Re: fake PC supplies

On Tue, 24 May 2005 16:21:48 -0400, Mike Monett <no@xxxxxxxx> wrote:


> Can you suggest a simple way to test the ov protection? If I
> understand correctly, most power supplies require a load to operate
> properly. So if you tried activating the ov protection by connecting
> an external supply to raise the voltage, the power supply may do
> wierd things.
>
> Is there any other way to see how well the ov protection works?

You can usually back-drive lower powered quasi-regulated auxiliaries
as you describe. The test is also partially effective in testing an
OVP circuit that is an independant thyristor crowbar. It may damage
simpler linear post-regulator circuits, however.

Some redundant devices are designed to be immune to
externally-generated faults. OVP may not be provided on some outputs.

Do not be surprised if the OVP thresholds are set higher than you
might feel comfortable with - read the spec before testing. The test
is incomplete if the unit is not demonstrated recover, on the removal
and re-application of input power.

If a main output has remote sense lines, the output voltage may be
alterable by disconnecting these and lightly loading them. This will
cause an increased output voltage on the main terminals due to current
flow in default-coupling resistive elements (usually internal 22 to
100R resistors couple the sense lines to the output lines when sense
lines are open).

Some supplies have a 'margin' terminal that can be used to alter the
output, but these may be idiot-proofed to limit the deviation made
possible.

If you have access to the regulation circuit, temporarily wiring in a
variable resistor, to parallel the shunting element in the voltage
divider, will allow the same effect as margining. Some products may
have an internal test point intended to induce an overvoltage
condition when grounded, for use in automated testing.

You can purchase fairly simple 2-terminal crowbar protection modules
that can be value-added-on. How effective these can be, on
higher-current low-voltage supplies, is questionable. OVP is intended
to prevent damage to hardware - not to protect data integrity - in the
event of a PSU single fault.

If lower-than nominal voltages also present a threat - it might be
worthwhile paying attention to how your computer handles the
power-good signal line. This is often configured to indicate
out-of-tolerance outputs, including undervoltage, but I have yet to
see a PC that uses them sensibly.

If you're really concerned about the quality of the power provided by
a PC supply, you should probably obtain devices that demonstrate the
desired quality and performance.

I've been blind-sided by simple failures in the my own PC's supply
signal lines - and power conversion is supposed to be my area of
expertise. Ignoring PSU quality really doesn't pay in the long run.

RL
.

Relevant Pages

  • Re: Surge Protection?
    ... low voltage is not hardware destructive. ... Hardware protection means 'whole house' protector ... ... Low voltage does not cause hardware damage. ... serious in third world power. ...
    (alt.security.alarms)
  • Re: power outage ? from a novice..
    ... due to utility provided power making quick, ... voltage and no voltage, ... Your numbers say a UPS will protect from ... UPS protection is that? ...
    (alt.tv.tech.hdtv)
  • Re: Kenwood VR-407 protection problem
    ... up no protection. ... same four power supplies. ... many times that just checking the pn junction voltage isn't enough, ... If the unit has 2 switched outlets, ...
    (sci.electronics.repair)
  • Re: Cheap/generic PSUs - any good? (was: Low power socket A)
    ... >switching power supplies) but, while I can't say I've seen 'every' power ... This was not a PC supply though, just a generic Switch Mode. ... >out excessive voltage. ...
    (sci.electronics.repair)
  • Re: Cheap/generic PSUs - any good? (was: Low power socket A)
    ... >switching power supplies) but, while I can't say I've seen 'every' power ... This was not a PC supply though, just a generic Switch Mode. ... >out excessive voltage. ...
    (sci.electronics.components)