It helped in two ways, first, I am able to use a H-Z ref for the (-) in and second, with the way it is being used as a bridge I am getting better linear response on the output. I wouldn't say it's perfect but it is better.
The op-amp I used is a Bi-fet and TC seems to be much better with this one. Also I have a much better gain response in the linear region.
Although CMRR does not seem to play a part in this, simply due to the fact that the total voltage dt is no more than a 150mv, I have ruled out any aggravating effects from that.
Voltage gain is actually less with the Bi-fet than what I was using however, when I plotted the response curve I was able to get a much better looking output.
The other issues are, it is very possible I was getting some demodulated RF with the monolithic front end, but that still remains up in the air with that. I'd prefer the use monolithic due to it being safer to handle but I could add ESD protection if needed.
We used what we had at the bench that was easy to test with. We still have lots of dip IC's to work with and they make good test jig models.
I was going to use a comparator for the application but since I want to limit BW to reduce the chance of parasitic noise or oscillations, I thought a Op_amp would work just fine with their inherent miller in them. I don't need to worry about phase error in this app :)
Jamie