I built and installed the AI Synthesis eurorack PSU and distribution board into my DIY Kosmo case. It is being powered by a Jameco AC-to-AC 12v 1000mA wall-wart.
Links:
I’m finding that often it will power on with only the -12v rail active. If this happens I can unplug it, wait for the LEDs to show the caps discharge, plug it back in … and it will work. At least for a while. Other times It’ll power up fine, but after 10 minutes or so it will die and only the -12v LED will remain lit.
Once, when it got into this “bad” state, I measured the voltages on the two regulators.
I could then power cycle it and everything came up with correct voltages again.
The left regulator is a 7812T and the right is a LM7912.
I swapped out the 7812T as well as the wall wart - neither had any effect on my issue. This is such a basic circuit, I’m sort of at a loss as to what could be causing this issue. I believe that the problem happens more frequently the more modules that I plug in, but I have seen it occur with only 2-3 1222 VCOs connected. I have gone through removing all modules, plugging them back in, swapping them around when a failure occurs - and am fairly confident that I’m not dealing with a bad module.
I’m at a loss. Looking for advice. Thanks in advance.
In the past I’ve seen that weird startup behaviour happen when the voltage on the output of the 7812 is above the input (caused by lingering charge on capacitors on the output side), or below the ground (maybe caused by the -12V rail powering up faster than the +12, idk) when the regulator gets powered up. D3 and D4 are supposed to prevent that though, maybe check that they measure good and are soldered properly. You could also try replacing them with schottkies to see if maybe that helps.
As for why it sometimes shuts off after 10 minutes, the only reason I can think of is that the 7812 overheats and shuts down. Have you measured the current drawn by your modules?
already ordered. I’m sort of surprised they didn’t come in the kit - but considering Abe @ AI Synthesis loves to tell me how “I’ve sold hundreds of these over the years and nobody has ever reported a problem” … I’ll just assume I just have bad electricity here and deal with it
I might end up relegating this PSU to bench-supply duties and put a meanwell in my case anyway.
I’ve never used one of the AI but it looks to be a pretty standard wall wart supply design similar to the FC Power and the MFOS one which I have used (with heat sinks). Aside from the positive regulator startup problem which the Schottkys fixed they’ve been fine.
Well, I have personally reported the same problem and got a similar response… surely wouldn’t be too hard for them to replace one diode for another in their kits?
that’s hilarious. He said he was updating his “knowledge base” with this info. Ultimately he REALLY just wanted me to send it in and pay the difference between the kit and a tested/working unit so I’d quit bothering him I guess. Glad I persisted to try and repair it myself since it wouldn’t have solved anything to send it in.
As I said it’s seemingly very similar to the MFOS Wall Wart Supply, and the Frequency Central FC Power and their others. Maybe other vendors have similar ones as well. They all have rectifier diodes like 1N4004 to ground on both the positive and negative regulators. As said in my video, from what I’ve read it’s not at all clear those diodes are needed to protect the regulators even though that’s what MFOS claims, based on my reading of the TI technical note. It may be cargo cult design: People put those diodes there because previous designers put those diodes there.
Anyway, presumably there are thousands of people who’ve used these designs, and MFOS and FC as well as AI have never changed them. (Yes, I know Ray Wilson is no longer around to change his.) The TI tech note that mentions using a Schottky to fix positive regulator startup problems goes back a long way, in fact Wilson refers to it in his writeup to justify the rectifier diodes to ground (which I don’t see that it does). Maybe these designs work just fine for a lot of people; it probably depends on things like what modules you are powering. But from discussions here and on modwiggler it’s clear a fair number of people run into this problem. So it’s puzzling the rectifier diodes are still in all those designs.
Added: Anyway I’ve bucked the trend in my adaptation of the MFOS design.
I use the PCB contributed by Arthur Saunier that has essentially the same design as far as I can recall, but never had any issues, despite the rectifier diodes throughout. One difference I see is that both rails come with indicator LEDs, which solves the issue of the -12V needing a load to start. Perhaps, somehow, using indicator LEDs on each rail solves the issue with the +12V as well?
By the way, you definitely need the heatsinks if you’re going to draw current from an entire case of modules. I never run TO-220 and TO-218 transistor packages (typically: regulators, amplifiers, motor MOSFET) without a heatsink. I often salvage them from old PSUs, or make them out of the heatsinks found in computer motherboards or even odd pieces of aluminium (curtain rails, etc) so I always have something available, and that’s better than nothing.
If the negative side comes up faster, it pulls the positive side down and the positive regulator won’t start up. My guess is that depends either on the modules that are powered by it, or just random variations in the regulators.
As a data point, the ones I’ve used (more recently at least) are STMicroelectronics, bought from Tayda.
I have two of those and used them over the last year with different module configurations but never encountered this problem. The discussion in this forum also suggests that different brand 7812s might be more prone to this problem.
Not sure how it isn’t. Felt pretty dismissive from my point of view. I do appreciate your policy and how in most cases it offers builders a solution - but you basically told me what I was seeing with my own eyes was so unlikely as to be near impossible. The reality is that if I had sent you my supply as you suggested I would be out a bit of money now that would not have been necessary to spend and I’d also be no closer to a solution. It also stung to hear that someone else in this thread had at one point spoken to you about the same issue in the past.
