Working on my prototype pcb. Im getting some undesired behavior from what I believe is ripple on the power rails? inadequate power conditioning? dummy mistake?
I have 3 3340s with an additional LFO on the same board. Even with the lfo signal completely disconnected from the fm inputs I can subtly hear it affecting the pitch. So it must be getting into the cv summers or pulling down 12v…Oddly when I turn the mod amount wheel down all the way it gets worse at the last 5%
I thought I had fixed it by putting the center note of each oscillator on the 5v regulator, but i noticed again it’s definitely there.
For crosstalk issues, it might be useful to post a board layout as well. One thing I don’t completely understand, is why you are powering everything from +/-5v rails. Generally, that’s the amplitude of the signals you want, and TL07x, not being rail-to-rail, will be unhappy and incapable of operating with that range.
Good luck, cheers!
[edit] I misread the schematic, discard this comment and read below
My prime suspect would be U9B. It being set up like a schmitt trigger ie having no negative feedback path, it’s output alternates between the supply rails. Or, as close to them as the opamp can get. This large swing might definitely be bleeding into your supply rails. You could check whether adding extra decoupling caps here might help.
I remember a case with on of my own modules where the noise on the supply rail was the inverse of the lfo-type so signal. When you set the mod amount to 5 %, you might actually be cancelling out the interference. A look with the scope might verify, but likely it isn’t the source of the problem.
That’s a pretty standard LFO design. But yes, I suppose it could be affecting the rail voltage. I noticed when I measured the current draw of my modules, the value for the LFO depended on whether the square wave was on or off. I didn’t look at the rail voltage, though.
(Reminded me of the old joke: “Hey, stand behind my car while I fiddle with the turn signal control and tell me if it’s working.” “OK. It’s working. Wait, no it isn’t. Yes it is. No it isn’t. Yes it is…”)
Long shot maybe, but were you using soft sync when you noticed this? If NOT, I was looking at Thomas Henry’s VCO Maximus design and his comments about it, which made reference to the datasheet saying:
His design therefore switches in a 100 nF cap to ground whenever you turn off soft sync, and the recommendation is to leave the switch in the hard sync position except when soft sync is actually in use. Actually he also has some non-datasheet components on the hard sync input as well, saying that’s susceptible to noise too.
I think I’ve heard the jitter he is talking about here! It was on my breadboard too and very random. However I think I have another issue somewhere.
I pulled the LFO opamp out. I am still able to hear osc3 modulating the others slightly if I mess with modwheel. This is with the mod switches OFF, which come after the modwheel…There’s no way any signal could get into the fm inputs. Has to be a power rail problem
I let the ref take the silkscreen from the schem and add all the values by hand at the end. Probably a source of a lot of errors…I should just change that layer to silk and take the dummy human out of the equation
I also end up with components with no value printed
(Yeah, I have R value and ref on the silkscreen in my custom footprint.)
And then there’s the resistor footprints on the MI Module Tester:
WTF??
Hmm, I was mostly following Kassutronics too but changed the input section and ended up with 1M for the HFT. Haven’t found it to have much effect although I haven’t gone nuts with trying to get great tuning yet. Maybe 470k would be better.
I got 5 octaves dead on (well with whatever resolution my rigol can read) using hand matched resistors on the 4.000 reference and kassus design on breadboard. Thomas Henry says you can get up to 10, but I imagine that it gets pretty fiddly on those last few octaves.
Yeah I would be plenty happy with those numbers personally. If you want perfect go digital. A little detune brings out the fatness.
How do you use your scope to measure frequency? I just look at that little counter in the bottom right but I have to readjust the scale a lot to get a good reading. Is there a better way?
My scope has two ways of measuring the frequency. Bottom left below you see a frequency measurement for each channel which as I understand it is done by something like looking at timing of zero crossings for what’s on the screen. Or something. Top right is a measurement of the frequency of one channel done by a hardware counter. You use the measure button and select the counter menu to select which channel. That should be a more reliable frequency measurement — I think. It has more digits anyway, but then again right above that they appear to be claiming to display the trigger delay (?) to a precision of 10 zeptoseconds, so maybe that’s not so meaningful.
