Strange results after switching tail current in a mixer

[microwaver]

Dear gays:

It’s weird thing happened when I simulate the NF of mixer. I changed the tail current MOSFET: one symbol split into two W/L remain the same but the results tell different. Can you tell me why? thank you in advance!

 

[safwatonline]

changed or unchanged

these two are not the same as the first is pseudo differential
to get better understanding think of the case when RF+ is high and RF- is low and then think of the current in each differential branch, u will see that the maximum current in one branch in case of split tail current source is half that in case of joined current source.

i.e. i think what u wanted to do is splitting the transistor into two transistors but keeping their drain connected

Added after 2 minutes:

just out of curiosity which one gives better noise figure? "i guess the joined one"

 

[microwaver]

Re: changed or unchanged

yes, thanks , indeed the joined one get better noise figure ,why? Can you expain? two legs get equal current before and after the change, the transistors size remain also ,which affect the noise figure?

 

[safwatonline]

changed or unchanged

well, i don't see the rest of the circuit but i can take a wild guess that they have the same DC Gain however in the first case u r switching the whole tail current while in the second u r switching only half of the current which probably means lower gain and lower NF.

however it has been a while since i looked into mixers so i am not sure

Added after 1 minutes:

BTW is the load active or passive?

 

[microwaver]

Re: changed or unchanged

thanks to safwatonline: it is passive load just resistor I adopt the conventional gilbert cell .the current/size unchanged leads to NF changed. surprising or not!

 

[safwatonline]

changed or unchanged

seems like u didn't understand me, the two cases don't have the same current during switching since the maximum current in the joined tail current source is completely steered during switching also since the amount of current is different the signal swing should be smaller (V=IR) in the case of splitt current source which should mean lower NF
"i guess"