Yeah, I built it on veroboard with a VTL5C3/2 and it worked as I had expected and is what lives in my rack at the moment, and I did some layout work in KiCAD but then got busy with work, so it never got finished to the point where I had professional boards made up.
2 Likes
I made one based on @EddyBergman’s schematic. It was waiting on a specific zener diode I had to order… I’m sure I have it somewhere, I’ll get back to it.
3 Likes
I’ve built the AI Synthesis LPG, putting it behind a Kosmo front panel — no (intentional) changes other than three of the pots and all the jacks are panel mounted and wired to the footprints on the PCB.
It kinda sorta works. But the depth switch has me baffled. Even with the depth knob turned all the way down, flipping the depth switch causes the amplitude (in VCA mode) or the cutoff frequency (in VCF mode) to decrease significantly. For instance at one point I had it in VCF mode, with the resonance turned up to self oscillation, and flipping the depth switch changed the frequency from 420 Hz to 160 Hz. The maximum amplitude or frequency with the depth switch on is low enough that I don’t consider the depth switch really usable.
Trying to investigate this, I did an LTSPICE simulation of the CV handling part of the circuit — and it seemed to behave consistently with what I was seeing; it showed a big decrease in the vactrol LED current with the depth switch on, even with the pot at maximum resistance.
Then I tried breadboarding the CV part of the circuit — and it too behaves in the same way.
“So that’s how it works”, right? No, because Abe (@AI-Synthesis) says with the knob turned down, the depth switch should not cause any change — and he sent me a video clearly demonstrating that on his module.
This is driving me nuts. I’ve checked my components, they’re correct. Checked connections, haven’t found any wrong. Measured resistances, they’re what you’d expect. Module, breadboard, and simulation are all in agreement and all, evidently, wrong. Abe hasn’t been able to come up with any explanation either.
Anyone here have any thoughts? Anyone built the AI LPG, or a similar design with depth switch; does yours behave like mine or like Abe’s?
1 Like
Isn’t that just what the depth switch does? It provides a path for the current to bypass the vactrol LEDs. I could check mine, which I think is a very similar design. Let me just get straight what the depth knob is on yours, does that set the resistance of the bypass path? On mine that’s just a trimmer on the back.
No! THUNDERDOME! ! !
1 Like
The portion of the schematic in question:
The depth pot is an A500k wired as a variable resistor. It’s a front panel pot. It varies the resistance of the bypass path. There is a trimmer, but it’s the 20k in the main feedback path, controlling the amount of gain.
Presumably the idea is that at the 500k end it is essentially too large for any significant amount of current to flow through it. On the face of it that seems reasonable, since the resistances along the main feedback path are so much smaller than 500k.
But the voltage drop across those resistances is much smaller for the main feedback path. For instance, in my LTSPICE simulation, with 5 V CV and both the level and CV pots at 50%, and with the trimmer at 10k and the depth pot at 500k, I get:
- Voltage out of the op amp ~ -6.6 V
- Therefore current through depth pot ~ 13 µA
- Drop across each LED ~ 2.2 V
- Therefore voltage after the second LED ~ 2.2 V
- Therefore current through 100k ~ 22 µA and current through 33k + trimmer ~ 52 µA
- So current through LED ~ 74 µA
The current through the depth pot is about 18% of the current through the LEDs.
The total current in the feedback, that is the current through the depth pot plus the current through the 33k + trimmer, must be equal to the current coming from the CV summing circuit, so turning off the depth switch should increase the 33k + trimmer current by 13 µA and correspondingly increase the LED current. That indeed is what happens in the simulation; the 33k+trimmer current goes to ~ 66 µA and the LED current increases to ~94 µA.
At smaller CV values the change gets relatively larger, since the op amp output is smaller and the LED voltage drops are larger relative to it. With the level pot at 0 and the CV pot at 25%, I get LED currents of ~21 µA and ~10 µA with depth off and on respectively.
None of that seems unreasonable to me at all. But it appears to be contradicted by the behavior Abe observes, and shows in the video he sent me. I can’t make sense of it.
1 Like
I am also stumped, but of course haven’t had my hands on the unit in question. This is the first instance of this behavior I have heard of from a builder. The PCB is the same as all the others, and the parts seem similar enough. There is wiring, so that is different, but if the wiring is correct then it should behave like all the others that have been built.
1 Like
The more I look at it the more I’m convinced there has to be a change in the LED current with the depth switch. Then if there isn’t a change in the output I’d think that means the vactrol LDR conductance is saturating, so you could double the LED current and it wouldn’t change the output.
But then there would be two things immediately noticeable: First, if you varied the level and CV knobs a little, nothing would happen, because the LDRs are saturating. Second, if you varied them enough to change the output, then at that setting you’d see a difference with the depth switch, because the LDRs are no longer saturating. I presume neither of these is observed.
I am just going to take a very wild guess here. Could the difference in behaviour explained by differences in LEDs and/or Vactrols used? The resistances in the circuit are easy to check and confirm that they are nearly identical, but the random LEDs that we buy here and there may have different specs, which the vactrols are known to have very wide tolerances.
Abe doesn’t think so. I’m less certain, but in any case regardless of the LED and LDR properties I’m baffled as to why it should work the way Abe says it should work and his evidently does.
My breadboard test used only one vactrol because I had only one extra, plus a random LED; didn’t behave substantially differently from my build with two vactrols.