolerance. In other words, it's going to be fairly relaxed.
00:04:15.731 --> 00:04:21.425
If it's 12.50, in other words, two digital sort of places past the digital point,
00:04:22.614 --> 00:04:25.702
it's going to be a tighter tolerance. If it's 0.30.
00:04:28.129 --> 00:04:32.529
So 0.500, it's going to be even tighter. So, again, we've gone on about this
00:04:32.529 --> 00:04:37.759
with respect to what is required on a particular sort of tolerance and what
00:04:37.759 --> 00:04:40.551
equipment can be used for each,
00:04:41.485 --> 00:04:47.645
tolerance that is being inspected or measured and what is required to actually report to.
00:04:47.946 --> 00:04:55.499
So, if your tolerance is actually 12.5000, that's a very, very tight tolerance.
00:04:55.772 --> 00:04:58.946
And you're not going to be measuring that with a steel rule.
00:04:59.218 --> 00:05:03.416
You're not going to be measuring that with a digital vernier or a caliper.
00:05:03.700 --> 00:05:07.569
You're going to need some specialized equipment like a three-year record measure
00:05:07.569 --> 00:05:08.640
machine or something like this.
00:05:09.312 --> 00:05:17.279
However, if it is 12.5, then your basic digital caliper will be okay.
00:05:17.289 --> 00:05:18.640
It's going to be probably okay.
00:05:19.638 --> 00:05:27.059
This is all about how repeatable that is. And again, we're going to come into this a little bit later.
00:05:27.309 --> 00:05:32.383
But the results can change with the equipment that's being used.
00:05:32.383 --> 00:05:37.218
So if I'm measuring with my digital caliper, Adrian, you're measuring it with yours.
00:05:37.693 --> 00:05:41.487
The guy down the road in the machine shop, he's got his own.
00:05:41.946 --> 00:05:45.603
Guess what? We're all going to get slightly different readings. Okay. Why?
00:05:45.968 --> 00:05:51.118
It's the way I use mine. It's going to be slightly different to yours. It's the way I hold it.
00:05:51.599 --> 00:05:55.663
It's maybe the temperature. Maybe it's the conditions, you know,
00:05:55.663 --> 00:05:58.873
of the actual part itself. Has it been cleaned?
00:05:59.407 --> 00:06:04.263
Yeah, has the vernier itself been cleaned, the faces? Were they just picked
00:06:04.263 --> 00:06:06.534
up off a dusty bench and used right away?
00:06:06.795 --> 00:06:11.203
So all of these aspects have to come into play with respect to repeatability
00:06:11.203 --> 00:06:15.497
and accuracy, you know, and this is why the dimensions can change.
00:06:16.211 --> 00:06:22.161
And you need to understand why that is. And one of those reasons is different equipment being used.
00:06:22.643 --> 00:06:25.954
We've all got digital verniers and callipers, okay?
00:06:26.251 --> 00:06:29.603
But mine's a different brand. Yours is a different brand. The inspector down
00:06:29.603 --> 00:06:32.590
the road is a different brand. The machine shop guy is a different brand.
00:06:32.920 --> 00:06:38.653
Yeah, different, you know, each of these pieces of equipment will have its own
00:06:38.653 --> 00:06:41.843
manufacturing tolerance built into it. So that's one thing.
00:06:42.354 --> 00:06:45.633
Again, I've talked about the way I measure it, the way you measure it,
00:06:45.633 --> 00:06:48.013
the way I handle it. Am I left-handed, right-handed?
00:06:48.893 --> 00:06:52.903
All of that comes into play and it's really, really important to understand
00:06:52.903 --> 00:06:56.842
that because we really should be setting down, and again, we'll get to this,
00:06:57.235 --> 00:07:02.012
a standard for how parts are inspected. And that includes...
00:07:02.776 --> 00:07:07.821
Handling, orientation, and methodology on how to inspect it.
00:07:09.410 --> 00:07:12.076
And again, we've already talked about, and I've quickly touched on,
00:07:12.355 --> 00:07:16.600
the environment. If I'm in a freezing room or a temperature-controlled room,
00:07:16.600 --> 00:07:20.592
that's going to be totally different to, or potentially totally different to,
00:07:20.743 --> 00:07:22.798
the hot and sweaty machines you have where it's actually made.
00:07:26.378 --> 00:07:31.150
Because materials do actually flex a little bit in different sort of temperatures,
00:07:31.382 --> 00:07:34.336
especially if they're extreme sort of temperatures.
00:07:34.737 --> 00:07:40.844
So this is one of the things we need to look at. It's, yep, don't take your first...
00:07:41.589 --> 00:07:47.547
Dimension as gospel okay i always when i've got a part in my hand i always measure
00:07:47.547 --> 00:07:51.417
it two or three times just to make sure i'm getting the same reading and i'll
00:07:51.417 --> 00:07:54.382
always make sure i'm trying to do it whether it's be a
00:07:54.817 --> 00:07:58.717
micromatic caliper or whatever the case may be try and hold it exactly the same
00:07:58.717 --> 00:08:03.717
way the way i've always done okay at least then i could see my own variation
00:08:03.717 --> 00:08:07.387
with respect to how much force i'm putting on whether i cleaned it or not etc
00:08:07.387 --> 00:08:09.561
etc so this is really important.
00:08:10.207 --> 00:08:15.950
And we've already talked about with respect to the different types of people that are measuring it.
00:08:16.205 --> 00:08:21.897
Again, this is super, super important when we come into a process in order to
00:08:21.897 --> 00:08:26.538
check all of this, which is called gauge R&R, or repeatability reliability.
00:08:26.695 --> 00:08:28.889
And again, cover that in a second.
00:08:29.202 --> 00:08:34.740
But there's a process that needs to be followed, which includes multiple people,
00:08:35.394 --> 00:08:39.156
measuring the same product with the same device.
00:08:39.968 --> 00:08:43.051
So we talked about a digital caliber, so I've got my digital caliber.
00:08:43.341 --> 00:08:45.037
Well, my digital caliber is going
00:08:45.037 --> 00:08:50.525
to be the benchmark for different people measuring the same component.
00:08:51.337 --> 00:08:58.097
It will measure that same dimension maybe 40, 50 times, and each different person
00:08:58.097 --> 00:09:00.728
will write down there, and therefore we've got a good spread.
00:09:01.397 --> 00:09:07.892
So this gives us a total variability on that, which is really important to understand.
00:09:08.153 --> 00:09:13.941
So different people measuring different things could potentially give you a different result.
00:09:15.323 --> 00:09:19.374
It's a small thing, but it's a minefield.
00:09:20.094 --> 00:09:24.937
Well, yeah, it certainly can be. It certainly can be. And I've got a couple
00:09:24.937 --> 00:09:27.995
of case studies later on that we can talk about.
00:09:28.697 --> 00:09:38.334
But one of those is not a digital caliper, this is an air pressure gauge for measuring a system,
00:09:39.197 --> 00:09:43.789
which I can sort of talk about the headaches we had with that one.
00:09:44.126 --> 00:09:44.337
Hmm.
00:09:45.994 --> 00:09:46.476
Okay.
00:09:46.668 --> 00:09:53.517
So, right. The next thing I want to talk about, Adrian, is measurement system reliability, really.
00:09:53.772 --> 00:09:57.158
And this is really all about making the right equipment for,
00:09:57.158 --> 00:10:01.673
or using the right equipment for the tolerances being checked.
00:10:02.488 --> 00:10:06.258
I quickly mentioned this earlier. I'm not going to expect anybody to measure
00:10:06.258 --> 00:10:08.807
a really tight tolerance with a steel roll, for example.
00:10:09.368 --> 00:10:13.427
So we need to make sure that we're using the suitable and correct equipment
00:10:13.723 --> 00:10:19.841
to measure the correct, or to give us the correct output on a particular dimension.
00:10:20.505 --> 00:10:25.008
Okay, so, and again, this is really important. And this is where we talk about
00:10:25.008 --> 00:10:30.518
the 12.5000, and this is all about the resolution of the actual measuring equipment.
00:10:30.518 --> 00:10:33.128
Can it actually measure that accuracy?
00:10:33.575 --> 00:10:38.388
Or is it, you know, down to the steel rule, if my steel rule says that it's
00:10:38.388 --> 00:10:42.258
broken up into half millimeters, well, the best stuff I'll probably get is a
00:10:42.258 --> 00:10:43.795
quarter of a millimeter accuracy.
00:10:44.033 --> 00:10:45.809
You know, it's somewhere in the middle of the line.
00:10:46.407 --> 00:10:51.148
No better than that, that's not exactly what we're looking for on some of the
00:10:51.148 --> 00:10:56.903
dimensions that are required, particularly on the 12.500, which is a high tolerance.
00:10:58.035 --> 00:11:02.763
Calibration is another thing we need to make sure that we're getting the correct
00:11:02.763 --> 00:11:06.825
equipment okay so if we go for our 12.500.
00:11:08.073 --> 00:11:11.173
So that's three decimal points okay past the decimal point we're,
00:11:11.949 --> 00:11:16.583
going to be using a 3nm okay a three-coordinate measurement machine okay now
00:11:16.583 --> 00:11:23.228
what's the last time we calibrate yeah because these things do go out of calibration now it's important that.
00:11:24.573 --> 00:11:29.363
The the equipment manufacturer specifies this piece of equipment should be calibrated
00:11:29.363 --> 00:11:31.865
every year, every six months, or whatever the case may be.
00:11:32.340 --> 00:11:38.823
But it's good practice that every day, right, someone comes in,
00:11:38.823 --> 00:11:43.006
you know, the operator of the 3MF, it's recorded on a measuring machine, will come in.
00:11:43.192 --> 00:11:46.203
They'll take a standard slip block, you know, which is a very,
00:11:46.203 --> 00:11:47.587
very high accuracy, you know,
00:11:48.505 --> 00:11:52.058
A piece of steel set to a certain dimension, very, very high accuracy,
00:11:52.383 --> 00:11:53.289
and now we'll measure that.
00:11:53.567 --> 00:11:58.885
And just to make sure that, yeah, we're getting that 12.500 accurately, no problems.
00:11:59.024 --> 00:12:03.679
Now, okay, now we're going to start the day knowing that every single dimension
00:12:03.679 --> 00:12:05.321
we measure is going to be correct.
00:12:06.925 --> 00:12:11.621
So again, it's calibration and making sure that is calibrated and recorded,
00:12:12.097 --> 00:12:15.284
okay? Because this is really important for an audit trail.
00:12:15.655 --> 00:12:20.520
So if you, as a customer, say, you know what, I'm building a medical product,
00:12:21.329 --> 00:12:25.860
and I need all of the paperwork associated with building the product,
00:12:26.174 --> 00:12:27.749
well, this is going to be part of that.
00:12:27.749 --> 00:12:31.949
This is going to be part of that technical pack that is required for the contract
00:12:31.949 --> 00:12:35.222
manufacturer to say, you know what, we measured your product,
00:12:35.881 --> 00:12:42.209
and we used this particular machine here, and it was last calibrated July 2026.
00:12:42.459 --> 00:12:45.737
And I go, okay, well, it's within the required specification.
00:12:46.596 --> 00:12:50.706
You know, from the manufacturer of the actual machine itself.
00:12:51.348 --> 00:12:55.669
So really important to understand and check because time and time again,
00:12:55.669 --> 00:13:00.853
I've seen you walk around machine shops and you see a Vernier laying on a bench
00:13:01.178 --> 00:13:06.211
and you turn it over expecting to see a calibration record sticker on the back.
00:13:06.403 --> 00:13:10.547
In other words, the identification of the actual product itself or the vernier,
00:13:11.023 --> 00:13:13.328
and the date it was last checked and calibrated.
00:13:13.734 --> 00:13:16.229
Some of these do not have stickers on our tool. In other words,
00:13:16.229 --> 00:13:21.403
they're not calibrated. It could be knocked around, dropped on the floor, who knows.
00:13:21.960 --> 00:13:27.339
And if the toolmaker is making your injection mold tool and is measuring it
00:13:27.339 --> 00:13:32.448
with this bit of kit, that's a bit of a red flag.
00:13:32.727 --> 00:13:36.860
So, it's important to understand calibration even down to the simple measuring
00:13:37.168 --> 00:13:39.815
pieces we put in such as your digital calibre.
00:13:42.412 --> 00:13:47.504
And the other one is really talking about defining the measurement method,
00:13:47.707 --> 00:13:51.532
the methodology on how you're going to measure something. You know what I mean?
00:13:54.614 --> 00:13:58.908
If you've got the tight tolerances, there's no question. You do need the specialized
00:13:58.908 --> 00:14:01.499
equipment, something like the three-quarter measure machine.
00:14:02.021 --> 00:14:05.548
And you would have jigs and fixtures potentially to hold something,
00:14:05.548 --> 00:14:07.670
you know, particularly orientation.
00:14:08.012 --> 00:14:11.774
And that's going to use time and timeout on this particular component.
00:14:12.389 --> 00:14:15.251
For example, if you've got a gear.
00:14:16.128 --> 00:14:23.719
Now, gears, I think it's just a gear. Well, a gear face or the face of the tooth is very complex.
00:14:24.125 --> 00:14:28.948
Okay. And it's very important that we actually get the correct measurement points
00:14:28.948 --> 00:14:32.096
that you know how to actually measure it and where to actually measure it.
00:14:32.531 --> 00:14:35.508
And this is the methodology. So all this is going to be written down,
00:14:35.508 --> 00:14:39.921
in other words, your standard operating procedure on how to measure something.
00:14:40.362 --> 00:14:44.418
Okay. So if we talk about standard operating procedures, we also basically think
00:14:44.418 --> 00:14:45.668
about the production line.
00:14:45.999 --> 00:14:50.848
Okay. and how part A goes with part B. You know, you screw it up to a particular
00:14:50.848 --> 00:14:52.698
tool, special standard operating procedure.
00:14:52.912 --> 00:14:58.048
Well, it goes along with inspection as well, okay, because we have to have that
00:14:58.048 --> 00:15:01.689
standard operating procedure with respect to how selection may be something
00:15:02.061 --> 00:15:04.598
and making sure that's defined every single time.
00:15:04.975 --> 00:15:10.528
Because if we've got that gear face, which is very complex, and we've got a
00:15:10.528 --> 00:15:15.788
different guy coming in there on a Monday, because the original guy's off sick,
00:15:15.788 --> 00:15:19.108
or the standard guys are sick, how's it going to measure it?
00:15:19.108 --> 00:15:21.348
It's going to potentially measure all over the place and say,
00:15:21.348 --> 00:15:25.298
yeah, it's all good, no problems. But the dimensions, if you look from batch
00:15:25.298 --> 00:15:27.482
one to batch two, they don't correlate.
00:15:27.644 --> 00:15:30.518
They're not the same. Why? Because it measures it in a different place.
00:15:30.814 --> 00:15:37.495
So it's really important that we have a clearly defined methodology on how to measure something.
00:15:38.743 --> 00:15:45.021
It doesn't necessarily have to be a 3mm. It could actually be use your digital calibrate, a caliber.
00:15:45.701 --> 00:15:50.910
To measure the outside dimension of a cylinder or a block or whatever the case may be.
00:15:51.178 --> 00:15:54.915
So, so it's written down, measure between these points, and this is what you're expecting.
00:15:55.621 --> 00:16:02.001
Anybody can actually do it, okay? And repeatability comes in here because I'm
00:16:02.001 --> 00:16:03.272
going to measure it the same way.
00:16:03.486 --> 00:16:05.623
Now you can measure it the same way.
00:16:06.581 --> 00:16:10.011
So whereas before, we said, what's the dimension? Well, we're just having a
00:16:10.011 --> 00:16:12.821
guess. You know, it's like, well, I'm going to measure it across this face.
00:16:12.821 --> 00:16:15.821
You know, you're going to measure it across a slightly different face or put
00:16:15.821 --> 00:16:20.029
it on the desk and use, you know, the point that comes down.
00:16:20.736 --> 00:16:23.781
It's a slightly different method. Yeah. Same result. Yeah, sort of.
00:16:23.781 --> 00:16:25.746
But it's not exactly what we need.
00:16:26.076 --> 00:16:28.172
Yeah, sort of is not good enough.
00:16:29.141 --> 00:16:30.231
Correct, yeah.
00:16:30.231 --> 00:16:33.452
This is not a surprise to me that in the world of manufacturing,
00:16:33.858 --> 00:16:37.573
even this relatively small process has to have
00:16:38.401 --> 00:16:42.421
a clearly defined method in order to make sure that we're getting the results
00:16:42.421 --> 00:16:45.444
that we expect. So, yeah. Correct.
00:16:46.281 --> 00:16:53.298
Yeah, we definitely need to know. And again, we're talking about inspectors. It's quality control.
00:16:54.314 --> 00:17:01.161
We really do need qualifiers, fully trained, certifiers, inspectors,
00:17:01.161 --> 00:17:02.801
particularly when we talk about
00:17:02.801 --> 00:17:06.446
running something like a three-coordinate major machine. Could I do it?
00:17:07.961 --> 00:17:11.461
Yeah, I could, but it would take me an awful long time. Would I make mistakes?
00:17:11.461 --> 00:17:14.165
Yes, I know I would make mistakes, but would I get the result?
00:17:14.565 --> 00:17:17.842
It's probably going to take me 10 times longer than someone's qualified to do it.
00:17:18.714 --> 00:17:21.355
It's always better to have the right person doing the right job,
00:17:21.483 --> 00:17:24.771
in my opinion. You know what I mean? Of course. You know, can I drive a forklift?
00:17:24.771 --> 00:17:26.401
Yeah, I can. Am I certified? No.
00:17:26.881 --> 00:17:30.045
So would I do it? No, I would not. You know what I mean? So,
00:17:30.172 --> 00:17:32.570
you know, it's getting the right person to do the right job.
00:17:32.721 --> 00:17:35.984
And clearly, you know, this is a specialized skill.
00:17:36.233 --> 00:17:39.741
So it's always best to have, and we do, you know, a gillion,
00:17:39.741 --> 00:17:46.056
you know, fully qualified inspectors and all the quality team are well trained and certified.
00:17:46.311 --> 00:17:49.736
Yes. Yeah, absolutely. Yeah, which is good.
00:17:50.789 --> 00:17:54.521
Right, the next thing is, this is basically leading through a nice bar,
00:17:54.521 --> 00:17:58.479
you know. I've already talked about gauge R&R, you heard me talk about that earlier.
00:17:59.158 --> 00:18:08.192
Now, this is all about whether a process is repeatable when measuring something,
00:18:08.808 --> 00:18:11.285
okay? And does it give consistent results?
00:18:12.173 --> 00:18:20.376
Let's take our, whatever it was, okay. I've got an earbud case, okay?
00:18:20.956 --> 00:18:24.584
So, right, this is the dimension we need to measure. Okay?
00:18:24.875 --> 00:18:27.606
So we're going to measure that 50 times. We're all going to measure that 50
00:18:27.606 --> 00:18:31.106
times. They're not going to pass this to you along with the measuring device
00:18:31.106 --> 00:18:35.852
and say, right, now, Adrian, you measure this 50 times and you record down your results.
00:18:36.200 --> 00:18:39.654
We're then going to do the same with the third person, fourth person, fifth person.
00:18:39.921 --> 00:18:45.507
Okay? Same product, same device. Right? Okay? We've now got a lot of data points.
00:18:45.762 --> 00:18:46.174
Okay?
00:18:46.418 --> 00:18:49.390
Five times 50, you know? And it's a lot of data points.
00:18:49.675 --> 00:18:54.246
Now, ideally, what we're going to do then is run some statistical analysis on
00:18:54.246 --> 00:18:58.406
that, see what our variation is. In other words, our sigma variation.
00:18:58.714 --> 00:19:05.077
Now, if we've got our gauge, in other words, in this case, it's our caliper,
00:19:05.571 --> 00:19:10.846
okay, is accurate, and we are following our standard operant procedure.
00:19:10.846 --> 00:19:14.671
In other words, we're measuring across this face, not this face,
00:19:14.932 --> 00:19:18.836
and we're using the roller wheel, not pushing it in. We're not doing it on the
00:19:18.836 --> 00:19:21.881
floor or the desk or anything like this. This is how we measure it.
00:19:22.532 --> 00:19:27.431
Write down or record in the spreadsheet, wherever your results are.
00:19:27.803 --> 00:19:32.226
From there, that's going to allow us to see whether the actual gauge itself,
00:19:32.488 --> 00:19:36.674
whether that be digital caliper, whether that be...
00:19:37.749 --> 00:19:43.641
And I'll talk about our case study later, whether that be your basic steel rule or,
00:19:44.221 --> 00:19:49.428
right up to your three-point measure machine, we will be able to see how repeatable
00:19:49.742 --> 00:19:54.687
that device can measure a certain product or dimension,
00:19:55.370 --> 00:19:59.831
given the same set of instructions over different operators.
00:19:59.831 --> 00:20:03.751
And this is where it's really important to understand because the actual device
00:20:03.751 --> 00:20:05.318
itself is the same device.
00:20:05.481 --> 00:20:09.492
It's not going to change between me passing it to you and you passing it to the other people.
00:20:10.009 --> 00:20:13.556
I guess it's the same products, the same piece of measurement equipment.
00:20:14.247 --> 00:20:20.181
But I may be a little bit more heavy-handed than you, and you may be a little
00:20:20.181 --> 00:20:21.956
bit more heavy-handed than the next person.
00:20:22.130 --> 00:20:24.133
The next person may be a really light touch.
00:20:24.522 --> 00:20:30.161
Who knows? Yeah, but this is all about getting consistent, repeatable results
00:20:30.161 --> 00:20:33.044
from a piece of equipment. This is your gauge R&R.
00:20:34.449 --> 00:20:39.934
So we're looking at the measuring equipment, we're looking at the operator itself,
00:20:40.480 --> 00:20:44.665
measuring the same product with the same methodology.
00:20:46.402 --> 00:20:50.384
Does it give us the same result time in, time out?
00:20:50.802 --> 00:20:56.319
If it does, then great. We know that this product and the process can give us,
00:20:57.207 --> 00:21:02.431
the same result regardless of the user. And I think those three aspects really
00:21:02.431 --> 00:21:04.642
come into what Gage R&R is all about.
00:21:05.716 --> 00:21:11.603
Before we go on, I just want to give you a quick two minutes on a case study, okay?
00:21:11.719 --> 00:21:14.726
It's had nothing to do with digital power.
00:21:15.132 --> 00:21:18.437
This was a product that we're doing for a particular customer.
00:21:19.011 --> 00:21:24.690
And in this particular case, we're testing a system which has a lot of components
00:21:24.690 --> 00:21:27.579
inside with air pressure, okay?
00:21:27.776 --> 00:21:31.242
Now, we were doing pressure loss over time.
00:21:31.718 --> 00:21:33.980
So we'll load the system up, hold
00:21:33.980 --> 00:21:37.976
it for a period of time at a certain pressure. If that pressure drops.
00:21:40.026 --> 00:21:46.520
The product failed. So how would we use the example of a gauge R&R?
00:21:46.520 --> 00:21:51.680
Well, our gauge in this case is the test equipment that loaded up the air.
00:21:51.680 --> 00:21:53.295
In other words, it pressurized the system.
00:21:53.655 --> 00:21:55.877
The gauge came up to a certain pressure level.
00:21:57.009 --> 00:22:00.470
We held that for a certain period of time. This is the piece of equipment,
00:22:00.470 --> 00:22:05.660
this is the gauge, and what we've done then is got one person to actually do
00:22:05.660 --> 00:22:07.900
that, recorded down some results.
00:22:07.900 --> 00:22:12.073
We've got a different operator coming in, doing exactly the same process,
00:22:12.334 --> 00:22:14.180
nothing changed, same product.
00:22:14.980 --> 00:22:18.952
So we didn't change the product in between, so we're still using our same product.
00:22:19.852 --> 00:22:24.701
They listed down this. The variability on this thing was insane.
00:22:24.950 --> 00:22:26.599
In other words, it was not reliable.
00:22:27.236 --> 00:22:28.675
It was totally not reliable.
00:22:29.784 --> 00:22:35.096
It wasn't until we actually carried that exercise out that we understood that
00:22:35.096 --> 00:22:39.486
the products that we were testing, or the gauge, in this case,
00:22:39.486 --> 00:22:43.239
the pressure system, we were getting failed units.
00:22:44.096 --> 00:22:46.159
We were failing the customer's products.
00:22:47.152 --> 00:22:50.796
But actually, it wasn't a fail. It was the actual piece of equipment that we
00:22:50.796 --> 00:22:53.287
were using. So we actually revisited that.
00:22:53.640 --> 00:22:56.298
We made some changes. We done some updates.
00:22:57.558 --> 00:23:02.415
And then we'd done that exercise again. And all of a sudden, bang, we've got it.
00:23:03.025 --> 00:23:07.186
We understood what the issue was. We addressed those issues because we were
00:23:07.186 --> 00:23:10.416
looking at a fairly complex sort of gauge in this case.
00:23:10.764 --> 00:23:15.333
We had to make some modifications. okay those implications meant that,
00:23:16.065 --> 00:23:23.611
we can now get a repeatable result and if a product did fail it was true fail and we went from,
00:23:24.122 --> 00:23:28.706
products we don't know whether it's good or not to right we know we've now got
00:23:28.706 --> 00:23:33.746
a 98 or whatever it was you know pass rate yeah and these two here are a faulty
00:23:33.746 --> 00:23:37.096
problem here whatever case may be change it out and we're going to actually
00:23:37.496 --> 00:23:41.508
repeat the test again until we've got virtually 100% pass rate.
00:23:42.222 --> 00:23:46.071
So without going through the gauge and R&R, we would have been lost in the dark.
00:23:46.262 --> 00:23:48.931
And the customer would say, my products are failing.
00:23:49.299 --> 00:23:52.736
Well, they're not. Well, how come you're putting them out in the field where
00:23:52.736 --> 00:23:54.323
you said they were good and they're failing?
00:23:54.687 --> 00:23:57.976
Luckily, that didn't happen, but it could well have done unless we actually
00:23:57.976 --> 00:24:04.676
went through this gauge R&R system. So here's a quick case study on why we say
00:24:05.056 --> 00:24:06.487
gauge R&R very seriously.
00:24:07.062 --> 00:24:08.666
Well, I think that's great. Good case study.
00:24:08.676 --> 00:24:11.491
Yeah, so that was a ruin.
00:24:13.062 --> 00:24:19.020
So, the next thing I want to talk about is the factory making decisions of using
00:24:19.020 --> 00:24:21.910
unreliable measurements, and I sort of touched on that,
00:24:22.730 --> 00:24:29.882
in the case study, because if you've got a system or a measuring system that
00:24:30.016 --> 00:24:33.040
does not give you acceptable parts,
00:24:33.733 --> 00:24:37.425
I'm sorry, acceptable results when measuring a part,
00:24:38.075 --> 00:24:45.372
you could actually fail a good product or get a quality escape on a bad part,
00:24:45.598 --> 00:24:49.273
which is exactly what we were trying to establish on this case study.
00:24:49.921 --> 00:24:54.590
This was in the pre-production run, so luckily it wasn't at production stage.
00:24:54.590 --> 00:24:58.150
We managed to find this, and that was a big.
00:24:59.783 --> 00:25:04.560
Not a relief, but a lesson learned with respect to, this is definitely worth
00:25:04.560 --> 00:25:08.770
doing on every single piece of equipment, particularly when it comes to the
00:25:08.770 --> 00:25:11.317
more complex inspection pieces.
00:25:11.870 --> 00:25:16.792
Like this specialized, put together, customized test tube.
00:25:17.221 --> 00:25:20.468
It wasn't necessarily your standard digital caliber, for example.
00:25:20.892 --> 00:25:26.672
So the engineers, when they start to understand this, when we have to have dedicated,
00:25:27.508 --> 00:25:35.130
pieces of equipment for testing, we really need to do this gauge R&R relatively early on in the process.
00:25:35.130 --> 00:25:41.310
We don't necessarily wait for the mistakes to happen because you will get bad
00:25:41.310 --> 00:25:44.312
parts being released or good parts being stopped.
00:25:44.742 --> 00:25:47.946
Again, simply because you haven't got a reliable system.
00:25:48.770 --> 00:25:54.760
Nothing's repeatable. So that's a serious thing that we really need to understand
00:25:54.760 --> 00:25:58.250
and you need to understand as a customer. It's like, have you done the gauge
00:25:58.250 --> 00:26:00.703
R&R on this particular piece of equipment?
00:26:01.324 --> 00:26:06.688
Ask the question because they should be able to say, yes, here's the data.
00:26:09.919 --> 00:26:14.372
And if not, you're building a castle on sand. So, yeah, I get it.
00:26:15.350 --> 00:26:20.670
Yeah, yeah, yeah, yeah, exactly. Okay, so that really, again,
00:26:21.310 --> 00:26:23.350
leads into the next thing. And as we're coming through this,
00:26:23.350 --> 00:26:25.517
get a little bit more deep, a little bit more technical.
00:26:26.619 --> 00:26:30.630
You've got to talk about statistical process and everything like that.
00:26:30.630 --> 00:26:36.897
So one of the things we look at is the CPK statistical process control.
00:26:37.411 --> 00:26:42.166
Okay, now this is really in the wheelhouse of our quality department.
00:26:42.630 --> 00:26:44.668
These guys know this inside out.
00:26:45.329 --> 00:26:50.608
CP and CPK help you show whether a process is consistent or not,
00:26:51.196 --> 00:26:53.994
and it stays within specifications and limits.
00:26:54.493 --> 00:26:59.040
Now, when do we use this? Well, it could be used in a number of different scenarios,
00:26:59.040 --> 00:27:02.276
but generally it's used in a machine job.
00:27:02.713 --> 00:27:12.153
Okay, so we've got a CNC machine, call it a CNC-5 maxis or a CNC-turning machine or a lathe.
00:27:12.704 --> 00:27:17.836
And basically, we've got a dimension being produced automatically, okay?
00:27:17.940 --> 00:27:22.702
That dimension is then measured on a periodic timescale, okay?
00:27:22.841 --> 00:27:28.190
Results put it down, put into the statistical data, and we've got a control
00:27:28.190 --> 00:27:31.728
chart, which is basically here's your upper limit here's your lower limit,
00:27:32.105 --> 00:27:36.068
here's your tolerance here's your standard sort of dimensions okay,
00:27:36.608 --> 00:27:38.681
what are we doing? And it's basically going to plot.
00:27:39.379 --> 00:27:46.651
The data points are in the chart. Do the line, align all of the dots.
00:27:46.907 --> 00:27:52.358
And if you see the line coming out of either the upper or lower limit,
00:27:52.665 --> 00:27:55.818
you know your process is going out of control.
00:27:56.375 --> 00:28:01.234
It's not out of control, it's going out of control. This is your control chart.
00:28:02.435 --> 00:28:06.040
So it's very, very easy to see because if it's along the center line,
00:28:06.040 --> 00:28:09.953
In other words, that's your 12.5, okay, dimension.
00:28:11.206 --> 00:28:17.679
If it goes to 12.55, okay, it's coming out 12.56, 12.57, hold on, we need to do something.
00:28:18.074 --> 00:28:23.101
Now, now's the time to go in and have a look at your parameters within your machine.
00:28:23.252 --> 00:28:26.820
Maybe you run out of coolant, yeah? Maybe your speeds have changed.
00:28:26.820 --> 00:28:33.722
Maybe your tip on your machining is broken or worn. Maybe it needs to be changed, okay?
00:28:34.414 --> 00:28:38.681
All of these things allow you to look at control, okay, and making sure that
00:28:38.681 --> 00:28:44.827
you've got consistent dimension being held by the process of being produced,
00:28:45.448 --> 00:28:47.779
or is producing it, okay?
00:28:47.999 --> 00:28:52.406
This is your CP that we process capability, basically, okay?
00:28:52.917 --> 00:28:57.391
Now, what you can have, now, there's a great, you can search the internet for
00:28:57.391 --> 00:28:59.725
this one. This has nothing to do with me, nothing to do with us.
00:29:00.015 --> 00:29:01.246
I don't know who'd come up with this.
00:29:01.554 --> 00:29:06.591
But it's basically a dartboard or a target where you've got your bullseye and
00:29:06.591 --> 00:29:08.276
then you've got a number of different wings around it.
00:29:09.414 --> 00:29:12.908
If you've got your bullseye here and you've got four of your marks here,
00:29:13.245 --> 00:29:16.081
is it consistent? Yes, it's consistent because they're really,
00:29:16.081 --> 00:29:17.715
really tightly grouped over here.
00:29:18.040 --> 00:29:22.266
Is it on target? No, your bullseye is over here. So it's under control,
00:29:22.417 --> 00:29:25.325
right? But it's out of tolerance. You need to shift.
00:29:26.061 --> 00:29:30.381
You need to shift that down until you're out of control. you've got your bullseye
00:29:30.381 --> 00:29:33.525
in the middle and you've got all your points scattered everywhere,
00:29:33.734 --> 00:29:35.881
more like a shotgun that's out
00:29:35.881 --> 00:29:40.799
of control you've got no idea what the hell you're doing so it's important.
00:29:42.361 --> 00:29:48.201
That on critical quality dimensions that we really should be looking at introducing
00:29:48.201 --> 00:29:56.420
statistical process control on these dimensions during manufacturing going back to our,
00:29:57.040 --> 00:30:03.382
air pressure system okay what we're using there we were using basic for cp and
00:30:03.382 --> 00:30:08.492
cpk we weren't looking at control charts we were just looking at consistency
00:30:08.492 --> 00:30:10.485
on the gauge r and r we can use,
00:30:10.932 --> 00:30:13.652
statistical process control during the gauge r on.
00:30:14.345 --> 00:30:19.452
As well to see how consistent everything is in other words you've got your bell
00:30:19.872 --> 00:30:21.305
uh you've got your curve like this,
00:30:21.874 --> 00:30:26.202
it was really wide it's out control it was really narrow it's really really
00:30:26.202 --> 00:30:30.185
in control okay so since you get a sigma variance,
00:30:30.661 --> 00:30:34.812
you know across the bottom here that gives you that idea this is all this is
00:30:34.812 --> 00:30:40.344
a proposal control i'm not going to go into too detail i can talk talk talk talk talk about this but
00:30:40.704 --> 00:30:44.782
i just want to let you know now that when you're talking about gauge rnr this
00:30:44.782 --> 00:30:49.760
is where that can be used with respect to making sure that you're getting consistency,
00:30:50.282 --> 00:30:53.115
and the accuracy is both aligned.
00:30:53.365 --> 00:30:56.029
It's consistent, but it's not accurate. Remember the bullseye,
00:30:56.279 --> 00:31:00.104
okay? It needs to be aligned, okay? So this is basically where we are.
00:31:00.522 --> 00:31:03.532
So you've got your center line, you've got your curve. If you've got a really
00:31:03.532 --> 00:31:07.032
steep curve, but it's offset over here, this is what we're talking about here.
00:31:07.032 --> 00:31:10.212
You've got your bullseye, everything's coming in over here, very,
00:31:10.212 --> 00:31:14.083
very consistent, but it's not on target. You need to bring it back in, okay?
00:31:14.681 --> 00:31:18.129
So again, very, very easy, but you need data points.
00:31:18.466 --> 00:31:22.632
You must have your data points in order for you to understand whether you've
00:31:22.632 --> 00:31:26.818
got repeatability and the accuracy. Okay?
00:31:30.347 --> 00:31:30.952
Reliability as well.
00:31:30.952 --> 00:31:32.292
Yeah, it's like we said at the top.
00:31:32.292 --> 00:31:38.952
Yeah, we're trying to avoid defects, right? But an inspector might find defective
00:31:38.952 --> 00:31:42.492
products, but by the time you've got there, the problem's already occurred.
00:31:42.492 --> 00:31:48.197
So what we're talking about here is being in the business of actually preventing those issues.
00:31:49.122 --> 00:31:55.239
Yeah, yeah, yeah. The last thing that the quality team want to be doing is all
00:31:55.239 --> 00:31:59.645
the analytics with respect to why did we get a quality escape?
00:32:00.069 --> 00:32:02.066
Why wasn't this caught?
00:32:02.978 --> 00:32:08.849
It's so much easier putting the work in up front in order to make sure the parts
00:32:08.849 --> 00:32:11.489
are correct as they hit the production line.
00:32:11.843 --> 00:32:14.797
They hit production line and we know the individual parts are okay.
00:32:15.198 --> 00:32:18.609
Then we've already done the work in the back end during the testing,
00:32:18.609 --> 00:32:23.069
make sure they go together okay, then there's a pretty good chance that the
00:32:23.069 --> 00:32:25.443
product is going to get out the door as it should be.
00:32:28.003 --> 00:32:30.836
That sounds sensible to me.
00:32:31.655 --> 00:32:34.896
Just a couple of things to wrap up there, Adrian.
00:32:35.129 --> 00:32:40.668
I've been talking about CPK, statistical process control, and you think,
00:32:41.040 --> 00:32:44.379
this is exactly what I need. That's going to solve all my problems.
00:32:45.376 --> 00:32:48.720
Because it's brilliant. It's fantastic. To be fair, I love that subject.
00:32:48.987 --> 00:32:51.299
A little bit of math, throw that in there.
00:32:51.879 --> 00:33:00.148
It's great. However, you also heard me say, if you didn't, go back and listen, critical to quality.
00:33:00.705 --> 00:33:08.083
I think this is where we need to understand what is required and what the requirements are in order to
00:33:08.367 --> 00:33:13.087
introduce something like statistical process control or your control chart or
00:33:13.278 --> 00:33:16.447
making sure your gauge r and r is measuring the correct,
00:33:17.259 --> 00:33:23.152
dimension okay no good dimensioning a hundred or you know on your 2d drawing
00:33:23.366 --> 00:33:24.731
a hundred different dimensions
00:33:25.148 --> 00:33:29.899
where only six of them are actually critical to quality in other words it's
00:33:29.899 --> 00:33:34.693
the bit and the function and the mating part it's really really really really critical,
00:33:35.176 --> 00:33:37.468
the outside It's just in fresh air.
00:33:38.310 --> 00:33:42.879
Does it matter? Oh, we're going to put some really expensive control charts
00:33:42.879 --> 00:33:47.557
and more additional workload on that. No, there's no need.
00:33:48.497 --> 00:33:53.037
On the two mating parts, whereas tight tolerance, on the wire part, the BMR part, for sure.
00:33:53.420 --> 00:33:57.501
Absolutely. That is where the money needs to go. That's where the effort needs to go.
00:33:57.617 --> 00:34:06.232
This is where your critical to quality dimensions need to be identified in the design stage.
00:34:06.557 --> 00:34:08.461
Okay? And your drawing output.
00:34:08.983 --> 00:34:15.797
In other words, when your inspector or your quality control guy gets a 2D drawing,
00:34:16.058 --> 00:34:17.962
he gets the part, what am I measuring?
00:34:18.638 --> 00:34:23.514
Well, I'm into everything, but what's really, really important are the critical
00:34:23.514 --> 00:34:26.521
to quality ones. So they're the top, they're the highlights, okay?
00:34:26.671 --> 00:34:30.811
This is your executive summary, let's say, on a report, okay?
00:34:31.159 --> 00:34:34.619
Are they critical to quality dimensions within specification?
00:34:35.066 --> 00:34:40.174
Are they under control? Yeah, what was the, and again, this is where I would
00:34:40.174 --> 00:34:43.454
go back and say, did we have any SPs?
00:34:45.055 --> 00:34:49.527
Statistical process control and control charts on the critical body dimensions
00:34:49.527 --> 00:34:52.746
as it was being produced. Give me the report. I want to see that.
00:34:53.427 --> 00:34:56.557
And again, don't be shy about asking for this information, you know,
00:34:56.557 --> 00:34:59.520
because you've got the right manufacturer in place, then they should,
00:34:59.915 --> 00:35:03.055
in theory, be able to have all this in place for you.
00:35:03.531 --> 00:35:06.926
If you think you need it, ask for it, then they will do it.
00:35:07.391 --> 00:35:12.697
Okay? But again, it's really understanding, do I need a statistical your process
00:35:12.697 --> 00:35:15.692
control on every single dimension, the answer is no, you're done.
00:35:15.970 --> 00:35:21.187
It's purely on the critical quality or those really important dimensions that
00:35:21.187 --> 00:35:24.242
matter is where we really should be looking.
00:35:24.567 --> 00:35:31.060
And again, going back to getting reliable data from an inspection point of view,
00:35:31.327 --> 00:35:32.697
a dimension point of view,
00:35:33.254 --> 00:35:36.957
that is the aspect of the,
00:35:37.538 --> 00:35:43.169
gauge R&R that really comes into play with respect to the important dimensions,
00:35:43.447 --> 00:35:48.877
they need to be right time and time again between batches and making sure the
00:35:48.877 --> 00:35:52.526
equipment that's being used to measure those is,
00:35:53.198 --> 00:35:57.027
under control or in control, is it calibrated, et cetera.
00:35:57.067 --> 00:36:01.337
So you can see the process as we've gone through that is really important.
00:36:02.143 --> 00:36:06.821
Good. Well, I think that more or less wraps us up,
00:36:06.821 --> 00:36:07.131
Paul.
00:36:07.611 --> 00:36:08.121
Okay?
00:36:08.272 --> 00:36:13.221
Yes. Yeah, yeah, yeah. I was just going to say, if anybody's got any questions
00:36:13.221 --> 00:36:17.081
on that, just come back to us. But basically, that's from start and finish with
00:36:17.081 --> 00:36:19.882
respect to the importance of understanding
00:36:20.214 --> 00:36:23.175
the dimensions, not just dimensions. There's a lot behind it.
00:36:24.674 --> 00:36:28.661
Yeah, we always welcome questions. Do email us or go to the Sophie's contact
00:36:28.661 --> 00:36:34.189
page and fill out the form and we will cover the topics that you ask if we can
00:36:34.795 --> 00:36:37.663
what i'm taking away from this paul right as
00:36:38.064 --> 00:36:42.841
you as the buyer you may not be responsible for taking the measurements of different
00:36:42.841 --> 00:36:46.905
components and things like that you've got your manufacturer slash supplier
00:36:47.212 --> 00:36:51.914
right but perhaps what you should be conscious of is rather than just
00:36:52.292 --> 00:36:56.271
blindly accepting that they said oh yep all the measurements are okay which
00:36:56.271 --> 00:37:00.971
actually sounds reasonable if they're saying it's okay well surely they're okay but maybe
00:37:01.571 --> 00:37:07.821
you should at least be aware that they are following a proper measurement system
00:37:07.821 --> 00:37:12.794
and that's actually in place at your supplier so then when they tell you that it's okay you can
00:37:13.431 --> 00:37:19.020
you can be more at ease with with with that sort of that feedback from them
00:37:19.837 --> 00:37:21.178
Yes yeah yeah definitely.
00:37:22.282 --> 00:37:26.281
So so now we know the system that they should be following and the questions
00:37:26.281 --> 00:37:29.491
that we should be asking so that's brill thanks paul
00:37:30.310 --> 00:37:31.541
All right hope that was good.
00:37:32.028 --> 00:37:36.471
Excellent and uh yep thanks for tuning in everyone we will be back next week
00:37:37.847 --> 00:37:41.515
Thanks again for listening to this podcast, brought to you by the Sophist Group.
00:37:42.038 --> 00:37:45.828
We're on a mission to provide you with everything you need to manufacture effectively
00:37:45.956 --> 00:37:48.278
in Asia, including inspections,
00:37:48.661 --> 00:37:52.336
auditing, new product development support, contract manufacturing,
00:37:52.591 --> 00:37:57.955
3PL warehousing and fulfillment, and much, much more across Asia's key manufacturing areas.
00:37:58.344 --> 00:38:05.403
Visit us at sophist.com, that's S-O-F-E-A-S-T dot com, to learn more and get help.
00:38:05.966 --> 00:38:09.884
If you've enjoyed the podcast today, please do rate, review and share because
00:38:09.884 --> 00:38:11.800
it will really help others discover us too.