How precise are measuring instruments?

[pete]

That's an interesting method I'd not heard of yet, thanks.

[thunderskunk]

Funny you mention it, I learned the other day about the Linearity and Bias study. Pretty much exactly like the incremental block you’ve got there: you measure several different parts of varying sizes with a calibrated size, then measure each of those parts several times.

I just did this for a 2D laser micrometer:
https://uploads.tapatalk-cdn.com/202201 ... f7c841.jpg

Bias is the difference between the true dimension of the object you’re measuring and what you get for a reading. Linearity is how consistent that bias is across a variety of sizes to be measured. Your incremental block would accomplish this perfectly.

In the study shown, I had one pesky flyer. This drove R squared to be pretty low, which is just a number that tells me how confident in my results I can be. I was able to get that up to 60% by changing where on the lens the measurement was taken.

[pete]

I think it's great to see this additional and mostly less well known information. Ron's original question was how precise are measuring instruments. By now he must be thinking he got a great deal more than he was expecting with that simple question. There's a whole lot more to it than just what the tool says the measurement is. And the more zeros to the right of the decimal point the higher the difficulty gets to be fully confident your numbers actually are what the measuring tool says they are. So far I haven't ever run across those specific linearity and bias terms in Thunderskunks post. My guess is there generally referred to with a less detailed and more generic term as uncertainty of measurement. Is that correct?

[thunderskunk]

I think it's great to see this additional and mostly less well known information. Ron's original question was how precise are measuring instruments. By now he must be thinking he got a great deal more than he was expecting with that simple question. There's a whole lot more to it than just what the tool says the measurement is. And the more zeros to the right of the decimal point the higher the difficulty gets to be fully confident your numbers actually are what the measuring tool says they are. So far I haven't ever run across those specific linearity and bias terms in Thunderskunks post. My guess is there generally referred to with a less detailed and more generic term as uncertainty of measurement. Is that correct?

Sort of. Bias is accuracy. Linearity is uncertainty of that bias. Repeatability/reproducibility is variation. Reliability is whether the engineer got his coffee and whether the machinist shows up on time.

Funny story: I went the engineer route because I wanted to build a live steam locomotive. Never had time to build it in school, so I bought all my own equipment. I’ve been a locomotive mechanic (best job in the world), a machinist, and a manufacturing engineer. Besides a few bearings and a whole bunch of steel plates, I still haven’t built an engine. Why? Because I overcomplicate EVERYTHING, haha. Hopefully that’s an indication to the value of doing gage studies in the home shop.

[pete]

Thanks for the clarification. One of the extras seldom mentioned about having a set of gauge blocks is they give you an easy method of checking not only the equipment your measuring with, but also the technique and level of skill used to get those repeatable measurements. I think it's best not to look at the tools vernier scale, dial or display as your doing so as that starts an automatic bias trying to get the measurement to match what the known size is. So your already modifying your usual methods and it then makes that test checking worthless.The first time I tried it that way I was quite surprised at the level of inaccuracy that can show up with multiple measurements on the same block. That repeatable technique is just as important as the tool itself. To + - .001" maybe not so much, + -.0001" then definitely.

"I over complicate EVERYTHING" Lol, I know EXACTLY what you mean. I suspect that's due to more knowledge, in the beginning in a home shop it's simple, I need to make it as close as possible to the drawing with what I have. Add some time and that extra knowledge, and then it's, well since I'm starting from scratch I may as well put in, re-design or add .......... Then if I use xyz material it will be stronger, better, last longer .......... Bearings are always better than bushings so..........That different, better machine, accessory, tool would then allow me to...........

[Lew Hartswick]

< well since I'm starting from scratch I may as well put in, re-design or add .......... Then if I use xyz material it will be stronger, better, last longer .......... Bearings are always better than bushings so..........That different, better machine, accessory, tool would then allow me to...........>
My old boss had a phrase for that but I can't remember the exact words. I was always guilty of it designing electronic "boxes" for use in the Chem Dept. Something like creeping generalization . Make it do extra things for the "just in case the future" .
...lew...

[mklotz]

My old boss had a phrase for that but I can't remember the exact words. I was always guilty of it designing electronic "boxes" for use in the Chem Dept. Something like creeping generalization . Make it do extra things for the "just in case the future" .
...lew...

Perfect is the enemy of good.

[John Hasler]

thunderskunk writes:
Bias is accuracy. Linearity is uncertainty of that bias.

(Non)linearity is not uncertainty. Like bias it is a systematic error which can, at least in theory, be measured and accounted for. Better to say that precision is uncertainty.

BTW ISO now deprecates the use of the word "bias", though I don't see that as sufficient reason to stop using it.