stabilt of beta reference

[analog_ambi]

i have designed a low power beta multiplier refernce (70nA) as shown. The opamp is a 5T diff pair.To check for stabilit i broke the -ve feedback loop and plugged iprobe from analoglib in cadence. stb analysis shows phase margin of 20 deg and gain margin of 14dB with C=2pf. I m not able to achive better phase margin even after shifting the mirror pole of the 5T diff pair. A phase margin of 60 requires C of 100p. The transient response with step Vdd (1us rise time) is shown as attached which is not quite stable. Can i expect it to work if tapped out?

Lastly, even if i achieve PM of 60 deg with C=100pF, the transient response is till the same which is shocking to me!! Pls help

 

[analog_ambi]

think i have found the problem!! The bias of the 5T opamp is derived form the beta multiplier as shown in attached figure. If the bias of the opamp is done externally then i get a first order rise of current in transient simulations. It forms a slightl unstable loop. It will be great if I get suggestions to improve the loop stabilit

I can use the self bias 5T as shown in attachment but i have to make aspect ratio of 1/200 to get 100nA bias current