[sdiy] A zero-offset buffer with a volt offset? Still debugging MOTM 310

[Aaron Lanterman]

Hi SDIYers (and Paul),

I'm still cranking on the MOTM-310. It's now oscillating from 47 kHz to 
142 kHz.

I had steered away from the buffer and towards looking at the expo 
converter... I noticed something that steered me back towards looking at 
the buffer.

The MOTM-310 is set up to drain a cap from 5 V to 0 V; unlike other 
VCOs that use FETs for reset, the 310 uses a BJT, which shorts one end of 
the cap (the raw saw waveform) to the other end (held at 5 V) when it's 
time to reset.

The saw at the cap runs through a 22 ohm resistor into a dual FET buffer. 
Out of deference to Paul I don't want to post the full schematic, but 
I'll tell you his buffers look a lot like Buchla's 291 buffers:

http://rubidium.dyndns.org/~magnus/synths/companies/buchla/Buchla_2910_150.jpg

Except paul uses 1.5K 1% instead of 150 ohm resistors. (I checked, they 
both measure 1.5K "in situ")

According to the scope, at the cap (or equivalently at the gate of the FET 
buffer), I see a sawtooth goinjg from around 5V to around  1V it seems, 
not ground, and the scope says it is averaging around 3V. This made me 
suspicious... maybe my 310 is resetting to soon?

So I checked out the output of the buffer (equivalently, the input of the 
comparitor). It shows a sawtooth running from 3V to around 0 V, and the 
scope says it averages around 1.6 V.

Most curious, I thought... is my FET buffer actually applying some sort of 
voltage drop? Some kind of attenuation?

The final sawtooth output is supposed to be 10 V peak-to-peak, but 
instead it's 7 V peak to peak, with an average around -1.4 V.

Twisting the coarse knob I get 42 kHz to 142 kHz.

Now here's the fun part: suppose I set the course knob to get 42 kHz. I 
thought (this is based on a clue from Paul for a long time ago) - what if 
I try to bypass the buffer? I took a wire and shorted from the input of 
the buffer to the output of the buffer. (Notice all the buffer circuitry 
is still there, doing who knows what, so it's not as if I really took it 
out.")

When I do this, my peak-to-peak sawtooth output goes from 7 V to 8 V, and 
my frequency drops from 42 kHz to something like 24 kHz.

I'm tempted, as an experiment, to try fully bypassing the FET buffer, but 
I'm hesitant to desolder the LS3954A dual fet.

Any ideas?

Oh, here's some other measurements - the LS 3954A is getting 15 V and -15 
V where it should. The resistor associated with the current-making part 
(the lower FET in the Buchla digagram) has a voltage of -14.5. On the DMM, 
the buffer input at the gate averages 2.8 V; the (source or drain? can't 
remember which is which now) of the main buffering FET averages 2.2 V; and 
the output of the buffer averages 1.6 V.

The 2.8 V to 1.6 V drop in the DC level of the sawtooth through the 
"zero-voltage" buffer makes that FET suspect... but I want to rule out 
other possibilities before I go trying to replace the FET.

Help!

- Aaron

-----------------------------------------------------------------------------

Dr. Aaron Lanterman, Asst. Prof.
and Demetrius T. Paris Junior Prof.    Voice:  404-385-2548
College of Electrical and Comp. Eng.   Fax:    404-894-8363
Georgia Institute of Technology        E-mail: lanterma at ece.gatech.edu
Mail Code 0250                         Web:    users.ece.gatech.edu/~lanterma
Atlanta, GA 30332                      Office: Centergy 5212