Gain in the common mode feedback

[shanmei]

design a fully diferential op amp with continuoous cmfb, dose the gain in the cmfb should be specified according to the gain in the main op amp ?
and how to test the gain in the cmfb ?

 

[yxo]

Finite gain in CT CMFB make error of the output common mode voltage. ABout testing CMFB, you can find in old posts

 

[abcyin]

ideally, the CMFB gain should be infinite to ensure the output level to be equal the reference voltage.

as to the simulation of the CMFB,
you can first break the CMFB loop at CMFB output point, and apply a voltage source with 1V ac magnitude in to the opamp at the break point ,
make sure that ac magnitude is 0V for the rest of the sources in the schematic,
run ac analysis,
result>direct ploct>AC gain and phase,
select the output of the CMFB,
then you'll get the the CMFB gain, at the same time, the phase margin of it can be got for stability verification.

hope it helps

edward

 

[surno]

To abcyin

I don't understand your expain. Please explain detail. And what's the difference your method and using "iprobe" method at the spectre simulation. Is the same method?

 

[abcyin]

To abcyin

I don't understand your expain. Please explain detail. And what's the difference your method and using "iprobe" method at the spectre simulation. Is the same method?

yes, you can also use "iprobe" to get the CMFB phase margin, the simulation results are the same after validation.

 

[iamxo]

ideally, the CMFB gain should be infinite to ensure the output level to be equal the reference voltage.

as to the simulation of the CMFB,
you can first break the CMFB loop at CMFB output point, and apply a voltage source with 1V ac magnitude in to the opamp at the break point ,
make sure that ac magnitude is 0V for the rest of the sources in the schematic,
run ac analysis,
result>direct ploct>AC gain and phase,
select the output of the CMFB,
then you'll get the the CMFB gain, at the same time, the phase margin of it can be got for stability verification.

hope it helps

edward

But if you break the loops, what you got is not exactly the loop gain, isn't it?

 

[surianova]

ideally, the CMFB gain should be infinite to ensure the output level to be equal the reference voltage.

as to the simulation of the CMFB,
you can first break the CMFB loop at CMFB output point, and apply a voltage source with 1V ac magnitude in to the opamp at the break point ,
make sure that ac magnitude is 0V for the rest of the sources in the schematic,
run ac analysis,
result>direct ploct>AC gain and phase,
select the output of the CMFB,
then you'll get the the CMFB gain, at the same time, the phase margin of it can be got for stability verification.

hope it helps

edward

But if you break the loops, what you got is not exactly the loop gain, isn't it?

The gain you get from that analysis is loop gain, GH. Basically, you need the gain for opamp about 60 dB is good enough, it give u the DC error of 1mV only. If you design the gain too high, the stability issue will arise.

 

[abcyin]

ideally, the CMFB gain should be infinite to ensure the output level to be equal the reference voltage.

as to the simulation of the CMFB,
you can first break the CMFB loop at CMFB output point, and apply a voltage source with 1V ac magnitude in to the opamp at the break point ,
make sure that ac magnitude is 0V for the rest of the sources in the schematic,
run ac analysis,
result>direct ploct>AC gain and phase,
select the output of the CMFB,
then you'll get the the CMFB gain, at the same time, the phase margin of it can be got for stability verification.

hope it helps

edward

But if you break the loops, what you got is not exactly the loop gain, isn't it?

what you can get from that analysis is the loop gain, and this method is used to measure the stability of the CMFB loop,
as to the gain of the Common mode feedback block, ideally, it should be infinite to reject the error between the Vcm and Vcmr.