How to simulate the gm of a differential amplifer?

[myblues]

I know the gm=dIo/dVin, and we can simulate it by ac analyse. It is a frequency analyse.
But I want to get the relationship of gm to the Vcm(common mode input voltage). How can I get it?

Thanks!

 

[philipwang]

Dear Myblues,
Are you designing a constant gm amp? As many paper mentioned gm versus Vcm can be plotted, I don't know how to achieve this result through simulation directly, why not try to achieve the result through simulating the relate between current and Vcm?
Welcome any better means!

Bg,

 

[myblues]

Hi,philipwang
As you say, I am designing a constant gm rail to rail amplifier. The gm vs Vcm might can't be achieved directly as you say, though the current vs Vcm can be got directly by simulation. But I do want the gm versus Vcm! How can I get it?

Thank you very much!

 

[philipwang]

Myblues,

If current can be achieved, just through dI/dVin to get the gm, then plot gm versus Vcm, as to how to get the result of dI/dVin you can study the wave-view software, it maybe supply this function! For example if runing simulation with Hspice and viewing the result with awaves, awaves can supply this function.
So pay attention to the function of wave-view software, just try it.
Good luck!
Bg,

Hi,philipwang
As you say, I am designing a constant gm rail to rail amplifier. The gm vs Vcm might can't be achieved directly as you say, though the current vs Vcm can be got directly by simulation. But I do want the gm versus Vcm! How can I get it?

Thank you very much!

 

[myblues]

Dear philipwang,
What you said is very right! But it is difficult to get the dI/dVin(though I know the derivative fuction in awaves), because my current is achieved at common mod input. dI/dVin should be got at different mod input and can't achieved in common mod. When in different mod, I get the dI/dVin, so I get gm vs Vid, but how can I achieve gm vs Vic(Vcm)? There is no Vcm yet!
Ples help me!

Thank you for your kindly help!

 

[xwcwc1234]

Can someone provide a spice file as an example ?

 

[philipwang]

Try another way, just sweep the Vcm and then probe di/dvin through defined function, in hspice parameter defined by myself is supported.

bg,

 

[sarit_g_t]

well one thing I do know is that the dI/dV thing usually gives a wrong picture in the way it is done in Cadence. I remember once having used Cadence to get the value of slew rate for an amplifier and got some crazy values bcos of the way Cadence calculates differentials. So use it with caution...