[sdiy] LFO VC Skew?

David G Dixon dixon at mail.ubc.ca
Wed Feb 19 09:03:12 CET 2020

Well, nevermind... 

I solved the problem of the 40dB.  All I have to do is bias the 2164 VC
voltage up by 660 mV (-20dB) or 1320 mV (-40dB) and decrease the amplifier
input resistor by a factor of 10 or 100.  If I want to use the preferred
30k, then I just have to adjust the LFO integrator capacitor accordingly to
keep the same frequency range, and the 2164 is operating in its happy place.

This is gonna work the first time, I predict.

Of course I could've just used THAT 2162s, which have 60dB of amplification
available, but I don't have any and they're only available in SMD, which I
don't use.  Plus, I've never worked out how to linearize those, because I've
never had to.  Most of my best ideas have come from just limiting myself to
the parts I have on hand and making them work.

-----Original Message-----
From: Synth-diy [mailto:synth-diy-bounces at synth-diy.org] On Behalf Of David
G Dixon
Sent: Tuesday, February 18, 2020 3:53 PM
To: 'Jason Proctor'
Cc: 'SDIY List'
Subject: Re: [sdiy] LFO VC Skew?

Well, the circuit is pretty simple.  However, there is one major snag, which
I will only discover upon trying it...

The maximum stated (amplification) gain of a 2164 VCA is 22dB.  My circuit
requires 40dB of gain.  Even at that gain, the VCA only has to process about
133uA of current, so it should be within the realm of possibility if the
gain limitation is really a limitation on output current (which I was led to
believe was closer to 300uA) and not on actual multiplication factor.  If
the gain is truly limited to 22dB, then the saw/ramp flybacks are going to
be fairly slowish -- possibly still useful, but...  If I can actually
squeeze 40dB out of the beast, then we're in business.  Either way, I'll
have a test circuit finished before I go to bed.

If it works well, then yes, I will give Danjel van Tijn first dibs on taking
the design for Intellijel (which he has already suggested he may want to do)
and will not be sharing the tricks with y'all (sorry -- my loyalties are
clear).  I won't discourage you from figuring it out yourselves though,
based on the hints I've given.

Of course, if the 2164 only gives me 20dB and I need 40dB, it's tempting to
just string two 2164s together sharing the same VC voltage.  However, this
won't work.  It will increase the frequency towards the middle of the morph
(where the triangle lives).  Because I'm working with logarithmic math, the
arrangement is pretty darn unforgiving.

-----Original Message-----
From: Jason Proctor [mailto:jason at redfish.net] 
Sent: Tuesday, February 18, 2020 1:48 PM
To: David G Dixon
Cc: Tom Wiltshire; SDIY List
Subject: Re: [sdiy] LFO VC Skew?

Will this mean a fancy new Intellijel LFO? :-)

fwiw, always liked the waveshaping on the venerable MOTM-320 LFO. Verrr

On Tue, Feb 18, 2020 at 12:54 PM David G Dixon <dixon at mail.ubc.ca> wrote:
> Ugh, gawd...   Mine's gonna rock.  I'm gonna put the finishing touches on
the design in about 15 minutes, once the fentanyl wears off just a little
bit more (although it didn't prevent me from practicing a Haydn sonata just
> ________________________________
> From: Tom Wiltshire [mailto:tom at electricdruid.net]
> Sent: Tuesday, February 18, 2020 4:24 AM
> To: David G Dixon
> Cc: ackolonges fds; SDIY List
> Subject: Re: [sdiy] LFO VC Skew?
> On 18 Feb 2020, at 02:37, David G Dixon <dixon at mail.ubc.ca> wrote:
> It's easy to get saw-to-tri-to-ramp with VCAs, but the trick is to
maintain the same frequency when you do so.  That's the impossible part.
> Many of the same problems you have when doing it analoguely (that's the
equivalent of "digitally", right?!) turn up in the digital version.
> I used phase distortion to produce the modified waveforms. Essentially,
you have a ramp core (an NCO) and you apply a wave shaper. To get the phase
distortion, you run the first half of the NCO's cycle at a different
frequency to the second half. To avoid frequency wobble when altering the
shape, the two sets of frequency increments have give the same overall
frequency as one steady increment. This means they needs to be pretty
accurate (how accurate depends on audio versus LFO and how much you care).
> This accuracy requirement also gets worse as you get to the extreme ends
(because you finish up with a division that approaches division by zero).
Using integer math, you reach a point where the results overflow the
available accuracy. For this reason, I limited the Distort CV on my chips to
4% to 96% duty cycle. This means you *can't* get a really vertical saw
waveform edge if you start by distorting a triangle wave.
> It's funny how the difficulties remain the same, even using a completely
different method and technology.
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy

Synth-diy mailing list
Synth-diy at synth-diy.org

More information about the Synth-diy mailing list