PC refuses to boot

In comp.sys.ibm.pc.hardware.chips keith wrote:
PSU's can certainly be flakey. L'Angel noticed this some
few years back and designed a tester (don't know how far
she got with the implementation) to stress PSUs. Actually
I was hoping that someone would pick this up and report.

IIRC, she was trying static loads. I like lightbulbs for
load, but where'm I gonna find 100W @ 3 V? I believe the
german mag c't tests PSUs. But IIRC they're also static,
and PSUs have to make static or they don't get UL/TuV ratings.

I believe most PSU problems are dynamic -- an inability of
the PSU to handle current slew (too much ESR). I did once
get started on `burnPSU` but the damn thing was a bit of
a nightmare bootdisk to write.

AMD may have made the HLT powersavings Northbridge/BIOS
configurable specifically at the request of bottom-end
mfrs who didn't want to have to handle the current slew.

-- Robert author `cpuburn` http://pages.sbcglobal.net/redelm

On Fri, 29 Oct 2004 02:02:25 +0000, CJT wrote:

keith wrote:
On Thu, 28 Oct 2004 13:51:54 +0200, Grumble wrote:


Grumble wrote:


I've been pulling my hair for 3 hours over this problem, so I thought
I'd ask in here, and see if people have had a similar problem.

Thanks Keith, CJT, George, and Tony! Your insight was very helpful.

The problem was the PSU indeed. In my (rather limited) experience,
PSU-funkiness is quite hard to diagnose, as there are many very
different possible symptoms. I'm thankful the PSU didn't take out my
other components when it died (which happened to a friend).


PSU's can certainly be flakey.

Amen. I had smoke pour out the back of one once. They take a lot
of stress.

Sure. The system (see: Cheap System in this group) I put together for a
friend was because a power surget let out all the magic smoke. I was able
to "replace" the system for a ham-sandwitch over what her insurance
company gave her for the smoking hulk. ;-)

--
Keith

On Fri, 29 Oct 2004 12:56:54 +0000, Robert Redelmeier wrote:

In comp.sys.ibm.pc.hardware.chips keith wrote:
PSU's can certainly be flakey. L'Angel noticed this some
few years back and designed a tester (don't know how far
she got with the implementation) to stress PSUs. Actually
I was hoping that someone would pick this up and report.

IIRC, she was trying static loads. I like lightbulbs for
load, but where'm I gonna find 100W @ 3 V?

That was L'Angel's original plan, but morfed into a dynamic tester. She
got some great tips from the folks on SED, though I don't know how far the
hardware went. Good resistors are hard to find (lightbulbs are *lousey*
resistors)

I believe the
german mag c't tests PSUs. But IIRC they're also static,
and PSUs have to make static or they don't get UL/TuV ratings.

Huh? UL is all about safety. They could care less about utility. What's
TuV?

I believe most PSU problems are dynamic -- an inability of the PSU to
handle current slew (too much ESR). I did once get started on `burnPSU`
but the damn thing was a bit of a nightmare bootdisk to write.

Sure, but a dynamic test shouldn't be too hard. A few resistors and FETs
should do it. I believe the reason it's not done is that PSU's are
Boorrringgg. THey don't make games go faster, rather can only bring them
to a halt. "If it works, it's good!"

AMD may have made the HLT powersavings Northbridge/BIOS configurable
specifically at the request of bottom-end mfrs who didn't want to have
to handle the current slew.

Ok... Why not slew the HLT? (oops!)

--
Keith

On Fri, 29 Oct 2004 12:56:54 GMT, Robert Redelmeier
wrote:

IIRC, she was trying static loads. I like lightbulbs for
load, but where'm I gonna find 100W @ 3 V? I believe the
german mag c't tests PSUs. But IIRC they're also static,
and PSUs have to make static or they don't get UL/TuV ratings.

I believe most PSU problems are dynamic -- an inability of
the PSU to handle current slew (too much ESR). I did once
get started on `burnPSU` but the damn thing was a bit of
a nightmare bootdisk to write.

Hmm, we started with static loads and light bulbs but I eventually
ended up with a dynamic design that was supposed to change loads in
20Khz speeds. Not sure if this is enough to qualify for your dynamic
definition. Though my friend helping me do the construction looked
quite dimly upon the idea... largely due to the cost and the fact
neither of us are that sure we can actually make it work despite the
helpful info from the folks here and seb PpPp


As a side trip, from your experience, is it possible to shutdown the
clocks to the processor on a P4 (or AhtlonXP since I'm at it) via
software instructions that be done with a very simple DOS based
program? I'm trying to do a reverse of your burning the cpu and trying
to freeze it instead pPpP
--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code

On Fri, 29 Oct 2004 22:27:55 -0400, keith wrote:
That was L'Angel's original plan, but morfed into a dynamic tester. She
got some great tips from the folks on SED, though I don't know how far the
hardware went. Good resistors are hard to find (lightbulbs are *lousey*
resistors)

Didn't go too far when we realized that measuring the stuff would cost
a hundred times much more that the tester if we made it work. An
oscilloscope isn't cheap. A digital one that can store or externally
output the digital data of the waveforms for 4 channels at the same
time is expensive. The equivalent DAQ for a PC is hideously expensive.


Though all in all, it was an interesting diversion project. Between my
online queries and his offline studies/hands on, we picked up a lot of
stuff we wouldn't otherwise.

--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code

In comp.sys.ibm.pc.hardware.chips keith wrote:
Good resistors are hard to find (lightbulbs are *lousey*
resistors)

If you consider "good"==constant. I was using "good"==disippates
lots of heat.

Huh? UL is all about safety. They could care less about
utility. What's TuV?

TuV is the rough German equivalent to UL. Both will test or
requires PSUs to be tested at full wattage for long enough
to heat up to show they don't catch fire. Voltage might
sag, but that's more likely to cause overheating, too.

Sure, but a dynamic test shouldn't be too hard. A few
resistors and FETs should do it. I believe the reason

Yep, those'll work.

it's not done is that PSU's are Boorrringgg. THey don't
make games go faster, rather can only bring them to a halt.
"If it works, it's good!"

Agreed.

Ok... Why not slew the HLT? (oops!)

That takes brains

-- Robert

In comp.sys.ibm.pc.hardware.chips The little lost angel wrote:
Hmm, we started with static loads and light bulbs but I
eventually ended up with a dynamic design that was supposed
to change loads in 20Khz speeds. Not sure if this is enough
to qualify for your dynamic definition. Though my friend

This is definitely dynamic, and in the right range.
I think mobos switch around 100 kHz, and the PSUs are
considerably slower.

helping me do the construction looked quite dimly upon the
idea... largely due to the cost and the fact neither of

I want to use software only. CPU as power load.

As a side trip, from your experience, is it possible to
shutdown the clocks to the processor on a P4 (or AhtlonXP
since I'm at it) via software instructions that be done

"HLT" as implemented in most idle threads is about
as low as I know going, although some mobile CPUs
(really mobos) have more severe measures (downclock
and down volt, ie. Intel SpeedStep).

with a very simple DOS based program? I'm trying to do a
reverse of your burning the cpu and trying to freeze it
instead pPpP

A two byte MS-DOS pgm will work CLI / HLT, but for
AMD, the Northbridge StopGrant bit has to be enabled.
Both instructions are priviliged, and run only in Ring0.
For the P4, you can do BACK:REP NOP / JMP BACK.
I haven't measured REP NOP vs HLT.

-- Robert