Need advice on opamp design

[richloo]

Hello, i am lack of experience in choosing different type of topology in opamp design. Practically, for 2 stage classical & folded cascode opamp, what is the highest gain or gain bandwidth product that they can hit? My application is high frequency filter, is it possible to use opamp-RC rather than gm-C?
Thanks in advance.

 

[ArchAngel]

the performance of opamp architechture depends on
process, try sc-filter for ur hf-filter maybe a better solution

 

[yeewong_su]

2 stage opamp can match you need. cascode opamp can give the high gain above 80db

 

[sunking]

both them you can use

and Gm_RC is better for high frequency

 

[richloo]

I have completed a 2 stage opamp with GBW of 330MHz, DCgain=60dB, GM=20dB and PM=54 degree. I found that i can't go beyond 3rd order unity gain 200MHz SallenKey LowPassFilter since the o/p signal is out of phase.
Can anyone give any comment/suggestion/solution?

 

[richloo]

BTW, the filter i m using is Butterworth... Thanks in advace...

 

[electronrancher]

richloo - that is a fast amp, what process did you draw in?

i would suggest gm-C instead of op-RC. if your overall amp is 330MHz, the diff pair is probably 500+MHz. Using gm's instead of opamp means you don't have the parasitic poles from an output stage.

 

[djalli]

I have completed a 2 stage opamp with GBW of 330MHz, DCgain=60dB, GM=20dB and PM=54 degree. I found that i can't go beyond 3rd order unity gain 200MHz SallenKey LowPassFilter since the o/p signal is out of phase.
Can anyone give any comment/suggestion/solution?

I see 300MHz bandwidth. What opamp did you use?

 

[richloo]

I used 0.18um CMOS, 2 stage opamp. but the transient response have slow rise time... can anyone tell me how to improve on this? thanks in advance...

 

[richloo]

electronrancher:
thanks for ur advice. i just wish to start gm-c but i can't coz i m running out of time. i am using 0.18um SCMOS. Since u said that it is a fast amp, from ur experience and with the simulated GainMargin & PhaseMargin, do u think possible to go to higher order (>3) filter? how to "tweak" it? thanks in advance.

 

[gevy]

You can download the useful E-book
"Inroduction to CMOS OP-AMPS and comparators" by R.Gregorian



It contains the answers to many practical questions.

 

[yeechyan]

What about you change the input differential pairs from PMOS to NMOS? NMOS is faster than PMOS. Anyway the noise will be higher.

 

[richloo]

thanks yeechyan! i have tried NMOS diff input pair n it is ok. but only little improvement...

i begin to try gm-c, if i wan to drive both resistive and capacitive load, how can i do that w/o introducing additional pole? an inverter?

 

[chency]

thanks yeechyan! i have tried NMOS diff input pair n it is ok. but only little improvement...

i begin to try gm-c, if i wan to drive both resistive and capacitive load, how can i do that w/o introducing additional pole? an inverter?

maybe you replace your first stage with a folded cascode

 

[richloo]

chency: do u mean that i have to replace only the diff pair stage?

regard

 

[qutang]

the structures of the picture is very useful.

 

[jiangwp]

The filter performance is function of the parameters of opamp and the value of capacitor and resistor.

On one hand, you should optimize the opamp , also do the value of resistor and capacitor.

The resistor and capacitor is load of the opamp . If their value is large , the opamp is work in the large and small signal region, then you can increase the slew rate of the opamp. If the filter process the small signal , the opamp work in the linear region , the bandwith is dominate .

I think you can improve the bandwith , slew rate and output impedance in the mean time.