Never used bipolar modes for true differential until now.
Using a 20 bit ADC (MAX11206) for strain gauge (Wheatstone bridge) measurements. The max looks like it takes an input range of +ref to -ref even when power source is not driven to -ref. So my common mode voltage just needs to be ground for measuring difference from Wheatstone.
The TI LMP8350 should do the trick. I need my common mode voltage to be what is 0 volts on the rest of my board. Does this make sense? I am not sure how to drive the ground voltage of the diff op amp to -X volts so that my common mode voltage can be ground relative to the rest of my board.
Please let me know if any of my description was not clear.

Thx for input.

 

Did you read pages 25 and 26 in the MAX11206 datasheet?
How much resolution (in mV, μV, or whatever) do you need?

Your input voltage cannot go below GND, or above AVDD, if you are using a single positive power supply. See attachment.

 

Are you sure? I thought Vref negative could go below ground when in bipolar mode. Look farther down the same page you quoted from. Yes I have read the datasheet about 10 times. And I definitely need the resolution. I have never worked with bipolar design below though, or had to work with negative voltages relative to my normal ground. How to people usually go about this?

 

Are you sure? I thought Vref negative could go below ground when in bipolar mode. Look farther down the same page you quoted from. Yes I have read the datasheet about 10 times. And I definitely need the resolution. I have never worked with bipolar design below though, or had to work with negative voltages relative to my normal ground. How to people usually go about this?

More documentation attached.
The second one is from page 10.
I asked you previously how much resolution you need. The datasheet suggests using the part without an amplifier if you don't need 20 bits of resolution, thereby avoiding the offset, gain and noise problems which come along with amplifiers.

 

I see. I am using a 2.048 Voltage reference. At 3000lbs load, I will output 31mV max. Note that this 31mV can be plus or minus depending on tension or compression. I need to measure the load with 1lb resolution, but correctly. I do not ever want to measure 1.5 lbs as 2. 1.9 wobbling around 2 and 1 is fine. That being said, I need to measure 31mV/30000 resolution (1.0333uV). Same as 62mV/60000 that I actually need to measure. If I just placed the 2.048 volts across the Wheatstone and placed the difference directly into the analog in neg and pos, I would be reading 1.95uV resolution. I have to us some gain or dividing to get the resolution I need, but am not sure about the best way to do it.

 

I see. I am using a 2.048 Voltage reference. At 3000lbs load, I will output 31mV max. Note that this 31mV can be plus or minus depending on tension or compression. I need to measure the load with 1lb resolution, but correctly. I do not ever want to measure 1.5 lbs as 2. 1.9 wobbling around 2 and 1 is fine. That being said, I need to measure 31mV/30000 resolution (1.0333uV). Same as 62mV/60000 that I actually need to measure. If I just placed the 2.048 volts across the Wheatstone and placed the difference directly into the analog in neg and pos, I would be reading 1.95uV resolution. I have to us some gain or dividing to get the resolution I need, but am not sure about the best way to do it.

Why do you need 0.1lb measurement resolution, but only 1lb readout resolution?

 

I see. I am using a 2.048 Voltage reference. At 3000lbs load, I will output 31mV max. Note that this 31mV can be plus or minus depending on tension or compression. I need to measure the load with 1lb resolution, but correctly. I do not ever want to measure 1.5 lbs as 2. 1.9 wobbling around 2 and 1 is fine. That being said, I need to measure 31mV/30000 resolution (1.0333uV). Same as 62mV/60000 that I actually need to measure. If I just placed the 2.048 volts across the Wheatstone and placed the difference directly into the analog in neg and pos, I would be reading 1.95uV resolution. I have to us some gain or dividing to get the resolution I need, but am not sure about the best way to do it.

You're asking for 3000 divisions (1 lb in 3000lbs). I've done 50,000 divisions (2g in 100kg) with no front end diffi-amp. The A/D wasn't the Maxim part, but something similar.

What you want to do is easy w/o analog front-end.

 

Also, you only need, internally, 1/3 lb. resolution (9,000 divisions) to ensure no instability whatsoever on the display at 1 lb increments (though I'd take it to 1/4th pound or better). This is equivalent to an ENOB of just over 15 bits.

 

Ron, I do not need to read 1/10 lbs for 1 lb resolution, but it would be nice. I could do with 1/3 or 1/4 as Joeyd999 mentioned. I always shoot for more resolution to leave room for filtering in software if need be. I think I found a nice solution. I will use the MAX11206 with the TI LMP8350 with a gain of 10. I will have some overkill in terms of resolution...but this is for internal use at our company so cost is not very important. If we need the resolution later, we will have it. Thoughts?