KPA aliasing experiment

The frequency graph wont show you much, you need to look at the spectrograph against frequency (I posted some in the other thread). All you can tell from those images is that there is distortion, a distorted waveform will contain many more frequencies than the input waveform, if it didn't then there would be no distortion it would be a perfect facsimile of the input and would sound identical, and as they show that then that's about all you can see.

What have been termed aliasing artifacts here and elsewhere are in fact compression artifacts or really limiting artifacts, as the waveform is compressed it goes from a sine to a square wave, thus gaining many more harmonics, adding a knee to compression or reducing compression attack would solve it if it were a simple compression algorithm (and upsampling can help a little bit), but it's an amp sim and doing that would probably damage some of the desirable properties and realism of the simulation, with a real amp the cab helps mask and cut the annoying higher frequencies. I've no idea if it's a limitation or if there's a solution, Christoph's the only one who knows what's possible with the KPA though.

A frequency graph is more than adequate. If you put in a single
frequency and you get out any frequencies that aren't multiples of that
frequency, or especially frequencies below that frequency, then you are aliasing.
From those graphs, it appears that the unit is aliasing heavily.

tayholliday - Here were some http://kemper-amps.com/forum/i…ad&postID=30423#post30423 the fillaments in there are clearer to see (and their relationship to the root notes), this is with a clean non distorted signal so it's easier to see the problem, you get straight harmonics of various orders and then interference harmonics/ and mirrored patterns which really sound terrible.

Quote from bigmuff

A frequency graph is more than adequate. If you put in a single
frequency and you get out any frequencies that aren't multiples of that
frequency, or especially frequencies below that frequency, then you are aliasing.
From those graphs, it appears that the unit is aliasing heavily.

That's not true in this case and far too rigid a definition. The OP is talking about high gain amps, and distortion results in many different frequencies (as well as harmonics) around the root note which are quite desirable even if they're not harmonics. It's true that amps and compression emphasise harmonics, e.g. a Class A producing produce desirable harmonics e.g. the 5th etc, but at the same time strong harmonics at lower orders can be undesirable and often mistaken when listening for aliasing artifacts (sometimes termed "ghost notes").

The trouble when looking at a static snapshot is that aliasing artifacts can also result in emphasis in harmonics, they are far easier to see when looking at a spectrograph with a time and frequency shift where you can clearly see the relationship between frequency pattern and root note, especially with weaker patterns in the middle of distortion noise. It can't so easily be masked.

You need to use your ears first then check the graph. There's a lot of noise in a good amp that you don't want to get rid of. Amps only sound alive if they're imperfect, first make sure you can hear the artifacts and that what you're hearing isn't an inherent part of the sound and distortion, after that find them on the graph.

Aliasing artifacts show up as more clear interference in a clear pattern that relates to frequency, while distortion shows up more as white noise (frequency all over the shop) or smearing of the notes in the frequency domain. If you raise the sampling rate during generation then the aliasing interference pattern should (depending on how clipped the waves have become) break down and just get absorbed into the general noise floor. High frequency noise as always being less disturbing within pattern than low frequency.

The KPA suffers from aliasing/compression artifacts on certain rigs only, the last firmware update was meant to fix the issue with stomps, but it makes me wonder if it was a part of the amps profiled in the cases where it's still audible (I've played tube amps that suffer similar ghost notes), or whether it was just the profiling process at the time, or if it's a more general problem in the KPA that just certain cabs mask better. I'm hoping that whatever the reason the next firmware update will lay the issue to bed once and for all though.