Anybody do any work in this field? I’ve spent some time looking into some things related to it, because I was looking at scales (“milligram” scales, 0.001g).
I’ve found some good resources, and one that discusses which class of weights one should use for what “general” activity or setting, and also for which level of “readability” (eg: 0.001). Also, I’ve seen some info about which class of scale might require which class of test weight.
I’m curious about the logic of choosing proper “test weights” or “check weights” for a given scale, in regards to which individual weights one should get (for a given scale, let say).
Eg: Say you get a milligram scale with a capacity of ~100g, a readability of 0.001g (1mg), an “accuracy/(or tolerance)” of +/- 5 “divisions” (which I think would mean +/- 5mg in this case?), well what actual test weights should you consider getting? A 5mg? A 10mg? A 200mg? A 1g? A 20g? A 50g? These things are incredibly expensive from any type of…known manufacturer. So what’s the logic of what actual weight(s)/mass(es) to choose? Because again, an actual “brand name” kit is very expensive in the higher class weights.
Eg: I’ve seen an F1 class test weight from one brand, where it was $55 usd for something like 1mg to 50mg iirc. And from 1g or 10g to something like 20g was $140 usd. That’s for a single test weight.
I’ve got several charts of the ASTM, OIML, and even NIST classes and tolerances, I’ve also been referring to sever glossaries and terminology pages, and the terminology in this field is all over the place. Two, and three different words all being used for the same “term” or meaning, and/or referring back to one of the others, or another term that then needs defining - and it’s defined by using one of the previous terms. Haha.
I don’t know that one would need 8 to 15 different test weight amounts for their scale (as in the 100g capacity, 0.001g readability example I gave above), or if maybe 2 to 4 specific test weight amounts might serve them just fine? Maybe there’s a “formula” or “rational” for determining this, like “you should have test weights that are 10%, 25%, 50%, and 90% or 100% of your scales capacity” - know what I mean? In fact, I might have glossed over something similar to that.
Sorry, my mind is absolutely fried from this. Haha.
@Gpaw@Northern_Loki (I searched “metrology” here and saw you guys mention it).
You should, at the minimum, have one calibration weight as specified in the balance/scale calibration procedure. The manual will (should) specify suitable weights that can be used when calibrating the balance. It should specify something like 10g, or 50g, or 100g, etc. You do not need to purchase a set of weights beyond nice to have.
You should not need multiple different weights although some scales may allow multi-point calibration but that is not common. In that case, you’d need a couple of different weights but most of the times those scales will also have a single point calibration as well. A good linear balance mechanism shouldn’t require multi-point calibration to remain within it’s readability specification, anyhow.
So, you need at least a single calibration weight. The class (tolerance) of the weight, along with the scale/balance readability, will now determine the scale accuracy. Note that precision and accuracy mean different things. Calibration is for accuracy.
The scale readability will serve to indicate the minimum calibration weight tolerance that’s acceptable to retain the scale/balance accuracy claims. You can use a calibration weight that does not meet the minimum but it translates into increase measurement uncertainty (such that the scale may not be working at it’s full potential).
As a general rule, you want a calibration weight (class) that has tolerance of less than half of the scale readability. But, go as good as you can swing. I try to go for 1/3 or less and seek out bargains, over-stock, etc.
And, yes, they tend to be expensive off-the-shelf. But, at 1mg readability, you have a relatively sensitive balance and to retain that accuracy you need to factor calibration cost into the overall cost (some balances / scale will come with an internal calibration weight). Or, you can ballpark it with a weight that has lessor class specification knowing that it’s tolerance now becomes your scales measurement uncertainty (possible skewed one way or the other).
Unfortunately better readability, better stability, wider range, higher accuracy, better repeatability (precision), robustness, brand (industry proven) tend to exponentially increase the cost of the set-up.
FWIW, with some care and time, you can usually find good deals via Ebay, auctions, etc to get your set-up into the scientific class if so desired. Occassionally, you’ll find new-old-stock at bargain prices.
I’ve surprised myself when I first purchased a good balance and how often I use it for a whole variety of tasks from uber-precise to close enough. From very low mass to heavy things. From nutrient analysis to cooking to weighing packages to formulating ingestibles to … Now we’ve collected several.
That’s a good question. I don’t really know. In practice, calibration using weights closet to the weights being measured in practice result in the highest level of precision (note precision). Although, the OEM of the balance should be specifying the readability and other specifications to account for differing calibration weights that will apply across the balance range. As such, it shouldn’t actually matter so much and with a good well-known balance/scale brand, you can usually rely on the specs to be what they say (otherwise they wouldn’t be a good brand). Additional check weights can be used to verify that is the case after calibration. So, maybe two weights if you want to check weight the calibration. One calibration weight and maybe one check weight probably on a far side of the scale range.
There is a whole lot of detail in the use of precision scales/balances. Some of it may not be clear or obvious. Some may or may not be of importance depending on the use case. For instance, static build-up for these very sensitive scales can cause measurement error. Heating of the scale/environment can cause error. Drift occurs in even the best scales/balances. Improper handling of calibration weight can lead to error. etc. etc.
Being aware of the tolerancing and sources of error is the important factor but much of this depends on how accurate and precise you need to be. For instance, measuring of 20g of a chemical for nutrient formulation probably doesn’t need the 0.1mg or 1mg accuracy for general use. For scientific or pharmaceutical formulation, then such things become much much more of a concern.
Also, as an example if the balance specifies 50g as a calibration weight with a 1mg readability, I’d seek out an ASTM class 2 or OIML F1 (or better) weight.
Those illustrations are new to me. Thank you so much for posting them!
That first one… That picture is most definitely worth a thousand words! A super clear display of precision and accuracy that one would be hard pressed to not understand.
From a manual: (“X” is the calibration weight can be set according to users’ requirements, we recommend a minimum weight of at least 50% of the Balance capacity)
This isn’t what I meant. But that’s interesting. Because to me, it seems like several of the “cheap” scales you’d get off amazon will use two of the same weight for calibrating. Eg: for a 20g scale, it will come with two 10g weights. It will tell you to place one of them, wait, then place the next. I don’t think this is what you meant though. I think you meant more like a pH meter that has “multi-point calibration”. Calibrating the scale at multiple different weights (which probably provide better results over the range of the capacity, like a pH meter).
This video helped:
So, you’re taking the 0.001g readability (he calls it display resolution) and dividing it by at least 3, is what you’re saying. He divides by 4, resulting in a astm class 0 weight being selected.
Yea, makes you wonder how the price range of “milligram” scales is $8 usd to over $6000 usd makes any sense. People are buying an $15 milligram scale to use for weighing medication, supplements, and even drugs (I read a review someone said they’re weighing dmt and stuff like that).
I “tested” four different amazon milligram scales in the last 48 hours. I returned all four. I don’t really have “data”/numbers written down, I didn’t even need to get that far into “testing”.
I have been looking there, but was looking for “sets” at the time (test weight sets). I’ll look again now, for singles, astm class 1, oiml class F1 (E2 is significantly better, F1 almost four times “worse” in tolerance .)
What’s the highest capacity 0.001g scale you’ve got?
When you say good brand, you mean Ohause, Sartorius, Adam Equipment, A&D, etc. though, right? Not “American Weigh Scales”, Smart Weigh, Weightman, Gram Press, Brift, Thinkscale…
Because I think most people (like myself in this case) will have never heard of the former, and will only be choosing from the stuff available from spots like amazon, in the $15 - 50 usd range.
EDIT: Maybe it’s also worth noting that these “amazon” scales we mostly tend to go for are “pocket” scale size, and that might have something to do with it too (even the ones with a wind shield).
Yea, I learned this very quickly when I dove in. And all of these challenges with use, as well.
So I followed the proper techniques as best as I could when I tested the scales (no fans on, slow moving, away from electronics as much as possible, even breathing in a different direction, wearing gloves, using a non-metal/magnetic pair of tweezers, etc.) I also didn’t know these types of scales should be warmed up (eg: in a manual it says 30 minutes warm up before calibration or weighing).
And people are gonna try to weigh up 20mg, on a $15 milligram scale, with their bare - unwashed hands, with no wind break, on a flimsy table, under a ceiling fan… finger print can weigh 0.5 milligram or something. Haha.
Sure. But I’d like to not have to round up/down the micronutrients amounts (let alone just have them be wrong and not know due to poor scales). “0.022g”, “0.247g”.
I’m likely over doing it though.
Thanks for your help Loki. Really appreciate your detailed answers.
Edit: That’s one chart I’ve been referring to as well, that’s a good one.
Edit 2: When I was testing the four scales, one issue was the “covers” / wind break (as they’re advertised) cause insane static electricity, that would physically move the small (mgs) weights on the platform if I tried to close it. And even when closed, putting a finer or hand near or on it would cause the readings to jump around.
