Anyone tried quiet power supplies bigger than 300W?

PSUs: The source of DC power for all components in the PC & often a big noise source.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee, Devonavar

Post Reply
deeseeel
Posts: 42
Joined: Tue Aug 13, 2002 9:07 am
Location: Sunnyvale, CA, USA

Anyone tried quiet power supplies bigger than 300W?

Post by deeseeel » Sat Aug 17, 2002 9:28 am

Has anyone ever tried any of the quiet power supplies (Q-Tech, Seasonic, Zalman, etc.) with a higher than 300W power rating?

Does the noise/temperature compare well with the 300W version? Is there much difference?

Thanks

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Aug 18, 2002 10:41 pm

A switching PSU is supposed to produce only as much power as demanded by components connected to it. The wattage rating only tells us the maximum power that can be delivered. So a 400W unit should produce no more heat than a 250W one if the system demands only 150W -- assuming the power efficiency is the same.

Preliminary testing of a Seasonic 400W shows this sample draws 76W idle and 119W full on the test system used for all the PSU reviews. The two 300W samples drew 71~73W and 114~117W, so the 400W sample seems very slightly less efficient, but it may be within the normal manufacturing tolerance range.

Not having examined many other 400W models, it is difficult to generalize, but at least for Seasonic, there should be a noise advantage when a system is running full tilt or with systems that draw more power than our current PSU test platform. This is because the thermal control circuitry in the 400 is calibrated to speed the fan up at a higher temperature point than in the 300. Seasonic's reasoning is that as the components in the 400 are rated for safe operation to a higher temp, they will let it run a bit hotter and allow the fan to stay at the minimal noise level through a wider operational range. (Of course, I have not really been able to veryfy this yet -- I can't push the PSU hard enough with my test setup yet. Need a bank of high power low impedance resistors to get the PSUs cooking.)
This is NOT necessarily the way most thermistor-controlled fan PSU lines work. My experience with 300, 350 & 430W Enermax, for example, indicates that they all follow the same fan speed curve. I haven't compared all 3 on a testbench, so this is an impression, but a fairly strong one. My belief is that PSUs are built to a price, and generally the only difference between lower & higher power models in a given line is the power rating of the primary internal components. Usually there is an absolute minimum of other design changes -- that is just too expensive & troublesome for assembly lines that churn out thousands of units a day. (For example Seasonics main plant in Taiwan has a capacity of 850,000 units a month.)

deeseeel
Posts: 42
Joined: Tue Aug 13, 2002 9:07 am
Location: Sunnyvale, CA, USA

Post by deeseeel » Wed Aug 21, 2002 8:07 am

Thanks for the info, MikeC.

Given what you say, it seems that the Seasonic 400W would be a even better choice than the 300W for silent operation (regardless of actual power needed by the computer system).

But I get the impression that the Seasonic is the exception rather than the rule.

But perhaps this would be an interesting topic for an article someday?

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Aug 21, 2002 8:41 am

The advantage is probably not evident unless you have a very power hungry system. Note that the power draw referred to in the PSU reviews is NOT the power delivered. If the idle power draw on the system is 70W, for example, the PSU is actually delivering ~50W (assuming 70% efficiency) in DC voltage. To deliver the rated 300W, it would have to draw ~430W from the AC line. At full tilt, the total power draw with the Seasonic was under 120W; this means it was delivering just 84W.

We'd have to get the delivered power to at least ~125W to make the noise advantage of the SS400 over the 300 evident. That's total AC draw of ~180W. Maybe a dual CPU system, or one with multiple hard drives and all the PCI slots filled... None of mine pull that kind of power.

Oz
Posts: 32
Joined: Sun Aug 11, 2002 3:26 pm

Post by Oz » Thu Aug 29, 2002 4:23 am

A related question: what would one need a 400W PSU for? Instead of a 300W'er, that is? More than 2 harddrives? An AthlonXP 2100+? GeForce 4? That type of thing?

Cheers,
Oz

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Aug 29, 2002 6:53 am

You're approaching the type of question that the industry doesn't really want you to be even thinking about. Truth is, except for high load multiple dirive & peripheral systems, there's NO single or dual CPU desktop system even with the latest processors that NEEDS a true 400W PSU. Note I said true 400W. Another truth is that the power rating convention of PSUs appears to be just that -- a convention, just a way of doing things. It does not appear to be a strictly defined, standardized parameter like horsepower in car engines, for example. So you can have a 300W PSU whose voltage regulation falters when it tries to deliver half its rated power, and another 300W PSU that WILL actually power an industrial-level fully loaded multi-processor server machine.

IMHO,
1) PSU makers generally overrate their products, most are not capable of delivering their full rated power for any length of time.
2) While the power requirements of loaded 24/7 servers can be very high, power requirements for desktop PCs are greatly exaggerated. It is profitable for the industry to have you believe that 300W is the minimum power you need for you P4 or XP system. As you know, SPCR's test platforms include a P4 as well as an XP and a T-bird. None of those in dual drive desktops can pull even 100W in DC voltage from the PSU.

A huge poorly understood issue with PSU stability & voltage regulation is the role of the cables and connectors that deliver the DC voltage. The cables & connects are defined by the ATX/PS2 standard, and it is what makes PSUs interchangeable. HOwever, the longer the cable, the higher the voltage drop through it. Combine this with a connector that has to be somewhat sloppy or loose to make installation and mating with other plugs simple, and the voltage drop can be enough to make a joke of even super-high quality PSUs. If power supply cables were permanently hard-wired to components using the shorts wire lengths I would venture to say 80-90% of the power-related problems in PCs would simply disappear.

If I am completely off the mark here, I'm sure one of the PS engineers that drops in occasionally can provide a good scolding. ;)

deeseeel
Posts: 42
Joined: Tue Aug 13, 2002 9:07 am
Location: Sunnyvale, CA, USA

Post by deeseeel » Thu Aug 29, 2002 10:25 am

So MikeC, what do you think about what AMD says in their Builder's Guide?

If you look at pages pages 7 and 8, it lists the power requirements as 182W for the "typical configuration" and 279W for the "high performance system".

(Yes, going by the maximums for all devices is probably a bit over engineered. The AMD document does recommend 80% of all non-cpu devices. But playing it very safe for the moment...)

I don't know where they got their current draws for each device. But assuming they're realistic, I would think a many users would be in 200W+ category. A power user could really hit 300W with the numbers in that PDF. Dual Athlon systems would certainly break that, right?

I guess I don't disagree with you that 400W is a overkill, but at the same time it seems not that far off either.

Back to the subject of testing power supplies; it seems that it would be nice to have the equivalent to dynometer for cars. There would probably be more graphs: voltage stability, power efficiency, temperature, fan speed, noise (all as function of power draw for each of the different lines: 3.3v, 5v, 12v, etc.).

Some of the spec sheets would provide some of that information, but they don't seem consistent. For example, the Zalman ZM300A-APF clearly indicates that the noise ratings are "at full load" (what does that mean?) with specific voltages. However the Seasonic power supplies just say 1%.

But it would be nice to be able to verify their claims too. :)

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Aug 29, 2002 11:05 am

deeseeel wrote:If you look at pages pages 7 and 8, it lists the power requirements as 182W for the "typical configuration" and 279W for the "high performance system".

(Yes, going by the maximums for all devices is probably a bit over engineered. The AMD document does recommend 80% of all non-cpu devices. But playing it very safe for the moment...)

I don't know where they got their current draws for each device. But assuming they're realistic, I would think a many users would be in 200W+ category. A power user could really hit 300W with the numbers in that PDF. Dual Athlon systems would certainly break that, right?

Back to the subject of testing power supplies; it seems that it would be nice to have the equivalent to dynometer for cars. There would probably be more graphs... some of the spec sheets would provide some of that information, but they don't seem consistent..... But it would be nice to be able to verify their claims too. :)
The 80% guide brings those numbers down to 162.47 and 241.91 -- well under 300W.

If you think about it, even their 80% rule is something of an overkill: you really can't have 100% CPU usage AND access all the various drives & perihperals. It's just not possible -- at least not in any version of Windows I have used. Perhaps Linux is better. Multitasking simultaneously with 100% usage in any version of Windows just bogs the system down so much you hardly get any productivity, so in normal usage, I never try to do that. Like if I am doing a major transform function in Photoshop on a large image file, there's not much point trying to do a search on the HDD. It's just slows everything down to a crawl. I'm better off to wait the 30 secs.

With the test PCs, at 100% CPU load with various testing utilities, I opened both CD drawers & inserted a data disk in each & closed it at the same time. Not very productive, as it took forever to read the CDs, but even then the momentary peaks never hit more than ~135W AC power draw. That's like 100W power delivered on the DC lines. Now it is very possible the kill-a-watt meter can't repond fast enough to catch really fast peaks that ran higher, but almost any PSU can deliver much higher than rated power for a few milliseconds, I would think.

I guess what I am saying is that if you look at PSU requirements from the POV of getting the least amount heat so that it can be cooled with the minimum of fan noise, and consider typical usage on typical desktop systems, current recommendations for PSUs aren't that sensible. Manufacturers have to be a bit like lawyers -- they look at worst case scenarios, and that's what most PSUs are designed for (although I bet many don't cut it anyway). In real world desktop apps, it just seems almost impossible to reach such high power draws.

It would make more sense to me if PSU makers made PSUs specific to desktop use, with very clear definitions about what this means, and other PSUs specific for "industrial" use -- ie, as they are currently made. While we're at it, why not also do a complete remake of all the critical power connectors in a PC for minimum losses. There have got to be better connectors that the ones used today.

That's about all for now, as I balance precariously out on this long thin limb, scanning the skies anxiously for the attacking crows soon to come... :wink:

ChiefWeasel
Posts: 262
Joined: Sun Aug 11, 2002 3:26 pm
Location: Worcester, UK

Post by ChiefWeasel » Thu Aug 29, 2002 12:08 pm

Not that im disagreeing with any of the points made here, theres a few points i thought i could add to this discussion...

Firstly on power draw from a PSU: Remember that at start-up lots of power is required to spin up the HDs etc. So you do need some power in reserve for tasks such as this.

Second on power draw: even if a complete computer system only draws say 150W from a PSU, a P4 for example draws a lot of this on the 12V line (This is why it needs the extra mobo connector - the standard ATX connector cannot supply enough current at 12V). It was recommended to me to find a PSU which can supply 18A (i think! dont quote me :roll: ) on the 12V line. So while you would think a 230W PSU would be fine to power the 150W draw from the computer, it is extremely unlikely that a 230W PSU is able to supply 18A at 12V - and hence the computer would have stability problems. you would need a good 300W PSU to supply 18A @ 12V. I hope i explained what i was trying to get at here :) You probably dont actually need 18A to power a P4 - i am just trying to explain that the total power output cannot just be spread out over the voltages as and where its needed.

Re having 100% CPU as well as HD access: I use my computer for music production, and this application can really push a computer to its limits! It is quite easy to have a high CPU load at the same time as high disk access, e.g. playing back several audio tracks simultaneously and applying realtime processing to all of them. When im making music my computer is normally at about 70% CPU and disk access varies upto about 50% or more. This is on a P4 2.4Ghz.

Just a few points i thought i should add :) I agree, there should be some kind of standards in PSUs, it makes it very time consuming choosing a PSU when you have to consider the power available on each line, as well as the efficiency and the % tolerance of the voltages etc. And thats before we consider the noise level!

deeseeel
Posts: 42
Joined: Tue Aug 13, 2002 9:07 am
Location: Sunnyvale, CA, USA

Post by deeseeel » Thu Aug 29, 2002 12:29 pm

MikeC wrote: is the role of the cables and connectors that deliver the DC voltage. The cables & connects are defined by the ATX/PS2 standard, and it is what makes PSUs interchangeable. HOwever, the longer the cable, the higher the voltage drop through it. Combine this with a connector that has to be somewhat sloppy or loose to make installation and mating with other plugs
I forget to ask about this earlier... Do you have any tests to quantify the losses caused by the ATX power connector and PSU cable length? Flaky connectors I can understand, but I don't think the losses due to wiring should be that significant. Wouldn't that make for hot wires if the losses were high?

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Aug 29, 2002 1:03 pm

There are detailed discussions I have read, can't recall exactly where at this point; will find & post later. But take the 5V line. There are many instances where PSUs with long leads (24") don't do well, allowing the 5V line to drop significantly -- say to 4.7V. This doesn't cause the system to fail or not boot, but under large current demand, as it sags further, you start seeing flaky behavior. The Enermax was accused of being a bad PSU for a while, due to the combination of long leads & -- probably -- a slightly off connector or two. I'll find the links & report here later.

deeseeel
Posts: 42
Joined: Tue Aug 13, 2002 9:07 am
Location: Sunnyvale, CA, USA

Post by deeseeel » Thu Aug 29, 2002 1:10 pm

MikeC wrote:I guess what I am saying is that if you look at PSU requirements from the POV of getting the least amount heat so that it can be cooled with the minimum of fan noise, and consider typical usage on typical desktop systems, current recommendations for PSUs aren't that sensible.
Who's recommendations? I thought your list is pretty sensible. :) I just think most recommendations are based on a different criteria. Are you looking to recommend 200/250W models?

So what is better? As a general rule, it shouldn't make a difference to run a 300W PSU at 200W and a 400W PSU at 200W. Assuming they're from the same product line and the only difference is one has higher rate parts, the heat/noise should be the same, no? In this case it's only a matter of not buying a 400W because you can save a buck or two. No heat/noise is added because someone bought the 400W model, right?

(Well maybe if they put a different fan in the models, but assuming that isn't the case.)

The only possible exception to this is the Seasonic 400W, right?

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Aug 29, 2002 1:47 pm

ChiefWeasel wrote:1. Second on power draw: even if a complete computer system only draws say 150W from a PSU, a P4 for example draws a lot of this on the 12V line (This is why it needs the extra mobo connector - the standard ATX connector cannot supply enough current at 12V).

2. It was recommended to me to find a PSU which can supply 18A (i think! dont quote me :roll: ) on the 12V line. So while you would think a 230W PSU would be fine to power the 150W draw from the computer, it is extremely unlikely that a 230W PSU is able to supply 18A at 12V - and hence the computer would have stability problems. you would need a good 300W PSU to supply 18A @ 12V. ...the total power output cannot just be spread out over the voltages as and where its needed.

3. Re having 100% CPU as well as HD access: I use my computer for music production, and this application can really push a computer to its limits! It is quite easy to have a high CPU load at the same time as high disk access, e.g. playing back several audio tracks simultaneously and applying realtime processing to all of them. When im making music my computer is normally at about 70% CPU and disk access varies upto about 50% or more. This is on a P4 2.4Ghz.

4. Just a few points i thought i should add :) I agree, there should be some kind of standards in PSUs, it makes it very time consuming choosing a PSU when you have to consider the power available on each line, as well as the efficiency and the % tolerance of the voltages etc. And thats before we consider the noise level!
1. I don't believe that's really the case -- I think the 2nd 12V connector is there to compensate for the losses with just the one standard ATX connector to the mobo. I think Intel discovered that the contact loss was too high too often & just doubled up to make sure there wouldn't be too much voltage drop. Most P4-connector outfitted PSUs appear to be exactly the same internally as the models the preceded them without that extra connector; it's just an additional set of wires going back to the same 12V source. Someone from Intel confirm/explain this one, please!

2. The question is why 18A on the 12V line?

3. You're right -- I did not consider that. Should consult with my PC musician friend on heaviest apps in this & include this as part of the test procedure.

4. Well I was going way out on a limb with that comment about standards. :oops: I am sure there are lots of good arguments against this.
Last edited by MikeC on Thu Aug 29, 2002 8:23 pm, edited 1 time in total.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Aug 29, 2002 2:00 pm

deeseeel wrote:Who's recommendations? I thought your list is pretty sensible. :) I just think most recommendations are based on a different criteria. Are you looking to recommend 200/250W models?

So what is better? As a general rule, it shouldn't make a difference to run a 300W PSU at 200W and a 400W PSU at 200W... (Well maybe if they put a different fan in the models, but assuming that isn't the case.) The only possible exception to this is the Seasonic 400W, right?
Referring to the AMD recommendations, for example, about minimum power requirements. I would recommend lower power units if they were cheaper, had quieter fans, and ran as well as higher power ones for destop systems. But of course you can't even find a 250W PSU these days, really. Certainly none designed for low noise, now that the Q-tech 230/250 are gone.

You're right about power draw from lower / higher rated PSUs, but because of the whole approval certification issues, the higher rated units have to be tested at full power, where they get hotter, and thus need a higher airflow fan to stay cool enough to pass the tests. A higher airflow fan reduced to min speed is usually louder than a lower airflow one at min. The fan in the Seasonic 400, for example, is an Adda rated at 0.3A, compared to the 0.25A fan in the 300. I haven't done a close comparison between the two fans, I suspect the difference will be minimal, but if there is a difference, the 0.3A fan will be louder.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Aug 29, 2002 2:05 pm

deeseeel wrote:Wouldn't that make for hot wires if the losses were high?
They do get hot. As we speak I am testing an Antec TruePower 380 loaded enough to draw 260W AC from the wall. It is delivering 210-220W in DC. Higher than I've been able to reach before -- thanks to a very clever solution from SPCR author John Coyle. Will tell you all about this later, but I tell you, those wires are HOT!

deeseeel
Posts: 42
Joined: Tue Aug 13, 2002 9:07 am
Location: Sunnyvale, CA, USA

Post by deeseeel » Thu Aug 29, 2002 4:35 pm

Speaking of loading power supplies, did you see the Overclocker's review of the Thermaltake Silent PurePower? They spliced two ATX cables together to run two motherboards off one PSU.

They didn't give much in terms of noise numbers, but I thought running two computers off one PSU was new.

jsw
-- Vendor --
Posts: 12
Joined: Thu Aug 29, 2002 9:24 pm
Contact:

Seasonic 400W PSU

Post by jsw » Thu Aug 29, 2002 9:36 pm

Hi MikeC,
I'm glad to see that you have tried the Seasonic 400W PSU. I had been curious about it. Perhaps you can answer a few questions about it? Does it have much bigger heatsinks than the 300W? I've also seen a web page that led me to believe that it has two fans. Is this true, and if so, how does the noise compare to the 300W model? Are the two fans both thermally controlled or does one run at a constant rate?

Thanks!

--Jeff

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Aug 29, 2002 9:59 pm

Jeff, the 400 and 300 look identical - no difference in HS size. One fan. I have not examined all the parts inside, but Seasonic states it has higher rated parts. The .3A fan is slightly noiser than the .25A fan in the 300, but the dif is not big. It does speed the fan up at a higher temp than the 300. A short review is coming soon. Short because it is so similar to the 300.

ChiefWeasel
Posts: 262
Joined: Sun Aug 11, 2002 3:26 pm
Location: Worcester, UK

Post by ChiefWeasel » Fri Aug 30, 2002 2:31 am

MikeC wrote: 1. I don't believe that's really the case -- I think the 2nd 12V connector is there to compensate for the losses with just the one standard ATX connector to the mobo. I think Intel discovered that the contact loss was too high too often & just doubled up to make sure there wouldn't be too much voltage drop. Most P4-connector outfitted PSUs appear to be exactly the same internally as the models the preceded them without that extra connector; it's just an additional set of wires going back to the same 12V source. Someone from Intel confirm/explain this one, please!

2. The question is why 18A on the 12V line?
Ive got some quotes from intel.com regarding the new ATX12V power supplies:

first quote:
The new +12VDC connector lies near the
processor's DC converter and voltage regulators,
enabling the delivery of more current, higher
power conversion and transmission efficiencies,
and future platform flexibility on Pentium 4
processor-based boards.


second quote:
The new +12V power connector enables the delivery of more +12 VDC current to the motherboard. The
presents of the +12V 4-pin connector indicates that the power supply is an ATX12V. The absence of the
+12V 4-pin connector indicates that the power supply is an ATX.

Increased +12VDC output is due primarily to the higher power conversion and transmission efficiencies
of +12VDC relative to +5VDC or +3.3VDC. Components on the motherboard having unique voltage
requirements are expected to be powered by DC/DC converters of the +12VDC power supply output.

An ATX power supply supports motherboards that rely on DC/DC conversion from +5 VDC and
+3.3VDC. However, an ATX12V power supply supports motherboards that rely on DC/DC conversion
from +12VDC, +5VDC, or 3.3VDC yielding more application flexibility for future platforms.


You can see from the first quote that the extra 12V connector is required to supply the processor with more current at 12V.

The second quote says that more components are expected to be powered by the DC/DC converters connected to the 12V line, rather than 5V or 3.3V. This also contributes to a P4 systems requirement of high current on the 12V line.

Didnt mean to start any arguments :oops: just thought that since no one had mentioned it yet, i should point out these interesting requirements of the P4, and that you cant just consider the total power draw of a computer system, you have to consider what is drawn at each voltage.

Ps: gotta love the spelling of 'presence' in the first paragraph of the second quote, nice one Intel :roll:

Oz
Posts: 32
Joined: Sun Aug 11, 2002 3:26 pm

Post by Oz » Fri Aug 30, 2002 4:34 am

ChiefWeasel wrote: Ps: gotta love the spelling of 'presence' in the first paragraph of the second quote, nice one Intel :roll:
LOL

Just trying to make sense of all the above, can I ask a simple question? Is it likely that a gamer (not really an overclocker) would need a 400W PSU in the near future? Say with a GeForce graphics card, 2 hdd's, AthlonXP 1800 etc.
I know it's a hard question to answer because of all the variables, but if the answer leans towards "you won't really need 400W in the forseeable future, even if you upgrade a few things", I'd be happy :D. I'm looking to buy a Seasonic model, and the place I found sells the 400W for just over $100, whereas the 300W is $59. Quite a difference...

ChiefWeasel
Posts: 262
Joined: Sun Aug 11, 2002 3:26 pm
Location: Worcester, UK

Post by ChiefWeasel » Fri Aug 30, 2002 5:04 am

As long as its on the AMD recommended list then i would say that you should be ok for the near future. 2 Hds and a geforce is nothing out of the ordinary so you should be well within the limits of a 300W AMD recommended PSU. Go for the 300W.

I did see a list somewhere of roughly how much power each component in a computer required, ill see if i can find it again cos i think it would be quite useful.

Oz
Posts: 32
Joined: Sun Aug 11, 2002 3:26 pm

Post by Oz » Fri Aug 30, 2002 5:32 am

Thanks Chief, that would be great.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Fri Aug 30, 2002 5:43 am

Presents indeed! :roll:
This has turned into a fun thread, btw. :D
You can see from the first quote that the extra 12V connector is required to supply the processor with more current at 12V.
I wholly agree with that. What I was suggesting is that if the losses at the ATX connector weren't so high, that second 12V connector *may* not have been needed. Conjecture, for sure, based on the +5V line problem reported so often in hardware forums, and explained so clearly by John Hoot in a link I thought I posted earlier.... http://www.overclockers.com/tips845/ The whole artilce is definitely worth reading, but here is a key section...

Now the main contributor to the voltage drop: The ATX connector itself.

Though the conductors coming to the connector may be large, their size is somewhat negated by being "funneled" down to the actual female pins in the connector and their associated male pins on the motherboard. Much like the surface of a rough finished heatsink mating to your CPU core, they do not necessarily make good contact.

The connector has to be loose enough to allow plugging and unplugging it without tearing the pins out of the connector and motherboard, so they are somewhat loose fitting. The more you plug and unplug that connector, the more of the mating surfaces erode and the looser they become. The looser the fit, the less contact area, the greater the resistance, the greater the voltage drop.

As these pins deteriorate, they embark on a downward spiral in performance. The less the surfaces are in contact and the resultant resistance, the more they heat up from the current passing through them. The more they heat up, the greater the resistance and the spiral continues. Some pins will get so hot as to melt the ATX connector.


BTW, that list of what components draw is available in the AMD system building guide someone referred to earlier. Similar docs available at Intel.

ChiefWeasel
Posts: 262
Joined: Sun Aug 11, 2002 3:26 pm
Location: Worcester, UK

Post by ChiefWeasel » Fri Aug 30, 2002 6:46 am

MikeC wrote: This has turned into a fun thread, btw. :D
It has indeed! And im sure TerryW will have something to add as well :)

Seems we were both coming at the same problem from different angles :P

That is an interesting article, not sure if ill be soldering extra power cables onto my mobo just yet thou :roll: - i didnt realise the ATX connector losses would be significant enough to cause problems for people.

There is a PSU you can get which has adjustable pots which you can use to adjust your voltages - i think it was the antec ones...

hyum
Posts: 90
Joined: Sun Aug 11, 2002 3:26 pm
Location: Santa Clara, CA

Post by hyum » Fri Aug 30, 2002 12:52 pm

the reason intel specified a new ATX power supply is because of the total current rating of the wires + ATX connector.

current rating of the ATX wires and especially the ATX connector, not voltage drop due to IR losses, is the real reason for the new 12V connector. voltage drop can be compensated with mechanisims like remote sense, although you lose effecientcy due to losses. however, with power requirements of the current _and future_ P4s, in a worst case scenario, the maximum rating for the wires and connector, especially, is violated. this is due to IR loses, which translates into heating and eventual melting of the connector, but as you can see at this point voltage drop is a moot point.

intel tends to be conservative with their ratings so this is worst case, plus this includes future P4s (3GHz+). therefore, in order to have transparent system upgrade, the new ATX spec. is mandated on system builders now. 12V was chosen so that less current is required to carry to the core regulators (which was powered from 5V before), which equals less new pins required for precious board real estate, (with the comprimise of less efficient regulator design).

hyum
Posts: 90
Joined: Sun Aug 11, 2002 3:26 pm
Location: Santa Clara, CA

Post by hyum » Fri Aug 30, 2002 1:37 pm

to give a little more background info:

the 5V wires delivered the most power so typical motherboard design used the 5V rail to supply all switching supplies on the mobo (e.g., CPU core, chipset core, system bus termination, and for DDR boards, DDR voltages and DDR termination) PLUS expansion slots (e.g., PCI, AGP) and peripherials (e.g., drives, etc.). due to the power hungry nature P4s, the 5V lines do not have the current rating to support current and future P4s. ATX specification recommends not exceeding 30A on the 5V rail (i believe it's 6A max per wire), which is 150W. current workstation and low end servers can exceed this, not to mention dual CPU systems. Add a few SCSI drives, a couple high end RAID UW SCSI cards, a few gigabit NICs, and 4GB DRAM, and you can see what i mean.

since the the ATX connector could not meet the power requirements, a new connector was needed. hence the ATX12V specification. the reason 12V was choosen over of 5V for the new connector is moving to 12V means less current versus 5V, which translates to less pins/less board space for the new connector. hence 12V was choosen to power the CPU core, taken the burden off the ATX connector.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Fri Aug 30, 2002 3:15 pm

hyum, that's good info. Was it in the Intel ATX12V spec docs?

hyum
Posts: 90
Joined: Sun Aug 11, 2002 3:26 pm
Location: Santa Clara, CA

Post by hyum » Fri Aug 30, 2002 4:32 pm

the data is from the ATX spec. i think you can get them at formfactors.org or the intel site. also you can get info from the intel p4 board design guides. not sure if these are published on the web or not.

the rational iteself came from an intel engineer at fall idf 2001 when they were intorducing p4s and the system/board ramifications it entailed (ATX12V was probably the largest system impact, heatsink guidlelines/mounting and core regulators were probably the largest board impacts).

fyi, i'm a hardware engineer designing OEM/ODM high-end embedded computing platforms, mainly for datacom/telecom equipment providers. this includes but is not limited to pentium designs. our group also does motorola & imb ppc, mips, network processors, and more.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sat Aug 31, 2002 10:54 am

There seems little question of the benefits of the ATX12V cable/connector; does anyone know whether this connector is used for ANY AMD mobos? Seems to me there is no reason why it would not benefit AMD based PCs as well.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Sep 01, 2002 10:15 am

Another point made by John Hoot -- http://www.overclockers.com/tips845/

To determine whether the +5V drop is the result of a PSU that has insufficient wattage for your system setup, or an ATX connector problem, all you need to do is take an unused Molex connector and measure the +5V on it while running your CPU at high load.

Since the Molex connector is connected to nothing, no current is flowing through it and hence no voltage drop is occurring. Under that circumstance, the voltage you read on it will be exactly what the PSU is putting out at the +5V common distribution point.

If that voltage is nearly as low as that read on your motherboard, then your PSU is not powerful enough. If it reads near to 5V and distinctly higher than what you read on the motherboard, then the ATX connector is your likely suspect


Obviously these comments apply to ANY of the voltage lines. On my main P4 test system, I've noticed the 12V line often fluctuates despite the additional 12V connector -- regardless of PSU used. I tried 2 PSUs this morn. Under a variety of work loads, the +12V line read by MBM5 ranged 11.48-11.84V. However, the reading directly from a 4-pin connecter was rock solid at 12.02 with one PSU and ranged 11.96-11.98V with another.

There is little question the voltage drop is occuring somewhere between the +12V common distribution point in the PSUs and where the mobo sensors are reading the voltages for MBM5. It seems odd that this variable drop is happening even with the extra 12V connector. I will have to fire up another P4 mobo I have on hand to see if the voltage regulators on the Gigabyte mobo are suspect.[/i]

Post Reply