Fully Differential Folded Cascode OTA With SC-CMFB design

[cdbular]

sc cmfb

Hello,
I have designed a fully differential folded cascode OTA with SC-CMFB.
I haver tested the circuit using a SC-CMFB network using Verilog-A macromodel switches. And the OTA seems to work very well.

When I replace the switches with transistors the clock injection is so bad that the CMFB voltage never is mantained at a fixed value and the performance of the amplifiers is terrible: low gain, the CM voltage is not the same as the reference value. The common mode open loop gain seems to be as large as the differential gain (~70dB) as well as the Unit gain Bandwidth (>125MHz).

What can I do to improve this desing. What should I do to reduce the effect of the charge injection produced by the CMOS switches. What is the "secret technique".

 

[EmbdASIC]

folded cascode ota design

Could you post the schematic and the output waveform of your SC CMFB simulation?

 

[cdbular]

folded cascode ota

I attached the images of the schematic and the simulations.

When using the MOS switches the CMFB voltage varies in a much wider range. As you can see in the simulation screenshots.

 

[mbyoussef]

cmfb folded cascode

Why don't you use dummy switches to minimize the effect of charge injection?

 

[yschuang]

folded cascode

Is it visible in the simulation if you add dummy transistors to get rid of the effect of chage injection ?

 

[jcpu]

cmfb switch

Dear Sir,

May we know how you perform CMFB simulation with ideal switch.
Did you keep feedback / sampling capacitors same as in MOS switch case?

 

[cdbular]

ota cmfb design

Dear Sir,

May we know how you perform CMFB simulation with ideal switch.
Did you keep feedback / sampling capacitors same as in MOS switch case?

I made a Verilog A model switch and replaced MOS switches with this model. Basically the ideal switches wor like a voltage controlled resistance, with very low resistance when activated (200 omhs), and a very high resistance when off (10Gomhs).