Designer’s Guide Community

Dipankar

Community Member

Offline

Posts: 59

fully differential OPAMP : common mode stability
Mar 09^th, 2010, 4:35pm

Dear All,

I was simulating common mode stability of an OPAMP inserting probe (vdc=0) at the common-mode feedback branch by "stb" aalysis.PM. To get a a good PM ( ~50-60 deg) I need to reduce the gain of my CMFB amplifier to such an extent that dc voltage of vout+ and vout- is off from the CM level by atleast 60-70 mV. OPAMP is folded cascode (pmos input diff pair) and CMFB is DDA-CMFB. Can someone help ?

Back to top

With Thanks and Regards,
Dipankar.

IP Logged

Manas Junior Member Offline Posts: 15 Bangalore,INDIA	Re: fully differential OPAMP : common mode stability Reply #1 - Mar 9^th, 2010, 10:58pm What u mean by DDA-CMFB ??
Back to top	IP Logged

Berti

Community Fellow

Offline

Posts: 356

Re: fully differential OPAMP : common mode stability
Reply #2 - Mar 9^th, 2010, 11:28pm

I have once implemented source degeneration to lower the gm of the DDA-CMFB (Manas: DDA=differential difference amplifier) and compensated the lower gain by a higher output resistance. In this way the bandwidth of the CMFB is reduced (unfortunately) and consequently the PM increased.

But maybe you can find a way to shift you first pole to lower frequencies?

Cheers

Back to top

IP Logged

subgold

Community Member

Offline

Posts: 97

Re: fully differential OPAMP : common mode stability
Reply #3 - Mar 10^th, 2010, 5:26am

hi,

i dont know your output range and desired cm level, but given an folded cascode output, 70mV of cm offset basically indicates your dda is just inactive, i am afraid you will suffer a serious CMRR degradation.

i think berti's suggestion might be helpful as long as you still have some freedom to manipulate the gain and speed of the main amp. you could also try to optimize the sizing of your cascode transistor and find a better way of feeding back your cmfb loop.

otherwise i see no choice but burning more current into your dda.

Back to top

IP Logged

Dipankar

Community Member

Offline

Posts: 59

Re: fully differential OPAMP : common mode stability
Reply #4 - Mar 12^th, 2010, 6:10pm

Thanks for the inputs. I'd like to discuss one issue �on the same line.

I use the following test bench (attached as TB_OPAMP.jpeg) to characterize all ac parameters. �Under certain bias conditions the
loop-phase plot starts from 0 - rises to �+90 and then falls where as the loopgain plot rolls of normally from some 90 dB(see "cmfb_stb_plot.jpeg"). �It occurs only at certain bias conditins. But at the same condition open loop gain or phase ( ( VF("/OUTP) - VF("/OUTN") )/ (VF("/inp") - VF("/inn")) ) looks normal. �If I do the same stb simulation with the feedback resistors removed I get reasonable stb-phase-plot (starting from 180 and rolling ogg accordig to pole positions) . So �I'm curious which signal loop is considered by spectre �while doing "stb". I �guess I Should remove the feedback resistors while doing cmfb loop analysis by stb. �Please advise.

Back to top

opamp_stb.rar

With Thanks and Regards,
Dipankar.

IP Logged

subgold

Community Member

Offline

Posts: 97

Re: fully differential OPAMP : common mode stability
Reply #5 - Mar 15^th, 2010, 8:19am

Dipankar wrote on Mar 12^th, 2010, 6:10pm:

Thanks for the inputs. I'd like to discuss one issue �on the same line.

I use the following test bench (attached as TB_OPAMP.jpeg) to characterize all ac parameters. �Under certain bias conditions the
loop-phase plot starts from 0 - rises to �+90 and then falls where as the loopgain plot rolls of normally from some 90 dB(see "cmfb_stb_plot.jpeg"). �It occurs only at certain bias conditins. But at the same condition open loop gain or phase ( ( VF("/OUTP) - VF("/OUTN") )/ (VF("/inp") - VF("/inn")) ) looks normal. �If I do the same stb simulation with the feedback resistors removed I get reasonable stb-phase-plot (starting from 180 and rolling ogg accordig to pole positions) . So �I'm curious which signal loop is considered by spectre �while doing "stb". I �guess I Should remove the feedback resistors while doing cmfb loop analysis by stb. �Please advise.

i think the position of your stb probe is wrong. it should include both CMFB loop and the resistive feedback loop. now the probe excludes the resistive feedback. basically the probe should be directly connected to either the input or the output of the main amp.

Back to top

IP Logged