• This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn more.
  • PLEASE: Read the FORUM RULES BEFORE registering!

Poly carbonate digital calipers

Bob Korves

H-M Supporter - Premium Content
H-M Supporter-Premium
#5
I have two Harbor Freight #47257 calipers that I bought 3-4 years ago. I use one every day for rough measuring in the shop, and it gets used on rusty stock, whatever needs a quick measure. The first one is still on its first battery, comes with a spare, and is still accurate to .001-.002", no issues. The second one is still new in the box. I bought them for $9.99 each, minus 25%, and including a freebie with each. I do not use them for work that needs to be accurate, more as an easier reading version of this:
 

Tozguy

Active User
H-M Supporter-Premium
#7
Thanks Bob, good article.
Being a target shooter I am fully aware of the difference between accuracy and precision.
But it was interesting to read about cheater circuits in digital programming...

When I compare a good dial calliper to a cheap digital one, the digital shows consistently .003'' to .004'' more than the dial caliper at 3 inches.
I have more trust in the Mitutoyo dial caliper and enjoy it more anyway.
 

woodchucker

Active Member
Active Member
#8
Do any of you have experience with plastic callipers like these or similar?
Check out the resolution of .0005''
http://www.kbctools.ca/products/MEASURING @@26 INSPECTION/CALIPERS/ELECTRONIC DIGITAL CALIPERS/5750.aspx

Or these
http://ftctools.com/polycarbonate-6-digital-caliper

How low can the price go?
Well how can it have a resolution of .0005 and an accuracy of only .008. Let's face it, there's something wrong with that. Those are at opposite ends of the accuracy /repeatability spectrum. The HF with res of .01 is probably about as accurate.
 

Bob Korves

H-M Supporter - Premium Content
H-M Supporter-Premium
#9
Well how can it have a resolution of .0005 and an accuracy of only .008. Let's face it, there's something wrong with that. Those are at opposite ends of the accuracy /repeatability spectrum. The HF with res of .01 is probably about as accurate.
Jeff, read the article in the link in my post above. Resolution, which sounds like a really cool and accurate thing, is only what the scales can show. If a ruler has 1/64th inch markings on one side and 1/8" markings on the other side, then the resolution of one side is 1/64" and the resolution of the other side is 1/8". If a caliper has digits on it that go down to .0001", then that is the resolution, even is the caliper is made of soft rubber. Repeatability is a completely different thing as well. If you keep closing that caliper to zero, but get different readings each time, then you have poor repeatability. All that stuff matters in getting good measurements, and more, such as ease of reading, and how comfortable the tool is to hold and use.
 

woodchucker

Active Member
Active Member
#10
Look,you can have .0005 resolution. it means nothing if the accuracy is .008.
end of story. The unit is just .008... posting .0005 is a trap. what's the point of having that resolution if you only have .008 accuracy.
I don't need to read it... having .0005 on accuracy of .001 is fine, you are now +- .0005, having a resolution of .0005 on a .008 means you are now +- .004 since you can't rely on the accuracy of .001 to begin with...
Yes/No?
 

Bob Korves

H-M Supporter - Premium Content
H-M Supporter-Premium
#11
Look,you can have .0005 resolution. it means nothing if the accuracy is .008.
end of story. The unit is just .008... posting .0005 is a trap. what's the point of having that resolution if you only have .008 accuracy.
I don't need to read it... having .0005 on accuracy of .001 is fine, you are now +- .0005, having a resolution of .0005 on a .008 means you are now +- .004 since you can't rely on the accuracy of .001 to begin with...
Yes/No?
You are correct. It is misleading advertising. Nothing new about that, whatever you are looking for. The only way to make it through the mine field is to understand well enough to know how to see through the spin...
 

Tozguy

Active User
H-M Supporter-Premium
#13
Actually the accuracy in the KBC ad is +/- .008'' for a total dispersion of .016'' but that might include hysteresis. :)

It is what it is.
 

Tozguy

Active User
H-M Supporter-Premium
#15
Look,you can have .0005 resolution. it means nothing if the accuracy is .008.
end of story. The unit is just .008... posting .0005 is a trap. what's the point of having that resolution if you only have .008 accuracy.
I don't need to read it... having .0005 on accuracy of .001 is fine, you are now +- .0005, having a resolution of .0005 on a .008 means you are now +- .004 since you can't rely on the accuracy of .001 to begin with...
Yes/No?
I think you are right. The problem with the KBC ad is that the word resolution actually refers to what Mitutoyo calls graduation;
http://ecatalog.mitutoyo.com/Dial-Calipers-Series-505-C1387.aspx
 

whitmore

Active Member
Active Member
#16
Look,you can have .0005 resolution. it means nothing if the accuracy is .008
Oh, yes, it DOES mean something; a simple pair-of-points divider has
very fine resolution, and no accuracy at all (there's no distance markings on
a divider). It's useful because it holds a setting. The inaccurate
calipers also hold a setting, and might be able to repeat it from
the numeric readout.

The metal points on a divider are a better choice for scratch-marking on
sheet metal, though, than the plastic.
 

Tozguy

Active User
H-M Supporter-Premium
#17
Wouldn't a pair-of-points divider be very accurate once it was set, locked and handled consistently?
Wouldn't its resolution depend on the width of the points (i.e. how pointed they are).
 

whitmore

Active Member
Active Member
#18
Wouldn't a pair-of-points divider be very accurate once it was set, locked and handled consistently?
Wouldn't its resolution depend on the width of the points (i.e. how pointed they are).
I'm uncertain how 'resolution' applies; the blunt points get a touch-up every once in
a while, but it's the scratchmark they make that holds the real precision of
interest. Rounded tips still make a narrow line in the Dykem.
Resolution should be high, if you think all settings in the opening range
are distinct and spaced according to the mark width.

The reason to use dividers, is to achieve symmetry. Same distance, on
all the scribed arcs... That characteristic, symmetry, is the virtue of that
instrument.

I think the 'resolution' and 'precision' concepts relate to the measurement marks
(resolution comes from the multiplicity of marks, measured in bits
as the logarithm base two...) and to the dimension that separates the marks
('precision' is the temperature difference that separates adjacent marks on,
for instance, a thermometer).
After a thermometer calibration, there is accuracy (in degrees C or F)
up to the level of the precision, in that the absolute numeric temperature
is well-known after consulting the calibration (table or curve).

There is also accuracy (lesser accuracy) in an uncalibrated thermometer,
and (if you read the data sheet) sometimes you know that the ice point
or triple point are maximally accurate because the factory did calibrate there.
 

woodchucker

Active Member
Active Member
#20
Oh, yes, it DOES mean something; a simple pair-of-points divider has
very fine resolution, and no accuracy at all (there's no distance markings on
a divider). It's useful because it holds a setting. The inaccurate
calipers also hold a setting, and might be able to repeat it from
the numeric readout.
Not sure I understand. A divider is not measuring to a number, it either transfers , or it divides (arc), so it is not measuring like a caliper , mic, etc. So it's an apples to orange comparison. Let's stick to measuring units.
Take a scale for instance. A scale marked in 8ths cannot measure to the 1 and 6/32nds yes I know it's 1 3/16 I'm trying to make a point. So this is the opposite of having a high resolution but low accuracy. If we determine that this scale is really accurate to the 8th. This scale is highly accurate to the 8ths but has a low resolution.

The ad in question has something measuring to 5 tenths, but the claim is we are only accurate to the side of the barn. So who cares about the 5 tenths, when this thing is measuring the barn rather than the your piece of steel.. :) I am sure you care about being close to the .001 and this measuring tool is only giving you really .01 so why would I care that it has fine graduations / resolution.. I missed the spec.. The customer wants 3.151 and I delivered 3.161... I lose. But I had a device that had .0005 resolution.. uh.. ok..

I am not making fun, but it's a serious issue. Having a high resolution, is garbage if you have no accuracy. Having accuracy is important. Because when you make parts for a friend , customer, yourself, you are basing it on a drawing, telephone call, whatever. You need to be accurate. Your .001 has to be matching their .001.. and thats all. If you have .0005 capability, but you are accurate .01 you cannot deliver.

The Harbor Fright unit has a resolution of .01, to me that unit is better than this .0005 unit. Why , because you really know what you are getting with it. That .0005 says I am more than I really am. It's a poser.
 

Tozguy

Active User
H-M Supporter-Premium
#21
Take two identical dial calipers that are mechanically capable of an accuracy of +/- .001'' and resolution of .001'' with graduations of .001'' on the dial, if one reads 1.236'' and the other reads 1.235'' on the same piece what do you make of it?
 

woodchucker

Active Member
Active Member
#22
Depending on the actual size, using a certified caliper, if the size is one of those readings, then the accuracy for both is in spec. If the actual size is 1.234, then only one is in spec.
 

Tozguy

Active User
H-M Supporter-Premium
#23
You are correct. It is misleading advertising. Nothing new about that, whatever you are looking for. The only way to make it through the mine field is to understand well enough to know how to see through the spin...
And this is what I see through the spin:

The number of decimal places in a readout is meaningless if the instrument itself cannot in fact resolve measurements to that number of decimals or to that fineness of graduation of the readout.
Just like how a speedometer that reads to 250 mph does not garantee that the car can go that fast.

The design of a sliding caliper does not lend itself to making measurements more precise and accurate than +/- .001'' (.002'' total). Plastic or composite calipers probably have special applications (i.e. measuring magnets or delicate materials) but might not be as durable, as accurate nor as precise as stainless ones.

We should be aware how easy it is to be lulled into relying on a digital calliper, that reads to four or five decimals, for accurate and precise measurements. In fact the actual size of the object being measured might be quite different than what the readout is showing.

For work to the last .001'' or less, a micrometer should be used.

Some good reading here: http://www.hobby-machinist.com/threads/metrology-101.22521/
 

woodchucker

Active Member
Active Member
#24
And this is what I see through the spin:

The number of decimal places in a readout is meaningless if the instrument itself cannot in fact resolve measurements to that number of decimals or to that fineness of graduation of the readout.
Just like how a speedometer that reads to 250 mph does not garantee that the car can go that fast.

The design of a sliding caliper does not lend itself to making measurements more precise and accurate than +/- .001'' (.002'' total). Plastic or composite calipers probably have special applications (i.e. measuring magnets or delicate materials) but might not be as durable, as accurate nor as precise as stainless ones.

We should be aware how easy it is to be lulled into relying on a digital calliper, that reads to four or five decimals, for accurate and precise measurements. In fact the actual size of the object being measured might be quite different than what the readout is showing.

For work to the last .001'' or less, a micrometer should be used.

Some good reading here: http://www.hobby-machinist.com/threads/metrology-101.22521/
I think there is a mistake in the link. The 1. should be +/- .5 I think, the decimal is not holding any precision other than units, since there is nothing after the decimal.
 

Ebel440

Active User
Active Member
#25
I see no reason to buy them really. You can get metal ones for 10$. With the plastic you would have more flexing and faster wear. Unless you need them for a specific use where plastic may be of some benefit
 

graham-xrf

Active Member
Active Member
#26
Precision? accuracy? And how to tell if there are games going on in the software!
It gets harder unless we doggedly insist on being hard-heads about it. I guess the temptation among those sales oriented folk is overwhelming, and gaming the software to give you a nice warm "Gosh - that must be accurate" feeling is only one small step away from going the whole 9 yards like Volkswagen did with the software "recognizing" that an emissions test was under way.

For me, it is often about angular resolution using encoder kit in satellite tracking Earth station antennas
Stuff like this --> http://www.renishaw.com/en/optical-encoders--6433
One of the products offers 26 bits, and for an extra £15, one can have 32 bits resolution, of a full circle.
No surprise then that a computer sampling at 40 times/sec (in between lots of other tasks) sees all the digits to the right of the true accuracy dancing about randomly.
Worse, the numbers are being crunched by a hard-working DSP (Digital Signal Processor), hard-programmed hardware to do trigonometric math on these useless extra numbers.
1/(2^32) = 1/4294967296 = 2.32830643654e-10 . Try 0.3 milli-arc-seconds! Is that about the height of a pencil in NY, if one is in London? I don't know. It is beyond the ability of the calculator trig function.
OK - so it strains credulity. Maybe there is some kind of very special application that requires it.. (might be?)
So I try the 26-bit product. Keep in mind these things are very expensive!

1/(2^26) = 1/67108864. Even then, the display will not allow those random digits to stand still.
We have to discard the last bit because of the +/-1 bit uncertainty about the state of the least significant bit.

Eventually, finding that the communication to encoder protocol to the servos can only do 22 bits.. Hmm the digits still dance about sampling noise!
1/(2^22) = 1/ 4194304 (of a circle), or 85.83 micro-degrees, or 0.309 arc-seconds. Better than the theodolite!

OK - we can have the system working fine, but that does not stop the sales guys bragging that it has (looking at the encoder) 32 bits accuracy.
In fact, the seismic vibrations of me stomping about on the concrete floor were making the last two digits jump!

Getting the terms used correctly may be near impossible amid the sales culture (yeah - mostly telling sort of lies!)