Help: Question about the rail to rail op amp

I'm designing a rail to rail amplifier. As anybody knows, the input total transconductance(gm) must be constant. There are several methods to accomplish it. Such as using 1:3 current mirror, level shift, tail current control, voltage control.
I use maximum current select circuit to achieve a constant gm. (see the attachment) By using the maximum current select circuit, the output current Io1 and Io2 is constant.(when simulate, two resistors are connected with Io1 and Io2 to VDD respectively) But I do not know how to simulate the tolal gm! Without the maximum current select circuit the input tolal gm is the sum of the two tail mos's gm. But with the maximum current select circuit I DO NOT know how to simulate the input total gm. In other words I DO NOT know which mos's gm is the total gm.

Thanks a lot!

The total gm is : Max(gmN,gmP) as shown in the figure.The highlighted line is the overall gm curve.
regards,

Thanks!
In rail to rail fig, I know the gmn is of M5, and gmp is of M6. But in the gm.jpg you post , the gmtotal is not the curve by simulation!
My question is with the maximum current select circiut how to simulate the total gm?

plot the drain current of M5 and M6 with VCM, then derivative .
it is impossible to simulation the gm directly.

asgg said:

Thanks!
In rail to rail fig, I know the gmn is of M5, and gmp is of M6. But in the gm.jpg you post , the gmtotal is not the curve by simulation!
My question is with the maximum current select circiut how to simulate the total gm?

Click to expand...

Hi,asgg
Now that you said"My question is how to simulate the total gm", How can you get that "in the gm.jpg you post , the gmtotal is not the curve by simulation"
Theoritically, gm_total=Max(gm_N,gm_P). So you can just simulate the gm of the N and P differential pair seperatly then you can get the total gm as shown in my figure.
Of course, current mirrors will have some mismatch. If you want to take this into account, you can simulate the output current curve (IO1 and IO2 in your figure) by varying the input voltage Vin+ and Vin-. Take the derivative, you'll get the total gm curve.

Hi,Alles Gute

Thank you very much!
Would I can get the gm_total from the equation below?
gm_total=(Io1-Io2)/(Vin+ - Vin-)

You're right, just one thing to remind: It should be a derivative, di/du cause it's a small signal parameter.