I thought switchmode supplies were pretty efficient? <clicky> Yes, linear supplies are 30-40% and switchmode can be 60-70% with a good design reaching 95% efficiency. Since these are little <cough> mass-produced-in-China </cough> supplies, I'd reckon around 70%ish (the one powering my OSD is stone cold).
For the power rating of the PSU, good guess. Here's a photo of the label:
(hosted on imageshack because somebody else's backup process has run my website over quota (long explanation, don't bother asking); it'll be sorted out soon but soon isn't right now...)
I like the part where it says a switchmode could fail catastrophically (as much as near dumping mains onto the power output). I've never in my life experienced a switchmode failure that was any more dramatic than exploding a 20-year-old capacitor.
On the other hand, I had a torrodial (sp?) linear supply in an MDFS server fail. Of the immediate aftermath, the fuse in the MDFS's plug (5A) had exploded. The fuse in the first extension lead (5A) had exploded. The fuse in the second extension lead (13A) had exploded. The ring main fuse (something like 30A) was toast, so much so that there was molten gunk and lots of smoke inside the fusebox. The master supply fuse (something like 60A?) survived.
Of the analysis... The MDFS was toast. The only thing that even started up was the SCSI harddisc. I don't know if it actually *works*, it's a weird format so while it sort-of talks to my SCSI card, I am not sure if it is returning correct data or gibberish. It *looked* correct, but it's moot now - there's not much call for a ~200Mb SCSI drive these days.
The line drivers in the BBC Micro were okay as it wasn't plugged in at the time. The SJ network bridge had three sets of line drivers, now two are dead. The A310's network card was junk. The A3000's likewise, and the A5000 too. Given that the only viable explanation is something went terribly terribly wrong and bridged live and earth, this means that 240VAC hit the earthing and as such it is probably lucky that the network cards took the brunt of the blast and that it didn't wipe out all of the computers. The network transient suppressor did exactly zero. It would protect me for a high voltage on one of the data/clock lines, but not for on earth as that's not ever supposed to happen.
There ya go, lovely story to end the week.
For those who have no idea what I'm talking about - we Acornites were running LANs with user level access and printer sharing back in the AppleII era (before the rise of the x86 PC), on similar hardware (6502s and Z80s). Take a peek at http://www.heyrick.co.uk/econet/
for some info. It wasn't really until Windows98 that desktop PCs caught up with the sort of stuff we'd taken for granted for decades. But then again, it wasn't until XP's ClearType that Microsoft managed to provide the level of anti-aliased on-screen text that we had sussed in '89 thanks to the RISC OS FontManager.
Here's an example lifted from a version of RISC OS dating to 1996; and remember this would be around the same time as Windows95 OSR2 and the very very ugly single-pixel-wide Arial font everywhere, looking really no better than did Windows 3.11...
Okay, that's my little advocacy spiel over. On final thing, RISC OS is being readied for the RaspberryPi. It's an old outdated OS (don't ask about inter-process memory protection!) but it's still kicking and it's the friendliest and most accessible system you'll ever come across. https://www.riscosopen.org/
PS: Connection? The OSD runs on ARM. RISC OS's core is pretty much pure ARM code. And the entire OS as a ROM image runs to 5-6Mb. This is why Arizona/OSDng overrunning a 16Mb Flash scares me. It's a real WTF?
moment. Of course, with the emergence of Android based phones and tablets, some people are saying "wouldn't it be interesting to see if we can have an ARM powered desktop computer". Duh. Where d'you think the ARM came from in the first place!?