Home > Community > Forums > Custom IC Design > Transient Noise Analysis | Noise File


* Required Fields

Recipients email * (separate multiple addresses with commas)

Your name *

Your email *

Message *

Contact Us

* Required Fields
First Name *

Last Name *

Email *

Company / Institution *

Comments: *

 Transient Noise Analysis | Noise File 

Last post Mon, Jan 6 2014 1:37 AM by Andrew Beckett. 3 replies.
Started by pitter 18 Nov 2009 01:36 AM. Topic has 3 replies and 8134 views
Page 1 of 1 (4 items)
Sort Posts:
  • Wed, Nov 18 2009 1:36 AM

    • pitter
    • Top 500 Contributor
    • Joined on Tue, Nov 3 2009
    • Warsaw, Poland
    • Posts 29
    • Points 485
    Transient Noise Analysis | Noise File Reply

    I am going to measure jitter of a VCO by means of Transient Noise Analysis. Simulation setup is not sophisticated, but I have some general questions though.

    I am using models of transistors that have been provided by PDK and they contain noise model too? In model setup dialog box (ADE) I can choose among several model sections. There are sections cmoswo (worst case one condition) and cmoswz (worst case zero condiction). What do they mean? Does the choice of the model section affects transient noise analysis results?  So, in general when simulating noise of a circuit, and considering the circuit only (without 'external noise sources') I do not need to add any other noise sources, for they are included in the componet model?

    The schematic contains a voltage source (vdc from analogLib), and in its properties there is 'noise file name' field. As far as I know, it contains frequency-magnitude pairs. What are the rules the file should be written? So, if I understand it correctly, by means of a noise file I can model noise of the voltage source that will be used to power the real circuit (after production). Can I measure the noise spectrum of a power supply device and describe it in the noise file to model the device noise in transient noise simulation?

    Quite many questions this time :), many thanks for your help in advance,

     Kind Regards,

    • Post Points: 20
  • Thu, Jan 7 2010 1:08 AM

    Re: Transient Noise Analysis | Noise File Reply

    Hi Pitter,

    Just realised that neither I nor anyone else had responded to this. So some answers:

    1. Often device models in a PDK will have noise modeled, but there's no guarantee. It's much more common place nowadays, but in the past foundries often left off the noise parameters if the models were primarily for digital simulation. You should consult the documentation. Doing a noise analysis (small signal noise analysis) in spectre, and then looking at the various noise contributions of the transistors should also reveal if (say) flicker noise has been modeled.
    2. The section names are specific to whatever PDK you're using. I don't know what they mean (I can guess, but it's likely to be in the documentation, or maybe comments at the top of the file).
    3. Since they will alter the bias of the transistors and the behaviour of the transistors, not only will the circuit behaviour change, but most likely the noise (things like shot noise is dependent on the current flowing through the device).
    4. If you're concerned with the noise arising from the circuit, normally you would not need to add any additional noise sources.
    5. The noise file name field on a voltage source allows you to represent an external noise source - e.g. from a block which the voltage source is representing (like a supply, say). It expects noise in V^2/Hz versus frequency pairs. This approach is fine, except you need to take a little care because the file is linearly interpolated (if my memory is right) - so make sure there are enough points in the file - and also you need to take care if you're representing something like a VCO, because that has noise correlated in frequency (i.e. the noise at one frequency may be correlated with another due to up/down conversion) and if that correlation is not taken into account you can end up double counting noise. See the info on cyclostationary noise here.
    6. Also, if measuring the jitter of a VCO, there are better ways to do it. Using pss/pnoise (particularly in "jitter" mode) is one way to do this - it's much more efficient than using transient noise. See the SpectreRF documentation for more information.



    • Post Points: 20
  • Mon, Jan 6 2014 1:29 AM

    Re: Transient Noise Analysis | Noise File Reply

    do you know how could i add a stochastic noise,such gauss noise, to vsource.

    thanks you very mcuh!


    • Post Points: 20
  • Mon, Jan 6 2014 1:37 AM

    Re: Transient Noise Analysis | Noise File Reply

    First of all, please read the forum guidelines (the first post in each forum) which tell you a) not to post on the end of old threads and b) not to post the same question multiple times. 

    If you want something other than a piecewise noise input, you could use a port instead of vsource. This has a built-in resistor and hence is a white noise generator. Other than that, you can use it like a vsource (except the amplitude can be specified in dBm).



    • Post Points: 5
Page 1 of 1 (4 items)
Sort Posts:
Started by pitter at 18 Nov 2009 01:36 AM. Topic has 3 replies.