How to do IMD simulation in LTspice!!!

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
"oversampling" in this context means more complete periods of your test signal - for my 1 ms repeat (from using integer times 1 kHz sines) this means Blackmann starts to look good for 5x full periods == 5 ms record for the fft

if you do a lower frequency IMD then you need 5x periods of your lowest frequency for good resolution of the spectral floor between close peaks

a tilted spectral noise floor is an indication of insufficient settling time of any ~DC transients
 
I hadn't done any IMD sims before using the SMPTE method instead of the CCIF.

I did this in the past few days and got good results that included a "legible" result from the .four function. It looks legit, as opposed to that ludicrous result from .four when trying this with the CCIF method.

Why would that be?

The method is a bit different, mostly because it's using a different set of frequencies.
Namely 60hz and 7k instead of the 19-20k or 18.5-19.5k or whatever.
And instead of the half and half ratio of the 19-20k, the ratio is 1:4 for smpte.

Still, it's 2 separate frequencies that are mixed up, causing IMD at some level, and the .four function can work right with smpte and gives bad results with ccif. Why?

To get the clean .four results with smpte, I used a notch filter on the output, to filter out the 60hz from the 7k, so all that's left is the distorted stuff.
I tried various ways with such filtering with the ccif method and .four never gave anything that looks legit.

If it works with smpte, there may be a way to make it work as well with ccif. But how?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.