Isolated Voltage Sense

[spec]

Further to post #6:

Better ADC (Texas Instruments ADS1115): https://www.digikey.co.uk/product-detail/en/texas-instruments/ADS1115IRUGT/296-24935-1-ND/2123299

Or a break-out board: **broken link removed**

Just add a couple of simple isolators in the S2C lines, download a bit of code and job done: high accuracy and good flexibility.

spec

 

[crutschow]

.....................
Back to the OPA2604 and common mode-input voltage range in general.
I'm still confused about the common mode-input voltage range. According to the sheet I need to stay 3V min off the rails. I'm not "using" the lower rail, its tied to ground, therefore I cannot be 3V above it (is this bad? already a violation and why it wouldn't work?).
I'm supplying 1.5V (24V divided) to the non-inverting input so 5-1.5 = 3.5, I'm 3.5V off the top rail, so I meet this requirement. I though this could have worked with no problem, but I'm having issues understanding it. Can you explain a little more please?

If the lower power rail is tied to ground then yes, the inputs cannot go below 3V above ground. Tying one input to ground will thus not work.
If you want to use a single supply voltage then you need to use rail-rail type op amp that will operate with the inputs down to ground.

 

[Mosaic]

If isolation isn't required due to dangerous voltages I'd suggest use an OPA as a differential instrumentation type amplifier. Then reference each voltage rail to ground and let the OPA do the differential result for you, scaled down using 100K+ input divider resistors and calibrated by a feed back trimmer.
I am doing this for a 36V and 12V pair of rails ref'd to a common ground. The output is scaled for a 5.12Vref uC ADC.

Got it done using a jellybean A spec LM324 ensuring that the 324 stays in class A mode. Added an 18V zener midway in the 4:1 voltage divider to ensure nothing past the 12V LM324 supply gets by (spikes etc.)

 

[drkidd22]

Hello,

So I ended up building a few board with the circuit below. The issue I have now is that the output at U20 varies between PCBs and I have to keep adjusting the scaling values in software to "calibrate" each board.

The only reason I used the IL300 is because the A and B ground are at different power rails, I figure if I tied them together I can just use a divider and Op-Amp.
Although the boards I have have the IL300 from same bin it seems the variation is still to great.

Any thoughts? I was thinking of replacing R77 with a mechanical potentiometer.

 

[KeepItSimpleStupid]

The feedback basically compensates the device for aging. It's not an absolute calbration. So, calibration is going to be required.

I can see two options:
1) A mechanical potentiometer
2) A digital pot with EEPROM capability. At least you can calibrate automagicly at production
3) If your just using it to measure the 24 V supply, there should be plenty of power supply monitoring IC's available. You MAY only need isolation on the I2C bus. See https://www.linear.com/product/LTC2945 for an example possibility.

 

[crutschow]

So you can or cannot tie points A and B together?

Certainly you could change R77 to a potentiometer to adjust the gain.

 

[drkidd22]

I can tie A to B, but wanted to keep the sources isolated.
If I do tie them, is there any specific way it should be done?

 

[crutschow]

I can tie A to B, but wanted to keep the sources isolated.

Why?

If I do tie them, is there any specific way it should be done?

You tie them at just one point, probably at the ground terminal for each of the supplies.

 

[KeepItSimpleStupid]

Do any of your supplies have sense leads?

 

[drkidd22]

No, no sense leads. These are just regular 60W AC to DC supplies.