how to calculate the ICMR of the full differential telescopic amplifier

[han6716914]

Lets say, a fully differential nmos-input telescopic amplifier, and I assume the supply is 3V, and the overdrive voltage of the M9, and M1, M3, M5, M7 is 0.5V, 0.2V,0.2V, 0.3V, 0.3V. that means the ouput swing is 1.5V for one side. and the threshold voltage for P/NMOS is 0.7V.

I want to calculate the ICMR of the OTA, I know the equation that

Vod9 + Vgs1 <= Vicm <= Voutmin - Vod3 + Vth1

and since Vod9 = 0.5V, Vgs1 = Vod1 + Vth1 = 0.2 + 0.7=0.9V, and then
1.4v <= Vicm,

but for the other half, how do I determin Voutmin ? if the Voutmin = Vod9 + Vod1 + Vod3 = 0.5 +0.2 +0.2= 0.9V, and then Vicm <= Voutmin -Vod3 + Vth1 = 0.9 - 0.2 +0.7=1.4V.

in all I get, 1.4V <= Vicm <=1.4V, isnt this weird ?

I need some help to clarify this for me. Thanks.

 

[erikl]

1. I think you mix up the overdrive voltage Vod with the drain-source voltage Vds.

The usual definition of Vod is

Vgs = Vth + Vod

(... where Vod would be negative for subthreshold operation),

whereas in your definition

the overdrive voltage of the M9, and M1, M3, M5, M7 is 0.5V, 0.2V,0.2V, 0.3V, 0.3V

... these voltage values actually are Vds values and haven't much in common with Vod values. In your next equation

Vod9 + Vgs1 <= Vicm <= Voutmin - Vod3 + Vth1

again, both Vod parameters actually are Vds values,

... whereas in your equation

Vgs1 = Vod1 + Vth1

... Vod1 is the real overdrive voltage in the usual definition.

In your next equation

Voutmin = Vod9 + Vod1 + Vod3

... these Vod parameters again actually represent Vds values, and so your next equation

Vicm <= Voutmin -Vod3 + Vth1

cannot be correct, and instead should read

Vicm <= Voutmin -Vds3 - Vds1 + (Vth1 + Vod1)

... where Vod1 is not necessarily the same as Vds1 (and could be much higher in a non-unity-gain configuration).

So I'd suggest to review and correct your ICMR min/max estimation.

2. From your (incorrect) definition

Vicm <= Voutmin - Vod3 + Vth1

... I take as your intention that the telescopic ota should work as unity-gain buffer. Is this correct?