Re: [Sheflug] Power fluctuations...

[Alex Hudson <nospam [at] york.ac.uk>]

Hi Craig.

Hmm. I would imagine that 200W should be enough, you're not running vast
amounts of gear of it. Generally, power supplies either work or they don't
- if there's too much load, or not enough load, they generally don't start
up. I'm pretty sure you'd know about it, anyway! One thing to try is to
put an RCD in circuit somewhere, see how much juice it's drawing. 

I would lay more money on it being heat related, to be honest. 3 hard
drives - they're bound to warm up, as is the gfx card. If you put your
hand near the fan on the psu, how warm is the air coming out? It should be
toasty (after leaving the machine running for a while), but not hot (this
is kinda hard to do, given the human body is completely unable to measure
temperature to any degree whatsoever ;). Anyway, random graphics / etc.
occurances are often due to heat: either the CPU getting a bit roasty and
getting a few bits wrong while it mops its brow, or the GPU doing the
same. Or both ;). Of course, it depends when these occur - if they occur
right from turn-on, it's not heat.

One other thing it could be, that's PSU-related but not power related, is
the spec of the PSU. It could be that the tolerance has moved sufficiently
such that the logic rails on the motherboard are a bit closer. This would
lead to random data rubbish also. Or, possibly some dried-up caps: the
smoothing on the power supply is essential: when large loads are applied,
the caps are there to just bump the voltages back up again. If the caps
are dry, they won't be smoothing and you'll be seeing the effects of
fluctuating power.

So, three reasons: two relate to a bad PSU, one to heat. If you measure
the heat of the system, and it's okay (or you get errors all the time),
it's probably the PSU.

Cheers,

Alex.

On Tue, 20 Jun 2000, Craig Andrews wrote:

> Hi all,
> 
> I have been suffering for a while with a rather annoying problem. Well,
> two actually.
> 
> First, whenever the hard drive gets a good thrashing (any drive, in fact,
> out of 3) the mouse cursor takes on a life of it's own. It can be annoying
> in Gimp to have a screen-long line just because of a flushed buffer! I
> think this is (partly) to do with either a) the PS/2 interface, or b) the
> IMPS/2 driver. Note that ALL IMPS/2 mice do the same thing. Not tried any
> other protocols.
> 
> Second, my RivaTNT2 seems to lock up whenever it has a lot to think
> about. Particularly involving alpha blending and the like. I thought that
> this was to do with the alpha glx driver I have, but the release Windoze
> driver does it too. It doesn't do it straight away, mind. It waits. In
> Tux- A quest for herring, it only locks when I press Space for the menu,
> and the transparent text appears. In NFS4 in Windows, it seems to be when
> water or lens flare is present.
> 
> What I put both these down to is a lack of power. In my machine, I have:
> 
> K6-2/500 w/ 96Mb EDO Ram
> NVidia RivaTNT2 32Mb AGP 2x
> SB Awe 64
> Modem (internal)
> CD-Rom and Floppy
> 15, 8.4 and 3.2 Gb Hard disk drives
> Genius optical mouse (IMPS/2 driver)
> 200W (old, cheap) PSU
> 
> What does anyone reckon to my theory, before I splash out on a more
> powerful PSU? (The old one cost 16 quid, with a case!)
> 
> Craig Andrews
> craig [at] fishbot.free-online.co.uk
> 
> - - - - - - - - - - - - - - - - - -
> Life would be so much easier if we 
> could just look at the source code.
> - - - - - - - - - - - - - - - - - -
> 
> ---------------------------------------------------------------------
> Sheffield Linux User's Group - http://www.sheflug.co.uk
> To unsubscribe from this list send mail to
> - <sheflug-request [at] vuw.ac.nz> - with the word 
>  "unsubscribe" in the body of the message. 
> 
>   GNU the choice of a complete generation.
> 
> 
> 

---------------------------------------------------------------------
Sheffield Linux User's Group - http://www.sheflug.co.uk
To unsubscribe from this list send mail to
- <sheflug-request [at] vuw.ac.nz> - with the word 
 "unsubscribe" in the body of the message. 

  GNU the choice of a complete generation.