Which one is the correct one?

[AMSA84]

Hi guys,

As I say in the topic, which of these two is the real NOR gate? Basic question but I have seen two different cofigurations:



Regards.

 

[FvM]

I don't see a difference except for flipped position of two MOSFETs which is meaningless.

 

[nitishn5]

Both are 'real' NOR gates, the difference comes when there are two types of inputs.

For example, lets say that you have two inputs,
1. A fast switching signal(like a clock),
2. A slow switching signal(like some control).

Both these go to the PMOS which are in series.

Having the clock input at the top PMOS would increase the delay to the output.
So it is better to have the control signal on top since it would be switching less.

 

[FvM]

Having the clock input at the top PMOS would increase the delay to the output.
So it is better to have the control signal on top since it would be switching less.

How much difference in propagation delay and rise tiime do you expect? I wonder if it's significant in relation to process variations?

 

[nitishn5]

It is just a 'good practice', I have not actually calculated out the difference...
And apparently it adds less jitter/noise on the clock than when it is connected the other way around.

But we do it for things like Muxes, Gates etc which use a control signal to control a Fast Clock.

That is the only difference between the two that I can see.

 

[AMSA84]

I see. Thanks for the reply guys.

 

[erikl]

Having the clock input at the top PMOS would increase the delay to the output.
So it is better to have the control signal on top since it would be switching less.

How much difference in propagation delay and rise tiime do you expect? I wonder if it's significant in relation to process variations?

Here are prop. delays from the Oct-2001 Artisan Std.Lib. for several NOR2 gates:

XL .. X4 are NOR2 gates with different fanout.

A is the input to the upper PMOS. As you can see, the A input's intrinsic delay - for the same output change - is slightly shorter than the B input's, whereas their load-dependent delays are practically the same.