[sdiy] Oscillator Help

[Ian Fritz]

At 12:13 PM 4/23/2008, Adam Troyer wrote:
>Hello everyone!

Hi Adam!

>My design is on page 4 of this pdf: 
>http://users.ece.gatech.edu/~lanterma/sdiy/datasheets/comparator/lm392_apnote.pdf,
>the 'Exponential V/F Converter for Electronic Music'.
>
>There are two main issues I'm having at the moment.  The first issue is 
>with the resistor and 5k pot on the left side
>of the schematic.  The resistor on the left doesn't have a value, so I've 
>been assuming that the text on the right of
>the pot (the '4.7k* select...') was referring to that resistor.  Does this 
>seem right?

Yeah, that circuit is hard enough to understand without having the labeling 
screwed up.

So for a sanity check, let's do a real rough ballpark estimate of what the 
mystery R should be.  You can do some little calculations to verify the 
following:

1.) We are told in the text that the 2uF cap has a signal level of 
~12mV.  Therefore the charge transfered each cycle is ~24nC.

2.) Let's assume that the oscillator will run up to 40kHz without 
saturating Q1.  (It's accurate up to 20kHz, so it should run higher than 
that.)  At 40kHz the current through Q1 is ~1mA.

3.) If the Vcb of Q1 is greater than zero (at 40 kHz) then the unlabeled R 
is less than 7k.

So 4.7k seems plausible. They say to select it, because its value depends 
on the reverse leakage current Is of the transistor, which can vary quite a 
bit between parts.


>The second issue is a little more complicated.  When testing the circuit 
>yesterday, I was getting some oscillations at the
>output pin of C1.  With the 5k 'Zero' pot turned all the way toward the 
>input, I would get a short time of about 0 voltage
>followed by a long time of constant voltage at this pin (it looked a lot 
>like waveform A on page 5, since I'm not describing
>it very well).

OK, that means it's oscillating.  You should be able to determine the 
frequency by measuring the time between the pulses.

>When I turned the pot away from the input, the oscillations would increase 
>in frequency some, but if I turned
>the pot too far the voltage at the pin would drop down to nearly 0V and 
>stay that way, and wouldn't oscillate again even
>if I turned the pot back.  I actually had to turn the circuit's power 
>supply off and wait a little while before turning it back on
>to get the oscillations to start back up.

It looks to me like if you set that resistance too small you could be 
pushing Q1 too far into forward bias.  Keep away from the end of the pot, 
or put a 2k fixed resistor in series with it.

>Even with the oscillations at this pin, though, I wasn't getting any 
>response at the
>inverting input of C1.

Impossible!  It's a very small signal, so you need to turn your scope 
sensitivity way up.

Sounds to me like its working! :-)

   Ian