Hi guys!
Long time not being active here, last summer I also started building my modular synth.
With your help, I got pretty far! I got 3 half-working VCOs, a DIY Power supply, Audio-Output-Module and a nice case:
But before I turn to building a VCF, I want to get my 3 VCOs based on AS3340 (2x) and CEM3340 (1x) fully working. And one big issue is panel-mount VCO/LFO switch that switches the timer capacitor from 1nF to 100nF:
My initial wish was to have an LFO-functionality too in my VCO. It kinda works (it shifts the frequency wayy down), but it also caps the frequency in VCO-mode at around 1661Hz, which is pretty annoying.
I asked myself, what could it be? And I fiddled around with the tuning trimmers, but that didn’t make a difference in terms of max. freq. Because the actual problem is the capacity that I bring in with the panel-mounted switch-component:
I tested it via unplugging the timer switch from the pcb and bridging it and it worked. (although I lost the LFO mode and the timer switch would be obsolete now)
Soo I need an alternative approach for the VCO/LFO switch!
A colleague of mine recommended a Mosfet pair, gates connected to the switch, but the timer caps need bidirectional current I guess, so that probably doesn’t work. Do you guys have a better idea?
wow thx for the fast answer! I’ll try it out, very elegant solution. To avoid the additional resistor, I probably replace the 100nF cap with a <50nF one lol.
You seem to be saying you’re somehow picking up on the order of 10 nF (>> 1 nF anyway) parallel to your timing capacitors from the switch connecting to the panel. That’s a lot.
For wires around 0.5 mm diameter and 10 cm long and a couple millimeters apart, the capacitance is on the order of 1 pF. So that’s negligible.
Your panel appears to be solid metal connected to ground. The capacitance of a similar wire to a ground plane a few cm away is also of order 1 pF.
Even with parallel plates, you’d need something like 5000 cm^2 half a millimeter apart to get around 10 nF. Or a smaller area with smaller separation, of course, but how would you get that?
(Is your panel grounded? Is it solid metal?)
Maybe I’m missing something but I’m skeptical picking up extra capacitance is your problem.
Consider doing this: Instead of switching between two capacitors with a SPDT switch, have the 1 nF on the board and a SPST switch connecting/disconnecting the 100 nF in parallel. (That’s not what you have on your PCB but I should think you could kludge it.)
Then in the high frequency mode the connection to the panel is broken and it seems to me it would be very unlikely any funny panel-related behavior could occur.
oh no sry people, the problem kinda resolved into thin air.
I tried to reproduce the problem (remove the bridge to 1nF timing cap and putting back in the spdt timing switch) and it works again.
Embarrassing story: I built the VCOs and put them to the side to work on Bachelors thesis, now I am looking at it again to do calibrating and stuff. On 2 of 3 VCOs, I accidently switched the timing capacitors of LFO and VCO mode. Obviously you cannot reach full freq. range with a 100nF timing cap. I didn’t know what I was thinking, spdt toggle switches are hard I guess.
In practice, I now have all the wished frequency ranges in LFO and VCO mode on all three modules, but I am still a little skeptical.
@analogoutput:
yeah right, its a solid aluminum frontplate, grounded through the CV IOs. Your calculation makes sense, I thought I were onto something xD.
Wow, 11 position switch?? Seems cool, but also hard to tweak and get right, isn’t it?
@craigyb: Huh that would be even easier lol, next time!
The thing is, the one VCO module where I replaced the timing cap switch with a bridge had its timing caps connected to the right switch position, I am a little confused. But I cannot really remember that, I just try everything out without documenting
Needs 11 resistors matched at the 0.1% level. Go through a few dozen 1% resistors and you can find them. Or now that Tayda has 0.1% resistors at much lower prices than Digi-Key or Mouser it’s probably better to just buy 'em.
After calibration most of the octaves are in tune within a few cents. The outermost ones are a little further out but I don’t particularly care if the dog whistle range is in perfect tune, and I really don’t care if my LFO range is in tune!
Maybe, one day one needs to build a digital VCO with perfect tuning, it seems so compelling. Put a Raspberry Pi Pico, a DAC and some OpAmps together and you have all this tuning kinda effortless. I really don’t like calibration , its like getting from the real world to a mathematical model as near as possible, but its never satisfying because its an asymptotical effort to get there.
I’ve already done that, I just developed a dual as3340 design with autotune. It’s posted in this group. Unfortunately some calibration is required, but I guess with digital pots then this could be automated to a point. But if you don’t calibrate to a reasonable degree first then any voltage you apply becomes inconsistent across multiple VCOs. So for example if you apply pitchbend to two non or poorly calibrated VCOs then they will be out of tune as they reach the destination pitch of the bend.
My only thoughts about auto calibration are the digital pots only have 256 steps which might not be enough resolution for calibration.
Well, by the nature of asymptotes, it’s never perfectly satisfying but is as close to perfectly satisfying as you want if you do it long enough.
Having a Mutable Instruments Module Tester makes VCO calibration an order of magnitude less tedious. It has a mode where it tells you the interval (in cents) between the two most recent pitches it’s seen, all you have to do is play (say) double octaves and adjust the trimmer until that interval is 2400. It’s also hugely useful in doing all sorts of other module testing and troubleshooting. Not a terribly cheap build, the DC to DC converter alone is an $11 part, but very worth it.
Just to add my two hap’worth, I recommend using a C0G type capacitor for the 1nF capacitor, for their stability over a wide temperature range and resistance to microphonic behaviour. They cost pennies.
oh okay, I’ve never heard of C0G type caps, I am mostly aware of tantalum, ceramic and electrolyte ones. In the VCO I’ve builtin a 1nF film one.
In the store website, they say, they only have a small temperature-dependance and eddy bergmans website says don’t use ceramic caps, but this is the only thing I know.
Thx for asking about the VCO-PCB! Its self-developed with the help of you guys, the AS3340/CEM3340 datasheet, LMNCs long VCO tutorial and eddy bergmans VCO guide:
The red square on the most right picture is the 1nF film timing cap and next to it is the 100nF ceramic timing cap (there, I don’t really care about stability). Other than the AS3340 / CEM3340 chips, I included a bunch of TL074 for signal normalization to 0-10V and offsetting to -5V to 5V to be filtered later easily with the LM13700 state variable filter which I still need to develop. With @analogoutput and @clarionut 's help, I also included a diode circuit to the triangle output of the 3340 to have a sine wave.
, its like getting from the real world to a mathematical model as near as possible, but its never satisfying because its an asymptotical effort to get there.
