Stats

  • Locked Locked
  • Replies 1
  • Subscribers 64
  • Views 1644
  • 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

Transimpedance Amplifier - Broadband Circuit - Cadence Noise Summary

growingmind

Hello,

I am simulating a broadband circuit, a transimpedance amplifier that runs to about 30 GHz.  SO I am using integrated noise under noise and selecting all the components from 1KHz to whatever frequency I want to integrate the noise too.  I then want to rank the top three contributors.

I am attempting to run a noise summary in Cadence, by doing a noise simulation in AC.

I find that when I run a logarthmic simulation under the noise simulation, ie.  I select logarthmic and pick 200 steps per decade, or 200 steps as the number of steps, I get a difference result under the noise summary, with different noise contributors (ie.  different top 3 noise contributors) versus when I select a linear noise analysis and pick say 100MHz step size or 200 steps for the number of steps.  This is with the same frequency ranges - ie.  1KHz to 30GHz.

Why is this ?  And what is the difference between a logarithmic and linear noise simulation and when would you use each in Cadence ?

For example, in some cases, mostly with less steps and less frequency resolutions in step size, I see the 1/f noise dominating.  With larger steps and less frequency resolutions in step size, I see the drain currents dominating - which is what I expect from this circuit.

Thanks.

  • By "in Cadence" I assume you mean using Spectre in the Analog Design Environment - there is no tool called Cadence (that's the name of the company, not the software). If you're using a different simulator or a different environment (you could have been talking about HSPICE in ADE, or PSPICE in the PCB environment) then please say.

    Whether you do a log or a linear sweep makes no difference to the simulation of a specific frequency; all spectre is doing is simulating the noise at a number of different frequency points and computing the output referred noise plus the output referred noise contribution of each noise source. The sweep simply controls what frequency points are picked for the simulation.

    If you are integrating the noise over a particular bandwidth, how accurate that integration is will be dependent upon the smoothness of the noise response in the frequency band you have chosen - which will be dependent upon the spacing and number of points you've simulated; the integration probably just uses the trapezoidal rule to integrate (I didn't check). This integration is done for both the output noise and each noise contribution, so it is quite conceivable that you will get different rankings in each case if there aren't enough points in the range being integrated.

    Without more detail (ideally seeing the data - so you might want to go to customer support) I can't give a more precise explanation for the difference.

    Regards,

    Andrew.

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.