The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> A Question aboutLayout Match
https://designers-guide.org/forum/YaBB.pl?num=1159284936

Message started by Rocky.Lee on Sep 26^th, 2006, 8:35am

Title: A Question aboutLayout Match
Post by Rocky.Lee on Sep 26^th, 2006, 8:35am

The frequence of my chip is around 2GHz and there are lots of differental pairs in my chip. So match is the most important issue to me.What confused me is that how to match a differental pair? For instance, there is a differental pair which contains two device A and B.Both A and B have a M=2. I should place them as the sequence of A B A B for interdigitaltated plan, just like a lot of chips does. But many books say that use the sequence of ABBA for a common centroid layout is better than ABAB to take out the effect of oxide and dopped grads.
So my question is that why lots of designers use ABAB style instead of ABBA? Is there anything I missed?
Thank you for your kind help~

PS: A B
B A --- I give up this cross type because of complexly wiring. Does it obligatory？

Title: Re: A Question aboutLayout Match
Post by ACWWong on Sep 26^th, 2006, 8:58am

In tha absense of dummies ABAB might be better than ABBA as the environment of each device is equalised for A and B.
Even with dummies, for RF layout, its true alot of designers prefer the DABABD rather than DABBAD because of the belief in localised process gradients in modern deep submicron process being more important for RF. It of course depends on the size of A and B.... and i've not seen measured evidence to support the difference.
Still for the best layout:
DDDD
DABD
DBAD (actually B and A here should be upside-down/back-to-front written)
DDDD
The use of D I have seen to give quantified improvements in mismatch. If your circuit is differential pair then with common source/emitter it shouldn't be hard to do the optimum layout with good symetrucal routing.
Anyway the best book on this topic is Alan HAstings "Art of Layout".
cheers
aw

Title: Re: A Question aboutLayout Match
Post by ccd on Sep 26^th, 2006, 11:15am

Without dummies, I don't understand why ABBA is better than ABAB when Bs see two surrounding As and Bs don't. So, different stress effects on the two groups.

Title: Re: A Question aboutLayout Match
Post by ACWWong on Sep 26^th, 2006, 12:07pm

ccd wrote on Sep 26^th, 2006, 11:15am:

Without dummies, I don't understand why ABBA is better than ABAB when Bs see two surrounding As and Bs don't. So, different stress effects on the two groups.

If you think of linear process/temperature gradient going across the array, then ABBA is better than ABAB. For simplicity, lets assume A and B are resistors and sheet resistance increase linearly left to right (either due to stress or temperture gradient) then the net value of both As will equal the B value better in an ABBA whereas the ABAB array will have B at a higher value.

Title: Re: A Question aboutLayout Match
Post by Rocky.Lee on Sep 27^th, 2006, 6:16am

I got it!
Thanks AW very much!