On Exponentials and 2164's

Christopher List Christopher_List at sonymusic.com
Wed Apr 24 16:14:53 CEST 1996

Hi DIYer's,

You may recall my asking about log amps and stuff a week or so ago. Well it 
turns out that I built first and thought later (luckily on the breadboard)! I 
realized that I didn't want my voltage to rise logorithmicly, I wanted it to 
rise exponentially! This is because the SSM2164 is ATTENUATING the volume with 
+ control voltages, and this attenuation is exponential! This means that the 
volume drops by 1/10th of it's current value for each .033 mV or so. At +.2V it 
is already at 50% of it's volume (what do you call this? "inverse 
exponential"?) I had a good time (from an educational point of view) building a 
log amp, and I got it to work and stuff - but I never even bothered using it 
with the 2164, because I started thinking about the problem (how novel!) and d
rawing graphs in Excel and stuff and I got that "hit on the head with a hammer" 

Anyway, the topic at hand. I've still got a little problem with it, and I've 
finally decided to cheese-out and go back to you guys for some advice. Here's 
the story;

The whole concept behind the "serge-style mixer" is the gain knob. From 0-50% 
it is used to attenuate the gain CV. From 50% - 100% it adds a voltage to any 
gain CV that might be present - this way you use only one knob, instead of two 
(one of "initial volume and one for CV amount) and you can use the circuit as a 
regular mixer w/o a control voltage if you like. This gives less flexibility 
than the two knob approach, but front panel space is always a premium and I've 
got the extra VCA on the chip, so I might as well use it.

What I'm done is run the gain pot between +/-15v and I have two diodes coming 
off the wiper. The one that lets the negative through goes to an inverting 
op-amp (that reduces it to a 0 to -3v range) and from there to the control 
input on one of the 2164 VCA's. The gain CV input from the front panel goes 
into this VCA, then comes out and gets summed with the positive diode output 
from the gain knob and this summed CV signal controls the audio gain. Nice and 
neat, and it works, but not as I'd like.

The problem is that the control of this gain CV needs to be linear, because the 
CV itself will have an exponential effect on the audio level. To get this 
linear action, the voltage out of the gain knob (the negative side only) need 
to decrease exponentially. Optimally, when the knob is at 0% (-15v) the output 
of the INVERTER should be 3v, at 25% it should be .2v, at 50% it should be 0 
volts. The rest of the pot's action (ignored by the inverter due to the diode) 
should be linear.

Following a tip from Bernie in electronotes, I used a small input resistor into 
the inverter to <<simulate>> an exponential output from a linear pot. He's got 
charts and stuff in the PCC that show these functions and I was able to figure 
out the math and plot these same charts in Excel. Unfortunately, when the pot 
is around 0v, there is a slight "linear bump" before the curve twists and 
starts following a more exponential path. - In other words I still wind up 
losing 50% of the amplitude of my gain CV in a VERY short throw of the pot. I 
was trying to figure out ways of level shifting the pot's output so that it 
would ignore this linear section, but haven't been too successful. Anybody got 
any other ideas?

I should probably just go with separate pots for initial level and gain amount, 
but you know how it is once you get that bug for something...

Sorry for all the verbiage on this tiny question :). I hope it makes sense.

- Chris

More information about the Synth-diy mailing list