I'm looking for a way to generate a bang when my LFO signal changes direction. Something in my distant memory leads me to think this could be accomplished using the second derivative of the LFO signal and looking for zero crossings.

I looked on here for "derivative" and found a couple posts about ten years old. I also seem to remember that a high pass filter is a "differentiator" but certainly the corner frequency must get involved in some important way. If I had a delay line with 2 taps coming out I'm pretty sure I could generate something suitable.

Any suggestions?

using [rzero~ 1] with [samphold~ ] might be interesting.
or rather, not samphold~ but i'm not sure what...

i think a lot of discussions about this either here or on the list end up concluding that it is preferable to just rebuild the LFO using vline~, so you know ahead of time when it will change direction. Are you using an LFO that has too complex a pitch modulation to build it with vline~?

as LarsXI hints [rzero~ 1] is a differentiator and a crude highpass filter, just to be clear (its output is the difference between successive samples)

Thanks for the responses.

Although I've messed with pd a bit in the past, I'm just picking it back up now with an actual concept of what I want to do. I did try the rzero object (real zero, FIR filter) and as you can imagine, the sample to sample difference of a 10 Hz sine wave is exceedingly small and when multiplied by 100000 or so, looks like we might be exploring some sort of quantization limit because "theoretically" I would expect that to be smooth, with a sine wave input. However, it is very spiky.

My LFO is built by summing two sine waves of (different) variable frequency and so, actually I don't know ahead of time when they will switch direction but I'll look at that vline object anyway.

I also thought about having something like two daisy-chained 5 msec delays and comparing those outputs, or subsampling the LFO some other way because I don't need audio sample resolution, I think 10-15 ms would be fine for this. I'm sure there's something obvious I haven't stumbled across yet.

Update - I cascaded 2 snapshot~ objects driven from a 40 msec metro object and subtract the outputs from each other - this appears to align zero crossings with the LFO peaks. Now I just have to detect zero crossings and generate a bang when that happens.

@Digital-Larry [edge~] will bang on zero crossings and give direction.
[change~] [xerox~] and [expr~] are also useful.
David.

@whale-av said:

@Digital-Larry [edge~] will bang on zero crossings and give direction.
[change~] [xerox~] and [expr~] are also useful.
David.

@whale-av I am not finding [edge] or [edge~], is that in pd-vanilla or pd-extended?

I have vanilla installed at the moment. Thought I read that extended wasn't being maintained? Is it possible to have both installed at once?

@Digital-Larry It was Pd-extended in the cyclone library, so it is very likely that it is available through Deken,..... and most likely available for 64-bit vanilla as well.....

I have extended installed, and I just run all flavours of vanilla from folders on my desktop.
Actually installing both could cause some confusion...... unsure?
You could download the "standalone" zip of extended for your system here ...http://puredata.info/downloads/pd-extended , and if you are running 32-bit vanilla on windows you should be able to just copy the executable for edge~.... edge~.dll or whatever..... into your patch folder for it to work. I imagine you could do the same for other os's if you can break into the installation file.
Deken would be best though.
David.

Thanks a lot for the recommendation, I got it working and am and onto the next challenge!