Reducing input parasitic capacitance

[giasis]

Hi,

I am reading some paper about Regulated Cascode for Transimpedance amplifier, and have some questions from it.

In the paper, there is a statement that "The CG and CD input configurations relax the effect of large input parasitic capacitance from the bandwidth determination better than the conventional common-source (CS) input."
(Title of the paper: "1.25-Gb/s Regulated Cascode CMOS Transimpedance Amplifier for Gigabit Ethernet Applications")

Basically it is saying that if input is connected to photodiode generating current, since photodiode usually has large capacitance, it will limit the bandwidth. But it says CG and CD amplifier reduces the input parasitic capacitance more effectively than CS amplifier.

Here I can understand that CG can reduce the input parasitic capacitance well, since CG has low input impedance (~1/gm). But how does CD amplifier reduce the input capacitance better than CS? Based on my understanding, since CD and CS input is gate of MOSFET, it has huge input impedance. So it doesn't make sense to me that CD and CS reduces the input parasitic capacitance.

Could anyone explain this for me? Your reply would be very appreciated.

 

[keith1200rs]

Common source has bandwidth reduction due to the Miller effect - common drain doesn't suffer from that effect so bandwidth is higher.

Keith

 

[giasis]

Common source has bandwidth reduction due to the Miller effect - common drain doesn't suffer from that effect so bandwidth is higher.

Keith

Thanks for your reply Keith.
Yes I am aware that due to the Miller's effect, gate capacitance(Cg) becomes much larger in CS than CD. But the thing is that, does it makes sense to use CD at all for connecting to photodiode which has large capacitance(Cpd) and large resistance(Rpd)? The time constant of CD + photodiode becomes Rpd*(Cpd + Cg), so it makes the time constant even slightly bigger.

But for me, connecting photodiode to CG stage makes much more sense because the impedance looking into source is 1/gm which can be small, and by connecting CG with photodiode, time constant becomes [Rpd||(1/gm)]* (Cpd) which can be made to be small since the parallel resistance can be approximated to be 1/gm (assuming 1/gm << Rpd).

In summary, my question is if using CD stage for transimpedance amplifier(For photodiode)(as described in the paper) even makes sense.

 

[keith1200rs]

I don't agree with your interpretation of their statement "The CG and CD input configurations relax the effect of large input parasitic capacitance from the bandwidth determination better than the conventional common-source (CS) input." as meaning "Basically it is saying that if input is connected to photodiode generating current, since photodiode usually has large capacitance, it will limit the bandwidth." To me it says that the CG/CD configurations reduce the effect of the transistor input capacitance.

Also, in a transimpedance amplifier with feedback low open loop input impedance is not required - the low input impedance comes in the closed loop due to the feedback. So, a CD stage works fine provided there is gain in the loop. Around half my discrete transimpedance amplifiers have had common drain input stages, although there are other reasons for doing that, not just input capacitance.

Keith

 

[giasis]

I don't agree with your interpretation of their statement "The CG and CD input configurations relax the effect of large input parasitic capacitance from the bandwidth determination better than the conventional common-source (CS) input." as meaning "Basically it is saying that if input is connected to photodiode generating current, since photodiode usually has large capacitance, it will limit the bandwidth." To me it says that the CG/CD configurations reduce the effect of the transistor input capacitance.

Also, in a transimpedance amplifier with feedback low open loop input impedance is not required - the low input impedance comes in the closed loop due to the feedback. So, a CD stage works fine provided there is gain in the loop. Around half my discrete transimpedance amplifiers have had common drain input stages, although there are other reasons for doing that, not just input capacitance.

Keith

Thanks for your reply, Keith. I think your interpretation about the statement is right. It makes sense now. Thanks!