Weighing system, load cell transmitter scaling for PLC

[parky]

Phrog30 is right.
I now realise that you want to know everything, however, in reality the things you need to know are the degree of accuracy for each of the components for example, load cell, amplifier & analogue conversion in the PLC. these should be obtainable in the spec sheets. Start by adding the accuracy of the load cell & amplifier this will give you some indication of the worst case error combined. This is not the true as on any particular weight it could be that the load cell is +0.01 gram out and the amplifier is -0.03 gram so the actual is less than the max error. by adding the stated accuracy that will give you the max possible error. The analogue card is a different matter, depending on if it is 12 bit or 16 bit conversion and the resolution for example a 12 bit A/D will only give you 2048 counts so if you were converting that to temperature of 0-200.0 degrees C then it is a little over 1 count to every 0.1 degree given that there is the analogue cards own conversion accuracy then you should be able to work out the total deviation possible & that is your accuracy (in a perfect world). However, there are many other factors, electromagnetic fields around the load cell, fixing type temperature all manner of things. Like I said earlier, if you are measuring 0-500g and the load cell gives out 10 microvolts per gram as the load cell is under load (your platform) of say 200g then your signal is not 0-5000 microvolts it's actually giving out 200 - 5200 microvolts or 200g to 5200g. so to choose a load cell to weigh 0-500 it needs to be capable to read 0-6000g as it is already under load. The 4-20ma will be 4ma at 200g & 20ma will be 5000g after all you will not sit what you are weighing directly on the load cell. I have been involved in many installations of weighing systems from pack weigher's measuring a a few grams up to tons. and also in the last 20 years done the 6 monthly calibration on vessels. I know the pitfalls of design & installation for example, one of the worst things you can do is use 4 load cells on a vessel, it is better to use 3 in a triangle, think of a chair with four legs and how it rocks, put it on three it's more stable. A lot depends on the design & the weight it is going to carry. I worked on a manufacturing plant where two of the vessels were on 3 load cells, however, the other was on a 4 load cell system. These were identical tanks & load cell mountings but the 4 leg would shift by a few kg so was not accurate enough perhaps down to engineering tolerances, however, they tried to shim it but it was not successful, They re-positioned 3 of the legs into a triangle & hey presto stable.
Just to re-iterate, concentrate on the accuracy it will work at and calibration.

 

[drbitboy]

The 4-20ma will be 4ma at 200g & 20ma will be 5000g

Yes, although that depends on exactly how the conversion is done e.g. https://www.sensorsone.com/0-20mv-to-4-20ma-signal-conversion/, so ensure you understand every step ((out - outLo) = (in - inLo)*(outHi-Outlo)/(inHi-inLo), or more generally y = mx + b)) along the weigh () from the scale at (2mV/V) / g to the analogue card digital count output.

 

[parky]

I was talking about the converter you scale it to read 0g at 4ma but the load cell will be reading 200g (the weight of the weigh pan when you calibrate the zero) & span say 500g (well span will depend on what you want the 20ma at).
In reality you calibrate it in total i.e. your scaled reading in the PLC is 0g at nothing on the weigh pan and say 500g with a 500g test weight.

 

[parky]

Just a note: over the last 30 years I have been trained by Nobel Weighing systems (used to be a division of Bofors Yes they also made arms), Mettler Toledo, AWS and a few more that I cannot remember. The last two companies I have worked for I was doing calibration on weighing systems temperature, conductivity etc. with the exception of check weighers (these were done by the suppliers) In both companies we had UKAS calibrated instruments like Dry Well, Load Cell, Water Meters Conductivity the list goes on. In my last position I was responsible for calibration of the vessel weighing systems temperature, conductivity and dosing systems, producing calibration certs for The British Retail Consortium audits all to UKAS standards. All 250 + calibrations were done every 6 months (company standard) but QA would do regular tests for example the check weighers were checked every 2 hours. I was also responsible for sending the test instruments to UKAS Accredited calibration companies. I have worked many times with the leading suppliers of weighing systems.

 

[drbitboy]

I was talking about the converter you scale it to read 0g at 4ma but the load cell will be reading 200g (the weight of the weigh pan when you calibrate the zero) & span say 500g (well span will depend on what you want the 20ma at).
In reality you calibrate it in total i.e. your scaled reading in the PLC is 0g at nothing on the weigh pan and say 500g with a 500g test weight.

right, tare weight. my bad!

 

[GaryS]

A few thoughts on this most weight system will produce 4 ma at 0 and 20ma at full span
and most plc system are setup to 4 - 20ma input as a standard
but when you do that any input less then 4 ma is displayed as 0
those same weight systems will transmit a signal of 0 -20ma remember 4ma = 0 2ma would be a neg weight
so I learned to setup my inputs for 0-20ma that way I show any neg value in the scale and can use it in my weight calculations
Just something to think about

 

[parky]

I know what you are thinking Gary, However, I have found on some systems 4ma is 0 but if it falls below it will go negative, at the rate of approx. 820 per ma based on a 20ma value of 16384. (it obviously depends on the A/D conversion). I used the 16384 but I know there are conversions of 12, 16 bit etc, I believe the 16384 allows for sign i.e. bit 15 is sign. Drbitboy, I can see my original explanation might be seen in a different way, sometimes when writing on the fly it can be confusing lol.

 

[GaryS]

parky
yes when the input is below 4 ma the scaled input will be neg weight that what you want
I have seen many scales systems where the empty vessel will be slightly neg but the plc see it as 0 so when you fill the scales you will have material in that not accounted for
on a 4-20 ma input
The value at 0 ma will be neg 25% of scale or the value at 20 ma
this way you account for ant neg value of the scales .
on a 100 lb span 0 ma would be -25 lb 4 ma would be 0 and 20 ma would be 100 lb
yes below about 2 ma the input would be unreliable and you should fag an alarm and not use it.
I have built many weighting systems ad have found this is the best way to do it with an analog signal from the scales, of course the best ay to read it in digitally
There is no reason to think if they are converting a mv signal from the load cell to a 4-20ma signal that they would double convert it with a A/D from mv and then D/A to ma

 

[parky]

Some of the systems I worked on would read up to -50kg reliably (these were 500-1500kg vessels so -50kg was only a small amount below 4ma and the Mitsubishi would actually go negative (if it did not you could have errors close to the 0, however, never checked it's accuracy as it was always tared before batch, we had an auto tare but with limits of +- 10kg around 0, anything outside of this would alarm and would not tare to zero. The vessel would not be set back to gross until transfer started (to avoid accumulated tares on product left in vessel), however, this would never happen as after every batch the vessel would run a CIP.
Remember, I have been doing this for over 30 years and calibration following UKAS & BRC standards.

 

[a062549]

Single load cell?

 

[parky]

Does not really matter, 1, 2, 3 ,4 you either use a combinator or just wire them in parallel Some years ago we had a combiner go down, wired the loadcells in parallel, re-calibrated & worked fine. in actual fact some of our systems were wired like this anyway by the supplier. But for 5000g I bet it is only one.

 

[theColonel26]

I will just chime in here having not read the whole thread.

Resolution is not the same as accuracy. If someone wants an more detailed explanation let me know.

 

[drbitboy]

I will just chime in here having not read the whole thread.

Resolution is not the same as accuracy. If someone wants an more detailed explanation let me know.

Good point (even though it's been mentioned, if indirectly).


Then there is also precision to be considered as well.


I think the OP has a reasonable understanding: obviously I could be wrong, but reading between the lines it appears they are aware that one way to look at it is that resolution is the limiting case for accuracy.

 

[GaryS]

Just a few thoughts
I see here a reference to a combiner we called them summing boards.
Early on they were required if you had more than 1 load cell the after a while they were no longer necessary and you could just connect the load cells to the instruments directly. And then later they came out with the C2 Electronic Calibration

https://www.hardysolutions.com/prod...dy-process-toolbox/weightless-calibrations-c2

on any weighing system I do a have a third party calibration service come in and calibrate all the scales that way everybody can assured that weights are accurate

 

[parky]

Yes I'm not sure what it was called but Nobel started to use an electronic system for calibration it was a dam site quicker than hauling test weights onto the vessel. The guy still had test weights delivered to site ready for calibration though.