I'm on a bit of a roll as far as fixing stuff goes.
The latest bit of gear to succumb to my endeavours is an old D-Link DE-812TP Ethernet hub rescued from the scap pile at work and which had failed in industrial use. It was being discarded as being "beyond economic repair". Had the unit been somewhat more youthful it may well have gone back to the company's repair shop but as it dated from 1997 it was regarded as "consumable" and should have been destined for recycling. In a previous life (or so it seems) it could well have been down to me to repair the hub myself!
Seeing an opportunity to acquire a potentially useful piece of equipment at no cost, and, frankly, fancying the challange, I opted to have a quick look at it, as having seen the fault symptoms I thought there was a reasonable chance of making a repair.
The fault symptoms were rapidly flashing LED indicators (ALL of them were flashing) which strongly suggested a power supply fault. There appeared to be no blown fuses either, a good sign.
Opening up the unit I discovered a small switched-mode power supply on a separate sub-board bearing a strong resemblance to a standard computer type supply, albeit a miniature version, without cooling fans and the like. The general arrangement looked very familiar so confidence was high.
The first step was to prove that the main motherboard was OK, so I disconnected the built-in supply and provided a temporary 5V feed to the motherboard which immediately sprang into life with "normal" indications on the LEDs. I was now more confident than ever that the PSU was the culprit so out it came for a closer examination.
Not having the circuit I had to make rough guesses about where and what the various circuit sections were, but I quickly identified what was probably the main capacitor filter section for the +5V supply and decided to test the reservoir capacitors with my Peak Electronics ESR60 Capacitor Analyser. Years of experience of repairing electronic circuits teaches one to suspect electrolytic capacitors before anything else! Strange readings were obtained with the capacitors in circuit (this is often due to other components connected to and across the capacitors and doesn't necessarily mean that there is a problem), so I unsoldered and removed them from the PCB in order to check them out of circuit.
See the photo below.
Once capacitor came out minus one of its legs (not a good sign!) and the other measured "Low Capacitance" on the meter. They were both branded as 1000uF 16V units. There was possibly some evidence on the PCB in the form of a small light brown stain resembling a drop of dried out spilt coffee that there may have been some electrolyte leakage, but it was clear that both of these capacitors were u/s. As you can see from the photo they were also branded as 105 deg C capacitors so it would have been reasonable to expect a long life from them. Still the hub had been in service 24/7 for around 12 years which I suppose is reasonable. Around 100,000 hours of continuous operation before failing. As the two capacitors are connected in parallel in this circuit one can ponder that perhaps they had been dying gracefully for some considerable time, but one never knows.
Happily, the capacitors were replaced with new units purchased cheaply on EBay, and the PSU and its parent hub are now back up and running, seemingly as good as new.