Octave down noise problem

[Fancy Lime]

Hi everyone,

so I breadboarded an octave down synth-sound circuit using a CD4024 ripple counter. It takes several ideas from Slackers Slacktave (mostly using the CD4024) and Valve Wizards U-Boat (mostly the comparator section). I changed the filtering before the comparator and turned the comparator into a Schmitt trigger, which improves behavior on quietly fading low notes on the bass a lot, at least with my setup.



All in all I'm very happy with the sound. Or I would be, were it not for the terrible high-pitched noise. It's a high screeching sound, much like a square wave that has been sent through intense highpass filtering. This occurs on the output no matter what the position of both volume pots are. By systematically taking parts of the board and sticking them back on I was able to identify the Schmitt trigger as the culprit. It must be introducing the noise via the power rails directly into the output (and probably other places too) because the noise is always there, even if all signal paths to the output jack are disconnected.

I tried buffering the bias voltage with an opamp, which only made things worse. The I tried to give the audio-path and the sub-synth-path individual bias networks, which did not change the noise problem but made the octave down stability a lot worse.

Any ideas how to tackle this are highly appreciated.

Andy

[ElectricDruid]

Sounds like the Schmitt trigger is oscillating, which implies there's some feedback from the output to one of the inputs. Well, there is - it's a Schmitt trigger, and they use positive feedback, if I understand it right. Trouble is, it's feeding back when you don't want it to.

I'd suspect that 100p cap. That will couple the sharp edges of the output transitions back to the input, and might be enough to kick the input polarity the other way...which then causes the output to flip, causing another sharp edge, which is then coupled back, etc etc at some fantastically high frequency.

What's the cap supposed to do? Why's it there?

HTH,
Tom

[duck_arse]

Merlin recommends using an LM833 for the comparator in the u-boat, or a TL0x2 with a pull-up(down, can't remember) resistor on one input. what oppy are you using? [I couldn't get the u-boat comp to stop fizzling on the bb.]

[Fancy Lime]

Hi Tom,

yes and no. The 100p cap sits in the negative loop of the Schmitt trigger, meaning it suppresses high-frequency oscillation. In fact I put it there to stabilize the tracking of the ripple counter and it does do that. It also actually helps diminishing the high-pitched noise but not very much. But you are also right about the Schmitt trigger. Not that it oscillates, it doesn't. That's the neat thing about Schmitt triggers. The positive feedback keeps them from oscillating. But the Schmitt trigger slams into the rails as hard as it can. I suspect that is causing high-frequency garbage in the rails. I am going to test what happens if I put anti-parallel LEDs in the negative loop of the Schmitt trigger. That should limit the swing of the trigger to the LED drop. I'll have to rearrange my messy messy breadboard a bit to try that but: good feeling I have about this. I'll be right back when I got it rearranged.

Cheers,
Andy

[Fancy Lime]

Merlin recommends using an LM833 for the comparator in the u-boat, or a TL0x2 with a pull-up(down, can't remember) resistor on one input. what oppy are you using? [I couldn't get the u-boat comp to stop fizzling on the bb.]

Oh, sorry for forgetting to label the op-amps. IC1 is a TL072, the others are NE5532AP. I had only 2 NE5532APs left when I started breadboarding this, hence the TL072. Now I have more, so I'll replace that one with a NE5532AP as well. Much less noise and better headroom to my ears (which may not be objective, I did not do any testing that would satisfy scientific standards). I considered going with a LM833 for the comparator but decided against it because I wanted to use as easily available, generic and universal op-amps as possible for maximum DIYability. I think everybody can get a NE5532, that may well be the most common op-amp, but LM833s may be hard to find in some places.

Cheers,
Andy

[anotherjim]

Do you have an oscilloscope Andy?

It looks like to should have sufficient AC bypass capacitance on your supply, chip state switching shouldn't really be that audible, but do note that breadboards add sufficient stray coupling capacitance between contact strips that signal can easily pass across to a high impedance input. When I've been troubled by this, I've separated trouble spots by grounding to 0v an unused strip between the suspects.

Your "Adaptive Schmitt Trigger" may be troubled by the op-amp behaviour when the inputs & outputs exceed their swing range. Op-amp types may not be ideal. We can't always win with the noise versus performance. Given the amount of filtering the signal gets and that you are using it to make a big fat square wave, I think low noise amps are the last thing to worry about.
Consider if a little self-noise in the amps were causing mis-trigger, then your circuit may be too sensitive in the first place?

Also, sometimes a fast amp is too fast for the circuit. Consider trying LM358, LM1458, RC4558 or some rail-rail CMOS.

[Fancy Lime]

Alright, so I took it all of the board and put it back on in a way that allows me to mess with the trigger some more. Needless to say, it stopped working. Completely. I just spent three hours debugging it but I got nothing. The comparator/Schmitt trigger does not say a word anymore. Not in the new configuration, not in the old configuration. Checked it 15 times or so, disassembled and reassembled it twice. Changed all the parts for new ones. Nothing. It may be time to move on and revisit this another day.

@anotherjim
I also suspected the breadboard crosstalk problem at first. Having a 9V square wave bouncing around the board might radiate to neighboring strips or even next.neighboring. But the problem also occurred when there was an entire vertical block between the squarewave and anything else. So I don't think that's it. Anyway, there is no way to check now since karma decided today is not the day.
As for the NE5532, I'm not married to using that one but I had built some nicely working comparators with it before. If I ever get this to work again I will certainly play around with different chips. Some research might be necessary beforehand to find something common that works well in this position. I really want to use an easy-to-get garden-variety chip. The goal here is to develop a good sounding bass octaver that is a bit easier to built and less picky in terms of component matching than the common ones as a sort of beginners-diy-octaver. Long way to go.

I'd love to hear from anyone who wants to try and build or breadboard this one. Successfully or not. If someone else is on a similar quest, maybe we can get something working together, cause apparently I cannot by myself.

Cheers,
Andy

[anotherjim]

That's bad news, but another day might be better.

I didn't finish before, but to clock CMOS can need a very fast switching drive that many op-amps can't deliver. The reason is that the clock circuit inside the CMOS can jitter if the input pin has a slow changing level. You get multiple rapid counts for one input switch. I think that to be sure, the clock should be prepared by a logic Schmitt trigger, even though you have an analogue one.

[Fancy Lime]

Hi Jim,

true, but the neat thing about the CD4024 is that it has an integrated CMOS Schmitt trigger on the counter input. The whole op-amp comparator/Schmitt trigger before it is basically just support for that. I even tried the whole thing with just a filtered bass signal into the CD4024 and it works. But its not very stable because the integrated Schmitt triggers hysteresis is relatively small and therefore harmonics cause additional triggering. The result is a synth sound noodeling wildly through harmonics of the down octave. Kind of fun but not exactly what I was looking for.

Andy