Load Cell calibration

HELLO,

I've been a member here for a while but this is the first time i've sasked a question; i usually get the answer or solution i am looking for from past Q @ A.

ANYWAY; HERE GOES. PLEASE NOTE; THIS IS NOT PART OF A REAL WORLD PROCESS; I HAVENT ELABORATED ON ALL THE FINER DETAILS AS I DONT THINK ITS NECCESSARY IN THIS INSTANCE; THE CONTROL PLATFORM IS MITSUBISHI AND THE SOFTWARE IS GX DEVELOPER.
Any help on this would be greatly appreciatted.

OK...I have 2 no. 50 litre tanks...both tanks are mounted on load cells with 4-20ma output (FROM TRANSDUCER).



What im stuck on here is....

• 0 liters = 135 in dataregister......should be 4ma....actual value = 5.1ma
• 10 liters = 241 in data register .....5.3ma
• 20 liters = 332 in dataregister.....5.5ma
• 30 liters = 424 in dataregister.....5.6ma
• 40 liters = 515 in dataregister.....5.75ma.....
• obviously the cells are rated for a much heavier load than is required here and i will loose resolution but in this case thats ok.....ma values are approx. but there is very little increase in the ma value.
• 50 liters = 615 in data register.....should be 20ma
• no matter what i do (zeroing etc.) i cant get the value at 0 to equal 4ma. It starts at 5.1 ma as you can see.
• S0; Can you tell me how i can scale these values into engineering units.....LITERS

Thanks.

The 5.1 mA is probably caused by the weight of the tank (tare)?
You are only getting about 0.016 mA per litre, obviously the load cell is far too large.
If you subtract 5.1 and multiply by 0.016 you will get EU.
You should try to change the load cells for ones with less span.
Roy

Thanks for replying Roy.

The 5.1 mA is probably caused by the weight of the tank (tare)? ...naturally; i was aware of this but i see what you mean.

Just to let you know what i was at (see below) and if you could elaborate on my efforts; basically can tell me how you arrived at the solution.
From y=mx+c where y = 635 and c = 135...the upper and lower count values.]
x = litres
m = 9.6...so 480/9.6 = 50 liters...and so on...is this way of going about it way off?
Thanks again.

counts @ 50L = 615
counts @ 0L = 135

615 - 135 = 480 counts for the span of 0L to 50L

50L/480 counts = 0.104 L/count

To determine liters from the measured counts:

(counts - 135)*0.104 = Liters

ex: (635-135) = 49.92L :close enough for government work

Dan

thanks dan,

looking at it after you showed me your angle i worked this one out; tried and tested it....its grand.
x = liters..y=mx+c...so..x=y-135/9.6..and y = the current count value.
so say for 40 liters...x=515-135/9.6=39.58..slight discrepency but i think its the way to go..count values per liter were not 100% accurate from my first post.

anyway, see what you think, maybe im still wrong; but genuinely thanks for the help.

same result.

liters = (counts - 135)/(9.6counts/L)

so it is.
good man,
thanks.
i've got these tanks done now and i've just finished the code.

i have some flowmeters to calibrate tomorrow on this same skid which will also be great fun i'm sure...basically the instruments were taken from much larger redundant plant and this skid is being set up as a test skid for mechanical technicians to train on...the flowmeter was probably taken off a 4" line and now its to be inserted on a 1" line.
Anyway i might be posting again tomorrow.
Thanks for the help.

A word (Or two) of caution. I have personal experience of a plant where the loadcells on a tank are vastly over rated.Try ing to measure 100 litres on 3 of 2 tonne cells (2X3=6T!).

The net result is that the plant is affected by temperature (As the metal frame they are on expands), operators walking on adjacent staircases etc.

A bit of maths tares off the tank weight but you can return 20 minutes later and the zero has shifted 2 Kg.

If your plant is for training, that's ok, but dont be suprised that you cannot use it in real life!

thanks for the input markie;
could you elaborate please with an example; this is the first time i've had to use load cells and i'm trying to soak up as much info as possible.... i wouldn't use oversized cells like this in industry; no matter how tight their budget is.

Although it looks like you have no problem with this
A thing to keep in mind is that anything attached to the tank can affect the weight or cause the tank to hang up or add weight to tank
resulting in error.

Dan Bentler

Hugh,
What type of flowmeter are they?
If the answer is Orifice plate/DP cell, provided they now have a line size plate they will be just as accurate.
If the answer is Mag flow or Vortex it probably won't because the jump from 4" to 1" is too great, the velocity will be 1/16.
Sounds like you have the load cell sorted but if you can find a label let us know what it says, perhaps there's a way of getting better resolution.
Good Luck
Roy

Hugh, in my application - which I didn't design - there are 3 loadcells each with a fixed output, zero weight = zero and max weight is max output. These are on a modbus system with a converter to get it into profibus, into a Siemens plc. As they are 0 - 2 tonnes each, the "Weight" is the sum of all three. As a previous post said, you also need to watch the connections to and from the tank - in my case they are stainless steel 50 mm bellows type couplings as this will affect the action and free movement of the loadcells. Also - as is my problem - if you are trying to measure small quantities with relatively large loadcells, they will have more of an affect.

The "Live" weight is fed into two data registers and the "tare" button (On a HMI) subtracts that amount from one of them to give a "Zeroed" weight.

As the tared zero is now set, I am measuring between this value and the 6 tonne maximum. Subtracting this tare, I am now trying to measure 100kg which is only 1.7% of the range. i am pretty sure a reasonable loadcell is only accurate to 1% in the real world!

Another problem is the hysteresis of the loadcells, where we add 100kg, to refill, then subtract 50kg. you can take out 20kg and this does not register!

I probably have not helped - rather suggested what not to do!!!

Sorry for not getting back to you sooner fellas; thanks again for the input.
Dan having seen some of the posts you were involved in the past i know to pay heed to what you have to say; apprreciate the advice. The tanks are connected (the inlet and outlet of each) with stainless steel flexible hoses; i know when the hoses are charged they may be heavier and so on. When i seen your post on Friday i decided to research applications of load cell use further. I underestimated the implications involved in their use and i have to say that rendered me unprepared; when "You fail to prepare; you prepare to fail". Apart from some weighbridges and chemical processes etc. i'm at odds as to why loadcells would be used because of the inherent inaccuracies of their use; can you (and the other lads) please tell me in your opinion where and when you would recommend load cells being used. Ta.
Roy thanks again. The flowmeter they had in mind was actually a 3" mag type but thankfully we were able to get our hands on the correct size for this application (1"). Thanks for the help on the load cells and your interest in the flowmeters.
I actually got sidetracked on Friday and i didn't get to the flowmeters...i was doing the manifolding between the tanks and pipng up the pump etc...i'm not back on this one until Monday and i know i'm going to have further probs; i'll pass the info on the rating plate of the cells to you then. After spending some time on this i asked an old mentor and this is what he recommended to me....

Hello Hugh
3 ways to solve.
1. y = m.x + C
2. Recalibrate A/D module when tank is empty = Offset, 1/2 fill tank and set as gain. See A/D manual. If you do this you wil get 0 for 0 L, 1000 for 25 L. and 2000 for 50L.
3. Replace the Load cells ( Oh No)

OK for 1.
For 0L
We want y to =0
y = m.135 +c Note at 0L you get x =135

Therfore c should equal -135
Now to get C
at 20 L it would be nice to have 200 in the y value, giving one decimal place ie 15.5 L would read 150.

200 = m.332 -135
m.332 = 335
Thefore m = approx 1

In the PLC read in A/D value say into d10, In D11 I will get tenths of litres

d11 = d10 - 135

Note D10 = A/D value
D11 = Tank tenths of Litre

So

M8000
---| |------------[ sub d10 k135 d11}

This should work but it will not be wonderful.

Option 2 Recalibrate A/D not easy to expalin in an email but it will do it all in the A/D.
Can you change the gain of the amplifier on the load cell?
....I like option 2 a bit better. His option 1 is smiliar to the path we were on the other day..i know your probably thinking of doing what he recommeds here in the transducer rather than the PLC but there is only zero and span pots...what do you think?

Markie; thanks man, you gave me some good food for thought here..Another problem is the hysteresis of the loadcells, where we add 100kg, to refill, then subtract 50kg. you can take out 20kg and this does not register!...would you elaborate on this please, specifically on the hysteresis...does this only apply where the tanks product is being heated. Again i value our input, ta.
Thanks fellas.

you should:
a) subtract 135
b) multiply result by 10
c) divide result by 96

if you want one decimal place, multiply by 100 instead of 10. this way you will read 367 for 36.7 liters.

Hello Hugh,

Load cell on a tank with 50 litres can be cause of error....to avoid the problems of added components/pipes/staircases....etc..we had put pressure transducer (Eurotherm PTE200 series) which was munted at the lowest point of the cone at the base of the tank.

Just a suggestion..

Cheers.