PLL Pitch Tracker Results

[gstopp at fibermux.com]

     Hi DIYer's,
     
     Finally I got the chance to test out the pitch tracker experiment that 
     I threw together a couple weeks ago. It works GREAT.
     
     The device consists of a mic pre-amp which drives the external input of 
     a phase-locked loop. The phase-locked loop uses an exponential VCO of 
     the Electronotes MAT-03/CA3080 triangle-square type. The correction 
     voltage from the PLL loop filter is brought out to control an external 
     synthesizer. The mic pre-amp also drives an envelope follower, which 
     provides a gate for the external synthesizer. Since an exponential VCO 
     is used in the PLL, the CV output to the external synthesizer is 
     intended to drive a 1v/octave synthesizer directly.
     
     I connected the thing to a Minimoog, and my wife played her flute into 
     the microphone. The synthesizer tracking was excellent over the entire 
     range of the instrument. Particularly surprising to me was the 
     tracker's ability to accurately track extremely high notes, such as 
     overblown harmonics and very high squeaks. Low end tracking was also 
     very accurate. Of course the volts per octave trimpot on the VCO had 
     to be carefully trimmed for this, but that didn't take long at all.
     
     Now for the bad parts. Well, maybe not so bad, because I would venture 
     to use the device AS IS on a recording because the "anomalies" are 
     rather novel.
     
     The device has a sample and hold for the CV output that is held open 
     by the gate from the envelope follower. When the input signal dies 
     out, the gate goes away, and the CV to the external synthesizer is 
     supposed to be held at the last valid voltage. Well as you can 
     imagine, if the input signal goes away rapidly rather than fading out, 
     the input pitch goes to zero at the same time the gate drops, causing 
     the sample and hold to latch a very low voltage. Upon the next rising 
     edge of the gate, the synthesizer will rise up from a very low 
     frequency to the current pitch. Staccatto playing will cause this 
     rising pitch attack on every note. Reminds me a bit of a CS-80.
     
     Even if the last note was played with a fade out, and the external 
     synth is holding the last note played very nicely, internal to the 
     device the PLL's VCO is sitting at a couple hertz. As soon as tne next 
     note happens, the sample and hold opens up, and the PLL slews up to 
     the new note taking the external synthesizer's pitch with it.
     
     Adding some attack time to the synthesizer's VCA envelope tends to 
     minimize or mask this rising pitch effect, so that the sound is quite 
     nice. It's a good idea to keep the release time to a minimum also, to 
     mask out the dropped pitch CV when the notes stop.
     
     Eliminating the attack pitch anomalies could probably be done by 
     delaying the rising edge of the gate with a one-shot. Eliminating the 
     CV-hold anomalies when a note stops would probably involve adding a 
     delay line to the input signal to the PLL, so that when the gate goes 
     away the PLL is still tracking an input. Note that since the input to 
     the PLL is a square wave (or should be), it can be dalyed with a 
     digital FIFO rather than an analog delay line. Note however that 
     attack tracking will also be affected by this delay time.
     
     Harmonic locking can occur, but in my design I added an input 
     conditioner that consisted of a crude R/C lowpass with adjustable lag, 
     to round off the input waveform before the square-wave comparator. This 
     knob has a great effect on frequency locking depending on the nature of 
     the input signal. I found that it needed to be set differently for 
     voice than for flute, for example. Frankly I did not expect this 
     quick-and-dirty control to be so effective. Although much improvement 
     could be made on the input conditioning circuit, the current design 
     seems to work well enough to provide hours of fun.
     
     Anyway, despite some quirks, this thing does allow you to get some 
     useful, musical synthesizer noises out of acoustic sounds. IMHO it 
     blows the crap out of the F-V converter on the MS-20.
     
     - Gene
     gstopp at fibermux.com