Electronics > Projects, Designs, and Technical Stuff

mg scale - interesting effect

<< < (4/8) > >>

coppercone2:
just keep in mind you can do all the electronic stuff right but you still need a stiff enough table. It does not look right.

For 50G a decent American scale would have about 3-4 times more material in the Part Being Weighed and the Load Scale. It looks like junk unless the industry changed (doubtful). It also uses the chassis as stiffener. The chassis of a real scale is wobbly compared to the 'anvil'. Your talking a heavy glass plastic part not some little sheet of whatever and some dull zinc assembly.

If you got a quality consumer scale and then put it in a hammond box it would be worth it, because they have similar EM problems.
Also I have NO idea of the quality of those load cells. It actually looks decent (mainly because its painted) but you better hope its not like a Bochen 10 turn pot. But seriously the load cell interface assembly looks like its not worth dealing with.

JBeale:
This scale is labelled for 200 g although I've only gone as far as 100 g.  My +/- 2 mg observation was for sub-gram weights, there is a little more variation at 100g.  I have two physical 100g test weights and just now I tried them A - 0 - B - 0 and repeat for six trials. The scale could tell them apart even though they are only about 10 mg different. So maybe the repeatability is +/- 2 mg plus +/- 0.01% of the reading.


--- Code: ---#   Weight A  Weight B
------------------------
1   100.108   100.123
2   100.105   100.118
3   100.104   100.118
4   100.106   100.115
5   100.105   100.114
6   100.106   100.121
--- End code ---

coppercone2:
thats pretty bad. A good one will be repeatable to like 1mg (and this is kinda shady) with a 50G test weight.

Are you letting it run for a while and putting a plastic bowl on top around the weigh plate (or a tupperware over it) to block air interference? did you level it correctly with a bubble center level?

between 5 and 6 test B you have more then 6mg

JBeale:
I did note the built-in bubble level and first levelled it on the granite countertop adjusting all four feet so there was no wobble.  I am just using the square windshield enclosure that came in the box, with the aforementioned aluminum foil.  I put on the weight, waited for the "stable" indicator flag, removed and waited for it to return to zero. It seems to have an automatic "tare to zero" feature if you're within say 10 mg of 0.  This feature (?) sort of hides the non-ideal behavior of the load cell, but to be careful I should probably wait much longer at 0 load in between weighing larger loads like 100g that have more effect on the zero setpoint.

For sure it is not a true mg-accurate scale, but for sub-gram weights their advertising is true ("repeatability: +/- 0.003 g") and for me it was a value at the $43 price.  It is noticeably better than the even-more-mislabelled "50g x .001" scale Amazon has around $18.

David Hess:

--- Quote from: Kleinstein on March 23, 2019, 10:50:58 pm ---I remember a precision scale having some +0 and -0. So the interval for 0 reading was twice as large as it should have been. So it was best used with some offset. One could also see this as a mild form of cheating.
--- End quote ---

Lack of + and - zero used to be a premium feature with integrated ADCs.  TI helped earn their bad name by releasing a new integrated ADC which was advertised to fix this problem ... except that it did not and they continued selling them for years after they were notified.


--- Quote from: ChristofferB on March 24, 2019, 12:44:48 am ---Some of the Mettler single-dish weights from the 50's would have a mechanical BCD lever system to lift and lower wights, and the last digit would be reflected trough a lens in the weighing beam as the torsion or flex in the beam.
--- End quote ---

I hated those things.  When I was going to Cal Poly Pomona, one year everybody in the quantitative chemistry lab failed because some jerk had deliberately twisted the weight selection control too quickly on the lab balances causing some of the weights to fall off inside the mechanism.  The teachers knew but elected to just fail entire classes of students (1) which is California schools and politics at its finest.

It got better.  I happened to work for a load cell company at the time so had access to calibrated weight standards.  Bringing the problem up with the teachers and proving it did not go well for me.


--- Quote from: T3sl4co1l on March 23, 2019, 10:43:58 pm ---A well designed one will be insensitive to AC, or well enough filtered to not mind ambient noise.
--- End quote ---

It is more like it takes a poorly designed one to be susceptible to interference.  RF decoupling is required to prevent the input circuits from rectifying RF to produce an offset and the input should be integrated over a whole number of power line cycles which is a good idea anyway to produces the needed noise free resolution.  A novice design can be identified by the lack of ratiometric operation.  The last digit should never show more than 2 digits or +/-0.5 counts.

In a good design, the load cells themselves contribute a majority of the error and there is really no excuse for more error than that.

(1) Quantitative analysis depends on accurate numbers.

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod