Pitch variation measurement

[zamproject]

Hi all

I want to evaluate pitch/tune variation of an oscillator over time.
Right now I have two options in mind but not that practical.
First is 30min audio recording on my DAW, then pitch analysis of the audio file, which draw some curve on a track layer, but difficult to extract that curve for graphical purpose nor to know the exact pitch value in Hz
Second, hooking my fluke 8060a (which have a decent freq measurement I suppose), siting in front for 30 min and manually collect value, let say each 30sec. It's probably accurate enough for my purpose and easy for .xls/graph processing, but so boring especially if I have to do it more than once

Any other ideas ?

Cheers
Zam

 

[JohnRoberts]

You could count zero crossings then divide by time to measure average pitch... for specific deviation data you might want to also collect min and max interval between zero crossings.

JR

 

[zamproject]

Hi JR
Thanks

I want to evaluate the time it take to the oscillator (analog syth) to stabilise within decent deviation, I'm not sure there is that much zero crossing before average, heat and temp stability (main factor?) is a curve, not an oscillation of pitch around target freq ?
So I suppose with the tools I get, a manual collection is unavoidable, either freq meter measurement, or tuner plugin with cents reading.

 

[rogs]

You could always stick a video camera in front of the Fluke meter and record 30 minutes of the dislay.
Download the video into a free video editor - like Virtualddub 2 - and then scrub up and down the timeline to see what happened.
You might find it easier to convert the video into an intraframe format using a codec like Cineform (already free in Virtualdub 2) to make the timeline 'scrubbing'. quicker and easier to read.

You say it's for an analogue synth... the amount of time for reasonable stability can vary a lot, depending on what type of synth?
The original Minimoog for example was notoriously unstable... and it depended a lot on the stability of the ambient temperature.

Later models were more reliable -- although they often relied on a chip inside being run deliberately 'hot' , and that could still take several minutes to settle down.
Even more recently some oscillators have become digitally controlled - even though they still claim to still be 'analogue' ..... Those are usually much more stable....

 

[zamproject]

tks @rogs that's an excellent idea ! look simple and probably enough accurate and I'm not sure I even need to edit the video.
I can just import the vid in my DAW and set the timeline/sequence in min:sec:frame.

The syth is way older than a minimoog, it's my all tube 50' Ondioline.

Cheers

 

[Bo Deadly]

A scientific way to do this might be to use code and a USB audio interface to generate a control signal specifically crafted to match the oscillator output as closely as possible. Then feed the oscillator and control signal back into two inputs of the USB audio interface. Now use code to determine the phase difference while simultaneously mixing the two signals such that they they cancel out. Use the phase difference to adjust the control signal frequency that maximizes the cancellation thereby determining the frequency of the oscillator. Record whatever software variables to characterize the variation of the oscillator frequency over time. Using this technique, it should be possible to detect even small variations in frequency over long or short periods. This is not completely like doing a discrete fourier analysis for a particular frequency but using feedback to adjust that frequency over time as necessary. It's also similar to the concept of a PLL but using software seems more appropriate for extracting the data.

 

[zamproject]

tks Bo, I think I understand what you suggest, but probably total overkill for my need

I just what to have a coars idea of the decent time the whole syth is ready to play or for adjust/fine tuning
For now I have doubt in the first 30min after power on, but maybe everything is fine after 10.
I think I'll go with rogs idea to check that.

Cheers
Zam

 

[Bo Deadly]

There's probably a synth module that uses a crystal oscillator and counter ICs to make an oscillator without any variation at all. That could be mixed with your oscillator to drive an LED and get a visual representation of how close the oscillators are. When the LED stops "beating", the synth is in tune.

 

[zamproject]

In my book I call this a tuner ?!?

I have some, hard and software that for sure I use to tune the synth.

By design the synth is somehow relative and need frequent corrections (there is a master tune pot) especially when you shift the octave register and RC network around oscillator tube. The more I play on it the more I think about it as wood string fretless instrument...

I just need to know when the instrument is -stable- (within inherent design tolerances)