danvales82

I'm trying to realize a FM feedback modulation and I need to have one sample delay of my carrier signal.

If I use the delay~ object in a feedback configuration (see thee attached patch) no sound is produced.

Otherwise, using the tapin~/tapout~ system I can't manage one sample delay because (in my opinion) the minimum appreciated delay time is 1 ms.

I'm trying to realize a FM feedback modulation and I need to have one sample delay of  my carrier signal.
If I use the delay~ object in a feedback configuration (see thee attached patch) no sound is produced.

Are there other solutions?
Thanks to everyone

one-sample-feedback-delay

delay~ specifies delays by samples, not sure if it goes down as low as 1 sample though.

if you have MAX 6 you could do this inside gen~ it allows single sample feedback connections

Unfortunately I'm using MAX 5 and I cannot use the delay~ object in feedback configuration because in that case the sound will stop.

Is there something similar to gen~ in the previous versions of MAX 6?

Unfortunately I'm using MAX 5 and I cannot use the delay~ object in  feedback configuration because in that case the sound will stop.
Is there something similar to gen~ in the previous versions of  MAX 6?

The minimum delay time of tapin~/tapout~ is one signal vector, and you're correct that delay~ does not permit feedback.

The easiest way to write an algorithm requiring one sample of delay is to use gen~, but if you don't have Max 6 then I'd suggest looking at mxj~ (or an external written in C).

doesn't send~ and receive~ create a 1 samp delay by default? or is it the non-msp-flavour send/receive?

doesn't send~ and receive~ create a 1 samp delay by default? or is it the non-msp-flavour send/receive?
anyway, this works:

nope, send~ gives you/needs one vector, which is why it works for loops.

what works, in theorie, is to do a vector brake under a vector size which is as small as one sample.

what could also work for an FM operator is that you delay the feedbackloop for exactly one cycle

of the oscillator (in case the oscillator type allows that)

what could also work for an FM operator is that you delay the feedbackloop for exactly one cycle
of the oscillator (in case the oscillator type allows  that)

That's because there's not a feedback loop in your patch - send~ / receive~ are clever enough only to introduce the delay when a feedback condition is detected. MSP passes audio between objects in vectors, so causality dictates that the minimum possible delay in a MSP feedback loop will be sigvs samples:

Christopher Dobrian is correct that C/mxj~/faust/whatever or a vector size of 1 (e.g. in a poly~) is the way forward in Max 5.

Thanks for the clarification Owen-- it's a good example of the advantages of Max 6 w gen (or more precisely the disadvantages of Max 5). I guess the OP could put my send/receive 'solution' into a poly~ with a vector size of 1 to fix their problem as you've suggested.

I was wondering if someone could help me out with introducing a delay of 1 sample using the precompiled Max 6 object GEN, or rewriting my patch around a sample vector-size >1 --

I am working with a powertrain model that uses Max 6, written around the += or accumulator object -- (F=MA and V=AT, and in this instance, T=1 sample, with the current acceleration in ft/s divided by the sample rate) the resulting values go into the accumulator , and that gives the present velocity. The idea was that the velocity calculated for the next sample (X wheel diameter X final drive X gear ratio = RPM) determines the torque value in the next sample, using a lookup-table (RPM x Torque). Even though there doesn't seem to be a noticeable difference when I try setting the vector size to 1, I'd like to have the feedback-loop be defined around a delay of 1 sample (otherwise, the RPM x Torque graph is not going to be accurate).

One sample feedback delay