2

First Time for me posting here. Recently I decided to perform a recap on an 20 y/o old SMPS, which is used in a digital mixer. The recap went without any complications. I followed the guide I found here: Recapping DDX3126 PSU I also used genuine low ESR Panasonic capacitors, just as in the guide.

There are also schematics provided for the whole Mixer including the PSU-Schematics

Also here is the datasheet of the Regulator, so you wont have to search :D

But 1-2 seconds after powering it up for the first time after the recap, the first of the three SMPS regulators exploded (The SMPS has three separate regulators). I desoldered that broken SMPS regulator, and tried again. Then the same thing happened with the second SMPS regulator too. Curiously the third SMPS regulator survived at first without any problems. The output voltages from that one were completely fine too.

I ordered replacements for the broken regulators. Soldered them in, and tested with 80 VDC applied to the input, which is currently the only method available for me to perform measurements, without risking being electrocuted with rectified 230 VAC. It appeared to work completely fine with 80 VDC input, even after 20 minutes. Then I re-adjusted the output voltage, because the output voltages were little too high after the recap.

Then I reassembled the psu, checked for any shorted solder joints, and solder splatter. Spotting nothing unusual. And performed the first test with 230VAC and the freshly replaced Regulators. Seemed also to work fine. Reassembled the whole mixer again, and powered it up. The mixer booted fine, and worked for a couple minutes. Then after re powering it again, the third regulator exploded. It was the regulator I didn't need to replace yet. At that point I thought, that it might have to do something with the little bit dirty solder job, since these Regulators were already replace once, by the previous owner. So I replaced that one too.

Again reassembled the mixer, powered it on with 230 VAC, and after 15 minutes it exploded again. I also made sure, that the body of the regulator is not shorted to the heatsink, by using PET and non conductive thermal compound.

These are the results of my measurments: Vcc at pin #3 is well within specification

Components in the soft start circuit are OK.

Feedback optocoupler and circuit is also OK.

Measured the the drain (pin #1) during operation with oscilloscope in respect to GND (pin #2) with 100 V/div but I cant remember the timing (See figure below: Lower channel)

Top channel: Secondary side; Lower channel: Drain (pin #1) Top Channel: Secondary Side ; Lower Channel: Drain(Pin #1)

Did anybody here have similar problems, or has some suggestions, on what to do? Because I unfortunately have no Idea, what causes this explosive failure :(

EDIT: These are the defect "exploded" ICS. I measured each pins with the component tester. Result is, that there is no difference to the new working ICS. My assumption is, that since the "explosion" always took place around the screw-hole, but always in a different spot, that the thermal compound eventually broke down, and initiated an ark, which would just bridge out the MOSFET, thus creating a short circuit, where the regulator would switch with its 70 kHz. After thorough cleaning of the damaged ICs, I realized, that the dammage is only done on the heatsink, and does not even penetrate into the silicon. Defect ICS Defect ICS from back

winny
  • 13,064
  • 6
  • 46
  • 63
anatol
  • 61
  • 5
  • I would recommend you use an isolation transformer 1:1 so that you can safely probe the converter when powered from the 230-V grid. Using a dc supply is also the way to go as you did at 80 V. The waveforms you show seem ok to me, what is the 3rd regulator you refer to? Check at 80 V dc, if the supply regulates ok, what is rectified voltage value biasing these regulators, making sure you don't exceed their max ratings. Does the primary-side switcher survives when the sec. side blows up? – Verbal Kint Jul 30 '23 at 15:07
  • "Does the primary-side switcher survives when the sec. side blows up?" - The secondary side never blew up or anything. "what is the 3rd regulator you refer to?" - There are three seperate Regulators (KA1M0880) in this PSU (fed by the same "bus"). At 80VDC every output Rail was regulated correctly – anatol Jul 30 '23 at 15:47
  • Yesterday I also ordered the components, to built the dummy load, specified in the schematics. Unfortunately I cant measure the PSU if its built into the mixer. And also too risky to fry the mixer while measuring :D Maybe this will provide something useful – anatol Jul 30 '23 at 15:57
  • I’d be replacing C6,26 and 48. When these caps dry out, you get ripple on the switcher Vcc which upsets the mosfet drive then kaboom. These caps are small and cheap. Depending on the age, quality of the caps ( brands such as Nippon chemicon, rubicon, nichicon are good) etc. it might just be worthwhile replacing all psu caps with quality ones. No use recapping a psu with crappy caps. – Kartman Jul 31 '23 at 13:09
  • @Kartman I already replaced those caps (as well as all other electrolytic ones)during the recap, with genuine Panasonic Low ESR Capacitors. Panasonic caps also have a good reputation, right? But nevertheless, I will measure, wether or not the Voltage ripples. Nice hint, thank you. This could also explain, why it works with DC on the input, since there wont be any 50hz ripple. I am not at home currently, but I hope i didnt accidentally switch up the C27(1uF for softStart) and C26(47uF for Vcc) caps, since they are near each other. – anatol Aug 01 '23 at 10:50
  • Nope. Sadly thats not the solution. The caps were placed correctly. Is it possible that it is something from the secondary side? In theory even a short circuit on the output shouldnt cause a smps to fail explosively, right? – anatol Aug 01 '23 at 19:23
  • 1
    I am 99% sure I found the problem (See the EDIT in the original Post). I thought that MX-4 ,which I had laying around, was not electrically conductive, which it in theory it should be. I dont know if thats because of its age or what. I will do some research, on how to properly isolate the body from the heatsink. If anyone got some quick tips, I'd highly appreciate it. Nevertheless thank you all for helping. – anatol Aug 01 '23 at 20:17
  • I’d normally use ‘SILPAD’ which is a thin, grey flexible film widely used for this application. No grease needed. You can buy it pre-cut for various package types. You want the TOP-3 size methinks. – Kartman Aug 01 '23 at 22:18
  • Yes, thank you. I ordered replacements. It came with these Silicone Pads from factory, but probably the previous repair-man was at it, and added thermal compound to the Silicone Pads, and so did I, but mine would break down under that voltage apparently. I wasnt sure if it needed thermal compound, so I repeated what I saw. But do you agree with my thesis from these images? – anatol Aug 01 '23 at 23:07
  • 1
    Silpad solved the issue! As of now the mixer worked reliably for 24h. Thanks so much for everybodys input on this topic. – anatol Aug 06 '23 at 19:41

1 Answers1

4

The cause for the short circuit was a thermal paste claiming to be not conductive. Unfortunately the dielectric strength of the thermal paste was not enough. It broke down eventually, creating a conductive path around the PET-foil, where the screw hole was.

Using silicone pads as thermal compound was the solution, because of its high dielectric strength.

Thanks to everybodys input on this topic.

winny
  • 13,064
  • 6
  • 46
  • 63
anatol
  • 61
  • 5
  • Thanks for updating. I'm really surprised that the MX-4 thermal paste was creating the short-circuits since in their website they claim it's non conductive – Ken Grimes Aug 07 '23 at 10:53
  • 1
    I think everything can be conductive after a certain voltage. The Property is called dielectric strength. Air itself is not conductive, but with a dielectric strength of 3kv/mm it will get conductive at a distance of 1mm at 3kv. And MX4 is typically used within Devices with voltages lower than 24V, whereas in the SMPS the voltages exceed 325V – anatol Aug 08 '23 at 13:55