If anyone is experimenting with DIY vactrols and wants to use them on a PCB, I’ve just published a design on Thingiverse for a 3D printed vactrol housing which will fit the KiCad vactrol footprint. It uses flat-top 5mm LEDs to reduce the length while keeping the LED/LDR alignment fixed. A 3mm LED (as in most of the cheap Chinese vactrols) can easily move giving variable results.
I read “a synthesizer module which needed 8 vactrols” and immediately thought, “Oh, like the Turing Machine Vactrol Mix?” Ha!
Anyone got recipes? Which LEDs and which LDRs?
Spot on! The original circuit specifies VTL5C10s which have a very long light->dark transition time (assuming the data I’ve found is correct), presumably to avoid clicks and pops as the channels turn off in the Vactrol Mix. I haven’t been able to find any LDRs with this sort of relaxation time unfortunately; in the kit of 7 types I bought the differences are in the relative resistances under dark and light conditions but they all seem to have similar relaxation times (20-30ms) according to the data sheet. I built a quick’n’dirty Arduino based test rig and plotted LED current vs. LDR resistance, and also resistance vs time in ms after 10 seconds of full brightness illumination - none of them were significantly slower to recover.
In the end I opted to use GL5528 LDRs and ultrabright flat-top red LEDs, which in preliminary testing seem to work OK on a single channel without any obvious switching noises. The resistance at full illumination is lower with green LEDs, though I’m not sure if this is due to better matching of spectral response or just greater brightness. There’s little or no difference in the relaxation time. I haven’t yet had time to finish assembly of the complete module with all four channels connected via input attenuator pots, so I still have some fingers crossed that the final module will work satisfactorily! I’ll report back when I’ve completed it - I’m just finishing off a new ADSR at the moment.
If the LDR doesn’t do it (the long transition), it’s probably possible to slow the LED transition from on to off with some capacitor… (depending on the driving circuit, some additional components may be needed).
This is the fallback solution to try if the vactrols do switch quickly enough to cause problems. Each of the 8 channels will need a diode so the capacitor doesn’t discharge straight back into the gate which drives it, and then a lower value LED ballast resistor to compensate for the voltage drop in the diode. It may be worth investigating anyway to get smoother transitions in the mix - I’ll see how it goes…
I have a design for a similar module but using a pair of DG412xJ analog switches instead of the vactrols, which would be considerably cheaper, but I haven’t breadboarded it yet and have been doubtful about it, whether their immediate on/off would lead to unacceptable artifacts. My meager past attempts at vactrol making have been unsatisfactory but they used some fairly random LEDs and LDRs. Your housing looks like a winner and maybe I’ll print some of them and go that route at some point.
Not sure what you have in mind, it sounds simpler than what I might think of for such a situation which is a precision rectifier: a diode in the feedback of an op amp. Kassutronics used those in their precision ADSR for just the same purpose, giving a one-way (dis)charge path to a capacitor.
Yes, I think rapid switching of arbitrary waveforms is likely to create nasty clicks, like cutting bits out of an audio file at a peak rather than a zero-crossing point.
A precision rectifier would be a proper solution - but would also require grafting in some extra op-amps on a daughter board. If there are problems I hope that a quick’n’dirty fix with a few components dangling off the existing PCB would do the trick. Of course I really hope that the thing just works…
This worked for me: ShedSynth LPG circuit
Nothing as precision as compensate for the voltage drop in the diode; didn’t think about that, it just lights up when the input is on.
Also a straightforward way to build a vactrol if that helps.
I found green diode + GL5528 worked out best from my (same?) kit of 7 LDRs.
You can find the “spectral response peak” wavelength listed on the datasheet for the LDR. The tl;dr is that, yes, most LDRs are more sensitive to light in the green spectrum (495-570nm).
It has to with the emission spectrum of cadmium, and specifically of cadmium sulfide, which is the compound used in most common LDRs. A green LED with a clear lens should give the best response since you’re getting unfiltered green light. After that, I would expect a higher lumens rating would give lower resistance in the on phase.
Perhaps - but it’s a bit more complicated. The response of a CdS LDR falls off very rapidly at shorter wavelengths, so some of the output of a green LED is effectively wasted. Ideally you should probably use a yellow LED as the LDR would respond to essentially all of its output - but of course yellow ‘ultrabright’ LEDs are much dimmer than ultrabright greens at the same current. The ultrabright red LEDs I used in my tests are also not quite as bright as the greens (though a lot brighter than the yellows and oranges I also tested), so the relationship between LED current and LDR resistance is probably more due to the intrinsic brightness than the colour. None of that really matters for my purposes, where I was trying to find an LED/LDR combination that responded more slowly, especially in the light->dark relaxation. That proved to be a fruitless quest, though if anyone knows of an LDR that responds more slowly than a GL5528 I’d really like to hear about it!
We should have taken this discussion of vactrols into a separate thread…
A quick (final?) update here though. I’ve now finished my home-brewed Vactrol Mix module with 8 DIY vactrols (GL5528 LDR / 3.5mm flat top ultrabright red LED) in the 3D printed housings I linked to above and it works perfectly. No switching artefacts that I can detect, so the 20-30ms switching time is (as I hoped - Phew!) long enough.
Does “documenting your builds” count as build progress?
Yes it does? Cool. Glad to hear. Happy we’ve just cleared that up.
Last few days I’ve been working on actually having stuff at the URL emblazoned on my PCBs. I’d like to have a more long-lived repository than info scattered on 10 different sites, especially as we see some of them like twitter going up in flames due to some bigot’s mid-life crisis.
I have the first test page up (without navigation or whatever), feel free to lemme know what you think: Sea Moss Raft—Aria's Honking Synths
While it’s using pretty modern tech, I wanted to have a more rustic, old-fashioned kind of design (it’s not really phone compatible-yet but will be eventually). It’s an occasion for me to keep up with the latest web tech. Those newfangled JS frameworks have, culturally, a certain stench about them, but if you remain critical about the ways you use them, they are incredibly powerful. I expect to have about 40 pages like that when it’s ready for use (probably sometime in september or so).
What is your preferred method for cutting PCB material? Hacksaw, tin snips, razor knife? Something else? It’s been so long all I remember from the last time I did it is making an absolute mess of it.
I use a hacksaw followed by sanding the cut edge for both SRBP (typical cheap stripboard) and glass fibre boards. Supporting the last bit of the cut is particularly important for SRBP as it’s likely to split. Glass fibre board is particularly hard on junior hacksaw blades, so I always use a full-size hacksaw for that.
Definitely want to start by saying I really love your design of your site, it looks really nice.
I don’t know how accessible it is though, especially if people are using a screen reader.
Testing the sizes for a daft build I dreamt up and couldn’t stop thinking of.
That’s a second-hand iPod Touch 5 (£30) married up to a 3D-printed piece that has a TRRS plug and an Apple Lightning connector mounted in the right places. (The little PCB you can see is what was inside the plug on the iPhone USB/SD card hub thing I picked up; the Lightning connector is on this.)
My daft idea was to make a module that shifts output levels, attenuates a mono input, provides USB for MIDI, and has some form of CV-to-MIDI (and possibly classic MIDI) circuitry incorporated. It’s ridiculous. All that circuitry will be in a Eurorack module with an 8-core cable running between that and this, so I can pick it up to peer at with my aging eyes.
This is yet another project to chip away at slowly…
More than corporate brands bother to, I can say for sure. It’s making use of semantics that automatically generate landmarks that jaws/nvda will use, wai-aria attributes where useful, skip to nav, additional visual descriptions in alt attributes, and semantics letting you use your browser’s reader-mode if you prefer. However the view-source will look nasty, in today’s js component frameworks the html is more of a compile target than source code.
I often thought about that kinda stuff!! There’s so much e-waste just waiting to be made useful again in such a fashion. It’s such a shame we have so many serviceable mini computers that no longer receive security updates or hold much charge, but would remain perfectly serviceable as airgapped appliances. On Android, Caustic is toaster-compatible, and would just be perfect if you could add some sort of CV in. I thought about the possibility of finding a generic software solution for a wide range of models + a generic printable cradle that fits a lot of devices
I’m glad it’s not just me! I had an iPhone 13 years ago which I’d thought about trying to use, but couldn’t find it in my junk box - a shame because the big wide 30 pin connector has stereo audio in and out (and composite video!) But this iPod Touch was a bargain and has far more power. I use Android everywhere else but I always liked Animoog
I’ve thought about doing this to a Nintendo DSi as well, for the Korg software, but there’s no way of interfacing to it in any meaningful way, despite it looking awesome