Hello all,
ATM, in Pd, I am struggling with creating a externally synched LFO receiving a clock from an external source. For instance a sawtooth (phasor~) .
For this LFO to complete its cycle, I need to calculate its frequency, therefore I think I need to know the time interval between 2 received ticks. The LFO also needs to start its cycle when receiving a tick( bang) from the emitter.
It also need to be quite steady : its frequency should not change at every tick.
But the emitter of the clock is of course is not perfect and the clock not completely steady. The time interval between 2 received ticks can change of a few ms.,
I tried averaging the value of the time interval with a running average but this is not satisfactory. The system does not respond well if too many points are taking into account.. leading to a desynch with the incoming clock and a slow respond to real change of the external clock (i.e. 120 -> 130 BPM).
I also wish not to change the average value of the frequency of the LFO every time a tick is received is received, but only when a significant change occurs.

Does any one have a solution for this ?

Thank you for your help.

@JMC64 Not sure you can satisfy all your criteria. Either you change when there is a difference or you don't. If there is jitter then you can eliminate it like this (if I have made no mistakes)....... tolerance.pd
But if you try to average then you just reintroduce the jitter.
I think you could just have to accept very slightly wrong tempos.
But maybe you could just accept very slightly wrong tempos (due to the jitter) and calculate a "catchup" or "slowdown" at each real tempo change so that in the longer time frame you are accurate some of the time.
David.

Thank you a lot.. It works very well. My mistake was also to use realtime instead of timer.

Does it help to reset your LFO phase with [phasor~] right inlet on every tick ? (or every time the frequency is due to change like in David's patch ?

@JMC64 said:

I am struggling with creating a externally synched LFO receiving a clock from an external source. For instance a sawtooth (phasor~) .

receiving a tick( bang) from the emitter

But the emitter of the clock is of course is not perfect and the clock not completely steady.

What clock-source is this?
An audio signal? A Midi clock?
Keep in mind that PD's control/message domain, including Midi i/o and most control-objects, (such as [snapshot~] ) have a resolution of 64 samples only.
Is this the cause of jitter here?

With [tabsend~] you can use every sample of a block~ in the control-domain.

I tried averaging the value of the time interval with a running average but this is not satisfactory.

I also wish not to change the average value of the frequency of the LFO every time a tick is received is received, but only when a significant change occurs.

Maybe try the median, instead of a moving average?
Together with a hysteresis detecting greater changes, as @whale-av has suggested.

leading to a desynch with the incoming clock and a slow respond to real change of the external clock (i.e. 120 -> 130 BPM).

Every averaging filter won't be exact by definiton.
And to be "on time" at a change, you have to "look-ahead".

Resetting the phasor~ is not really an option, as the phasor is used to read a cos~ or to be further processed ( for making a triangle or inverted saw tooth). therefore, reset creates discontinuities in the calculated LFO..

The patch is to be run on a Bela platform or Rpi in a Eurorack modular synth. The main clock is provided by a dedicated module called Ornament and Crime.