[sdiy] Resistor matching in Bergfotron complex VCO

[Aaron Lanterman]

Howdy gang,

One of my students, Greg, is working on building the Bergfotron  
version of the Buchla 259 triangle core. He's using the Linear Systems  
LS358 (free samples they kindly sent) for the current mirror, and  
LM394s for the diff pair in the OTA and the expo converter (I had  
gotten a bunch of those from Futurlec a while back, IIRC).

The Bergfotron uses 5.1 k resistors. Our stockroom didn't have those  
so I told him to use 4.99Ks (I figure that's what's used in the  
original Buchla anyway).

After an afternoon of debugging, we got it oscillating with expo  
control. (I assist mostly by asking the students to talk me through  
the circuit, showing me things on the scope as the go, and they most  
often figure out the problem on their own while I pretend to be  
insightful... there's something about explaining your circuit (same  
thing works with computer programming) to someone else that makes it  
easier to find bigs.)

At higher frequencies, it looks pretty triangular, but at low  
frequencies it's more like a 20% triangle. (Also, when twisting  
various knobs to get the lowest frequency, it seems to only get down  
to 600 Hz, but I suspect that will be fixed once we get the triangle  
more triangular.)

The Bergfotron webpage emphasizes matching the resistors at the top of  
the OTA - it says "Buy a belt of 100 resistors and find two that give  
the exact same reading on a 4 1/2 digit ohmmeter. Any error here will  
give a non-symmetrical triangle wave." The resistors Greg is using  
currently measure at 4.963K and 4.972K, which are a far cry from  
measuring exactly the same to 4.5 digits. But, before I send Greg off  
to spend an afternoon at a DMM measuring a truckload of resistors, are  
there other sources of nontriangularness we should be looking at first?

Is that circuit really that sensitive to the resistor matching? That  
seems OK for a one-off like the Bergfotron, i.e. a design a person  
makes to build just one or two of - but it seems to me that would have  
severely limited the manufacturability of the original Buchla 259  
(although perhaps that shouldn't surprise me.)

- Aaron