Current Mirror Layout with Interdigitation

[goatmxj666]

Hello.

I am currently doing the current mirror layout.

The sizes of each are as follows.

I would like to layout with interdigitation method for matching. I think common centroid layout is not possible as it is too time consuming.

Also, I made some of the mosfets 8um because it was too long to do all the mosfets in 4um.

I was thinking of doing something like "D D D D D D D D D D D D D D C C C C C B B A F E G A B B C C C C C D D D D D D D D D D D D D", is that okay?

I really need some advice from layout experts.

Any advice would be greatly appreciated.

Thanks.

 

[dick_freebird]

You're probably already overthinking it but one recommendation is to make every "finger" a finger-pair to make full low level symmetry, then distribute those pairs as you see fit.

Until you get to very large W*L, random mismatch is likely to still dominate any gradient driven "anecdotally systematic" (thermal, alignment) offset contributions.
Especially at finer nodes with RDF now a thing.

 

[goatmxj666]

You're probably already overthinking it but one recommendation is to make every "finger" a finger-pair to make full low level symmetry, then distribute those pairs as you see fit.

Until you get to very large W*L, random mismatch is likely to still dominate any gradient driven "anecdotally systematic" (thermal, alignment) offset contributions.
Especially at finer nodes with RDF now a thing.

Hello. Thank you so much for your answer.

What I understand is that create finger pairs from "D D D D D D D D D D D D D D C C C C C B B A F E G A B B C C C C C D D D D D D D D D D D D D" and then arrange them in the order of DCBAFEGABCD. Is this correct?

 

[sidun.av]

What I understand is that create finger pairs from "D D D D D D D D D D D D D D C C C C C B B A F E G A B B C C C C C D D D D D D D D D D D D D" and then arrange them in the order of DCBAFEGABCD. Is this correct?

1. Your array is probably too big for a single-line interdigitation pattern (the aspect ratio of the array would be too large which is bad for gradients);
2. Keep at least 2 fingers for each device;
3. Keep the same unit width for each transistor (use multipliers to achieve target total width);

My idea of device sizing and matching is attached. It's somewhere in between of common-centroid and interdigitation, but it passes the gradient test, so it should be okay. Consider using Layout GXL for faster device placement (you can enter your pattern there).
Good luck!

P.S. Even though the array looks massive, when you will stack the neighboring MOSFETs, I think it should be fine

 

[goatmxj666]

1. Your array is probably too big for a single-line interdigitation pattern (the aspect ratio of the array would be too large which is bad for gradients);
2. Keep at least 2 fingers for each device;
3. Keep the same unit width for each transistor (use multipliers to achieve target total width);

My idea of device sizing and matching is attached. It's somewhere in between of common-centroid and interdigitation, but it passes the gradient test, so it should be okay. Consider using Layout GXL for faster device placement (you can enter your pattern there).
Good luck!

P.S. Even though the array looks massive, when you will stack the neighboring MOSFETs, I think it should be fine

View attachment 193102

Thank you for your very detailed and thoughtful response. It really helped me a lot.