the problem with the preamp

hello guys,

I have a problem with the small signal analysis with the preamp circuit. You can see the circuit in the picture. At the beginning I thought

the gain Av = 2*gm*Rd. But when I do the simulation, the result is difference with my analysis.

can you help me?




https://obrazki.elektroda.pl/85_1346598160.png

Are you counting the outputs individually? Or are you bridging across them?

Each output swings within a volt range which is the same swing as the input. This is because mosfets are voltage-controlled devices.

However notice that you can put a load across the outputs, and get 2x the voltage gain.

My simulation shows this amp can serve as a power amp. The topmost resistors can be as low as 10 ohms.

ql396387901 said:

At the beginning I thought the gain Av = 2*gm*Rd.

Click to expand...

It depends how you define gain.
If Vref1 = Vref2 = constant, and you define gain as: gain = (Vout+ - Vout-) / (Vin+ - Vin-), then gain = Rd*gm, where gm is the value for a single device.

BradtheRad said:

My simulation shows this amp can serve as a power amp. The topmost resistors can be as low as 10 ohms.

Click to expand...

With idling currents = 100uA? :-?

I wrote: My simulation shows this amp can serve as a power amp. The topmost resistors can be as low as 10 ohms.

Click to expand...

godfreyl said:

With idling currents = 100uA? :-?

Click to expand...

The 100 uA bias resulted in the output staying at the upper supply rail.

I found I had to increase the bias to several mA, just to bring the output level to a sensible area of operation.

For a while I tried high values for the upper resistors. However the outputs were mostly a straight line, except at input peaks.

I could not obtain linear response until I made the resistors around 330 ohms.

Hi,
if Vref+ and Vref- were tied together, then set them being equal to the common mode voltage of Vin+ and Vin-. the gain of the circut would be 2*gm*Rd, like simple differential pair.
The Cin(Cgs) could be halved at cost of twice power and addtional load cap. I don't think this circut work well as peramp, but it can be a good buffer. because it is difficult to get high gain and low noise.
Ziqi

Yes, I tend to agree with Ziqi, because my result (hand calculation) is as follows:

(Vout+)-(Vout-)=-2gm*[(Vin-)-(Vin+)+(Vref+)-(Vref-)].