[SOLVED] GSRR vs PSRR ... should they be identical?

[diarmuid]

Hi all,

PSRR = rejection of disturbance from power line
GSRR = rejection of disturbance from ground line

In my simulations both PSRR and GSRR results are identical.

Is this correct and if so why?

Thanks,

Diarmuid

 

[erikl]

For a totally symmetrical design they can be identical, but not for most practical designs, as the rejection ratio depends on the architecture of design (impedance to the considered line, e.g. cascode blocking - or not).

Results may depend on your test bench setUp.

 

[dick_freebird]

In regions where the amplifier open loop gain and feedback
determine the output's rejection of supply and ground, the
results ought to be similar. The question of "which ground?"
might come into play. At higher frequencies where the amp
and loop have lower gain, you start to see path and
sensitivity differences come up that can no longer be
covered up by the larger loop.

You would expect a single ended ground referred loop
to directly follow that ground's perturbation, but if you
also refer the output measurement to that same point
it may all rinse out. The relation of this ground, to the
ground reference of whoever's criticizing, remains a question.

 

[diarmuid]

Thanks for the responses. So Ive attached the circuit I am seeing PSRR = GSRR at DC.

Its a standard V2I circuit which takes in 1V and generates a current from this across a resistor. The OTA
is folded cascode with NMOS input diff pair and vssa referred CMFB. PSRR / GSRR measurements are
made at the output of the 2nd stage ("vfb" in the attached).

Circuit is far from symetrical so why at DC am I seeing PSRR = GSRR?

Thanks,

Diarmuid

- - - Updated - - -

quick correction ... typo in the attached ... ground line should read vssa and not vdda

 

[erikl]

Circuit is far from symetrical so why at DC am I seeing PSRR = GSRR?

I don't think both RRs should be identical here, at least not for higher frequencies (s. dick_freebird's argument above).
Could you show your test bench(es)? Where/how did you inject the vdda resp. vssa interference?

 

[diarmuid]

Yes you both are correct (there was a small bug in my tb).

At DC both are the same but as frequency increases they start to deviate by a couple of dB.

Thanks for the help,

Diarmuid