Load cell / ina125 amp / pic 16f887 - question regarding gain

Hi everyone,

I'll be honest, I'm rubbish with op-amps, had some theory at uni and very little practical classes, so not much knowledge from that period stayed in my head. I'm currently doing a project which uses a load cell and I need to use an amp for the load cell signal to get anything reasonable on the ADC of the PIC 16F887.

Load cell is 1mV +/- 0.15mV / V and it's 1kg capacity. I'm planning to use 5V to power it, which will give me 5mV at full weight load (1000g). I'd like to get 0.1g precision, so I need quite a lot of gain from the INA125 to get there. With standard ADC on the PIC, I have 5V/1024 = 4.9mV per step, with the load cell and no amp I have 5mV / 10000 = 0.0005mV per 0.1mg. To read 0.1g as a single step on the ADC I need a gain of 10000 which is the maximum the INA125 can do (I think it's an Rg = 6ohm) and here's where things get a bit complicated, at least for me. A gain of 10000 means good precision, but surely at full weight load of 1000g I will get 5mV which is amplified and so ends up as 5mV * 10000 = 50000mV = 50VDC on the input of the PIC. This is the place where uni knowledge kicks in and suggests that this is not doable. ie. the maximum output voltage is limited by the amp and will flat out at a some sort of lower level.

The main question is, is it possible to get a full 0-1000g linear range with 0.1g precision? Most of the time the load cell will be used to weigh the contents of a container. The container weighs anywhere in the range of 300-400g and the contents would be 10-30g, so really I wouldn't mind a low range with high precision (assuming a gain of 10000, a single 0.1g is 5mV, so 10-30g would be 500mV - 1500mV), but I'm worried that the container will kick the signal out of range and overload the ADC input of the PIC.

Regards,
T.

hi dsc,
You should think in terms of resolution , not accuracy when using a 10bit ADC in a PIC, ie: 0 thru 1023 counts.

If you get 5mV at full load scale and you amplify that by 1000 to get 5.00V, you can resolve to only 1023 counts.

E

Thanks Eric,

indeed I should've used the term resolution rather than accuracy. As I said above, all I need is 0.1g resolution on a range of 0-30g, so I could theoretically use 10000gain, although I will be using various containers which weigh way more than 30g. I could simply revert to using a 100g load cell, but then I will overload the cell if I use a container weighing 300-400g. Ideally I'll stick with the 1000g cell, I'd need to use a tare option, so use a container, then tare and weigh 0-30g with 0.1g resolution. I'm quite sure how to go about this though, as with a high gain the 300-400g container will bring the INA125 output signal to max levels. Or so I think.

I've seen tutorials with people using 1000kg cells and getting 0.1kg resolution, so somehow that was doable.

Regards,
T.

hi,
Using 1000Kg load cells can give a 0.1Kg resolution.
If you consider you need at least 10000 counts for +/-0.1Kg , shows a 14bit , 16384 count is required, you should consider a ADS1100 A2D converter with a I2C connection to your PIC.
I use MCP3202 dual A2D 12bit converter , it gives me 4096 states.

With regard to the Base and Tare weights, you could offset these before the amplifier.

E

Thanks again Eric, I will go with the ADS1100 and I2C into the PIC.

Regards,
T.

I read the INA125 and get excellent resolution with the humble PIC18f4520..

The reference on the INA125 and the ADC module is set at 1.25v...( this is directly from the 125, If you don't, you bottom out around 0.6v )

I then oversample 24 times and then average down to 12bit... The result is VERY accurate...

Thanks for your input Ian, I've now switched to a 3000g load cell (better quality, but got one off ebay, so couldn't choose a lower range) and so need an even bigger ADC resolution. With 16bit I will have roughly 0.05g resolution per count, which is good (I only need 0.1g), but I don't think oversampling is an option here as I'd have to do 1024 samples in one go and need to complete the whole operation in less than 100ms.

Regards,
T.

If this problem is consistently troubling you, I guess it's the best time for you have a new load cell device. A load cell will offer high-end, digital output that is reliable for heavy duty measurements.

I have a bigger 3kg 2mV/V load cell now, although that doesn't make much difference as the range is higher even though the sensitivity is higher. I'd need more like a 600g 2mV/V, but it's hard to find stuff like that on ebay

Regards,
T.