I’m using AS3340, you mentioned CEM, might this make the difference?
As noted over there, I’ve just built a MFOS (non 3340) VCO that uses a comparator on the ramp wave to generate the pulse wave, and after a bit of a modification on it I’m getting a pulse width range of about 0% to 98%… adequate, I suppose!
tbh I can’t remember which brand they were I will check in tut morning and keep you posted!
aah nice!!! yeah comparator seems the way to go! its odd because it is a comparator in the 3340’s very strange goings on… but indeed im plagued with a never ending dream to keep component count down haha.
Yeah, I think Ray Wilson had the opposite approach! For the voltages on the PW pot instead of just using a fixed division of the rail voltages he essentially used a pair of sample and holds to set them to something based on the minimum and maximum of the ramp wave amplitude… What problem that was supposed to address and how much it was worth worrying about I have no idea, but in my testing it was producing too small voltages and thereby limiting the width range (to “only” about 5% to 90%…) so I bumped those voltages up. If you compare his VCO and the YuSynth one there’s a big difference in the parts count. 7 trimmers including 3 in the sine shaper alone! But he put years of work into it and it so far seems to work very very well.
hi so I just did some more tests. im using AS3340 and CEM3340 each do the same with me. i popped it on middle C and im getting a drift under 1hz which is nothing. (and findings seem to compliment the notes to hz response quite similarly up and down the range)
im quite surprised by your findings which makes me think maybe its something else with your setup. what power supply are you using? even in none modified mode these should not be fluctuating as much as your readings are saying… unmodified you cant notice the pitch drift with your ears… my findings below sort of match that.
I did the mod to 4 oscillators and all had a similar response to the mod 20M is more than adequate. (sweet s[pot could be anywhere between 18 and 25M for absolutely zero drift but 20M is more than adequate).
So I just did a bit more checking im running around the C at around 260hz. my findings are
no modification… drift of under 1hz (not noticeable)
modification of R13 of just the 200k a drift of 3hz (too big)
with the 20m mod it now has a drift of 0.5hz near enough (not noticeable)
in the Lower register Lower C 130hz this finding is less (hence why my readings dont match yours) and higher registers its a tiny bit more but this matches up to the note response and all in all throughout the whole audible range im not noticing anything.
so yeah it might be worth someone else giving it a go! if they want more pulse width from the square out but not at a loss of pitch drift.
my only suggestion is mess with the 20M value if the note is pitching up increase resistance, if note is drifting down decrease resistance… or find a 5m trimpot or something ha.
Im very pleased with this mod personally and has vastly reduced the component count on the design of mk2…
I dont know about you but an unspoken rule in DIY synth fun is a magic ratio…
that ratio being KNOBS:COMPONENT COUNT
too many knobs to components the knobs do less fun things… but ooh lots of knobs to twist!
too many components to knobs! arse ache to build and go big…aaah not enough knobs to twist haha
also could you send a pic of the back of one of your modules??? I just want to check something, because yeah your drift that your saying is awfully high now I think about it and do the tests of my own
I’m currently using a Meanwell RT-65B PSU, but already ordered PCB from Frequency Control to switch to a more stable and powerful PSU.
I have the same drift on all VCOs using two different PCB versions:
And tbh the black PCB sucks when checking solder points
did you try increasing the ohmage? 30M even? it’ll totally give you more pulse width and remove pitch drift, along as the 20 or 30M is coming from the outside leg of R13 to the leg pointing away from the Arduino on the R26
I dont know about the mean well. I have tested this mod running it on a befall excalibus and also Frequency central and the mod works on both.
like I said if its pitching upwards add another 10m to the mix
I would say you would get a very stable and accurate supply from the Mean Well unless your overloading it.
I tested with most modules unplugged (except for the VCO, a buffered multiple and headphones amplifier I needed for testing
I spent some time on the drift topic, finally!
The module I used for testing seems to have different issues. (the black edition). It already comes with a drift of 3Hz without modification and octave switch troubles in the middle position (F instead of C).
However, I replaced it with another module (yellow edition).
The results are:
| modification | drift at | drift at | drift at | | | 130Hz (C3) | 260Hz (C4) | 520Hz (C5) | ---------------------------------------------------------- | none | -1Hz | -2Hz | -5Hz | | R13 200k | -4Hz | -8Hz | -17Hz | | R13 200k + 5M | +1Hz | +2Hz | +4Hz | (positive drift!) | R13 200k + 6,6M | -0Hz | -0Hz | -1Hz | (10M and 20M in parallel) | R13 200k + 10M | -1Hz | -2Hz | -5Hz | | R13 200k + 20M | -2Hz | -6Hz | -11Hz | | R13 200k + 30M | -3Hz | -6Hz | -13Hz | RT-65B psu | saw wave used as reference | V2 centered no CV signal
Practically, 200k // 6,6M even reduces the drift compared to no modification!
It also nice to see the drift is scaling.
EDIT: updated table with more measurements.
EDIT2: replace 7,3M with 6,6M. (my brain went off for a minute!!!)
Very strange that it went from negative at 0M to positive at 5M to negative at 10M. Would be curious to see more points between 0M and 5M.
Here comes another benchmark. Hopefully things were calculated and measured correctly. You never know on Saturday evening
R13 swapped with 100k, results in high PWM bandwidth.
Some more points between 0-5M for @analogoutput
| modification | drift at | drift at | drift at | | | 130Hz (C3) | 260Hz (C4) | 520Hz (C5) | ---------------------------------------------------------- | none | -1Hz | -2Hz | -5Hz | | R13 100k | -4Hz | -8Hz | -17Hz | | R13 100k + 1,5M | +14Hz | +28Hz | +56Hz | (positive drift!) | R13 100k + 2,5M | +7Hz | +14Hz | +28Hz | (positive drift!) | R13 100k +3,75M | +4Hz | +6Hz | +13Hz | (positive drift!) | R13 100k + 5M | +1Hz | +2Hz | +5Hz | (positive drift!) | R13 100k + 6,8M | -0Hz | -1Hz | -1Hz | | R13 100k + 7,5M | -1Hz | -2Hz | -4Hz | | R13 100k + 10M | -1Hz | -3Hz | -6Hz | RT-65B psu | saw wave used as reference | V2 centered no CV signal