[sdiy] Pitch detector Envelope follower!

[Harry Bissell]

LOL well that says it all !.

Analog pitch tracking absolutely works 100% for amplitude stable regular waveforms (such as sine, triangle, square,
ramp with 2nd harmonic less than the fundamental.  Hey, no problem :^)

It also helps a lot if the input frequency either does not mode or cannot move quickly. Or if the input frequency is
very high so that any ripple will be outside the human hearing range.

Tachometer circuits suffer from long acquisition times and high ripple. You can maybe beat the ripple problem
by sampling the CV out at some regular point in the waveform (like the peak value) then  filtering that result.

Even the ramp/hold P/V converters have trouble with ripple, consecutive wave conversions can differ in voltage
making a stairstep effect. Most designs dealt with this by comparing the current and past cycles and rejecting the data if they don't match within some arbitrary amount, or by selectively filtering a small step more than a large
one.

The ripple / response time tradeoff is the worst issue...

which is handled quite nicely in the MS-20 (the queen of tachometer circuits). They use three sections of a four-gang pot to make a low pass filter on the input. The more you filter the input, the slower the response but the more accurate. Less filtering is faster but less precise. The fourth section is the ripple filter, you guessed it... the
faster the thing tracks, the less ripple is rejected.

So you have the choice (imho), a smooth CV too slow to track anything, or a fast CV with too much ripple to use.

(A little)Ripple might be tolerable driving a VCA but WILL be very noticeable if you drive a VCO (the most common
use) or a VCF (the 2nd most common). Ripple is an AC modulation which generates IM distortion.

Of course if your customer makes bug music, this might be a feature, not a bug...   :^)

I think that Tom Gamble (EFM) was on the right track in trying an MS-20 clone substituting a VCF for the
lowpass filter and the lag circuit. What I heard then of his experiment (audio) convinced me his design was
not yet working, but it should have.  I'd suggest the MS-20 is the closest to a working 'simple' circuit
that has ever been done.

And although it might be dsPICable, the digital option might actually be a good idea....


H^) harry




----- Original Message -----
From: KD KD <pic24hj at gmail.com>
To: synth-diy at dropmix.xs4all.nl
Sent: Wed, 02 Nov 2011 03:04:58 -0400 (EDT)
Subject: Re: [sdiy] Pitch detector Envelope follower!

2011/11/1, Harry Bissell <harrybissell at wowway.com>:

> I'm not sure I understand. Do you need to have Pitch tracking or envelope
> tracking ?

Both!

> Envelope tracking can be quite simple if the input signal is (somewhat)
> predictable.

Yes indeed it is.

> Generating a 'filter sweep' can be done with a peak detector very easily.
Just what i did with the diode+cap+ resistor. ;)

>This works well for  percussive sounds (syndrum dooooo....).  The trouble is detecting
> a sudden  drop of the input  signal, if you need to track that.
dooooo? here we say piiuuuuu.

>Often a filter sweep completing with no  input can reveal  a lot of unacceptable noise..

That's if resonance are set high and decay curve very long.

> If pitch detection ~is~ needed, how fast does it need to respond ?

No idea really, but there are trade-offs to make as we know.

>Do you need a slow response  to control (let's say) the general range of filtering,
>or  are you trying  make a VCO follow a pitch input ?

> I don't know why, but it seems in these posts that no one wants to say (in
> any detail) what their  application will be.  I'd be happy to discuss the benefits
>and shortfalls of  the various methods either on-list (preferred) or in private.

Well we have discussed pitch trackers in the past several times and come to the
conclusion there are many bad ways to do it since the whole bussiness are tricky
in analog domain. But here is the catch:

I'm not trying anything besides trying to convince a customer who's
stubborn enough
believing that analogue pitch trackers actually works 100% do not , so
better to fool the
dude with some simple stuff that works a bit and convince him that
that's how well it can
be done for the amount of money he says he's willing to pay me for my
time spent on a
hopeless endeavour. :)  I assume he's gonna use it for anything else
but accurate pitch tracking!

LM2907 with some filtering would that do? If people are happy with
MS20 pitch tracker
what else can be done similarly but with less amount of components?
Maybe i just take dsPIC and some soft and fool the dude its all
analog? :)

Reg
KD
_______________________________________________
Synth-diy mailing list
Synth-diy at dropmix.xs4all.nl
http://dropmix.xs4all.nl/mailman/listinfo/synth-diy

-- 
Harry Bissell & Nora Abdullah 4eva