Electronics 2000

Hi Nick,
I have bult the cicuit succesfully and the summing circuit is working as it should although it dosnt appear to be inverting the DC signal as the multimeter is showing positive. The third scales monitoring the output seems happy though so its not a problem.

As the instrumentation amplifier has a gain of 5 I have reduced the output of the summing op amp using a potential divider to bring it back down to a similiar level as the output of each scale but the output is not quiet as linear as the input. I can adjust for one weight say 50kg but then a lighter weight say 20kg is out by a factor of 20%. I have used mostly 10 ohm resistors in the summing amp and on the output. I have tried larger resistors but it just seems to kill the output.

Could you offer any sugestion of how I could control the levels from each weighing scales effectivly and cure the non linear gain. I can draw up the circuit I have built if that helps.

Regards
Nevjc

Hi Nevjc,
It's great to hear that you were able to build it all and get broadly the required functionality.

As for the non-linearity, a few things come to mind:

The first is that 10 ohm resistors sound very low, on the whole you are dealing with small currents so I would imagine that bigger resistors would be better. On the whole, as long as the required ratios are maintained the actual values shouldn't make much difference to operation, but may cause output stages to be over loaded if the values are too low, or excess noise if they are too high.

Secondly, what tolerance are your resistors? I would suggest using 1% throughout and if you can, go for even tighter than that. Though cost will start to go up pretty quickly so there maybe a tradeoff here. I would start with 1% and see how things go.

The third thing to consider is input bias currents. These have caught me out a number of times and for a long time I never even understood their significance until it was pointed out to me. Most opamps source a very small ammount of current from their input pins. Normally in the nA range. Now if the input source is low impedance for example a 350ohm strain gauge then this current is taken care of and effectively shorted to ground. If however the input signal is very high impedance say from a piezo sensor, then there is nowhere for this current to go so it starts to mess with your signal. Often a 1Meg resistor from the input pin to ground helps to route this current to ground without loading your source.

Of course the problem you are seeing may not be related to to any of the above lol. If you could post a schematic then I would be more than happy to look over it.

Kind regards
Nick

Hi Nick,
Thanks for that. I did try larger resistors and it seemed to stop the circuit working but maybe the resistirs I tried were to high.
I have dropped a PDF of the circuit here https://www.dropbox.com/s/2s7zkcse6qm0i ... %20Amp.pdf

I would be grateful if you could have a look and give your opinion recomendations.

Thank once again for the help

Nevjc

Hi Nevjc,
I'm sorry for taking so long to reply, things have been a little manic here.
I've had a quick look at the circuit, when you say that the output for 20kg is 20% out, is it 20% higher than you are expecting by any chance? According to the datasheet for the INA122, they can source a maximum of 3mA on their output pins. Therefore the maximum output voltage you can expect from the INA in your circuit is 30mV (0.003A * 10R).

I'm wondering if the output is saturating at say 30 or 40kg and as such throwing off your calibration.
Have you tried plotting the output of the instrumentation amps over the required weight range? My gut feeling at least is that you will find the graph flattens out at some point. If this is the case, then I would increase R1 and R2 to > 1k ideally 10 or 100k. That way they should be having little to no effect on the output signal of the Inst. amps.

My apologies if you have already tried this

Kind regards
Nick

Thanks Nick I will have a look and try your suggestion and try and plot the graph. Not that easy to do as the scales turn themselves off faily quickly and then lock when the weight is static.

I will come back to you

Kind Regards

Nevjc

Hiya,
I was thinking more plotting it at different weights with a multimeter connected across the output of the instrumentation amp. At this stage I'm thinking we should probably make sure the first stages are giving us the results that we expect, then move onto the summing amp itself.

Kind regards
Nick

Ok I will give it a try and come back to you
Regards

Hi Nick,
I have changed R1 and R2 to 15Kohm and measured the output of the instrumentation amplifier over a range of weights from 5Kg to 70kg and it appears that there is a small log type hump around the bottom end between 0 and 20kg and then it is fairly linear after that.

I have disconnected one side of the summing amp until I get one side correct. I have changed R3,R5 & R6 to 15K Ohm also but it just kills the output of the summing amplifier.

Do you think perhaps my supply voltage needs increasing so the amps are nearer there operational centre?

Look forward to hearing from you

Regards
Nevjc

Hi Nick,
I have been looking at R1& R2 and have found that with just 50K ohm load the output of the Instumentation amp is now reasonable flat over the range.

I am now moving on to the summing amp.

Regards
Nevjc

Hi Nevc,
That's great news, I'm glad to hear that the response has improved.

Hopefully the summing amp should be a little easier to debug now, you will probably find that you can increase the resistor values there as well now that R1 and R2 aren't loading the instrumentation amps. Keep in mind that 3mA max on each output pin though when experimenting with the resistor values in the summing amp.

Just one more thing to consider, from the schematic, it looks like you have designed an inverting summing amp (there are non-inverting designs available, I think there is one further down on the page I linked to last time). So the fact that it appears to be working as a non-inverting amp causes me a little concern. Even if it does appear to be a happy coincidence hehe.

Kind regards
Nick