PGA117: accuracy of current sense measurement

Hi Brendan,

There are a few things I notice about this circuit. Let's start with your input stage...

It looks like you are trying to measure a 0V input with the OPA325. I cannot see the negative supply because of the label but your positive supply (pin 5) is connected to -7.5V. I assume you simply mixed up the supply pins but even with -7.5V supply on the negative supply terminal, OPA325 has a maximum 5.5V supply which means the positive supply would have to be at most -2V. This puts a 0V measurement far out of the common mode input range of the device.

What power supplies do you have available? Do you plan to use separate supplies for each OPA325 on your board?

One option is to use the same bipolar supplies for both OPA325 at +/- 2.75V but that would put your 3V high-side full scale output out of the common mode range.

Another option is to use the single 3.3V supply you have available along with the LM7705 which provides a -232mV bias as your negative supply. The LM7705 is very convenient for allowing near-ground level voltages to be measured when bipolar supplies are not available, although it does come with some switching noise at 91kHz. What is your desired measurement bandwidth?

The PGA117 is a single-ended input device with a MUX that can be switched to measure different input voltages at different times. Do you plan to measure the high-side of your shunt resistor and then switch the mux to measure the low-side, and then use post processing to realize the voltage differential? Instead, let me suggest an additional differential-to-single-ended OPA325 stage before the PGA. see below.

I have attached this TINA simulation for your reference. Depending on your bandwidth requirements, you can add filtering capacitors at the U3 stage to reduce noise.

Brendan Hayes-Oberst said:

The programmable gain amplifier might not be able to magnify the sense resistor voltages at very low current levels or even higher ones even with large gain

Yes at high current levels you will need to adjust your gain setting to accommodate the output swing limitations of the PGA117, see below. That is the advantage of a PGA!

For the low current levels keep in mind that the minimum output swing to the rail is 50mV. Your low range of 0.5mA results in an ideal input voltage of 250μV to the PGA, and the highest gain setting of 200V/V results in an output of 50mV which is right at the saturation point. Keep in mind that there are also offset voltages from the PGA and the OPA devices and other non-idealities that will combine with the ideal input voltage. You may need to increase your shunt resistor to accommodate the lower current range with less error.Current_Shunt_OPA325_PGA117.TSC

Regards,

Zach

Hi Zach, 

Yes, I connected the bottom of the op amp to -7.5V. This can be changed because it is wrong. 

I have a few supplies available, I have ground, 12V and -7.5V from two switch mode power supplies coming into the board. The -7.5V is swappable with a -3.3V.  I feed the 12V into a linear voltage regulator that outputs both 3.3V to a digital and analog side of the board. The analog voltage regulator supplies the positive rail of the op amps (opa117s). They all are powered from this voltage regulator with bypass caps before each component.  So, I do not plan on separating each of the OPA117 power supplies on my board.... is this ok? I assume it is as long as the voltage regulator can supply the amount of current needed by the op amps. 

I like the idea of a small negative supply voltage. The only drawback is my ADC has an absolute maximum rating on the input pins of -0.3V. The recommended minimum is -0.1V. So I have to be careful of not putting in anything less than that to the pins. Since what it will be measuring will never be less than 0V as long as it does not have some negative gain I assume this will be ok. let me know if there is way with hardware to protect this negative voltage condition. 

My desired bandwidth, not sure but it is pretty low. The max amount of samples that my ADC can take at that oversampling ratio is 4800 per second. I expect to be making 100 samples a second max. 

Do you plan to measure the high-side of your shunt resistor and then switch the mux to measure the low-side, and then use post processing to realize the voltage differential? Yes. I have 8 channels on this ADC so I could do both and settle later if one is superior. 

I am debating whether to have two constant current resistor banks, one of higher resistance that is selected when needed to measure very small currents. I might incorporate this in current or future design upon testing. 

Thank you for that simulation. I will simulate it with TINA and most likely comment again about it soon. I was also thinking of using a instrumentation amplifier with the OPAx325 series op amps. https://en.wikipedia.org/wiki/Instrumentation_amplifier. It seems very similar to your circuit. 

I really appreciate your help and insights. 

Best,

Brendan  

Hi Zach,

Thank you for that link. There are loads of good information and designs for measuring current in this situation on that link. I have one last question though about my original screenshots. I was using the opa325 to buffer the input signal into the PGA117. I don't think this is necessary and I believe the voltages ground and the top side of shunt resistor can go straight into this PGA without filtering. Would you agree? 

Thanks for the help I will mark the chat as this resolved my issue when I hear back from you about that opa 325 being not needed. 

Best,

Brendan 

please let me know if you think those are suitable choices. 

To clarify, I would get rid of opa 325 and pga 117 and just use those two op amps. Most likely together, since the INA 190 does not have a gain of 1 but the INA 851 does. The INA190 is a current sense op amp and INA 851 is an instrumentation amplifier with output clamping. 

Hi Zach, 

I just got finished with the schematics that I hope will provide some clarity. I just put in the INA190s for now. There are three of them. They have 200, 50 and 25V/V gains to measure different current levels. They all sense across the same sense resistor bank .  For the larger currents I was planning on using a similar configuration with a gain of 1 or the INA851. The ADC that I am using is a sigma delta ADC with 8 channels. They can be either 4 differential or 8 single ended. The ground on that will be the same ground as the INA190s in the schematic.

I chose the INA because they are high precision current sense op amps. If there is a similar current sense op amp with a 1V/V gain then I can use that for high currents. I was hoping that this multiple gain configuration would give me great resolution across the whole spectrum of 1mA to 6A. 

I do not have time to simulate tonight as I have never used TINA before. Please let me know your thoughts. 

Thank You,

Brendan 

the 1V/V current sense op amps all needed I2C bus and because of other reasons in my design I would rather avoid this. 

will do.