The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> the linearity of the source follower
https://designers-guide.org/forum/YaBB.pl?num=1298720100

Message started by yishuihan on Feb 26^th, 2011, 3:34am

Title: the linearity of the source follower
Post by yishuihan on Feb 26^th, 2011, 3:34am

Hi guys:

we know that for the NMOS source follower, when Vbs is not zero, the linearity is low because of the body effect. it is not suggested using the source follower when the linearity is the main consideration.

however, we also know that the body effect can be regards as a gmb conductance and behaves like an source degenerator. to some extent the linearity can be better.

who can tell me why there is this paradox?

thanks in advance.

Title: Re: the linearity of the source follower
Post by Rakesh on Feb 26^th, 2011, 4:36pm

For the Source follower you have negative feedback in the gm loop, So the swing vgs is small, However if you think of vbs swing is higher and hence the gmbs is not constant for all range of the voltage. Hence this will result in non linearity.

Correct me if i am wrong
Rakesh

Title: Re: the linearity of the source follower
Post by yishuihan on Feb 27^th, 2011, 2:54am

Hi Rakesh:

that's true there is a negative feedback loop via the degenerator (resistor or current source), quick conclusion would be the larger loopgain, the better linearity.

i realize my confusion might be the working conditon of the transistor. if the AC signal is samll, body effect is not a severe issue. for the large signal, the source will not track the gate voltage any more due to the dependence of the Vth on the Vbs.

anyhow, for the small signal, when Vbs is none zero, the equivalent gmb is paralled with others, loopgain will decrease more or less, thus the lineariy.

Title: Re: the linearity of the source follower
Post by Rakesh on Feb 27^th, 2011, 6:57am

Even for large signal if u have sufficient loop gain then i think source will track gate voltage.
Basically there are two paths from input to output. once is via gm and other via gmb. The path via gm has negative feedback and hence it is a linear path. However path via gmb is a open loop system and hence u ve a large swing . So u have non linearity.
The change in vth with signal is included by variation of gmb with signal

Title: Re: the linearity of the source follower
Post by carlgrace on Mar 3^rd, 2011, 3:50pm

yishuihan wrote on Feb 26^th, 2011, 3:34am:

There is no paradox. Try simulating a source follower with a non-constant source degeneration resistor. It would not improve linearity. The gmb conductance is not constant across the output swing of the source follower.

Title: Re: the linearity of the source follower
Post by RobG on Mar 7^th, 2011, 2:50pm

To give another viewpoint... Degeneration helps stabilize current mirrors so that a "large" variation in gate-source voltage causes a "small" change in current. By the same principle, a "small" current change in a degenerated fet will result in a "large" voltage change. This "large" voltage change is exactly what you don't want in a source follower!

As you know, the effect described above isn't the normal problem (it is the Vt dependence on bulk-source voltage); however, I hope it helps answer the question on why it doesn't help.

Title: Re: the linearity of the source follower
Post by Pravesh Kumar Saini on Mar 12^th, 2014, 4:57pm

Hi Guys
What if we make Vbs=0 by using IPW(deep-nwell)?

Title: Re: the linearity of the source follower
Post by raja.cedt on Mar 13^th, 2014, 4:08am

Hi,
in that case there is no Question of body effect, hence it is irrelevant here. But with deep-nwell source will loaded with huge capacitance, may be limited frequency response, i guess this is the reason why people go for pmos source follower than deep-nweel case (even if their technology allows).

Thanks,
Raj.