Stats

  • Locked Locked
  • Replies 7
  • Subscribers 127
  • Views 14285
  • Members are here 0
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

tdnoise overestimation ?

analogy

So I have a basic inverter circuit and the input signal is 1 GHz clock.

I ran PSS + PNoise analysis (shooting engine) for time domain noise.Being a clock signal I should have used full spectrum but we use a rather old version of Spectre (7.0.1.179.isr16).

So I plot my integrated noise in frequency range from 50 MHz to 7GHz.But I noticed that integrating over this bandwidth the noise voltage was in order of Volts !

And so I plotted noise spectrum (Pnoise>tdnoise) and found that I have a huge spike (spur) at 5 GHz.For reference :

Integrating noise from 50MHz to 1050MHz yielded noise of ~5.1 mV

Integrating noise from 50MHz to 7GHz yielded noise of ~170 V

What might be the problem ?

Parents

  • With tdnoise (and PMjitter which uses tdnoise underneath) there is an ideal sampler at the output of the circuit which takes care of strobing the noise at the PSSfund frequency - this is strobing the noise at the particular instant during the period. Because of this ideal sampler, all noise gets aliased into the bandwidth from DC up to PSSfund/2 - so you should never sweep the frequency beyond half the PSS fundamental (so not beyond 500MHz in your case). Otherwise you end up double counting the noise. Also, if you happen to go very close to an exact multiple of the fundamental, you'll end up with an "infinite" (it's not really infinite, but very large) noise contribution because of the up conversion of the 1/f (flicker) noise.

    So change your sweep to be 50MHz->500MHz and you will capture all the noise - the sampler is doing that.

    Regards,

    Andrew.

Reply

  • With tdnoise (and PMjitter which uses tdnoise underneath) there is an ideal sampler at the output of the circuit which takes care of strobing the noise at the PSSfund frequency - this is strobing the noise at the particular instant during the period. Because of this ideal sampler, all noise gets aliased into the bandwidth from DC up to PSSfund/2 - so you should never sweep the frequency beyond half the PSS fundamental (so not beyond 500MHz in your case). Otherwise you end up double counting the noise. Also, if you happen to go very close to an exact multiple of the fundamental, you'll end up with an "infinite" (it's not really infinite, but very large) noise contribution because of the up conversion of the 1/f (flicker) noise.

    So change your sweep to be 50MHz->500MHz and you will capture all the noise - the sampler is doing that.

    Regards,

    Andrew.

Children
No Data

Community Guidelines

The Cadence Design Communities support Cadence users and technologists interacting to exchange ideas, news, technical information, and best practices to solve problems and get the most from Cadence technology. The community is open to everyone, and to provide the most value, we require participants to follow our Community Guidelines that facilitate a quality exchange of ideas and information. By accessing, contributing, using or downloading any materials from the site, you agree to be bound by the full Community Guidelines.