Stats

  • Locked Locked
  • Replies 2
  • Subscribers 125
  • Views 14520
  • 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

how to save transient data with arbitrary step size?

VahidDR
I am facing this problem with cadence:

I am running a transient noise simulation for a Super-Regenerative Receiver (SRR). Despite knowing the theory behind the SRR it requires to run a VCO centered at around 1 GHz. I've also included noise components up to 50 GHz. It resulted to have a step size as small as 2 ps. Actually the Spectre itself uses around 200 fs (in moderate state) step sizes. But I can tolerate the error due to this amount increment in step size.

on the other hand, I need to run this simulation for at least 1ms (1k bits at 1Mbps) to check for 0.1% BER. it means that there would be more than 500G samples!! and I only have saved 2 differential output nodes' voltages and it has been as large as 130GB of data.

The problem is although the simulation runs and completes without any error, the final results cannot be shown (or even written to a *.dat od*.txt file using ocnPrint command in CIW). The Cadence tends to say the server has gone out of memory (it has 40 GB of RAM).

my question is if there is any way that I can run the simulation with the same fine precision as 2ps steps but when it comes to saving the data, I can down-sample the result before writing it on the hard drive. Because although I need the simulation to be run with fine resolution to show the effect of noise, at the end what I want to look at is the 1Mbps data.

Regards,
Vahid

  • Vahid,

    A better solution would be to use the jitter mode of pnoise. This can give you BER based on a small-signal analysis of noise over a single period. All that running in transient noise does is make it much slower and doesn't really tell you any more (in general). The only time you need to use transient noise is either if the circuit is non-periodic or has a large signal response to the noise.

    If you really do want to run with transient noise, you could use either skipcount (to output every nth timestep) or strobeperiod to output on a regular timestep - these would reduce the amount of timesteps. Not sure what the benefit of of running with such a short timestep would be if you are only going to throw away most of the timesteps...

    Regards,

    Andrew.

  • Dear Andrew,

    Thank you for your kind help. strobe period seems working for my purpose.

     my design, as I briefly tried to explain, is a super-regenerative receiver. It is widely used in ultra-low power designs, though the sensitivity and bit rate are somehow limited.

    The basic of the receiver is based on the transient response of an oscillator. imagine an oscillator that is not biased at its maximum bias current (the current at which its output voltage saturates), but still starts an oscillation. This oscillation is basically due to the noise in the circuit. Now, suppose that somehow we have coupled that oscillator to a CW signal which is exactly at the center frequency of the oscillator. In this case again the oscillation happens, but faster. by measuring this time difference you can say that the oscillation started whether from pure noise or a couple input. This is why I need to simulate my design with such precise time steps (to include high frequency noise), but once it is done, I only need the data at the rate of the input (which is simply an OOK signal) bit rate.

    Having that said, I think jitter mode in pnoise simulation cannot yield accurate results. I would appreciate it if you can imagine of any other possible way to simulate such a system with less simulation time. Now it takes me more than a week to check a BER of 0.1%.

    Regards,

    Vahid

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.