12V Power supply gets a revamp, part 1...

Doesn't time fly... I have been doing stuff at home... just not as much as I'd like!

I decided it was time to get a NAS and get a proper storage solution sorted out for 2013... so ended up going for one of the Synology boxes in January.  I wanted highish performance (>50MB/sec) and lowish power consumption (<10W for the CPU idle) in the same box... I was hoping to hold out for the Atom SoC based stuff but it seemed to be taking forever to materialise.  Of course, now they have been announced.

While it is relatively power efficient (I measure it at about 9W before the HD spins up, at which point it sits at around 14W) for the speed of it, I was concerned out what would happen in the advent of a power cut.  They are not renowned for coping well in such a scenario.

So a UPS seemed like a good idea.  An off the shelf inverter-style one would be not very efficient, so wanted to go for something that could output 12V cleanly and easily.  Something that uses a cheap 12V lead acid battery and is designed for efficiency.

The first approach was to try modifying the switching supply that the Synology came with to output 13.8V.  It's a standard block power supply, efficiency rating V so not bad at all in terms of waste power.

After finally breaking my way into the case (they really don't want you to open these things!) and removing some extra metal shielding, I was greeted with this...

 

I took a look at the electrolytic caps inside and didn't like what I saw.  These might be a perfectly reputable brand (indeed, I was able to find a datasheet) as far as Korean caps go but felt more comfortable replacing them with more familiar Panasonic units of a 25V rating... particularly as I intended to up the voltage output.

 

A look at the underside reveals that the supply is based on an LD7578 PWM controller.  The output section is based around a neat UM603 which contains a dual op-amp and a 2.5V reference wired to one of them.  Logically, there had to be a voltage divider near the chip... indeed there was.  As an added bonus there was an unpopulated part which had pins in parallel with the resistor in question, so I could do the mod quite neatly - you'll see a diagonally placed resistor near the main chip in the picture below.

 I now had a perfect 13.8V output.  The problem was that the supply was now out of kilter.  While it will happily provide 13.8V at very light load, any significant loading of the supply was triggering protection circuitry and the supply would go into "hiccup" mode, which was audible through the iron.

I'd had enough of this hassle before with switched mode supplies, so decided to go back to the original 12V supply design with a linear first stage, though I went a bit over the top...