just put 2 Vsource with different frequency sine in series for the input waveform, otherwise same considerations as any .tran for a fft plot - do adjust the amplitudes so the sum fits the supplies, gain of the amp
a trick is that the 2 frequencies should have a "nice" common multiple - use n, m kHz and then some exact few milli-second tran/fft length 19 + 20 kHz 1:1 amplitude is popular, the n, m are often selected to be relatively prime to keep their harmonics from overlapping with, confusing the IMD sum and difference products
since I also like Blackman window in the fft the length should be > 4x the period you want to resolve to separate the adjacent frequencies which the Blackman window spreads to 4 bins - typically I use 5 to 10 ms
a trick is that the 2 frequencies should have a "nice" common multiple - use n, m kHz and then some exact few milli-second tran/fft length 19 + 20 kHz 1:1 amplitude is popular, the n, m are often selected to be relatively prime to keep their harmonics from overlapping with, confusing the IMD sum and difference products
since I also like Blackman window in the fft the length should be > 4x the period you want to resolve to separate the adjacent frequencies which the Blackman window spreads to 4 bins - typically I use 5 to 10 ms
Last edited:
SMTE low/high 4:1 doesn't reveal as much in sim as in the real world
some of the things it probes - long time constants associated with device heating aren't simulated in default Spice models, often power supplies aren't modeled all the way to the AC mains for full ripple/load current interaction
it can show some bias modulations though
I like high loop gain feedback so I assume there's enough loop gain to clean up anything low, I really want to probe mid to high audio frequency performance where the loop gain is rolling off
some of the things it probes - long time constants associated with device heating aren't simulated in default Spice models, often power supplies aren't modeled all the way to the AC mains for full ripple/load current interaction
it can show some bias modulations though
I like high loop gain feedback so I assume there's enough loop gain to clean up anything low, I really want to probe mid to high audio frequency performance where the loop gain is rolling off
- Status
- Not open for further replies.