Hi,

I have often seen discussion about use of slew recovery devices, especially for folded cascode opamps to prevent the transistors from turning completely off (Razavi, Johns & Martin). However, most of the schemes suggested in the texts seem to offer little help when actually used.

Maybe my opamp is not really suffering very badly from the problem, and so I see very little change on adding these devices (they are not really called into action). Or on the other hand, maybe I have implemented the slew recovey devices wrongly.

Can anyone provide some insight into this topic? The standard methods I have seen are (for NMOS input opamp with PMOS folding nodes):

1. Clamp the folding nodes if they should fall too far below VDD.
2. Clamp the folding nodes if the difference between them exceeds a certain limit.

Thanks & Regards
Vivek

Hi Jerry,

Thanks for your comment. However, the problem is not really solved this way. If I don't design the clamp correctly,
then I may see no improvement in performance or even worsening.  So, this question cannot be answered unless I have
a good working clamp, and if I had that, I wouldn't be asking the question.

Is there a good design guideline or some particular fact to keep in mind when inserting the clamps? I just did some rough
calculations treating the clamp device as if it would be a diode. The body effect was something I took a look at as well.
However, I don't know if there is anything else to watch out for. I am not using very fine line technology, and the amp is fully diff.
so that I don't really expect a drastic cutoff on the devices.

Any comments?

Regards
Vivek

Hi chase,

The ratio of supply to threshold is about 3x-4x in my case. Could you post a copy of your elaborate clamp or any references to it?
I found the same effects with my clamp circuit. However, I did not build my clamp to have a small VGS drop, reasoning to myself that
the clamp should be OFF unless there is a large transient. Otherwise, it would have become too big.

Regards
Vivek

Vivek,

  Do you need to use a clamp at all? If you bias the pull-up current sources
at 2.X times the tail current, assuming n-ch input pair with p-ch loads, then
cascodes should not turn off during transients. As long as the pull-up current
is greater than the drain current of the input pair, the cascodes stay biased.
There is a trade-off of static power for dynamic performance and you will
need to be careful when optimizing, you will need to account for the effect
of device mismatch.

                                                                          Best Regards,

                                                                             Sheldon