I have question about the common centroed matched array. I have this array for example to match two transistors A and B
ABBA
BAAB
ABBA
BAAB
or I can make it
ABBA
BAAB
for sure the first array is supposed to be the best but during work I see it is consuming a lot of wiring which I think it will add considerable parasatics.
The second array is working fair but it has a problem with big transistors in which dividing them by 4 is not enough. Therefore I have two options in which I want to take your opinion on
the first is extending the array horizontally like this way
ABBA ABBA
BAAB BAAB
or I keep using the
ABBA
BAAB but using two gate per unit transistor in the array
In my opinion, the second placement is good enough. It's simple for wiring and still get matching for A, B transitor.
There will be trade off when we choose too complex placement. Like wiring complexibility, parasitic resitor and capacitor...
The second is seems better for matching when it has better common centroid.
One more thing is the number finger of A and B in the second configuration should be even for abut.