What PSU rating for Athlon 2400 with 6 HDDs?

[kony]

No, its not. As soon as I posted I thought perhaps on reflection it was not
their nf2 board.


I will if I can find it. I specifically remember a review of an Albatron
board where they were questioning the sanity of the designers not including
the P4 connector on a 12V board. Perhaps it was the KT600 or KT880 board, I
can't remember.

Ah, but the thing is, reviewers are often kids who barely
know what they're doing. Being "in print" doesn't make it
factual... something you will understand better when you are
older.

Not so. See my comment above.

To be kind, you have assumed something that isn't true, or
trusted someone (a reviewer?) who is clueless. Some
reviewers I've corresponded with just don't have enough
experience to realize that even though AMD's later design
guides recommend 12V for CPU power, that doesn't mean all
motherboard manufacturers do so. They are headstrong and
foolish to never bother testing their theories, or bothering
to take voltage measurements.

There are no brand name boards having _ONLY_ 20 pin ATX
conn. that use 12V for CPU. If you do find a clueless
reviewer that claims differently, be sure to keep looking at
other reviews of the same motherboard before trusting a
reviewer who is in error.

I seriously doubt that even an unknown very poor generic
board manufacturer would have an engineer competent enough
to get any motherboard working, but would make the mistake
of trying to power CPU by 12V without 2nd 12V input, whether
it be from a 24 pin connector or 2nd, 4-pin connector or
something less common/proprietary.

 

[kony]

When you say "clearly", how can you be clear that its 5V CPU power? I can
see how you can deduce whether its two or three phase, but how can you tell
its a 5v design?

Because of the amperage needed plus the 2 phase design,
there is not enough V margin for using 3V, and there isn't
sufficient supply line for 12V. There are no other
high-current rails on a PC power supply.

Then again there's experience, I get bored with same old
plug-part-A-in-slot-B, and hack away at boards just for the
heck of it, haven't had a board in years that I didn't take
voltage measurements on. You start to notice these things
after a while, and then there's overclocking... for extreme
overclocking on an Athlon board that uses 5V for CPU it may
be necessary to suppliment the power plane with a jumper
wire (among other things), as example:
http://69.36.189.159/usr_1034/M7NCG_5V_Mod.jpg

 

[larrymoencurly]

"Qtec" - for example - are a very popular low-end

brand. You might refer to them as a no-name, but arguably they are one step
up from that. They produce a very popular 550w PSU.

The first thing to note is that by 550w, they mean *Peak*. Which basically
means nothing at all.

I didn't believe it until I visited their website. This has got to be
the first time that a company has admitted that its most prominently
advertised power rating was for peak power. Is it a sign of greater
honesty, or of lower standards?

Are you sure that Q-tec is a step above no-names? This picture of the
insides of their 550W model: www.bit-tech.net/images/review/123/7.jpg
Makes i seem more like a decent 250W.

 

[Lionel]

What rating PSU should I use on a system which has an Athlon 2400+
with a modest graphics card (old GeForce2 MX 32MB) and 768 MB of SD-
RAM. Unusually, it will have *SIX* IDE internal hard drives.

All the rest of the system is pretty normal with no overclocking on
the cpu.

Would a decent 350W PSU be enough? I am thinking of this Nexus PSU.
http://www.nexustek.nl/nx3500.htm The specification chart shows
current delivery at min load, normal load, max load:

+3.3V 0.3A 14.0A 21.2/28A
+5V 1.0A 12.7A 30/25.5A
+12V 0.2A 4.5A 16A

The correct way to calculate this is to look up the maximum power
consumption spec's for each device in your system, (which should be in
the sheet/manual for each device, or look on the manufacturers website -
don't forget to include the CPU & any extra fans), add up the currents
for each voltage, add a safety factor of about 20%.
Compare them to the continuous rated current output from the PSU. If
the requirements are less than the continuous rating for the PSU, you
should be fine for normal operation.
*However*, there is one complication: Hard disks need much more 12V on
startup, while the spindle motor is coming up up to speed. Calculate the
same figures as above, using the startup current figures from the disk
drive spec's instead of the continuous rating to give you the startup
load. If the 12V load is still under the continuous rating of the PSU,
everything's fine. If it's within the peak (temporary) load of the PSU,
you should be okay in theory, but in practice, the cheaper PSUs are
likely to fail earlier. If the startup load is greater than the PSU peak
rating, get a heavier PSU.

 

[CBFalconer]

I didn't believe it until I visited their website. This has got
to be the first time that a company has admitted that its most
prominently advertised power rating was for peak power. Is it a
sign of greater honesty, or of lower standards?

Are you sure that Q-tec is a step above no-names? This picture of the
insides of their 550W model: www.bit-tech.net/images/review/123/7.jpg
Makes i seem more like a decent 250W.

5 Va.c. _________\|____ 1 k ohm ___________________ 5 V out
/| | |
diode | |
1000 uF 5V zener
|________|
|
gnd

should provide 5v regulated with a 600 or more peak power output
rating. Depends only on the quality of the capacitor. Maybe I
should go into business.

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

5 Va.c. _________\|____ 1 k ohm ___________________ 5 V out
/| | |
diode | |
1000 uF 5V zener
|________|
|
gnd

should provide 5v regulated with a 600 or more peak power output
rating. Depends only on the quality of the capacitor. Maybe I
should go into business.

600 or more _what_?

Where do you get 5VAC?

 

[JW]

600 or more _what_?

I'm assuming he meant watts, but...

Where do you get 5VAC?

More importantly, if he did mean watts, I'd like to see how it'll get
through that 1k resistor

 

[David Maynard]

I'm assuming he meant watts,

"Peak" watts.

but...




More importantly, if he did mean watts, I'd like to see how it'll get
through that 1k resistor

Easy. It charges the capacitor back up, over time.

 

[Tom MacIntyre]

"Qtec" - for example - are a very popular low-end

It means that they must have been manufacturers of computer speakers
before they started with power supplies?

Tom

 

[kony]

It means that they must have been manufacturers of computer speakers
before they started with power supplies?

Tom

LOL, that's exactly what i was thinking.

 

[CBFalconer]

I'm assuming he meant watts, but...


More importantly, if he did mean watts, I'd like to see how
it'll get through that 1k resistor

It doesn't have to. I said peak, which is supplied by the
capacitor alone. The point is that the (flawed) specification can
be filled with about two bits worth of parts. The requirement, for
500 W. peak at 5V, is that the capacitor can supply 100 A. for any
period, no matter how minute. You can use as small a capacitor as
you like, also, provided it has low internal resistance.

 

[Michael Brown]

Watson A.Name wrote :

I'm assuming he meant watts, but...


More importantly, if he did mean watts, I'd like to see how it'll get
through that 1k resistor

Err, I think he was making a joke The above circuit would indeed supply
600W of power at something around 5V for a very brief period of time if the
capacitor was up to it. Assuming we're talking ATX tolerances (5% on 5V
IIRC), the supply voltage would be allowed to drop to 4.75V before it became
out of spec. If the "power supply" was allowed to be on for a while before
any load was applied (ie: it charges up the capacitor) then a 1000uF
capacitor should be able to hold it within spec at a 600W load for about
2uS. The resistor and power source don't have a big impact while the load is
applied precicely for the reason you noted: the resistor limits to
satisfying a ~6mW load at best.

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

"Peak" watts.


Easy. It charges the capacitor back up, over time.

There is no time factor in watts.

 

[David Maynard]

There is no time factor in watts.

But there is in charging a capacitor through a fixed resistor.

 

[Franklin]

"Qtec" - for example - are a very popular low-end

I didn't believe it until I visited their website. This has got
to be the first time that a company has admitted that its most
prominently advertised power rating was for peak power.

Does anyone have any 'real' specs for this PSU? You know what I
mean, the normal sort of data in a chart.

I can't find any at their website. I am overlooking a small but
significant link?

http://www.qtec.info/products/group.htm?pc=POWER-SUPPLY-UNITS

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

But there is in charging a capacitor through a fixed resistor.

As a matter of fact, you can't store a single watt in that capacitor.
You might store so many coulombs of charge at a certain voltage, but not
watts.

Wattage is dissipation, and if you put a low resistane across the
charged capacitor to get a lot of current, there would be little
dissipation, even for a short time. To get 600W or 120A at 5VDC, the
resistor would be about .04 ohms, and that is probably not much more
than the ESR of the capacitor.

 

[Franklin]

Absolutely correct. "Qtec" - for example - are a very popular
low-end brand. You might refer to them as a no-name, but
arguably they are one step up from that. They produce a very
popular 550w PSU.

Poor old Qtechnology - who make some half-decent PSUs. They must be
sick of having a name so silimar to that of Q-Tec.

http://www.qtec.info/products/
http://www.qtechnology.net/products.htm

And don't forget to throw in a hyphen into Q-Tec but not into
QTechnology. Omit the hyphen from both their URLs. The potential
for confusion keeps on growing!

 

[Franklin]

Any/all of these can vary current. One board may default
memory to different voltage, another may be running at
higher FSB speed, then the obvious things like chipset or
processor count too... these differences could offset each
other or add up to a signficant difference in some cases.
Even so, the difference between one motherboard and another
will often be much less than the difference between one CPU
or another (if large enough frequency or voltage, core
change), or comparing a budget video card to a high-end
model, or number of hard drives.

Which compenents in a PC tend to use the +3.3V supply? Is it
*typically* the cpu, mobo and memory?

 

[David Maynard]

Does anyone have any 'real' specs for this PSU? You know what I
mean, the normal sort of data in a chart.

I can't find any at their website. I am overlooking a small but
significant link?

I never found anything 'real' on their site either. The only thing I came
up with earlier (but don't have handy) was a customer 'review' on a site
selling them and he claimed to have received an email from them, and posted
the data. The best recollection I have is that the 'real' numbers were 20
to 40% less, depending on the power rail, than the amps put on the sticker.

 

[David Maynard]

As a matter of fact, you can't store a single watt in that capacitor.
You might store so many coulombs of charge at a certain voltage, but not
watts.

I didn't say anything about 'storing watts'. The operative word was "time."
As in "there is no time factor in" ... "but there is in."

Wattage is dissipation, and if you put a low resistane across the
charged capacitor to get a lot of current, there would be little
dissipation, even for a short time.

Correct. A real SHORT period of time. But for that short period of time you
could hit '600 watts', under the right conditions.

Then a lot longer to charge the cap back up before you could get another
jolt of "600 peak watts." But, by golly, you can get "600 peak watts" out
of it.

To get 600W or 120A at 5VDC, the
resistor would be about .04 ohms, and that is probably not much more
than the ESR of the capacitor.

You've missed the whole point of his demonstration diagram. The context was
Qtec rating their PSU with "peak watts" and he's showing, with an absurd
circuit, that a "peak power" rating means absolutely nothing except,
perhaps, that it has a low impedance output.