I cracked the mystery of Chinese precision jewelry scales

[magic]

This scale has a drug dealer's mode

How to enter:
- prepare a 15.432g reference weight
- enter cal and place the weight instead of 100g
- switch display units to grains
- multiply readings by 10 to get grams
- (yes, it would be easier to cal with 10g instead, but my scale doesn't accept that)

Right off the bat, noise and short term repeatability are down to ~1mg with a bit of mental lowpass
It's looking like the ADC might be the bottleneck in normal operation
Drug dealing mode probably kicks some frontend PGA into higher gain and overcomes it.

I will test it tomorrow. And how cheating works in grains, because I have no idea.
 

[magic]

Nevermind, this POS is utterly hopeless and unfixable.

In any unit other than grams it still lies, and...
wait for it...
it still rounds to full grams

I have seen youtube reviews of similar scales by guys using them to measure bullets and powder charges (in grains).
Well, hate to be the bearer of bad news, but...

[Kleinstein]

There is even a chance to find some kind of cheating with more expensive scales: It sometime comes with +0 and -0 reading. So the bin for showing 0 gets about twice as wide. I don't remeber the brand were I have found this, but it was a lab balance from one of the big players.

The + and- zero reading can also effect some cheap DMMs (with dual slope ADC) - in this case not a SW problem, but a HW weakness.
Such things may also happen by accident, when rounding and converting to decimal.

[HighVoltage]

There was a time that I liked KERN scales from Germany a lot.
One day I bought a broken KERN lab scale on ebay and thought I could at least try the repair.

That scale was so badly built that by opening the scale, the stiff cables got ripped out of display PCB.
Everything was built for a one time assembly and not to be repaired.
This scale most likely was built in China but hat the "Made in Germany" label on it.
Since that time I stayed away from KERN scales.

So, not only Chinese scales are bad!

This is also true for calibration weights!
I bought a 1kg calibration weight on ebay a few years ago and it has a M4 set-screw in the bottom. So, naturally I took it out and wondered why there was a screw. Well, I found sand grains behind the screw to adjust the mass of the weight. And this weight was of the F1 class!
Unreal!
 

[PlainName]

sand grains behind the screw to adjust the mass of the weight

Is that particularly bad? I mean, so long as it doesn't drift and actually weighs what it's meant to, seems a cheap and quick way to get it done. Also, if you drop and chip it, it's not the end of the world.

[gamalot]

sand grains behind the screw to adjust the mass of the weight

Is that particularly bad? I mean, so long as it doesn't drift and actually weighs what it's meant to, seems a cheap and quick way to get it done. Also, if you drop and chip it, it's not the end of the world.

It gets heavier when it absorbs moisture and vice versa.

[PlainName]

ok

[CatalinaWOW]

I'm more with Dunkemhigh on this.  Much sand is largely silicon dioxide which is going to absorb very little moisture.   A grain of sand is on the order of 10-15 mg.  So the few grains of sand would have a mass of something like 50 mg.  The tolerance on an F1 1kg mass is 5 mg, so the sand would have to change its mass on the order of five percent to put it out of tolerance.  And that assumes the set screw didn't seal the cavity.  Of coarse it is possible that this sand is beach sand which in some cases has a high proportion of calcium carbonate.  Bigger problem then, but the point is that this weight may be an honest weight.  The presence of this trimming technique doesn't seem to me to prove that it is junk.  It seems to me the much larger volume and surface area of brass would be a larger problem than the few grains of sand in this weight.

I would be worried about a sand filled weight where the sand provided a large fraction of the total mass.  Surface absorbtion on the very large resulting surface area would be a likely problem and even modest amounts of bulk saturation would matter.
 

[gamalot]

I'm more with Dunkemhigh on this.  Much sand is largely silicon dioxide which is going to absorb very little moisture.   A grain of sand is on the order of 10-15 mg.  So the few grains of sand would have a mass of something like 50 mg.  The tolerance on an F1 1kg mass is 5 mg, so the sand would have to change its mass on the order of five percent to put it out of tolerance.  And that assumes the set screw didn't seal the cavity.  Of coarse it is possible that this sand is beach sand which in some cases has a high proportion of calcium carbonate.  Bigger problem then, but the point is that this weight may be an honest weight.  The presence of this trimming technique doesn't seem to me to prove that it is junk.  It seems to me the much larger volume and surface area of brass would be a larger problem than the few grains of sand in this weight.

I would be worried about a sand filled weight where the sand provided a large fraction of the total mass.  Surface absorbtion on the very large resulting surface area would be a likely problem and even modest amounts of bulk saturation would matter.

Even though sand doesn't really make a big enough change in weight, filling the weights with sand makes consumers feel cheated. I don't know how much HighVoltage spent on eBay for his weights, a 1kg F1 grade weight in China costs about $40 and doesn't have any screws on the bottom.

[HighVoltage]

Even though sand doesn't really make a big enough change in weight, filling the weights with sand makes consumers feel cheated. I don't know how much HighVoltage spent on eBay for his weights, a 1kg F1 grade weight in China costs about $40 and doesn't have any screws on the bottom.

The screw was not even tight or even real good fitting.
It was one of those threads that was far too loose to call it a fitting threaded screw.
In other words, shaking the weight a lot would probably loose the screw over time.
I don't know, to me that looked like a really cheap way to make a calibration weight.
The opportunity to make a picture was missed!

I have now a really good quality F1 class weight, that came with a note to not touch with bare hands.
Actually a real good calibration weight has a soft cloth within the box to protect the weight.

[Gyro]

This is also true for calibration weights!
I bought a 1kg calibration weight on ebay a few years ago and it has a M4 set-screw in the bottom. So, naturally I took it out and wondered why there was a screw. Well, I found sand grains behind the screw to adjust the mass of the weight. And this weight was of the F1 class!
Unreal!

At least it's a metric screw! 

[HighVoltage]

Here are some examples of really good calibration weights with certificate.

No screw, no sand and no fingerprints.
Both came with a nice cloth for handling.

[jfiresto]

Here are some examples of really good calibration weights with certificate.

No screw, no sand and no fingerprints.
Both came with a nice cloth for handling....

Have you taken your weights to the local Eichamt to check their accuracy? Ours is very friendly but can only compare to no better than F1-class weights.

[HighVoltage]

The "Eichamt" here is not so friendly, they are a profit center and like too much money.

But a local scale repair service tested them for me.
The Sartorius 2kg F2 weight was measured to be:
2 kg + 1,6 mg with an uncertainty of 2,0 mg

So, its mass is 2.000,0016 Gram

[Slartibartfast]

What surprises me is that so many fall for these cheap Chinese scales.

What do you mean by that? Who "falls" for those scales? I guess you cannot mean all those who buy them? It is as with other tools: You need to take into account their limits.

I have one. I use it mostly to balance the relative amounts of binder and hardener when mixing epoxy. I do not need a scale with 50000 counts (or whatever it is that mine has), and in particular for that application I do not need any accuracy, just a little resolution! And the resolution of these scales is plenty for mixing epoxy. Why should I buy a 500€ scale when a 8.50€ scale does the job?

You can ask how many people fall for (believe) the specs of these scales. That's an entirely different thing.

Cheers  Peter

[magic]

Correction: "resolution" is meaningless if it's not backed by any accuracy

As for the 8068, the worst error I found so far is 3% at 1g, so probably good enough for mixing epoxy, but it is also the case that anything would work.

[Slartibartfast]

Correction: "resolution" is meaningless if it's not backed by any accuracy

As for the 8068, the worst error I found so far is 3% at 1g, so probably good enough for mixing epoxy, but it is also the case that anything would work.

I do not quite understand your first remark, but I suspect I see what you're aiming at. In fact, I'd better used the probably more common word "precision" for what I had in mind (in the old times when analog meters abounded these words were synonyms). Precision is what I need for mixing epoxy, accuracy does not matter, and in fact can be lousy if trueness is bad. If a scale has good precision but no trueness and hence no accuracy, I can still compare weights, and find accurately whether they are equal or not, or which one is larger. If the scale shows both as 1000 tons even they are only 1 gram it does not matter for the comparison.

It is true though that I indeed greatly appreciate that there is not too bad linearity. Without linearity it would be quite cumbersome to mix something like 100:40, but even that could be done.

[mzzj]

What surprises me is that so many fall for these cheap Chinese scales.

What do you mean by that? Who "falls" for those scales? I guess you cannot mean all those who buy them? It is as with other tools: You need to take into account their limits.

I have one. I use it mostly to balance the relative amounts of binder and hardener when mixing epoxy. I do not need a scale with 50000 counts (or whatever it is that mine has), and in particular for that application I do not need any accuracy, just a little resolution! And the resolution of these scales is plenty for mixing epoxy. Why should I buy a 500€ scale when a 8.50€ scale does the job?

You can ask how many people fall for (believe) the specs of these scales. That's an entirely different thing.

Cheers  Peter

I use mine also for mixing epoxy and to cook various chemical experiments. I need hardly ever better than 1% accuracy but mg resolution comes handy time to time.

Unlike the noobs here I checked mine with E2 calibration weight set  :

[magic]

Unlike the noobs here I checked mine with E2 calibration weight set  :

There is a sucker born every minute

[HighVoltage]

What surprises me is that so many fall for these cheap Chinese scales.

What do you mean by that? Who "falls" for those scales? I guess you cannot mean all those who buy them? It is as with other tools: You need to take into account their limits.

I have one. I use it mostly to balance the relative amounts of binder and hardener when mixing epoxy. I do not need a scale with 50000 counts (or whatever it is that mine has), and in particular for that application I do not need any accuracy, just a little resolution! And the resolution of these scales is plenty for mixing epoxy. Why should I buy a 500€ scale when a 8.50€ scale does the job?

You can ask how many people fall for (believe) the specs of these scales. That's an entirely different thing.

Cheers  Peter

Well, that was bad choice of wording on my side.
I meant falling for the ridicules specifications !

For some purposes these cheap scales are perfect and I have a few of them myself.

A few years ago I needed to test lots of Magnets in a cheap way and made a small device to hold a scale and a micrometer with steel plate attached. The adjustment was made that the magnet lost 50% of its weight at the correct setting. Precision was not required and accuracy was good enough with this 200 gram scale.

For this purpose a small China made scale was perfect. See picture


 

[magic]

It gets better: my negative feedback on this item already has two downvotes. As I scrolled down to even find it, I encountered another report of this anomaly from past year, also downvoted.
 

It reminded me of a high school buddy who made money working for a "public relations" company, doing roughly that sort of things. Gotta love those Interwebs.

[thm_w]

It gets better: my negative feedback on this item already has two downvotes. As I scrolled down to even find it, I encountered another report of this anomaly from past year, also downvoted.
 

It reminded me of a high school buddy who made money working for a "public relations" company, doing roughly that sort of things. Gotta love those Interwebs.

Can you link the listing?

[Conrad Hoffman]

I use an old Mettler. The only electronics in it is the light bulb that illuminates the scale of the optical lever. The thing resolves 10 ug and the accuracy isn't a lot worse. Here's a DIY trick for very small masses. Measure a length of good wire and weigh it on a good scale. Now you know the mass per length. Cut pieces as needed. If you start with fine wire you can easily create accurate ug masses. Naturally it should be uninsulated, or magnet wire at worst. You can calculate the error if you know the uncertainty of the cut.

[miegapele]

Bought same scale ~2 years ago. At the time it cost ~15USD. There were like 5 models with 100g/1mg resolution on Ali for that price range. Did extensive research and already then there were reviews for other models that they are rounding. Choose this hoping it will not, but to no avail.
But the scale in general is not good, or I would like it to be better. I have 500g/0.01g resolution scale from 5 years back or so, and that looks very good with like 0.02g repeatability. I sort of expected the same from the new scale, but unfortunately repeatability seems to be around 10mg, so hardly better than 0.01g resolution scale.
Bough another one 500/0.01g scale in 2021. which looks the same. However, that might be rounding, just not that aggressive .
I guess availability of calibration weight sets for $3 made manufacturers do this.

This scale round very aggressively, like X,970 is already rounded to X,997. So, there is easy way to test this rounding. Calibrate scale with 101g pretending to be 100. Then weighting 1g it will show 1.000g exactly, weighting 2g it will show something like 1.998g. But weighting both together you will get something like 2.960g . However if manufacturers were smart and did not round that aggressively it would be way harder to prove.

For multimeters its way easy to gradually increase voltage so maybe it will not happen. But who knows. Most buyers don't know any better, buying cheap AD5584 and cheap multimeter showing 10.000 might be satisfying.
Also for resistance they already round shorted leads to zero. I wonder what  is logic in that implementation.

[miegapele]

This is now a thing in on other scales. Got 10kg scale recently (there is at least two similar variations of it, one with grams in upper right corner and milk, and the one with grams in lower right corner). It weights nicely and seems accurate, but it rounds to every 100g. Rounding is such that anything more than 95g is rounded to 99g and anything more than 97g is rounded to 100g. similarly it rounds down above 100g.

For more, see here
Scale also have calibration, and at least two secret menu, one to set scale capacity, one to set something for (60,120,180), maybe idle timeout, and one to set L00,L01,L02m but not sure what that is.