Multiple Current Mirror for Biasing

[melkord]

I need to generate 8nA and 70nA from 2.4uA.
Here is what I did.
Is there any better way to do this, especially to generate 8nA?

 

[arthorios]

You need a factor of 2.4uA/8nA = 300, which is a quite large current ratio. You can get to this by using a series-parallel association. There are many papers on literature addressing this (https://ieeexplore.ieee.org/abstract/document/309905, https://ieeexplore.ieee.org/abstract/document/1683892). But you can also refer to Vadim Ivanov's ISSCC tutorial (print attached), showing how unity transistor (same W/L for better matching) series-parallel association can be used.

Then you can have only one current mirror to get your 8nA, which would improve the matching of your current mirror compared to what you currently have.

I hope this can help you.

 

[melkord]

You need a factor of 2.4uA/8nA = 300, which is a quite large current ratio. You can get to this by using a series-parallel association. There are many papers on literature addressing this (https://ieeexplore.ieee.org/abstract/document/309905, https://ieeexplore.ieee.org/abstract/document/1683892). But you can also refer to Vadim Ivanov's ISSCC tutorial (print attached), showing how unity transistor (same W/L for better matching) series-parallel association can be used.

Then you can have only one current mirror to get your 8nA, which would improve the matching of your current mirror compared to what you currently have.

I hope this can help you.
View attachment 174412

Thanks!
Is there a way to tune the nominal value other than changing the input current?
I use 2uA and expect 2uA/256 = 7.8nA but get around 9.5nA.
I worry about this value since after fabrication, I am not sure whether it will be monitored.

 

[arthorios]

Thanks!
Is there a way to tune the nominal value other than changing the input current?
I use 2uA and expect 2uA/256 = 7.8nA but get around 9.5nA.
I worry about this value since after fabrication, I am not sure whether it will be monitored.
View attachment 174423

Your current ratio is off by too much; something is not correct. Check if your current ratio is really 256 and if your transistors have similar VDS (and are saturated). For accurate current copying, use cascodes.

 

[melkord]

Your current ratio is off by too much; something is not correct. Check if your current ratio is really 256 and if your transistors have similar VDS (and are saturated). For accurate current copying, use cascodes.

I tried with cascode, I got more or less the same result. Thanks, though.