[sdiy] pitch tracking

[Scott Gravenhorst]

David Moylan <dave at westphila.net> wrote:
>Another note:  I built a circuit posted at generalguitargadgets.com
>
>http://www.generalguitargadgets.com/index.php?option=content&task=view&id=175&Itemid=200
>
>which is based on 4046 PLLs.  I was surprised how well and how fast it 
>could track.  Yes, there's a little glitch at the start of a note, but 
>you get a long of bang for the buck with the 4046.  

This is what my experience is as well.  In my case, I use a PLL (4046 in fact) to track a synth VCO
output (not after the ADSR, VCF and VCA have affected it) and even this glitch is non-existent.

With a PLL, it really depends on how and where you apply it.  As you say below, guitar is not the
easiest signal to track, multiple zero crossings per cycle don't help and an ouput that varies a
lot is also a problem.  Reading the 4046 specsheet, you'll see that for best performance, a 4046
likes to see a rather constant rail to rail input (seen from the 4046's perspective, which is why I
put mine right on the output of my VCO).  In my particular case, I don't use the resulting CV
generated by the PLL.  I cared only about multiplying the input frequency by 2 and by 3 to get
octaves above and perfect fifths above.  But the performance of the system is quite nice, excellent
in my opinion.  

It is also good to be mindful of the loop filter, mine has adjustable damping which helps to "tune"
it's tracking performance.  Choose the right caps (I chose mine by cut and try) and I suggest a pot
for damping.

>The circuit 
>basically LPFs the input then runs it into a opamp with positive 
>feedback to get a nice square which feeds the PLL (same requirement of 
>ramp/hold pitch trackers).  It also uses the square to gate a sample and 
>hold (not filtered, gates at audio frequency) in the CV loop of the PLL 
>- the idea is as long as there's signal the PLL will track, then hold 
>when the signal stops.  In practice, this doesn't work to well on messy 
>guitar signals; noise or glitchy decays set off the S/H and you lose 
>your note.  Would probably work better with a separate gate circuit to 
>hold when the input drops below a certain level.  The circuit works much 
>better (int terms of note length) with a compressor/sustainer in front 
>set to max sustain.
>
>In general, the circuit sounds pretty dull, very akin to an octave pedal 
>or bad organ, but could be made more interesting in a number of ways.   
>VCF would help a lot.  Or have PLL number 2 not track, but use the CV 
>from 1 plus a manual tuning voltage (would probably need to trim the 
>scale. Often).  That would also free up the XOR in PLL2 which could be 
>used to do psuedo ring mod.  Grant Richter's woggle bug schematic shows 
>how to grab a sawtooth off the timing cap of a 4046 which would let you 
>do PWM and all sorts of fun things.
>
>Dave
>
>
>amokan wrote:
>
>> Thanks. I'll dig up the schem for that and check it out.
>>
>> On 12/18/05, *harrybissell* <harrybissell at prodigynet 
>> <mailto:harrybissell at prodigy.net>> wrote:
>>
>>     Pitch tracking might be one of the most difficult processes to make
>>     work.
>>     There are two types... Tachometer (like the MS-20) that always take a
>>     large
>>     number of pulses to get the correct result... and Ramp/Hold types that
>>     usually
>>     work in one, or two cycles.
>>
>>     The Tachometer circuit is probably the easiest. The MS-20 is a
>>     VERY good
>>
>>     example of how to do this. I would clone the circuit if you are
>>     interested. It should
>>     be easy, excpet for the four-gang potentiometer which will be hard to
>>     find.
>>
>>     Three of the potentiometer stages form a low pass filter on the
>>     input,
>>     the fourth is
>>     a low pass filter on the output (lag).  You set the filter to the
>>     expected input frequency
>>     range. Higher frequencies can use less delay ... low frequencies need
>>     more, much more.
>>
>>     The idea is a pulse is produced for every input cycle...and these
>>     charge
>>     a capacitor.
>>     The more pulses, the higher the voltage.  There will be ripple in the
>>     output, something
>>     that makes driving a VCO and tracking pitch almost impossible.
>>
>>     The ramp/hold types are much quicker, but MUCH harder to design and
>>     build. I made
>>     a board of this type some years ago, which was distributed by EFM. It
>>     was based on a
>>     design by Bob Moog (although he was not the first or last to use the
>>     technique). This
>>     board was intended for use with the Etherwave theremin, and could
>>     give a
>>     V/oct output
>>     that can be used to slave a VCO.  OTOH, the theremin has a continuous
>>     wave output.
>>
>>     Your 'drum' is probably untrackable. The non-harmonic tones in a drum
>>     would almost
>>     guarantee that is IS not pitch to track. You might perceive a
>>     'pitch' to
>>     the drum but it proabably exists only in your mind :^P   A circuit is
>>     unlikely to read that (oh how I wish !!!).  You will probably get
>>     garbled gook out of the converter... or maybe you will read
>>     a voltage proportional to the number of drum events (a drum roll
>>     would
>>     be a higher voltage than a single hit).
>>
>>     Tom Gamble (retired of EFM) proposed a clone of the MS-20 using active
>>     filter (OTA)
>>     stages to replace the four-gang pot.  I don't know if he ever
>>     built it.
>>     The MS-20 design is really very clever, and worth your effort in
>>     trying.
>>
>>     H^) harry
>>
>>     amokan wrote:
>>
>>     > Is there a modern version of the pitch tracking circuit in the
>>     MS20? I
>>     > know the MS20 wasn't all that accurate, but I actually like the
>>     > inaccuracies the most for things like drums and whatnot.
>>     >
>>     > Just wondered if there is a module out there that I'm overlooking.
>>     >
>>     > Thanks.
>>
>>
>
>

---------------------------------------------------------
- Where merit is not rewarded, excellence fades.
- Hydrogen is pointless without solar.
- What good are laws that only lawyers understand?
- The media's credibility should always be questioned.
- The only good terrorist is a dead terrorist.
- Governments do nothing well, save collect taxes.

-- Scott Gravenhorst | LegoManiac / Lego Trains / RIS 1.5
-- Linux Rex         | RedWebMail by RedStarWare
-- FatMan: home1.gte.net/res0658s/fatman/
-- NonFatMan: home1.gte.net/res0658s/electronics/
-- Autodidactic Master of Arcane and Hidden Knowledge.