What?
While doing some work on a device, my Siglent power supply suddenly rebooted. I just ignored it the first time, but the duration between the individual reboots became shorter and shorter, ending in a dead instrument.
Switching it off and waiting a couple of minutes helped, but then this process just started again. And it was independent of the current which was provided on the channels – it failed even while sitting idle.
So time for a side project fix…the thing you need most, when you are busy with something else.
First ideas and searching
The behaviour smells like a thermal problem, so my first intuition was to look for some hottie components on the circuit. Unfortunately there a way too many hot components on the three boards of the supply.
There are several 79xx and 78xx voltage regulators, all without heatsinks and also other components on the main control PCB. Since no schematics are available (the service manual does not contain them) I had to partially reengineer at least the function blocks.
The three channels of the supply are fully isolated and floating (you can see the optocoupler banks in the picture). This is also the reason for the various support-voltage regulators on the board, since every channel is independently sourced. For the MCU another separate transformer winding and voltage regulator is providing the typical 5V and derived from this another regulator for 3.2V.
Measuring these voltages while the supply was failing, it turned out, that the 3.2V are slowly going down….because the 5V were slowly going down, but the transformer winding delivered the correct voltage.
Fault identification
After some poking around, I suspected the bridge rectifier (under the heatsink in the picture) or one of the three caps in the direct neighbourhood since this was the hottest area on the whole board.
After verifying and searching around a bit on the web, I found some evidence that the BR is either a bad batch or not sufficiently dimensioned for the current and often fails. Siglent changed the rectifier in subsequent versions of the supply. Find all the details in this blog post.
Fix it
Since just replacing the rectifier, which was indeed bad, will not help in the long run, I used an off-PCB-rectifier and mounted it somewhere in the supply.
This was inspired by a YouTube video I also found. However I did not change all the components as proposed by Thomas on the video.
I also checked the three caps – the two big ones were still fine, but the smaller one was already loosing capacity, so I preventively changed that one as well.
Conclusion
This was an easy fix and some burn in testing proofed that the fault is really gone.
I was a little bit disappointed that one of my newest instruments (which was unused when I bought it) failed….but since it works again…it’s okay. Now I can continue with what I really wanted to do in the first place.
