I want to simulate the Noise figure for such a setup that first a band pass filter, then sample and hold, and then follower by amplifier, and ADC and so on.
Can i get the noise figure from source to each stage with Pnoise?
I understand that if the output is specified with voltage of two node but NOT specified by a port or probe, it won't exclude the noise from next stages in the NF calculation. So eventually, this is not the noise figure exactly upto such stage.
But how could I get the correct noise figure for each stage. And what i want to see is that the NF can be validated with just accumulate from one stage to the following. like NF04=NF01+NF12+NF23+NF34, where NF04 means NF from source to 4th stage.
Thanks a lot.
I don't know if this is an unusally question. But it should be common to ask what's the NF from stage two to stage three, isn't it?
The spectreRF manual suggest not use noise summary for Pnoise. Calculate by hand with result browse is not recommended as well. But there should be a simple way to get the NF from a non-port input to certain output node i guess.
Does any body have the same issue?
Or someone tell me this is not possible within cadence, so that i could move on.
Thanks in advance.
In reply to zcam:
If you have an urgent question, you should go to customer support. This forum is a community forum, and is answered by people (including the Cadence folks, like me) in their spare time. Personally I've been rather busy the last week and have had very little time to respond here.
Anyway, you'd probably need to isolate each stage from the next. That could be done with a voltage controlled voltage source (say) - this will allow you to measure the noise at the output of each stage (prior to the voltage controlled source) whilst still allowing the noise and signal to be transferred to the next stage. Of course, the problem with this is that you have an ideal buffer in place - and hence the intermediate outputs would not be seeing the load of the subsequent stage.
In reply to Andrew Beckett:
Thank for the quick reply.
You're right! add VCVS can isolate each stage. I will try that.
But as you said, load noise will be isolated as well.
In almost ideal unilateral case, this is not an issue. But not so accurate otherwise.
That promote me another related question:Can we turn the noise of part of the circuit off?
I know most modern device models have such feature by specifying "isnoisy=no". However, what if i want to turn off the noise of a group of devices (e.g. in a subckt). Do you have any suggestions?
In Eldo, it is supported! I guess spectre has the similar feature, isn't it?
I hope this could help more people than myself. :-)
Thanks a lot
On Simulation->Options->Analog you can turn off the Noise Contribution of particular blocks (on the "Main" tab, at the bottom).
I realized that the NF is not accurate with load noise turned off. Since the baseband noise is also neglected which shouldn't be.
Is there a way to specify where the load is using the similar way as specify a probe? Equivalently, can some subckt be treat as a whole probe?