PSU Replacement Problem

[John Leddy]

I am stumped.

My older emachines T2200 was and is working fine. I, over the last year,
added an extra drive, an audio card, a graphics card and maxed out the
memory with very little cost due to rebates. I felt that the original power
supply (Bestec Model #ATX 250-12E) might not be powerful enough and was
getting old, so I ordered a Thermaltake Purepower 500W supply. I heard that
Thermaltake power supplies are excellent.

When I installed it and turned the machine on, the power supply fan started
and the cooling fan started and the led on the front of the machine went on.
The power button wouldn't light up and the computer wouldn't boot up. I
uninstalled the unit.

I then borrowed a friend's PSU PCPWR & Cooling 370W RT unit and installed it
and the same thing happened. However, I could get the machine to boot by
playing around with the power button, pressing in on and off and holding it
in for some time. Once the machine booted up, everything seemed fine.
However, the boot-up problem was still there every time the machine was
started up.

I sent a message to PC Power and Cooling and they told me to return the unit
for a replacement. I also sent a message to Thermaltake, but I haven't
received a response as yet.

Is there some type of compatibility problem? I would like to have the
Thermaltake unit in my machine as the power is considerably more than the
Bestec unit.

My machine is working fine right now with the original Bestec supply
installed again. Is there such a thing as a proprietary power supply for
this machine?

It only takes me about 5 minutes to make the change now.....4 screws and 6
connections.

Suggestions please!

 

[Luis Alfonso]

I am stumped.

My older emachines T2200 was and is working fine. I, over the last year,
added an extra drive, an audio card, a graphics card and maxed out the
memory with very little cost due to rebates. I felt that the original power
supply (Bestec Model #ATX 250-12E) might not be powerful enough and was
getting old, so I ordered a Thermaltake Purepower 500W supply. I heard that
Thermaltake power supplies are excellent.

When I installed it and turned the machine on, the power supply fan started
and the cooling fan started and the led on the front of the machine went on.
The power button wouldn't light up and the computer wouldn't boot up. I
uninstalled the unit.

I wish I had a loonie for every Bestec PSU I've replaced.

I then borrowed a friend's PSU PCPWR & Cooling 370W RT unit and installed it
and the same thing happened. However, I could get the machine to boot by
playing around with the power button, pressing in on and off and holding it
in for some time. Once the machine booted up, everything seemed fine.
However, the boot-up problem was still there every time the machine was
started up.

Rather than reconnecting the power switch to the motherboard header try
shorting those two pins with a flat screwdriver or a length of wire.
Have the Thermaltake installed at this time.


Does the motherboard use the 4 pin P4 power connector?

 

[Paul]

I am stumped.

My older emachines T2200 was and is working fine. I, over the last year,
added an extra drive, an audio card, a graphics card and maxed out the
memory with very little cost due to rebates. I felt that the original power
supply (Bestec Model #ATX 250-12E) might not be powerful enough and was
getting old, so I ordered a Thermaltake Purepower 500W supply. I heard that
Thermaltake power supplies are excellent.

When I installed it and turned the machine on, the power supply fan started
and the cooling fan started and the led on the front of the machine went on.
The power button wouldn't light up and the computer wouldn't boot up. I
uninstalled the unit.

I then borrowed a friend's PSU PCPWR & Cooling 370W RT unit and installed it
and the same thing happened. However, I could get the machine to boot by
playing around with the power button, pressing in on and off and holding it
in for some time. Once the machine booted up, everything seemed fine.
However, the boot-up problem was still there every time the machine was
started up.

I sent a message to PC Power and Cooling and they told me to return the unit
for a replacement. I also sent a message to Thermaltake, but I haven't
received a response as yet.

Is there some type of compatibility problem? I would like to have the
Thermaltake unit in my machine as the power is considerably more than the
Bestec unit.

My machine is working fine right now with the original Bestec supply
installed again. Is there such a thing as a proprietary power supply for
this machine?

It only takes me about 5 minutes to make the change now.....4 screws and 6
connections.

Suggestions please!

I see a comment, that the T2200 uses an ECS L7VMM2 V1.0 motherboard, and
the manual is here. Not that this really helps with your problem, but
maybe someone else will spot an issue.

http://download.ecsusa.com/dlfileecs/manual/mb/eng/l7/l7vmm210eng.zip

One potential difference I see, is the old Bestec power supply probably has
a wire and pin installed for pin 18. That carries -5V to the motherboard
on a 20 pin connector. It could be, that the other two supplies don't
have the -5V wire, and pin 18 is left blank.

The ATX power supply specs (formfactors.org) removed the -5V wire
some time ago. Motherboards were not supposed to be dependent
on it any more, because there isn't any circuitry that really needs
a voltage like that. For example, if a designer needed to run a
GD75232 RS-232 level shifter, for a serial port, they could feed that
via +12V and -12V. Similarly, if they wanted to implement some
op-amp circuit, they might power that from the same two rails.
Or, if they wanted, use a 7905 regulator, fed from -12V, to make
a local -5V if that is what they wanted.

So there is no reason to have a dependency on -5V.

(The last time I needed -5V, was for some ECL circuitry in the lab )

That leaves a timing problem of some sort, with Power_Good
or the like. The fact that fiddling with the power button
made it start, suggests the problem is not with -5V. If
it really needed -5V, it would never have started.

That is not a lot to go on.

This would be one way to make a -5V supply, from the -12V
wire on the power supply. You need to know a bit about
electronics, to build this, and then connect the output to
where pin 18 is on the ATX power supply connector. If
a lot of current is drawn from the regulator, it gets
red hot before it shuts down. To draw the maximum
possible current from it, it needs a heatsink bolted
to it. I have a circuit similar to this, sitting beside
me at the moment, so I have used this technique for
making bench supplies.

http://www.national.com/mpf/LM/LM7905.html

In terms of buying a supply with -5V on it, as a test,
I can find a supply here, but the reviews aren't good.
Rosewill probably did not make the supply themselves,
and the power supply industry does a lot of outsourcing.
This one supplies -5V at up to 0.5 amps. Also, I note in
the reviews, that the customers are complaining about
getting a "new version", and when that happens, sometimes
the dopes remove the -5V from pin 18 :-( In other
words, it is possible the "new version" doesn't even
have the same specs.

+3.3V @ 30A, +5V @ 45A, +12V @ 26A, -5V @ 0.5A, -12V @ 1A, +5VSB @ 2.5A
http://www.newegg.com/Product/ProductReview.aspx?Item=N82E16817182009

Another possible source is here. This appears to still have
-5V on it, at 0.3 amps. The fans on these can be a bit on
the loud side. (OMG - the price is $189!) Oh well, so much
for that idea.

http://www.pcpower.com/power-supply/turbo-cool-510-atx.html

http://www.pcpower.com/downloads/specs_3830.pdf

Paul

 

[Franc Zabkar]

I see a comment, that the T2200 uses an ECS L7VMM2 V1.0 motherboard, and
the manual is here. Not that this really helps with your problem, but
maybe someone else will spot an issue.

http://download.ecsusa.com/dlfileecs/manual/mb/eng/l7/l7vmm210eng.zip

FWIW, I have the same issue with an ECS L7S7A2 motherboard and an
Antec 350W PSU. The original fake 400W PSU (really only 200W) also had
the same problem.

When the machine starts up with a black screen, I hold down the start
button until it turns off, and then I press the same button to restart
it. That seems to work every time. The machine then runs reliably, and
can be put into standby mode and restarted from standby mode, until I
power it off at the mains. This machine has behaved like this from
new. Nothing is overclocked.

- Franc Zabkar

 

[John Leddy]

Paul.

Thanks Paul for your response to my problem. At lease my machine is working
fine with the original Bestec PSU.

Maybe finding a power supply with ONLY a 20 pin motherboard connector would
do the trick..........they're probably not around any more.

John

 

[John Leddy]

GMAN.............On this motherboard, there is no 4 pin 12V CPU socket!

Thanks for trying to help!

 

[~AlicGinnis~]

my nepthew built a pc and couldn't get it to boot.
spent a week on it.
he didn't plug in the 4 pin connector, so yeah,
that happens.

 

[w_tom]

Maybe finding apower supplywith ONLY a 20 pin motherboard connector would
do the trick..........they're probably not around any more.

Maybe this and maybe that. Your replies are only speculation
because the facts provided are so minimal and do not provide numbers.
You have two problems.

1) Bestec power supply (may be) too small. You don't know. A
defective or too small supply can still boot a computer. But it takes
only a 3.5 digit multimeter (and many times less labor compared to
swapping a supply) to know immediately if the supply is sufficiently
sized. The meter means you have an answer so definitive that you
don't even consider the problem any more.

2) Other supplies don't work. Again the meter and less than 2 minutes
labor would have answered that question OR provided numbers to the
better informed. Those who could have best answered your question are
silent because you did not provide those numbers.

For both questions, only existing answers are speculation. Replies
are only as good as information you provide. That means numbers from
a multimeter both before and when the power switch is pressed
(question 2) and with the supply under maximum load (question 1).

After all those posts, you did not learn anything useful except why
-5 volts does not exist; may or may not be needed on your motherboard.

Is a Bestec power supply too small? The 'watts' game is alive and
well. A power supply rated by a computer manufacturer at 250 watts is
also rated by the third party suppliers at 360 watts. They don't
lie. They just know many computer assemblers only arbitrary facts and
do not understand more important numbers. Far more important are
current numbers for each voltage. But that is too complex for many
computer assemblers who only understand vague numbers - dollars and
watts. Well, this paragraph suggested one (unlikely) reason why those
other supplies would not work. And like so many other replies, this
paragraph says little useful because important (useful) numbers are
not provided.

 

[Paul]

Maybe this and maybe that. Your replies are only speculation
because the facts provided are so minimal and do not provide numbers.
You have two problems.

Yes, maybe we should go back and examine the threads you've
contributed to. For example, where you asked an original poster
to make a measurement for you. The OP complied. What kind of an
answer did he get from you in turn, based on the measurements ?
A "plate of waffles". He was no further ahead than he started.
So I would not be so quick to hand out "fatherly advice".

The thing is, some problems need an oscilloscope, and
a schematic of the motherboard, to understand what is
going on. For example, when there was a problem with
Antec supplies and Asus motherboards, there was some kind
of timing problem, where the Vcore regulator seemed to be
shutting off, in response to some slow reaction of the
Antec ATX supply. How would a person detect a subtle timing
problem, using only a multimeter ? A multimeter is not suitable
for debugging many problems. The ones it would be useful
for, probably had a cloud of smoke around the unit anyway,
as a hint as to what happened.

Some problems can only be properly diagnosed in lab conditions,
with a bit more equipment than most hobbyists will have at their
disposal. A digital storage scope for example, would be
indispensable for working on problems like this. It can capture
slot transient responses, such as the first 50 milliseconds
after power up. It can also capture higher frequency behaviors,
such as ripple on the rails, or step response to transient load.

There are very few schematics available for motherboards, and
certainly none for retail motherboards. I have a couple
Intel reference schematics, and you can learn a little
bit from them. Retail motherboards use a variety of solutions
for onboard power regulation, which means no two designs
will be alike.

If you really wanted to make progress on debugging the problem,
you won't be doing that on a kitchen table. And deriding posters,
because they don't have a laboratory at their disposal, isn't
helping them at all.

Paul

 

[Franc Zabkar]

There are very few schematics available for motherboards, and
certainly none for retail motherboards.

This web site has a large number of circuit diagrams for pre-1999 DTK
motherboards and PSUs. It may be good for reference purposes ...

DTK Mboard circuits:
http://www.dtk.com.tw/tech/circuitm.html

DTK Mboard ECNs:
http://www.dtk.com.tw/tech/ecn.html

DTK PSU circuits:
http://www.dtk.com.tw/tech/circuitp.html

Critical FAQs:
http://www.dtk.com.tw/tech/dtkfaq.html

- Franc Zabkar

 

[w_tom]

Some problems can only be properly diagnosed in lab conditions,
with a bit more equipment than most hobbyists will have at their
disposal. A digital storage scope for example, would be
indispensable for working on problems like this. It can capture
slot transient responses, such as the first 50 milliseconds
after power up. It can also capture higher frequency behaviors,
such as ripple on the rails, or step response to transient load.

A schematic and digital scope are necessary if doing an advanced
design analysis and if timing problems existed. Based upon facts
provided, there is no reason to assume a timing problem or any other
speculation. If a timing problem existed, then numbers from the meter
would demonstrate a timing problem or related problems exist. But
without those meter numbers, then we could easily speculate 100 other
failures.

Without that meter, then every suspected problem still exists as
speculation - nothing has been elminated as a suspect and nothing is
learned.

Less then two minutes with a meter would have eliminated most of
those above wild speculations immediately. For example, even a doctor
does not perform an MRI without first learning simple stuff such as
body temperature and heart rate. Unforuntately we don't even have the
simplest of data - numbers from a multimeter.

Multimeter - a tool sold where hammers are also sold and for about
the same price.

A discussion last month was "What's your best guess?" in
alt.comp.hardware.pc-homebuilt from September thru 8 Oct at:
http://tinyurl.com/6khcnf
Larc had maybe 50 posts before finally using the meter on 7 Oct. Very
next post identified shotgunning as a waste of time and the reason for
so much confusion:

These numbers says the power supply is perfectly good in this
computer. ...
The relevant power supply functions are working fine. So the
power supply controller is not telling the CPU to operate, or
some other motherboard function has failed.

What waffle? Finding a new motherboard proved harder than using a
meter. Meter 1) correctly identified the only problem immediately
and 2) 1) identified the power supply as perfectly good despite so
much wild speculation that said otherwise.

Another example (in Feb 2007) where a multimeter eliminated wild
speculation immediately. OP's reply to others who then mocked the
multimeter solution:

2. The offered suggestions, even by people who meant well, were all
in the area of 'try this and try that'. There was even one to start the
computer with a DOS floppy ...
Asus will not send me a replacement board, just because you and your
'upgrade' friends _think_ it's 'obsolete'. They require proof. I had that
proof in my hands when I made the call to get an RMA. What kind of
proof did I have? The voltage test results as suggested by w_tom.
Like it or not, he knows this particular area, most of you don't. I was
willing to _learn_ something new, ...

Best solution starts with a tool sold where hammers are also sold,
that is about as complex to use as a hammer, for about the same price
as a hammer. Without numbers, then every speculated problem -
including a possible timing problem - remains only speculation.

BTW, we never spend big bucks obtaining a storage scope. We did far
more complex timing analysis quickly even with a conventional (non-
digital) 20 Mhz scope. But none of that need be done here. So much
time, money, and speculation replies averted immediately by simply
reading and posting numbers from a multimeter.

 

[Mike Tomlinson]

[usual hand-waving waffle and lies from w_tom flushed]

Now, w_twat, how about answering the part of Paul's post that you so
conveniently snipped? I'll quote it here for your benefit:

"Yes, maybe we should go back and examine the threads you've
contributed to. For example, where you asked an original poster
to make a measurement for you. The OP complied. What kind of an
answer did he get from you in turn, based on the measurements ?
A "plate of waffles". He was no further ahead than he started."

I expect no more than your usual waffle, lies and evasions by way of
reply.