Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
I'm waiting well-grounded explanation.
It's seem to me that separated symmetry centers for A and B more sensitive to gradients than kind when center is common.
ABBA ABBA is a common centroid pattern which provides an axis of symmetry , so any process related gradients are cancelled off if u lookm from the axis of symmetry.
AABBAABB on the other hand is a case wherein u just interdigitize them uniformly following a pattern, in cases where utmost matching is required like in a diff amp, this is not recommended, as the current in 2 branches would differ.
Hi Brittoo,
Can u explain the concept of differing of current in two branches for the pattern AABBAABB.how the current flows for the pattern ABBA ABBA.
When it comes to differential amplifier, we expect the current in both the differential pair tx to be the same, that is what it would be designed for in fact. Hence matching becomes critical here. To obtain optimum characteristics from this design, in the layout certain things are to be followed as thumb rule
- common centroid matching for symmetry
- identical routing
-guard ring connected to the corresponding potential.
-adding dummies to achieve poly symmetry as well as continuous diffusion.
Now if u look at AABB--AABB pattern, although there is a uniformity maintained, there is no symmetry, hence process gradients are not nullified.Now you do want your diffential amplifier layout to serve its true purpose right!! Take a look at the ABBA--ABBA pattern and lemme know why thiis is not better than the former combination.
tips to identify better pattern
1. godd dispersion of devices,
2. centroids closer to each other, ideally coincide
3. spread in both dimensions if possible
4. symmetry in
1. sizing of devices
2. position os devices
3. when you look into the inputs and outputs, identical routing structure makes better matching
hi Brittoo,
Thanks for the explaination..
I have another doubt.If the pattern consists of two transistors with even number of fingers then we can go for good matching pattern.
But what abt if we have more number of transistors with different number of fingers(either even or odd) then how we can get a good matching pattern..
pls tell me what constraints we should check out for getting matching patterns for such transistors..
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
