Recently I went back to looking at the AS3340A on a new breadboard setup, and at first I thought I’d have to eat many of my words on the subject:
Well crap… the pulse width frequency shift is back, with -5 V regulator power. I don’t know why, I thought I’d pretty much ruled it out before but now I’m definitely seeing it. About 40 cents (almost half a semitone) across the full pulse width range. Also unlike before, I’m seeing a good pulse width range with only 3MΩ or even 1MΩ between pins 4-5… but they’re not suppressing the falling edge oscillations under all circumstances. Hello, consistency? Will you be coming around any time soon?
And then a couple hours later I went back and… the pulse width frequency shift was gone, and the oscillations were being suppressed for wave frequencies all the way down to about 12 Hz using a 1MΩ resistor. Which, however, was still not messing up the pulse width range anywhere near as badly as before. It reduced the range slightly, but only slightly. Maybe by 5%.
And I haven’t seen the frequency shift or the falling edge oscillations since then. Also I haven’t had the chip powered down since then for more than a few seconds at a time. Maybe the problems occur just while warming up, and it takes more than a few minutes to warm up? I don’t know.
(Actually I thought I was seeing the frequency shift at one point, but only at a level of about 3 cents, so who cares, and right at the moment it seems to be gone completely.)
This new breadboard setup is going to be an end to end realization of my full module design. Right now it only goes from inputs through the AS3340A chip; the output stages and sine shaper I haven’t done yet. But I did do the sine shaper previously and it worked, and the output stages are just simple op amp gain and offsets, so how can they not work? He asked.
I’ve tuned it to 1 V/oct within a couple parts per thousand over eight octaves, I could probably push it to nine. I’ve checked that my hard and soft sync work, more or less — maybe a little flaky but that might just be the breadboard. Good enough for me.
I mentioned the 1M resistor to suppress the falling edge oscillations. Oddly enough, the CEM3340 datasheet recommends such a resistor, but to improve the fall time, and in Thomas Henry’s writeup of the VCO Maximus he says the same. And he says he added a 1 nF cap from pin 5 to ground for the falling edge oscillations. But what I see is that the resistor stops the oscillations but appears to have virtually no effect on the fall time, and the cap seriously worsens the fall time! I guess there must be something else going on. Maybe it’s a difference between the Curtis and Alpha chips, or the AS3340 and AS3340A. I don’t feel like chasing after it, though. What I’ve got is working. For now!