[sdiy] VCO Tuning goals
Jerry Gray-Eskue
jerryge at cableone.net
Fri Oct 2 15:26:27 CEST 2009
David,
A few observations, that you may already be aware of,
<<Indeed, the HF current is only about 1 nA at 1400 Hz, while the
total CV current is about 5.5 uA. However, at 3000 Hz, the HF current is 70
nA and the total CV current is 15.6 uA. Simulations also suggest that the
HF correction should be virtually perfect with this setup. >>
Keep in mind that when you build the physical circuit your integrator cap
and threshold(s) will not be precisely the value in your simulation. This is
key to the I to F curves. You may have the perfect ratio in your simulation
but both currents will have to be adjusted to fit the actual cap and
threshold values and (the tricky part) preserve the ratio.
<<calculate the
necessary CV correction current, draw an I-V curve, and then set the zero-CV
frequency such that this curve coincides with a diode's I-V response as
closely as possible. >>
This involves yet another design decision, you have to pick a Temperature at
which you are matching the curve as the diode is yet another temperature
sensitive component.
- Jerry
-----Original Message-----
From: synth-diy-bounces at dropmix.xs4all.nl
[mailto:synth-diy-bounces at dropmix.xs4all.nl]On Behalf Of David G. Dixon
Sent: Friday, October 02, 2009 12:50 AM
To: 'Dave Leith'; 'David Ingebretsen'
Cc: synth-diy at dropmix.xs4all.nl
Subject: RE: [sdiy] VCO Tuning goals
Dave, I'm finally building my version of Sowa-Hoshuyama '2164-expo VCO this
weekend, and I've arranged it so that a CV of zero volts corresponds to a
frequency of 1000 Hz. I've also designed in a high-frequency trim circuit
which takes the CV summer output, inverts it and sends it back to the CV
summing node through a 1M resistor and a diode. An 18-turn trimmer on the
feedback loop of the inverter serves as the HF trim.
My simulations suggest that the setting of this trimpot should have no
effect on the frequency below 1 kHz, since the diode prevents any current
from reaching the CV summing junction unless the CV is positive. This
should, of course, allow tuning below 1000 Hz independent of the HF trim
setting. Indeed, the HF current is only about 1 nA at 1400 Hz, while the
total CV current is about 5.5 uA. However, at 3000 Hz, the HF current is 70
nA and the total CV current is 15.6 uA. Simulations also suggest that the
HF correction should be virtually perfect with this setup.
This leads me to think that perhaps some VCOs set their zero-CV frequencies
too low, and are therefore being over-corrected over the mid-frequency
range, where they may not require any compensation at all. Now, granted,
the '2164-based circuit should require less HF correction than a VCO with an
expo based on transistors, because the only source of error should be the
integrator reset, and not the transistor bulk resistance. But even so, the
general principle should still apply.
Perhaps I'm deluding myself here (it has been known to happen!) but it seems
to me that one should be able to predict (through simulations) just how
"off" a saw oscillator will be with increasing frequency, calculate the
necessary CV correction current, draw an I-V curve, and then set the zero-CV
frequency such that this curve coincides with a diode's I-V response as
closely as possible. Then, scaling the two curves to the same magnitude
should determine the feedback resistor value. Is this how the gurus do it,
or is it all just one big thumb-suck?
> ..........
>
> Should add if the low range corresponds to low notes one can let this
> section be a bit less accurate if it is more important to get the mid
> frequencies in tune.
>
> This depends on what if any the octave switching circuits are.
>
> Say the critical zone (middle of piano keyboard ) is 2 to 5 volts, the
> one could accept a bit more detuning between 0 to 2 volts. and the hi
> track should pull in 5 to 7 volts.
>
> hope this helps!
>
> Dave
>
> On 10/1/09, Dave Leith <dave.leith at gmail.com> wrote:
> > Hi David
> >
> > It's hard to be general here but way back in the 70's I worked as a
> > tech and tuned a lot of analog synths. If there are no instructions I
> > would usually tune the 0 - 3 or 4 CV volt range with the scale trim
> > then tune the higher octave(s) with the hi track. I always first
> > earball the initial octave settings rather than use a tuner as it will
> > get things in the right range faster (see note below about it getting
> > worse with each adjustment)
> >
> > Usually pin bottom note then play octaves getting higher and higher
> > until the upper notes are too flat to be pulled in reasonably with the
> > scale. I would not re-tune the VCO when returning to the pinned note
> > but re-adjust the tuning device. This is usually just easier and
> > faster. Then use the hi-track. Finally reset the basic VCO frequency
> > which will have shifted during the tuning process. This shouldn't
> > affect the scale or hi-track.
> >
> > Sometimes due to circuit design one needs to pin a hi note then use
> > the scale to track a lower note. This would apply if each time you
> > make a tracking adjustment the tuning gets worse rather then better.
> > (I.e. an ARP pins lower and I believe the Mini Moog is reverse)
> >
> > For a tuning device I use a Yamaha PT4 which is a very accurate strobe
> > tuner made for tuning acoustic pianos.
> >
> >
> > Dave
> >
> >
> > On 10/1/09, David Ingebretsen <dingebre at 3dphysics.net> wrote:
> > > Ian,
> > >
> > > Thanks for the suggestion. I know there are many designs out there
> and have
> > > read the information on your site and looked at your designs as
> well, very
> > > nice.
> > >
> > > Do you have any suggestions for my question though? I have several
> different
> > > oscillators, each with different features and behavior. Some better
> > > tracking, some less which is fine. I don't necessarily always want
> dead
> > > accurate tracking. My question was really meant to be more to the
> general
> > > "best" way to approach tuning to get the most musically useful VCO.
> > >
> > > Thanks
> > >
> > >
> > > David
> > >
> > > David M. Ingebretsen M.S., M.E.
> > > Collision Forensics & Engineering, Inc.
> > > 2469 East Fort Union Blvd. STE 114
> > > Salt Lake City, UT 84121
> > > www.CFandE.com
> > >
> > > 801 733-5458 Office
> > > 801 842-5451 Cell
> > >
> > > dingebre at CFandE.com
> > > dingebre at 3dphysics.net
> > >
> > >
> > >
> > >
> > >
> > > ~~ -----Original Message-----
> > > ~~ From: Ian Fritz [mailto:ijfritz at comcast.net]
> > > ~~ Sent: Thursday, October 01, 2009 4:02 PM
> > > ~~ To: David Ingebretsen; 'SDIY diy'
> > > ~~ Subject: Re: [sdiy] VCO Tuning goals
> > > ~~
> > > ~~ If you used an accurate VCO to start with you wouldn't need to
> ask these
> > >
> > > ~~ questions. I have several designs that track to better than 0.1%
> up to
> > > ~~ 30kHz - 40kHz. The largest error is always in the top octave.
> These
> > > ~~ designs are freely available.
> > > ~~
> > > ~~ Ian
> > > ~~
> > > ~~
> > > ~~ At 01:07 PM 10/1/2009, David Ingebretsen wrote:
> > > ~~ >In brief, I just finished building a trio of MOTM 300 ultra
> VCO's into a
> > > ~~ >Frac Panel
> > > ~~ >
> > > ~~ >http://www.xmission.com/~dingebre/page8.html
> > > ~~ >
> > > ~~ >As I have been tuning them, I have a question on the concept of
> tuning a
> > > ~~ >VCO.
> > > ~~ >
> > > ~~ >I understand all too well the scale will always have errors, and
> my
> > > ~~ >understanding is that an error of +/- 0.2% in the frequency is
> likely
> > > not
> > > ~~ >detectable, BUT... as one plays, it seems most successive notes
> are
> > > within
> > > ~~ >the same octave and often within a few notes. The question then
> is:
> > > ~~ >
> > > ~~ >Is it more important to have better accuracy octave to adjacent
> octave
> > > or
> > > ~~ >between 4, 5, or more octaves? That is, when I tune the
> oscillators,
> > > should
> > > ~~ >I be more worried about a 0.2% difference between C0 and C1 or
> C2 and C3
> > > ~~ >than between C0 and C5? I hope this makes sense.
> > > ~~ >
> > > ~~ >I've been making a chart, and have been playing with Paul
> Haneberg's
> > > high
> > > ~~ >frequency trim modification
> > > ~~ >
> > > ~~ >http://www.wiseguysynth.com/larry/mods/VCO_track.htm
> > > ~~ >
> > > ~~ >or
> > > ~~ >
> > > ~~ >http://dragonflyalley.com/images/MOTM300/VCO_track.pdf
> > > ~~ >
> > > ~~ >(but I am using a multi-turn trimmer in series with a fixed 1M
> resistor
> > > ~~ >instead of swapping resistors).
> > > ~~ >
> > > ~~ >I've noticed there is some interesting interaction between the
> scale and
> > > ~~ >high frequency trim and I just want to make sure I'm approaching
> the
> > > tuning
> > > ~~ >in the best musical sense. I'm curious musically/theoretically
> if it is
> > > ~~ >better to worry about the error between several octaves and let
> the note
> > > to
> > > ~~ >note difference be what it is, or work on minimizing the "octave
> to
> > > adjacent
> > > ~~ >octave"/"note to note" error. Reducing the high frequency error
> between,
> > > say
> > > ~~ >C0 and C5, increases the inter-octave error between C1 and C2,
> or C3 and
> > > C4
> > > ~~ >somewhat.
> > > ~~ >
> > > ~~ >The accuracy of the 300 makes some of this argument moot, but I
> am
> > > curious
> > > ~~ >about the whole zen of VCO tuning.
> > > ~~ >
> > > ~~ >David
> > > ~~ >
> > > ~~ >David M. Ingebretsen M.S., M.E.
> > > ~~ >Collision Forensics & Engineering, Inc.
> > > ~~ >2469 East Fort Union Blvd. STE 114
> > > ~~ >Salt Lake City, UT 84121
> > > ~~ >www.CFandE.com
> > > ~~ >
> > > ~~ >801 733-5458 Office
> > > ~~ >801 842-5451 Cell
> > > ~~ >
> > > ~~ >dingebre at CFandE.com
> > > ~~ >dingebre at 3dphysics.net
> > > ~~ >
> > > ~~ >
> > > ~~ >_______________________________________________
> > > ~~ >Synth-diy mailing list
> > > ~~ >Synth-diy at dropmix.xs4all.nl
> > > ~~ >http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
> > >
> > >
> > > _______________________________________________
> > > Synth-diy mailing list
> > > Synth-diy at dropmix.xs4all.nl
> > > http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
> > >
> >
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
_______________________________________________
Synth-diy mailing list
Synth-diy at dropmix.xs4all.nl
http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
More information about the Synth-diy
mailing list