EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: philpem on July 09, 2023, 08:49:30 am
-
Alas, I did a silly thing with my Amrel EL-1132. I was discharge testing a pair of 100Ah lead-acid batteries and wasn't paying attention to the SOA limits on the EL-1123. It managed a 5A discharge, a 10A discharge, ... then I mistyped and entered a 40A discharge.
End result, a kilowatt of energy rammed its way through the power stage and blew up at least one transistor with a loud bang and a flash.
I've cleaned up the PCB and fitted a new transistor, but it's still playing possum: there's a load
I'm slowly digging through the power stage and trying to figure it out but in the meantime -- does anyone have manuals or schematics for these?
Notes from past repairs would also be nice to have.
Cheers
Phil
-
I've finally got this one fixed. Problems were, roughly in order ...
- All but one of the power transistors were blown (IRFP250s, so not terribly expensive)
- All of the zero-ohm links protecting the power transistors were blown
The transistors mostly failed dead-short (as MOSFETs are apt to do). They're easy to replace but the PCB quality is awful, the tracks lift at the slightest provocation. One of the transistors in my unit had been factory-replaced and they'd apparently had the same problem.
The "fuses" are tricky to find replacements for. They're zero-ohm links, but you need decent ones. I expect you could use actual N-amp axial fuses instead, but Amrel for some reason didn't do that (maybe they considered it but changed their minds later on)
I tried cheap (okay, Amazon-special Chinesium) metal film quarter-watt zero-ohm ones, and they all failed when I tried to do the high-range constant current calibration. Prior to that the load current was quite unstable. In hindsight the stupidly thin legs should have been a hint that they weren't going to work.
After that I picked up a pack of 50 quarter-watt "wire-link style" style zero-ohm resistors which look like carbon-film resistors (just like the originals) and work fine.
Voltage calibration is still dead on. Current calibration is (as usual) tricky on these. A cheap Chinese 5V 60A switch-mode tin-box power supply is fine. A lab power supply and a China-special boost converter (up to 60V output) will get the voltage cal done (needs 60V at about an amp). Otherwise if you have two identical 30V bench supplies or a dual (ideally tracking) PSU which can have its outputs wired in series, that'll work too.
If you want to improve the design I guess you could fit ~10A axial wire-ended fuses in place of the resistors. If a transistor dead-shorts, that should blow the fuse, then the rest should cascade as the remaining transistors take up the load.
The power stage has eight transistors in total, against the 60A maximum load means about 7.5A a piece and 5V = 37.5W dissipated each, with a 5V/60A load.
There's no overvoltage, overcurrent or (critically) overpower protection on these... the later Amrel units have software protections in place to stop you doing stupid things.
In lieu of software protection, I've put an SOA graph label on the top of mine, and two very obvious yellow SOA warning labels with guideline max voltage/current combinations. Hopefully "future me" will pick more reasonable settings!
And now onto doing a C/20 load test on my PV array batteries.
-
Free gift for anyone working on one of these ... dumps of all the ROMs and GALs. Turns out the GALs were unprotected - or at least they didn't read back as all-1's.
The I/O board is the small plug-in board with all the ribbon cables plugged into it and the single small edge connector.
The Analog board is the big plug-in board with two huge edge connectors.
Analog board processor is a Temic 80C31
I/O board processor is a Zilog Z180.
Looks like most of the UI stuff happens in the I/O Board processor. The Analog board processor ROM only contains a couple of hundred bytes of code, so I guess it's mostly for analog interfacing, maybe some power-stage regulation.