Bipolar regulated PSU latching?

[Guille]

Hi, I'm trying to fix a regulated bipolar power supply from a Behringer - Ultra curve pro deq2496. The problem is the diode between -15V rail and GND gets really hot and burns up. A couple of things to consider:

• This happens with the supply unloaded
• Visual inspection didn't reveal any other problems ( images here: https://imgur.com/a/DQJyqfU)
• Both the L7915 and diode were replaced and the problem persists
• The circuit ran ok once after replacement (measured 15V at both outputs but did it in a hurry and was not sure on the polarity. i.e.  15V or -15V? idk just 15...    ). The diode then burnt up at a later power cycle
• Double checked that the diode is correctly polarized

Sadly I couldn't find any schematic on the web for the PSU and their support team ignored our request for one. Closest thing I could find is the  application suggestion stated in L79s datasheet https://www.mouser.es/datasheet/2/389/l79-1849327.pdf which apparently Behringer engineers followed closely.



The diode is reversly polarized and so it shouldn't be draining current under normal operation, however the schematic notes that they are there to protect "against potential latch-up problems". What does this mean? I tried reading on latch up but it's not clear to me how it works and how it applies to this case in particular. Does it cause the -15V rail to go positive? How can does this happen in this particular case? I read many general explanations for latch up but I still don't understand the problem or how would the diodes mitigate it.


If this is not a latch up problem what else could it be?

[fzabkar]

See page 14 of this document:

https://web.archive.org/web/20121119221950if_/http://www.harrisonelectronics.co.uk:80/datasheets/78M05.pdf

See this thread:

https://www.badcaps.net/forum/showthread.php?t=45001

[Kim Christensen]

There is a partial schematic of the power supply on this site.
Notice that the pinouts (input, output, & ground) of the 7815 vs the 7915 are different. So make sure you've got that diode the right way around.

The diode then burnt up at a later power cycle

Did it burn out that time when the power supply was unloaded or when powering the "Ultra curve"?

[Guille]

Hi, yes we measured at the correct pins, good point. We always tested the supply unloaded.

The whole story is the machine suffered a short between the heatsink (tied to the regulators input pin) and ground. The diode started to smoke so my collegue changed the L79 and diode, however he soldered the diode backwards, tried it and no luck, I believe the new diode was probably stressed at this point. I pointed out the fault and resoldered the same diode correctly this time. Switched on the supply and everything measured correctly plus the diode was cool to the touch. Later on we swithced it on again and the diode immediately got hot and burned. Same exact conditions though, the supply was never loaded for any of the tests.

We are in process of changing the diode and testing again.

[wraper]

Ensure you get vregs from reputable source. 90% of 78xx and 79xx on Ebay and alixepress are counterfeit not up to spec junk.

[fzabkar]

Try a 1A Schottky diode instead of the 1N4001. The lower Vf might circumvent the latch-up problem. That seemed to work in that badcaps.net thread. That said, the other circuit was not unloaded.