# Pitch to CV, and related.. TECHNICAL!

Posted: 4/15/2009 10:49:09 AM

From: Eastleigh, Hampshire, U.K. ................................... Fred Mundell. ................................... Electronics Engineer. (Primarily Analogue) .. CV Synths 1974-1980 .. Theremin developer 2007 to present .. soon to be Developing / Trading as WaveCrafter.com . ...................................

Joined: 12/7/2007

.

Posted: 4/15/2009 5:40:52 PM

From: Croxley Green, Hertfordshire, UK

Joined: 10/5/2005

So let's see if I've got this...

The higher a note, the quicker we can determine its pitch, which is good. So let's build a hypersonic BFO (Beat Frequency Oscillator), and then once we have its pitch as a CV we can reduce the voltage by a fixed amount to put it into the audible range. (Probably not actually a separate operation, but integrated into to the Pitch to CV conversion.)

Let's conjoin this with a regular BFO (to produce an audible tone), so they share a common variable RF oscillator, and their fixed RF oscillators oscillate hand-in-hand, albeit the frequency of one a fixed amount higher than the other. This fixed amount is related to the amount by which the CV is reduced, so that the CV corresponds directly to the pitch of the audible tone.

Questions.

Am I right in presuming that the linearity of the two BFOs must be the same because they are joined at the hip, so to speak?

I recall discussing the idea of a Voltage Controlled BFO previously. Is there a reason to prefer the conjoined arrangement over a CV only theremin (utilising a hypersonic BFO) controlling an audible VCBFO?

In the conjoined arrangement I understand it is preferable to have similar RF oscillators for the audible BFO as this avoids getting a musically uninteresting sine wave output, but for the hypersonic BFO I presume that a sine wave is preferable to a harmonically complex output for the purpose of conversion to CV. So would its fixed RF oscillator be of a different type to the variable oscillator, or would that make it harder to keep its frequency at a fixed distance above other fixed oscillator's frequency?

Would it be un-PC to call this a Siamese Theremin?
Posted: 4/16/2009 12:06:20 AM

From: Eastleigh, Hampshire, U.K. ................................... Fred Mundell. ................................... Electronics Engineer. (Primarily Analogue) .. CV Synths 1974-1980 .. Theremin developer 2007 to present .. soon to be Developing / Trading as WaveCrafter.com . ...................................

Joined: 12/7/2007

.

Posted: 4/16/2009 1:52:29 AM

From: Croxley Green, Hertfordshire, UK

Joined: 10/5/2005

[i]"by which the CV is reduced" - I dont really understand what you are saying here[/i]

It's OK, I do. You phrased it as "removing the offset" --- Say we are generating the audible tone A4 (440 Hz) - In volt/octave CV, 4 volts equates to A4 - if the hypersonic tone generated is 20 octaves above that it gives a CV of 24 volts, so we need to reduce the CV by 20 volts to get the CV corresponding to the audible note.

(NB. Of course, when I wrote "hypersonic" I meant "ultrasonic.")

Thank you for the clarification. As ever your ideas are fascinating.
Posted: 4/16/2009 3:50:31 AM

From: Eastleigh, Hampshire, U.K. ................................... Fred Mundell. ................................... Electronics Engineer. (Primarily Analogue) .. CV Synths 1974-1980 .. Theremin developer 2007 to present .. soon to be Developing / Trading as WaveCrafter.com . ...................................

Joined: 12/7/2007

.

Posted: 4/16/2009 5:48:11 PM

From: Escondido, CA

Joined: 2/6/2008

Pardon my ignorance here, but it would seem that if all you are after is the most accurate pitch to CV possible, it would seem that you would want to just digitize the direct RF output of the pitch oscillator and calculate the desired pitch vs CV value.

The other stuff (ultrasonic oscillators, multiple LC filters, etc) would seem to be the most beneficial to generate the analog signals to be demodulated and fed to the audio output to offer a variety of timbres.

It would seem that you could alter the pitch-to-CV algorithm to whatever you like to control whatever it is you want to control from it (external synthesizers, for example) without messing with the timbre and dynamics for your "live feed" ... sort of like the recording feed and house feed of a venue's PA system might be setup as separate paths with different intents.

Don
Posted: 4/16/2009 8:01:22 PM

From: Colmar, France

Joined: 12/31/2007

Dear friends, I'm sorry but you forgot one thing: In order to measure the difference of 1Hz (which is much larger than 1ct in the bass register) it makes no difference if this Hz is between 32 and 33Hz or between 285032 and 285033Hz, you'll need a gate time of 1 second. So all this up- and down mixing will not solve the problem. But there are other methods... ;-)

@Fred: In the time when I learned programming microprocessors, we always said that floating point operations were only for girls. A "real" developer would use techniques as bit-shifting and subsequent additions in order to calculate a logarithm without run-time problems... ;-)
Posted: 4/16/2009 10:22:40 PM

From: Eastleigh, Hampshire, U.K. ................................... Fred Mundell. ................................... Electronics Engineer. (Primarily Analogue) .. CV Synths 1974-1980 .. Theremin developer 2007 to present .. soon to be Developing / Trading as WaveCrafter.com . ...................................

Joined: 12/7/2007

.

Posted: 4/17/2009 12:15:57 AM

From: Eastleigh, Hampshire, U.K. ................................... Fred Mundell. ................................... Electronics Engineer. (Primarily Analogue) .. CV Synths 1974-1980 .. Theremin developer 2007 to present .. soon to be Developing / Trading as WaveCrafter.com . ...................................

Joined: 12/7/2007

.

Posted: 4/17/2009 2:21:13 AM

From: Colmar, France

Joined: 12/31/2007

@Fred: The last year task was not an exponential question, but doing an y=1/(ax+b) conversion, which would not have been a problem. The problem was that a and b were almost not known, because the values given by design could vary (due to the usual environment conditions) by more than +/- 50% which made the real task - auto-calibration - impossible because of the extreme gradient of the curve.

With your calculations above you seem to be right, as long as you do a linear f to v conversion. But you risk to loose this precision when doing the logarithmic conversion:

Let f_top b = 3750Hz
f_top - 1ct = 3747.83454053 Hz
t_top = 266.667us
t_top -1ct = 266.821us
delta_t=154ns => f_tick_min = ca. 6.5MHz

Let f_top b = 303750Hz
f_top - 1ct = 303747.83454053 Hz
t_top = 3.292181us
t_top -1ct = 3.292204us
delta_t = 23ps => f_tick_min = ca. 43.5GHz

The latter being rather difficult to implement...

You must be logged in to post a reply. Please log in or register for a new account.