OTA noise simulation in Cadence

[borodenkov]

Hi all!

I've got a folded-cascode OTA with SC CMFB and wanted to look at its output noise. The clock sampling frequency is 64MHz, input signal bandwidth - 2MHz. I used PSS (beat freq 64MHz, 0 output harmonics) and PNOISE (1Hz-32MHz, maxsideband=30, no input noise source) for it.

The results that I am getting are quite suprising:

The output noise is 6mV/sqrt(Hz) at 1Hz, then it quickly goes down. I plotted both x and y axis in log scale. There is a noise pole somewhere around 100kHz.

I would expected that there should be a folding point where the flicker noise is equal to thermal - but I can't see it.

Also 6mV/sqrt(Hz) at 1Hz seems to be too huge!

Has anyone got any ideas about it?

Also if I would like to calculate the total noise - until which frequency is reasonable (signal_bandwidth, signal_bandwidth*2, sampling_freq)?

Regards, Alex

 

[CTT]

Hi,

Isn't it 1/f noise that often generates unbounded noise value at low frequency ?

CTT

 

[intuition]

I also have the unreasonable huge fliker noise at low frequency, while using spectre pnoise analysis. Had the issue been solved so far?? 8O

 

[nxing]

I guess that Spectre just use some sort of formula to calculate this 1/f noise and then get some unreasonable value. But I am not very sure. Can anybody give a clear answer?

 

[sunking]

does the model is right for noise?
please contact the foundry.