phaser glitches out when reaching over 189-512 Hz in modulation rate

[K0rrid0r]

Describe the bug
This is a very old problem existing already in surge 1.5.2 but actually still exist in the latest nighly. When the phaser reaches over 189-512 Hz in modulation rate it completly glitches out and creates all sorts of infact "dangerous for the speakers" problems, i dont actually know the terminology for it even, but it sounds dangerous and i remember doing it on a big sound system once and scared the shit out of everyone (this was when i found this bug in a club environment during a gig)

The only way to solve it is to take the phaser modulation rate all the way down to 0 or down to the safe area between 0-150 Hz and re-trig surge via restart of the DAW with stop and play. If this is not done the phaser stays in this glitch mode so to say.

It would be nice to be able to go all the way up even if its extreme in nature, but that is the essence of surge i think, that it should be able to do such things, I mean the LFOs goes up in audio range without problems.

Please let us know your surge version
This bug exist as a very old problem all the way form 1.5.2 to latest nightly and also 1.6.6 both in VST2 and VST3
Surge-NIGHTLY-2020-05-09-9569d16

• Bits (32/64)

To Reproduce
Hold a note and just drag up the phaser and check what happends (use a limiter) with the modulation rate and the sound when it comes between somewhere at 189-512 Hz

Expected behavior
To not glitch out between 152-512 Hz

Desktop:

• OS: [e.g. windows XP, Win 7, Win 8.1, Win 10]

Additional context
It should be a milestone to fix this before release of 1.7 because users who are not aware of this might end up actually getting really scared or crash a project or be close to destroy the speakers because the glitch introduces low end frequencies that shouldnt be there and also loud peaks etc.

If it cant be fixed those high frequencies should not be there imo.

[baconpaul]

Link with #1706

[baconpaul]

So I just spent some time on this

basically yeah the APFs aren't stable under a modulation that extreme. Depending on the depth and the input signal the just explode and don't go back until after the ringdown and phase reset

and with modulation on modulation you can easily push well outside the operating range

I think the solution is to just clamp the rate to 150hz max or so

I will do that but if there's a better plan around I'm all ears.

[baconpaul]

Oh ha no that's not it
Well that is it
but it's for a different reason
it's the same bug as FM2/3 and many other spots
I have a fix

[baconpaul]

So here was the problem basically

rate = (command to get the rate)
lfophase += rate
if( lfophase > 1 )
   lfophase -= 1
lfovalue = triangle(lfophase)
omega = base + depth * lfovalue

where triangle is the typical 1 - |2 - 4 * phase| thing

this all happens at samplerate / BLOCKSIZE, BLOCKSIZE is 32, so the control nyquist frequency is 48k / 32 / 2 = 750hz

so you see the problem right?

if rate gets really big then that lfophase can go above 1, sure, but also above 2 once we are at and around that nyquist range

and if goes above 2 the lfovalue is no longer a [-1,1] triangle because that definition assumes in bound [0,1]

and that drives the apf unstable

but the fix was to make lfophase -=1 become lfophase = fmod( lfophase, 1 )

and then the phaser is stable all the way up to 512hz

pushing now.