[Fixed 9.01] Speaker Imped squeal

[yyz67]

Just installed 9.00 and played with Speaker Impedance on my first preset with a DR. Increasing the value to a certain point starts creating a background squeal artifact (independent of input) that changes with the parameter value. I then tried it on a variety of amps (mostly Fenders) and get a similar but unique squeal for many of them. On the Princeton I get a squeal with a value as low as 1.8 but my high freq hearing isn't great and I think it begins earlier.

Attached is a simple preset with all amps set to default. The Speaker Impedance is set above 1.0 and on my unit 5 out of 8 Fender amp models have a squeal/noise artifact. Varying Negative Feedback and Transformer Matching affects the frequency and intensity of the squeal. Strangely, if I remove one of the amp blocks it gets better overall but I can still find settings that have artifacts.

Also attached recording of preset stepping through each scene (level adjusted +24 dB in DAW to make it very audible). Scenes 1-5 have squeal while 6-8 don't.

Can others confirm?

 

[yyz67]

Also happening on other amp models like AC-20 and D-30 (preset attached), although sometimes instead of a background squeal it can be a hum and also lead to a random farty tone with guitar input (scene 2).

 

[phil92]

Ok, so here goes the squeal, but where’s the whapoosh? (sorry, had to)

 

[FractalAudio]

Fixed for next release. However you shouldn't turn Negative Feedback all the way up and also turn Speaker Impedance way up. This causes instability in the power amp just as it would in a real amp.

In you second example you took amps with no negative feedback and turned the negative feedback to maximum. This completely changes the character of those amps. The whole reason an AC-20, DC-30, etc. sound the way they do is because they have no negative feedback.

There may always be a combination of negative feedback, transformer match and speaker impedance that causes instability just as would happen in a real amp. The Axe-Fx lets you play with things but it can't always protect you from doing things that break the amp just as if you did these same things in the real amp. If you put an 8K transformer in a Plexi, change the negative feedback to the 16-ohm tap and connect a 16-ohm speaker to the 4-ohm output it's going to oscillate. So will the model. I've tried to prevent that the best I could but with great power comes great responsibility.

 

[yyz67]

Fixed for next release.

Thanks!

... with great power comes great responsibility.

Sure does.

There may always be a combination of negative feedback, transformer match and speaker impedance that causes instability just as would happen in a real amp. The Axe-Fx lets you play with things but it can't always protect you from doing things that break the amp just as if you did these same things in the real amp. If you put an 8K transformer in a Plexi, change the negative feedback to the 16-ohm tap and connect a 16-ohm speaker to the 4-ohm output it's going to oscillate. So will the model.

Cool, yeah as a curious AFX'er but also a previous software developer/tester, I check out the limits of parameters to see what happens sonically and also from a software robustness standpoint. The release notes didn't mention possibility of instability/oscillation, but given it's a CLR network (with possibly other nonlinear factors) that makes sense for some combinations. I wanted to make sure.

Is there a way that the firmware can determine mathematically getting into unstable/oscillating regions (e.g. via Routh-Hurwitz, Nyquist)? Or is it a matter of just what the virtual circuit does which has to be assessed real-time? If either of these is detectable generally for any amp configuration, it would be awesome to alert users of the stability of settings, say with a little colored indicator (green/yellow/red) next to the mini-tuner. Of course this is probably not trivial!

 

[yyz67]

Both cases above CONFIRMED FIXED in 9.01!