which one is faster - source follower or common gate amp?

[sutapanaki]

which one is faster?

Here is something that came out of a discussion not long time ago - what is faster - source follower or common gate amplifier? Assume the same conditions for both - same bias current, same load capacitance, same size of transistors. Any opinions?

 

[jacneo]

Re: which one is faster?

Here is something that came out of a discussion not long time ago - what is faster - source follower or common gate amplifier? Assume the same conditions for both - same bias current, same load capacitance, same size of transistors. Any opinions?

I think the source follower will fast.
I think the gate why call it the gate, because it need a break power.
It take some time, so I think the source follower will faster than common gate.

 

[gauiver]

i also think that souce follower will be faster becaz the becaz the output is just across the resistor, but u can check that. make the circuitry (with the desired bias point) and then take the two outputs from the single transistor. using oscilloscope ucan check. do tell what u found. i hope i am right

 

[sutapanaki]

It is for IC design - analog design, not that much for board applications. And it is more of a principal question, not something that I need badly. Yes, I can simulate, but I believe one should not answer questions like that through simulations.
As for the answer, well I think it is a bit ambiguous. You see, source follower indeed charges the gate and the gate-drain capacitor, but there is no Miller effect for the gate-source capacitor - on the contrary it is bootstraped-out. The output capacitor on the other hand is chrged through a low output resistance.
On the other hand, in common gate the input source should charge the gate-source capacitor, but the impedance level is low there. At the output it only has to charge the output capacitor and the gate-drain capacitor (which we assume equal to the source-follower case) but through a high drain impedance of the transistor.
So perhaps both configurations should have comparable speed. What do you think?

 

[flatulent]

another factor

Another factor is the gain bandwidth product. This makes it dependent on the load resistance. If the common gate has high gain the bandwidth will decrease.

 

[sutapanaki]

Which means that if you design common gate stage with low gain like 1-2 then it may become even faster than source follower. Which perhaps means that the load resistance is small and the time-constant at the output is smaller.

 

[electronrancher]

sutapanaki you take a good stance toward your question! no answer is easy, which is why you ask here!

yes you are right, GBW will be very high with common gate - the gate is held strongly at a potential, it is AC ground. this guy only acts like a resistor, to instantly (or very fast) pass potential to the other side.

But the trade is this - you have to dump so much current into this, it can't be used in sensitive areas. Think how you would disrupt the delicate 1/2-1/2 balance of a diff amp if you add a power-hungy resistor to the output side!

We know CMOS source follower uses capacitance to modulate the channel, so why worry about a little? 15fF can be driven VERY strongly by many signals, so our miller cap only hurts our input if we make the gain large enough to have a VERY strong signal on the output! Plus 15fF still means 2GHz from sub-micron CMOS..

You don't seem to need much gain (although 10-100x is always nice) so make a gentle souce follower if you are taking a precision output and moving it to another stage.

If your input is a power RF stage or something strong and very hi-frequency, 15fF will matter a lot. But to trade, you have a lot of power to give. Common gate is now best. You waste a little power in this "resistor", but you have a nice 1:1 buffer for your bipolar or GaAs output amp..

What application are you working on? Those necessities will tell you when to use each case.

 

[sutapanaki]

I don't have an application right now that needs the choise between source-follower and common gate. As I said this came out of a discussion. That's why I specified that the two stages are under similar conditions. But it is true that these are different kind of bufferes - source follower is a voltage buffer - high input impedance and low output one. Common gate is a current buffer - low input impedance and relatively high output one and being a current buffer it takes current as input signal.
There is one other aspect when we consider the high frequency operations. From the output the source follower exibits inductive behaviour (under certain input conditions) and this sometimes can be nasty. I don't remember well but I think common gate also has some inductive behavior.
Anyway, it was just an interesting question.