Fast analog sw. as VC resistor

Mark Smart smart at
Fri Jan 3 18:56:07 CET 1997

> Hi Mark.
> I still don't understand why you need such complicated a circuit.
> Anyhow, switching resistors are an interesting topic. Look at this
> It's based on a switched capacitor and is not more difficult than yours.
> V1 -------------o\  o---------------V2
>                   |
>                   |
>                 -----
>                 -----   C
>                   |
>                   |
>                  GND
> Bye.
> --
> +-------------------+----------------------------------+
> | Paolo Predonzani  |  email: predo at |
> +-------------------+----------------------------------+

That's a pretty good idea. I saw a couple of references to the switched
capacitor style filter in my Electronotes. The reason it didn't seem to fit
what I want to do is that I was looking for a way to control a LOT of
voltage-controlled three-terminal pots with only one oscillator. To
generate a complementary resistance (the other half of the pot) with the
switched-capacitor filter would require another oscillator whose frequency
was proportional to 1/F of the first oscillator, right? There would have to
be two voltage-controlled oscillators and 1/x circuitry for every function 
to be adjusted.

With the circuit I drew, one master sawtooth or triangle oscillator can
control all the pots in the system. A separate comparator would be used for
each funtion to be adjusted, generating one pwm'ed clock to control each
set of three-terminal pots.

Message 1/63  From Tim Cockram                         Jan 3, 97 08:09:24
am PST
Date: Fri,  3 Jan 97 08:09:24 PST
Subject: Re: Fast analog sw. as VC resistor
To: Tim Cockram <tim at>, Mark Smart 
<smart at>
Cc: synth-diy at
Sender: owner-synth-diy at
Precedence: bulk

>>I realized last night that it would be better to use an analog switch in
>>PARALLEL with a fixed resistor, rather than without a fixed resistor like
>>in my previous post. If you use an anlog mux like in the 4053 and two
>>resitors, you can make a three-terminal pot very easily:
>>      |      R          |        R           |
>>      |                 |                    |
>>      +---------------o/  o------------------+
>>                    1/4 4053
>>This way, you can vary the resistance to any fraction of R, and just use
>>one PWM clock to control a pot.

>The pot law could get interesting?

It's almost totally linear as long as R is a lot bigger than the on-resistance 
of the analog switch.  I tested this circuit out last night with a couple of
100K resistors and a 4066, which has Ron of about 130 ohms. It worked great once
I got a good pwm'ed clock for it.  The envelope generator circuits and filter
resonance which I need this for are all either 1M or 100K pots, so it should
work fine for them.  The basic idea is that the overall resistance is the
time average of the resistances in the two states.  Like if it spends 20% of
the time being on and 80% being off, then the resistance is 0.2(130//100K) +
0.8(100K).  The first term is small compared to the second term, so the 
resistance is close to R*(1-(fraction of time switch is on)), and the 
resistance of the other half of the pot is close to R*(fraction of time switch
is on).  If this still wasn't linear enough, two or more analog muxes could be
hooked up in parallel to reduce the on-resistance still further.

Experimentally, I tried measuring the resistance with a Fluke meter, and
making a voltage divider with the vc pot.  The Fluke meter is so slow that it
reads exactly what it should.  I was able to get the voltage divider output to
look like a pot by sticking a 0.1 uF cap between the 4053 output and ground.
I am pretty psyched about this circuit.

The function generator wouldn't do pwm properly, so I had to make a comparator
circuit also.  Two things I learned from doing this:

1. Never use a 1458 op-amp for this application. Its slow slew rate makes it
unusable as a comparator above about 10 Khz.  I had to settle for this to test
my circuit, because it was the only op-amp handy which worked.

2. Never use an LM3900 for a comparator.  I tried this: for some reason its
input did not like having a voltage divider pot hooked to it with no feedback
resistor.  I think I blew it up! Anyone know the reason for this?

I'm going to get some LM339's and try them.

>Thats right the good old Jupiter 4. Roland use an OTA for resonance and 
>separate oscillators to drive the ADSR.  I think I might have the circuit some
>where, I'll check at the weekend.

I would greatly appreciate that!

>If You can get copies of the Maplin synth kit schematics the VCFs used PWM
>techn ology although I think the sweep range was limited.

I have not heard of this.  Where could I get that from?

Thanks for all the suggestions and information, dudes!!

*     Mark Smart                               *
*     Network Technician                       *
*     University Communications Inc. (UCI)     *
*     smart at                      *

More information about the Synth-diy mailing list