Tim Lloyd

I've noticed what appears to me to be unexpected behaviour with [lores~] in Max 7.0.1.

- resonance around the cutoff frequency increases

This gets very extreme with high cutoff frequencies. Even with a resonance of 0 (or close to 0 such as 0.0001), a cutoff frequency of 10kHz essentially results in a resonant peak filter rather than an LPF. Two named images are attached, showing the output of [lores~] with white noise input.

I ask mainly because I've also tried a custom filter external that was created with using the [lores~] example in the SDK as a starting point, and it exhibits a similar issue.

- level in the stopband increases
- resonance around the cutoff frequency increases

lores-breaks-with-high-cutoff-frequency

Hmm, don't remember when/how my new account showed up ... yes, probably password related. Same old(er) Tim, back to asking lots of questions probably :p 2 threads are due to the 'pending approval' message not showing up after the first submission ... thought it didn't work. Seems impossible to delete the other one.

Just to clarify the OP -- the custom external isn't based on the lores~ DSP -- the DSP is different. I didn't write it, and can't post the code since it includes protected IP. That doesn't make it easy for anyone else to help :-/

The similar unexpected behaviour might be coincidental, or perhaps might be related to how arguments/messages are passed in and/or dealt with internally.

well, without seeing the code, you're right - there is no way to answer your question.

but, do you at least can point out what topology is used for your resonant lowpass?

but from pictures it seems that implementation is reaally weird.

just a guess that unit delay stages are mixed badly, and if this is a DF filter it could be even more problems there

but from pictures it seems that implementation is reaally weird. 

just a guess that unit delay stages are mixed badly, and if this is a DF filter it could be even more problems there


Problem solved btw. A sneaky little *= had been optimised out ...

Too bad about the odd behaviour of [lores~] -- definitely seems strange, but probably unsolvable due to legacy.

Yes, lores~ is essentially unchanged since MSP 1.0 back in the days of Max 3.5.8.

[lores~] breaks with high cutoff frequency