[sdiy] LM13700 crossfade query

David G. Dixon dixon at interchange.ubc.ca
Wed Jun 15 15:42:42 CEST 2011

> I've got a question on summing together two LM13700 stages -- I can't 
> clarify if I'm being clever or dumb!!
> (I know, I should probably just test it out...)
> Here's a little sketch: http://bugbrand.co.uk/images/LM13700_query.jpg
> (just showing the skeleton..)
> The idea is to have two VCAs with inverse control signals so as one 
> comes up, the other goes down. (I've done this part before 
> and it works 
> well) but I'm stumbling on whether I can seriously simplify 
> the summing 
> stages at the end.
> I'm using external TL07X buffer stages rather than the 
> transistor pairs 
> on the LM13700 -- typical simple VCA solution is achieved as 
> drawn but 
> leaving off the 2nd LM13700 stage (ie output connection from 
> pin 12 does 
> not join the net from LM13700 pin5 to TL07X pin2)
> What I'm wondering is whether the summing sketch with the 
> outputs summed 
> as drawn is a good way or a total idiotic no-no. In which 
> case I'd have 
> to do two TL07X buffer stages (one for each half of the lm13700) and 
> then have another summing stage.
> I'm sure there was a project that did similar but I can't 
> track anything 
> down now. The nearest I've found is the Crumar BIT1 which 
> sums the pins 
> of multiple LM13700 outputs together and then buffers in a 
> similar way 
> to my drawing but with one slight difference - there's now an input 
> resistor before the TL07X (ie. after summing the LM13700 outputs).
> Thanks for any pointers here!

You have effectively described the OTA-based version of my Multiplying
Modulator, except that the two signal feeds were the inverse of each other.
This would work well as a crossfader, and in fact, we are using the
linearized 2164 VCA-based version in a couple of Intellijel modules.

My only advice would be to bias the CC+ and CC- signals, since OTAs don't go
off until the current is zero, and the control voltage must be within a
couple diode drops of the negative rail for that to happen.  This is easier
to do with linearized 2164 pairs.  In the Multiplying Modulator, it didn't
matter because the signals would cancel each other out, but in a crossfader
you would never entirely eliminate one side unless CC were biased properly.

More information about the Synth-diy mailing list