Why we need to reduce miller effect in LNA design?

[GDF]

We often add cascode device in common source LNA design for several purposes.
One of them is to reduce the miller effect. Why do this? What does it matter if
we don't reduce the miller effect?

 

[Grig]

Hi
You will have gain slope with Miller effect.
If You don't need the wide bandwidth but narrow high freq bandwidth
You can use inductance to compensate Miller effect

 

[firewind]

To avoid miller you should always use Cascode or CG setup. Miller only happens in CS amplifier because you have that Cgd cap and it can really limit your bandwidth at high speed. w3-db = 1/RC and C=Cgd(1+K). You see C is getting amplified by its own gain, the denumerator gets really large and that will reduce w3-db greatly.

If you use Cascode and CG instead of CS, that Cgd cap will not short your input and output. Thus, you don't have Cgd problem, that is no more miller effect and you will get more bandwidth. The tradeoff is more noises with Cascode and CG, and lower input impedance with CG. (sorry I had to change that I mistyped earlier).

 

[GDF]

To avoid miller you should always use Cascode or CG setup. Miller only happens in CS amplifier because you have that Cgd cap and it can really limit your bandwidth at high speed. w3-db = 1/RC and C=Cgd(1+K). You see C is getting amplified by its own gain, the denumerator gets really large and that will reduce w3-db greatly.

If you use Cascode and CG instead of CS, that Cgd cap will not short your input and output. Thus, you don't have Cgd problem, that is no more miller effect and you will get more bandwidth. The tradeoff is that you will loose gain and output swing with Cascode and CG.

Does the miller effect affet the Q of Zin of CS amplifier?

 

[dingjingfeng]

this can increase the stablilty and prevent the LO leakage!

 

[firewind]

Does the miller effect affet the Q of Zin of CS amplifier?

Of course it affects the Q, if you included all the two main caps Cgs and Cgd at HF then write your transfer function. You will get a second order system, you will get denumerator as S^2 + w0/Q x s + wo^2. If you include that Cgd of the CS into the transfer function you will see that not only will it adds a zero into your system but also change the Q value. Just write out it out on paper and play around with a few number and see it for yourself. Realistically, for RF you should include more cap/resistivity but that just makes hand analyzing brutal.

It will also affect Zin because at HF, Cgd is shorted so Zin will see everything all the way to Zout in your small sig model.