problem of source degenaration

[noiseless]

For the below two topology, I thought the linearity of (b) should be better than (a) for there is no body effect. However, the Specture simulation results show that (a) is better than (b). Anyone know why? Thanks

 

[safwatonline]

what is ur test setup?
if u force the same current while they are having the same Vg , then the one in (b) will have lower Vth =>> larger Veff =>> lower gm
and since the equivalent Gm=gm/(1+gmR) becomes more linear (i.e. nearly equal 1/R) when the gm is larger so the one in (a) will be more linear.

 

[noiseless]

Saf,

I used self-biased cmos to bias gate voltage. The self-biased cmos also configured as resistor degenarate to overcome the mismatch of bias voltage. If the body connected to the gnd directly, so did the self-biased transistor. So I guess urs is not fit to my case. thanks anyway.