guitar synth?

[Hairy Harry]

Hi Scott: (et al...)

I think the PLL solution for guitar is a big problem. In your
"fatman" case, you can count on the PLL being locked to an
input frequency that is always present, so the loop is already
in lock whan you start. And the waveform is very "clean" so the
phase comparator does not have to deal with a flakey waveform.

I'd say that the guitar would HAVE to be divided by 2 in order
to guarantee a consistant wave. It would be possible to have a
short (100uS) pulse at the fundamental frequency using the method
I'm using. Can you lock direct to a pulse (I think for phase comparator 1, 
you might be able to do this...)

You would have to try to S/H the last valid input frequency to keep
yourself in the ballpark with the PLL, or the loop filter has to acquire 
from zero (or some minimum... it could be set to respond to
a two octave range (per string) only.

Adding the dividers would slow things down also. I think I can make
a 2X wave from a square wave, with a lag time of a single cycle of
the fundamental. Once that cycle is passed, slow changes in the frequency 
would cause duty cycle errors... so no pure waveform would
be available.

I've checked out the "frequency shifter" idea (Bode, JH, Electronotes 
etc...) and even then the "dome filter" will have quite a response lag.

How slow is too slow... many think that the 1-2 cycle delay of most
guitar synths is "too slow" as that could be 12-24mS minimum. More than that 
likely.  I'm not dissing the PLL I just think it will be in the "too slow" 
category also...

I agree that I would rather use 6 PLLs than 6 Analog Pitch Shifters
(or even digital shifters).  The harmonic intervals are useful for sure, 
even though you will have beating between the perfect fifth of
the 3/2 ratio, and the equally tempered fifth of the guitar (a little 
flatter...) This would only show up in polyphonic play (which most
guitarists insist on...)

The fatman may also have a higher input frequency than guitar, the
lowest note is 80Hz or so. The lower the note the worse everything gets !!!

Hey its a Holy Grail. I'd even BUY one if tere was one that really
worked....

PS: do you have any samples (.wav or .mp3) of your toy in action.
I'd like to hear it. For the keyboard application it might work
very well.


>From: Scott Gravenhorst <chordman at earthlink.net>
>Reply-To: chordman at earthlink.net
>To: synth-diy at node12b53.a2000.nl
>Subject: Re: guitar synth?
>Date: Wed, 19 Jul 2000 08:06:47
>
>harrybissell at prodigy.net wrote:
>
> >Guitar to square wave is really hard to do. The harmonics of the strings
> >keep the square wave from ever being 50%.  Thats why most
> >guitar synths have a divide by 2 circuit included. Only way (I) know
> >to guarantee a square wave.
> >
> >If you get a square wave (at 1/2f) then how to multiply it... Plase 
>locked
> >loop ?  That's slow...
>
>Y'know, I keep seeing this statement and I would agree that it's
>response to pitch can never be instantaneous, however, with an
>adjustable loop filter, it can be *very* fast.  I discovered
>this when I built my PLL freq mult for my FatMan.  People always
>say it's slow, but how slow is too slow to be useful?  Whenever
>the input pitch changes, the PLL will lag behind (or overshoot
>depending on how the loop filter is designed and adjusted)
>causing some interesting phase stuff when mixed with the dry
>signal from the pickup.  Much depends on the interval
>over which the pitch changes as well.  For 2 octave jumps, I can
>adjust my filter so that the PLL is close enough that this effect
>is simply part of the attack sound.  For smaller intervals,
>the effect can be adjusted almost completely away (from a human
>perception standpoint.  You'd still see it on a scope.)
>
>A PLL freq mult is an extremely simple solution.  My main point
>here is that of usefulness.  It can be, in fact, quite useful,
>and given the level of usefulness, it's a darn good bargain
>both in price and in effort.  Imagine building an analogue
>perfect octave pitch shifter FOR EACH STRING as opposed
>to six 4046 ICs and some 4027s...  Peanuts in comparison.
>
>And let's not forget the fact that bumping the pitch by a
>perfect octave is not the only possibility.  Counters other
>than divide by 2 can produce musically valuable intervals
>as well.  Divide by 3 (I have a schemo for anyone interested)
>makes perfect 5ths!  Very useful.
>
>I just want to make sure that the PLL gets a fair shake and
>reasonable consideration.  After all, do these things have
>to be technically/purist perfect?  What in nature is?  I mean
>to say that if you are considering making a guitar synth, you
>obviously don't want it to sound like a guitar...
>
>-- Scott Gravenhorst : On The Edge, but the Edge of What?
>-- Linux Rex, Linux Vobiscum  |  RedWebMail by RedStarWare
>-- FatMan: www.teklab.com/~chordman
>-- NonFatMan: members.xoom.com/_XMCM/chordman/index.html
>-- The 21st century does NOT start in the year 2000!!!
>
>

________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com