[time-nuts] Spice simulation of PSRR and phase noise
kb8tq at n1k.org
Sat Oct 28 11:08:16 EDT 2017
The “fun part” of harmonic balance is making sure you are not off in a corner
case where the results are not as good as they might otherwise be. Maybe not
as much an issue for a VCO as for some other structures.
> On Oct 28, 2017, at 7:36 AM, Rafael Gajanec <rgajanec at vercet.com> wrote:
> Hi Attila,
> On 27-Oct-17 8:25 PM, Attila Kinali wrote:
>> Hi Rafael
>> On Sun, 22 Oct 2017 17:20:52 +0200
>> Rafael Gajanec<rgajanec at vercet.com> wrote:
>>> you haven't specified what sort of circuits would you like to simulate,
>> Simplified, they are differential amplifiers driven into saturation.
>> A bit more detailed, I am looking at ring oscillator stages and sine-to-square
>> conversion circuits and their behaviour regarding various key factors
>> (note: I am not sure what the key factors are, yet)
> Oscillator design - that's what I found HB simulation particularly useful for. It gives you almost instant results, compared to the transient simulation, say 10 seconds instead of 5 hours! Just imagine what it means if you are trying to tune several parameters of an oscillator... The only other reasonably fast and accurate way I can think of is to build the bloody circuit and measure it using some expensive equipment.
>>> but maybe the answer is Harmonic Balance.
>> Hmm.. I didn't know about Harmonic Balance. I have some reading up to do.
>>> HSPICE from Synopsis and ADS from Keysight (which I use) also have the
>>> HB engine.
>> I am mostly using ngpsice, because it's very easy to script (I have a bunch
>> of perl scripts that feed simulations into a Grid Engine cluster, extract
>> data and analyzse it). Is there any big advantage of the commercial spice
>> engines that would make them worth considering? And would the license alow
>> to run hundreds of instances in parallel?
>> (Yes, I am doing crazy things :-)
> Attached are some results of a simple transient simulation using Hspice M 2017.03, BBspice A/D 5.2.3 and ADS 2016.01. It's basically *V1 1 0 SIN 0 1 1Meg *and then *.FOUR 1Meg V(1)* in Hspice, VspecTran in ADS and spectra computed using postprocessor in BBspice and ADS. As you can see, there are some differences... To be fair, possibly there are some simulator-specific settings/methods that could improve the results and you should figure it out yourself what's the way to get the best results from your spice. See http://www.audio-perfection.com/spice-ltspice/distortion-measurements-with-ltspice.html
> Commercial spice engines may have lower computational noise and shorter simulation times. For example my out-dated BBspice (which is commercial too by the way) crashed several times before I got some results, while it used little RAM and only about 10-12% of available processor resources... I intended to get you Pspice results of this simulation as well, but I gave up after half an hour and about 1% of progress.
>> Attila Kinali
> Best regards,
> Rafael Gajanec
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts