OPA376-Q1: What is the criteria to calculate the unity gain negative feedback resistor value in Sallen Key LPF?

Well Samrat, 

The intent for that resistor would be to match the source impedance on the non-inverting input for bias current cancellation. however, with CMOS parts like the OPA376 that does not work anyway, so adding such a high value only hurts noise and phase margin. I would either get rid of it, setting it to zero, or if you have a place for it in the layout maybe a 100 ohm R699. 

Well actually Tom and Samrat, the stability issue here is also very real. 

The attached shows that in a few easy steps where only 4deg phase margin results with the 43kohm back into the 20pF input parasitic C. 

I would be curious where the original schematic came from - someone is shipping out designs assuming bias current cancellation is required (the high feedback R) that might in fact be oscillators. Not uncommon when you neglect the op amp parasitics in your design flows. The feedback R needs to go way down or to zero in this design. 

OPA376 model testing in SKF filter for phase margin.docx

here is the LG file, 

OPA376 SKF stability with high Rf.TSC

Thanks Michael for enlightening me about the issue it can create .

After some more analysis ,If needed then i will again get back to you.

Thanks & Regards

Samrat Pushpesh

Thanks Thomas for the valuable suggestion.

Thanks & Regards

Samrat Pushpesh

Hi Samrat,

I assumed that the LP filter design was validated and the original question was merely about the additional resistor in the feedback path. Since Michael took it one step further and found the filter design results in insufficient phase margin, then that is another and different issue.

Sallen Key filters synthesized using anything from filter handbooks to software should result in a design that has sufficient phase margin. Certainly, this may not be the case when the user select particular response conditions that produce low phase margin, or apply certain op amps that by virtue of their complex open-loop output impedance (Zo) lead to low phase margin. However, most active filters that one develops using commonly available filter synthesis tools and the majority of op amps will results in an inherently stable active filter.

Since Michael determined the particular Sallen Key low pass you provided has a very low phase margin (4 degrees) it should be reconsidered before being put into use. I suggest you use an active filter synthesis program such as TI's Filter Design Tool with inputs that concur with the filter response requirements. You can find the tool here:

http://www.ti.com/design-resources/design-tools-simulation/filter-designer.html

If the response requirements are reasonable, I would expect that the resulting filter would have considerably more than 4 degrees of phase margin and it would be stable. If you find you end up with very similar Sallen Key component values to what you are currently using, then we would need to examine your response requirements more closely.

Regards, Thomas

Precision Amplifiers Applications Engineering

If I follow the TI filter designer link and do a gain of 1 Butterworth at 50Hz, I get this - no feedback R, which is good, but still using exact C on the feedback without standard value snap - I think there might be  a place to force that, but not seeing it now,