Determining Weighing / Counting Scale Tolerance - No manufacturer specification

[etdiyas]

What would the tolerance to be used in the absence of manufacturer's specification. I'd like to ask for any publication stating the applicable computation for tolerance. I've read something about 0.5 of scale division and i was confused because how can my digital counting scale 300g capacity with readability/resolution of 0.001 can read 0.0005g tolerance if i have to follow 0.5(0.001)- 1/2 of scale division. And is it possible to use 2 times of scale division as a basis for my tolerance because i'm working on weighing cal procedure because I'm anticipating QC staff in asking where and what is the basis of my computation of tolerance. thank's in advance

[Marc]

I haven't seen anything written. Maybe one of the others can help.

Are the scales calibrated? Is uncertainty stated?

[etdiyas]

the scales are calibrated but the uncertainty is not stated.
how about the NIST Handbook 44 and 105-1, some people i know says that these could help, where can i purchase these references thus it cost much?

[tomvehoski]

There was a freeware uncertainty calculator I had about a year ago that gave a good example on calculating uncertainty for a weighing scale. I can't seem to find it right now, but I believe there is a reference to it somewhere in the archives here.

The 1/2 of the least significant digit is only one part of the uncertainty for a scale. For example, if a scale has a resolution of 1 gram, a reading of 100 grams could be 99.5 or 100.4, so there is a rounding error.

Other components include the uncertainty of the calibration of the weight standard you are using (this should be stated on a calibration certificate for the standards), buoyancy due to air, operator methods, and others.

It's been awhile since I worked on a scale project, but I will see if I have something in my files I can upload.

Tom

[M Greenaway]

Another question on weigh count scales.

Do weigh count scales need to be calibrated against known standards of weight ?

The reason I ask is that we use our weigh count scales by hand counting a certain quantity of product, entering that count into the scales, and then tipping in product up to a certain count. As we sell product in a counted quantity and not by weight do we need calibrate these scales against standards of weight. After all what we are really doing is using the scales as a comparitor between the known sample size and the quantity of parts tipped into the scales.

What is important however is the linearity of the scales, but couldnt I determine linearity using the product itself ?

Any thoughts....

[David Mullins]

MG:
Yes, known mass', over the range of the scales (or the range used by you - as you don't need to calibrate a range if you don't use it). One mass doesn't do it, because of linearity. You could calibrate a point (1 calibrated mass), and then use that to check the range using your product, e.g. if a 5kg mass used, and that equated to 100 component parts, then you could check 10kg accuracy with 200 component parts, and so on - this doesn't calibrate the range (given component part weight variation) but it provides documented levels of confidence that you can deem adequate for your process.


National testing authorities usually dictate tolerance levels of lab measuring equipment. In Oz we have Australian Standards that also provide tolerance levels based on capacity, graduation, etc. At the end of the day, not enough information on your process and how this is being applied.

My understanding of the half scale division thing, is that you have a reading error which equates to half the smallest graduation, so if you are using a ruler with 1 millimetre graduations to measure something, the READING ERROR is 0.5 mm + 0.5 mm (you are measuring at each end of the ruler after all) = 1mm.
This is not to be confused with tolerances!

That's mythoughts anyway.

[M Greenaway]

David

My point was that I do not need to know a relationship to a known standard (or standards) of mass as I am counting parts, not weighing them.

[Ryan Wilde]

MG:

What you are proposing is a 1:1 calibration. Are there any variations to the weight of your parts? Of course there is, and let's say it is 0.1% variation. I could be off by a count per thousand units. From what I've seen, that is actually very good, as things such as fasteners tend to vary by around 0.8-1% of weight per item, and that is just what I've observed. If your weight variation of your part is sufficiently small, and the allowable error of your count is sufficiently large, then using your parts as "standards" is just fine, and you should write a procedure and use that method.

If I were your customer, however, I would be very uneasy with the method, and I would ask for your proof that it is adequate. You are, in fact, doubling the possible error of the count, whereas calibration against calibrated standard weights does not.

Ryan