I have been struggling lately with buying the CEM or AS chips, as they are really expensive to ship into my country (as in almost 100 times more expensive than a TL07x opamp).
Therefore, I’m designing some alternative circuits using only opamps, transistors and some other components easily found on aliexpress or similar so it should be equally available for everyone across the globe
I thought of making a thread with these designs in order to have them more easily accessible.
I’ve already made an ADSR, based on renè’s 555 design, but using one TL074 instead (can use one TL072 if you don’t care about buffering), to have a bit more control over the parameters. only thing still missing is looping capability, I have an idea on how to implement it, but any suggestion would be appreciated (and I wanted to document it as it is now first, will post the schematic as soon as it is done!).
An early simulation result, with red being the gate:
I also have a Pulse/Sawtooth/Triangle/Sine wave VCO coming up, centered around the CA40106, as I can get all 4 waveforms simultaneously (and also have 6 voices in the same chip!), and also an opamp based one, with a couple variants. These are still a bit early on so I won’t post them for now (still waiting for the parts to arrive so I can actually test the circuit instead of just trusting the simulations).
VCAs are a bit tricky to achieve without matched transistors (and I want a bit less distortion than a ohmic region JFET gives), but there are apparently some ways to achieve with opamps that I’m going to try eventually.
Most definetelly. I’m looking into some papers about signal multiplication though, and there are also some designs you can achieve good (or at least decent) using unmatched transistors, which I’m going to try to look into.
I’m trying to avoid having to match transistors manually though, as it can get costly (and boring!) too
Ideally I want cheap and effective circuits, even if it means they get a little larger or complex (as long as this doesn’t make them more expensive, of course haha!)
Here is a somewhat cleaned up schematic!
Still missing the looper function(though I have this rough idea of switching the inputs to a comparator that will send a trigger everytime the output voltage is close to 0 or sustain level).
Fixed up some things since this morning to have more accurate sustain level tracking and some other minor things…
The two diodes (D8 D9) in the positive feedback on U2 were to provide a voltage drop so the fliflop could be reset. But when I simulated the circuit it was not exactly enough. You could remove them if you want, but it will require a higher value for R22.
very interesting… the 2n3904s usually go for 0.80 to 1BRL (which is still cheap, compared to others ive seen) but buying 100 would not be reasonable. Luckly, they are extremelly cheap on aliexpress(100 for 4BRL ahaha) so I just bought 100 3904s and 100 3906s to experiment now
thanks for the tip!
(edited for clarity regarding which currency I used)
I know, it was 4-5 bucks in my currency, which is a whopping 0,72USD for 100 2n3904s
it looks like a pretty good deal for me hahaha
Also, tayda/mouser/non-chinese sellers usually have a prohibitivelly expensive (albeit quick) shipping cost, starting at around 37 USD (for comparison, the minimum wage around here is just below 200USD, so paying almost 25% of a month’s wage in shipping cost alone feels really bad hahaha)
You might not need R8 as the attack charge current is already limited by R2.
Edit: more comments:
I don’t know if it’s just to get short simulations, but in the real circuit you will need a much bigger C5 capacitor and maybe also bigger potentiometers.
Could you explain what purpose does C6 serve in your SR flip-flop circuit?
Matching is pretty easy, just time consuming. I spent a nice couple of evenings matching transistors for a couple of filters a week or so ago. I used the circuit by Ian Fritz as shared by Kassutronics here: https://kassu2000.blogspot.com/2015/10/transistor-matching.html
But I messed up the breadboard the first time and didn’t catch it until after I wasted an evening. So the next night I drew up a simple PCB so I’d have both the NPN and PNP versions ready at any time:
Unfortunately I picked a less than helpful for me pinout for the transistors…(and I only had 4 pin headers on hand so I just pulled one pin out and shoved them in.) But that worked out kind of nice since I used some extension cables to adjust the pinout and so I could position the transistors in front of a fan to help them stabilize temp quicker:
Funny thing…both types of transistors I had to match the first two I pulled out were a perfect 0.00mv match Of course then it took a few more before finding any other matches! I just left the first transistor in and kept swapping them out - then built piles sorted by mv. Once I had a pile with enough in it I swapped to one of them and re-matched them to make sure they were really matched and I hadn’t been sloppy or rushed.
Biggest problem I had was every time my AC came on it threw off my readings. Thankfully I think we’re FINALLY past AC weather now
I included some gerbers which I test-uploaded to JLC and they didn’t flag or reject…but I just etched my own so no guarantees on the gerbers being 100% ready to use.
yes, the values for C5 and the pots are for shorter sims, the values listed on the sch are not necessarily the ideal for the real deal
Good catch! I absolutelly meant decay and not delay! haha just a brain fart that slipped through
also, the C6 was to avoid some circuit instability that I saw sometimes in the simulations (possibly due to the model I was using at the time) that made the FF latch imeediatelly after resetting… forcing the R pulse to ramp up a little slower fixed it. I do believe it could be replaced by a 10n or even 1n cap
Also yes, R8 is not necessary, its from the remains of the previous, buffered version, hehe
No problem. I’ve been meaning to make a post about it but was being lazy about making the gerbers. Seeing you had a bunch of matching ahead of you I finally decided to get it done and share
I’ve never done this myself. I’m thinking that rather than setting up one transistor, waiting for it to stabilize, measuring it, and then repeating, could one do something like this: Set up 20 or so transistors along the length of a solderless breadboard, wait for them to stabilize, then measure each one. It’d be quicker if that worked. Of course if you depend on a small fan it might not be able to act equally on a whole line of transistors, but maybe there’s a workaround for that. Perhaps just a line of several fans.
That could work - just use a m-m jumper instead of a m-f so it could plug into the breadboard.
You’d still want to give me a few seconds to stabilize after putting power to them. But it would probably be quicker than waiting for them to stabilize after touching them.
It’s still not THAT time consuming. I was able to get through about 25-30 transistors in an hour and a half or so while doing other tasks. I’d just pop one in…work on something then a few minutes later check on it and if the value wasn’t changing bin it and move on to the next.
.
Of course then it took a few more before finding any other matches! I just left the first transistor in and kept swapping them out - then built piles sorted by mv. Once I had a pile with enough in it I swapped to one of them and re-matched them to make sure they were really matched and I hadn’t been sloppy or rushed.