KK: We'll use heat pipes and a custom cooling solution - PSU could sold separately

[Carl B]

It depends on the quality of the power supply, the actual power circuitry barely takes up any space, but the cooling devices do. I've noticed the "higher quality" power supplies generally look the same as cheapie ones, but with huge, heavy heatsinks on them. Like my near silent 480W Antec Neopower. Then to contrast that I have a no-name 550W power supply (which actually cost a fair amount of money, though not as much as the neopower, and does deliver higher amounts of power than the Neopower does) which is extremely light but has two loud fans to cool it. Then I can pick up a generic 400W (say, one of those $10 or $15 ones) power supply which is again super light, has quiet fans, and delivers decent amounts of power in comparision to the Antec, but is likely to die a very quick death due to lack of cooling.
I'd imagine Sony's isn't going to have the kind of passive cooling the Antec does due to cost, so in order to keep the thing from frying, we're probably looking at a relatively loud fan on it. (for its power output)

Well, they say they're going to try and match the noise signature of the slimline PS2, so on those grounds alone I'm not sure about a 'loud' fan. I suspect your 550W power supply is being cooled via two 80mm fans as opposed to some that are cooled by a single 120mm. I just dealt with a 500W Ultra V-Power that had a single 120 and carried 28 amps on the 12v rail, and I was very pleased with it.

Since they're using heat pipes, I wouldn'tbe surprised if fans period are kept to a minimum, with the piping over the RSX and CPU leading towards the rear of the case were perhaps one larger or slower RPM fan pushes the heat out, and then maybe another fan located towards the opposite end of the case to assist with airflow.

I've dealt with Antec PSU's before as well, and a lot of their appeal is the level of active PFC they have. I think that from the way KK is talking, this power supply is going to be quite strong in the size : power efficiency department, especially if it does incorporate that FPA tech.

Anyway we'll just see in the end. I'm looking forward to Anandtech or someone (maybe PSINext?) busting a PS3 up when it launches and a thorough innards examination.

 

[Rolf N]

5. Why not an external PSU like Microsoft did?

Because it's a fuckin' stupid idea? Externalize a power supply that dissipates enough heat to require some level of cooling?

This isn't a ~40W PS2 or Gamecube PSU, this is a significant PSU, and it won't work if you throw it into a thick black plastic case and lay it on a carpet without any ventilation.

Damn myself for explaining it because some MS rep will read this and finally start to understand the issue ...
The power supply was moved into the unit deliberately. The innards of the console have a cooling concept in place, which the old-fashioned (lower power) power bricks lack, with a certain amount of heatsink surface area and airflow. The PSU simply reuses this. The alternative would be power brick with a fan or a horrible design failure.

http://arstechnica.com/news.ars/post/20051125-5625.html

The culprit, or one of them at least, is the power supply. Apparently the power supply brick easily overheats, which in turn causes power line problems which render the box unstable. We've heard that improving air flow around the supply helps. For some this means merely moving it off of their shag carpeting, while others have found the propping it up on a open box or suspending it with a cord or string does the trick.

 

[Dave Baumann]

I think there is some flow through the 360 power blocks - at least there are a set of holes either end of the brick, of which one side constantly attracts dust.

 

[SugarCoat]

Because it's a bleepbloopin' stupid idea? Externalize a power supply that dissipates enough heat to require some level of cooling?

This isn't a ~40W PS2 or Gamecube PSU, this is a significant PSU, and it won't work if you throw it into a thick black plastic case and lay it on a carpet without any ventilation.

Damn myself for explaining it because some MS rep will read this and finally start to understand the issue ...
The power supply was moved into the unit deliberately. The innards of the console have a cooling concept in place, which the old-fashioned (lower power) power bricks lack, with a certain amount of heatsink surface area and airflow. The PSU simply reuses this. The alternative would be power brick with a fan or a horrible design failure.

http://arstechnica.com/news.ars/post/20051125-5625.html

Actually having it outside and seperate does alot more for cooling. Stuffing it into a black shell just wasnt a smart move. Hasnt anybody opened one yet? Little bit surprised at that, i thought a good chunk of it was empty.

By putting a PSU into a confined area that cools off already hot air from other components you essentially require it to have multiple fans and problably an aluminum casing. It is not an optimal solution for keeping the PSU cool by any means. Niether is stuffing one into a black shell. Microsoft does have the advantange of packing smaller power bricks with the consoles down the road as the parts get smaller and thus requiring less power. Sony will problably release another design which is a bit more expensive in terms of R&D which i imagine played a part in Microsofts decision as well. Sony is after all releasing the biggest console, in terms of size, ever.

.
.
.
.
.
.




I had to do it!

Seriously though its quite large.

 

[OtakingGX]

Not necessarily. A heatpipe or thermosiphon is the thermal equivalent of a really really good heatsink, which means that it can spread heat quickly over a much larger surface than a metal heatsink. The larger surface means that more air is touching the heat dissipator.
So a heatpipe can work at higher temperatures without a fan.

Heatpipes are great for moving a lot of heat, quickly. At the other end of the heatpipe, though, that heat must be "discharged."

Imagine you had a heatpipe with a dinky little radiator (heatsink) at the other end. You turn on the piece of electronics it's attached to, and at first it works great. The fluid in the heat pipe has a huge thermal capacity and will take a lot of heat to increase in temperature. Once you reach steady state though, heat in = heat out. With the dinky little radiator you have a small surface area. You must increase the temperature of the radiator (compared to that of a larger one) to reject the same amount of heat. If the radiator is hotter then the heatpipe is hotter and so is the piece of electronics. So a smaller radiator means a hotter chip.

You're right, the surface of the pipe itself will convect heat. The bulk of your heat transfer though, will be through the radiator. If you had no heatsink, your heatpipe would cool things very poorly, as it only has the surface area of the pipe to convect heat.

And has anyone taken apart an Xbox 360 PSU? I mean, it's not the smartest thing to do, considering you could get shocked pretty badly, but I'm interested to know if there is a fan in there, or if it's cooled by buoyant convection.

Actually having it outside and seperate does alot more for cooling. Stuffing it into a black shell just wasnt a smart move. Hasnt anybody opened one yet? Little bit surprised at that, i thought a good chunk of it was empty.

Black is the best possible color of shell Microsoft could have stuffed their PSU into. Black not only absorbs the most heat via radiation, but it radiates the most heat. Assuming that the power supply is not sitting in direct sunlight, it's safe to say that being black only helps the heat issue.

And Sony and Microsoft will both spend the same on R&D to reduce the size of components inside their consoles. Reducing the exterior dimensions would cost more in retooling production lines than R&D. Towards the end of the console's life that would be a good thing to do to such a big chassis, as you can save money on materials too.

 

[Rolf N]

By putting a PSU into a confined area that cools off already hot air from other components you essentially require it to have multiple fans and problably an aluminum casing.

I don't understand you. There already is a need for a cooling system inside the main unit case. There already is a need for proper airflow, and a fairly large heatsink that's shared between the major heat emitters is a pretty logical next step. E.g. C1 and Xenon share a single heatsink and fan. Why shouldn't it be possible to load the PSU's heat output off onto such a shared heatsink as well, in turn saving a dedicated heatsink and a dedicated fan?

The PSU's heat dissipation which I'd wager will lie in the 15~30W ballpark can simply piggy-back on that bigger cooling infrastructure. It's nowhere near the heat output of RSX or Cell (80%+ efficiency isn't unreasonable to expect from a modern PSU IMO). It would only be a problem if it sits in its own case, outside of the primary airflow. Hence I really don't understand why you say the internal PSU leads to a new requirement of "multiple fans and probably an aluminum casing". That's approximately what you'd use for the hot main components of the machine anyway. The PSU won't make it that much worse, just more cramped, that's a given, but then there needs to be a reason why the PS3 case is bigger than the XBox360 case.

It is not an optimal solution for keeping the PSU cool by any means. Niether is stuffing one into a black shell. Microsoft does have the advantange of packing smaller power bricks with the consoles down the road as the parts get smaller and thus requiring less power. Sony will problably release another design which is a bit more expensive in terms of R&D which i imagine played a part in Microsofts decision as well. Sony is after all releasing the biggest console, in terms of size, ever.

And again I don't understand you. Microsoft can shrink the power brick, yes. They could also move it into the main unit if they find it to make more sense.

OTOH Sony can just as well make their own shrinks and power reductions. They've done it before at least. They might even decide to move the PSU back out of the case, into a power brick, if its heat dissipation becomes manageable after some die shrinking etc.

The external PSU doesn't make sense now IMO, obviously. But I reserve the right to change that opinion when the power reqs go down, and I'm sure the respective system engineers reserve the same right.
I just don't see how the external power brick is supposed to put Microsoft at an advantage here. Sorry.

I had to do it!

Seriously though its quite large.

It has to be. There are simply more things inside in comparison to the XBox360. But that's okay I guess in exchange for a device that keeps itself and its PSU cool with a single, reasonable fan -- okay, I'm speculating, slightly.

 

[patsu]

I had to do it!

Seriously though its quite large.

To be fair, footprint-wise, XB360 will be much larger than PS3 if we include the external power brick (which has to be placed/suspended somewhere) and the external HD-DVD player. Due to the concave casing and heat signature, you may not be able to stack them.

 

[Shifty Geezer]

I don't understand you. There already is a need for a cooling system inside the main unit case. There already is a need for proper airflow, and a fairly large heatsink that's shared between the major heat emitters is a pretty logical next step. E.g. C1 and Xenon share a single heatsink and fan. Why shouldn't it be possible to load the PSU's heat output off onto such a shared heatsink as well, in turn saving a dedicated heatsink and a dedicated fan?

Basically, a heatsink can get rid of x amount of heat, and if you add more heat to it beyond what it can cope with, components will warm up. If the sink used for RSX+Cell is just enough to run them properly, you can't piggyback an extra PSU's worth of heat off it. It's more a cost/simplicity thing to combine them all into one if you can. One larger heatsink + one bigger fan is going to cost less than one heatsink for chips and another for PSU, both with fans. Depending on the materials you use, airflow, and amount of heat you need to get rid of, determines whether you can run three devices of one large heatsink or not.

 

[SugarCoat]

I don't understand you. There already is a need for a cooling system inside the main unit case. There already is a need for proper airflow, and a fairly large heatsink that's shared between the major heat emitters is a pretty logical next step. E.g. C1 and Xenon share a single heatsink and fan. Why shouldn't it be possible to load the PSU's heat output off onto such a shared heatsink as well, in turn saving a dedicated heatsink and a dedicated fan?

Well i'm thinking in terms of the PS3 cooling essentially acting like that of a laptop or small desktop computer. In order to remove heat sufficiently, especially with everything internal, they're going to require a more robust cooling solution, be it alot of copper and aluminum, or small fans at high RPM. For example if there was room in the Xbox360 for what is contained in the power supply, you cant expect the PSU to stay cool passively just by whats already within the unit, its going to need its own cooling as well which will add cost and perhaps noise. Otherwise you'd be right back to where you have your complaint, which is that the power supply is overheating by itself. Now add the atmosphere of the inside of a console and its going to be worse, not better. Seriously imagine it like a computer for a moment. Whats cooler for the PSU, the temp of the air outside of the case or the temp of the air inside the case? Why do you think all modern computer power supplies are equiped with their own exhaust fans?

The PSU's heat dissipation which I'd wager will lie in the 15~30W ballpark can simply piggy-back on that bigger cooling infrastructure. It's nowhere near the heat output of RSX or Cell (80%+ efficiency isn't unreasonable to expect from a modern PSU IMO). It would only be a problem if it sits in its own case, outside of the primary airflow. Hence I really don't understand why you say the internal PSU leads to a new requirement of "multiple fans and probably an aluminum casing". That's approximately what you'd use for the hot main components of the machine anyway. The PSU won't make it that much worse, just more cramped, that's a given, but then there needs to be a reason why the PS3 case is bigger than the XBox360 case.
And again I don't understand you. Microsoft can shrink the power brick, yes. They could also move it into the main unit if they find it to make more sense.

See comments above for most of what you brought up. Atmosphere inside a console is going to be worse to the PSU then the atmosphere on the floor behind your entertainment center. As far as the Xbox360, its cramped already, so my point about the shrinking external PSU was that they could, for example, at 65nm chips in a ~year, reduce the cost of the PSU and shrink it without touching the actual box. Should make for a very easy and fast way to reduce costs of at least that component.

To be fair, footprint-wise, XB360 will be much larger than PS3 if we include the external power brick (which has to be placed/suspended somewhere) and the external HD-DVD player. Due to the concave casing and heat signature, you may not be able to stack them.

The picture was ment as a joke . The actual console is bigger then the origonal Xbox by just over 1inch in height and length. Its not too bad at all but its still the largest console to date.

 

[Rolf N]

Well i'm thinking in terms of the PS3 cooling essentially acting like that of a laptop or small desktop computer. In order to remove heat sufficiently, especially with everything internal, they're going to require a more robust cooling solution, be it alot of copper and aluminum, or small fans at high RPM. For example if there was room in the Xbox360 for what is contained in the power supply, you cant expect the PSU to stay cool passively just by whats already within the unit, its going to need its own cooling as well which will add cost and perhaps noise.

I think you're overestimating the significance of PSU heat output. It's important to note that PSU (peak) heat dissipation is directly proportional to its rated power output*1/efficiency.
99%+ of the electric energy going into the console will have to leave it as heat as well.

A PSU for a console that draws 100W (reasonable guess) max with 80% efficiency (conservative guess) will draw 125W from the outlet and dissipate up to 25W of heat, while the console it powers will consume/dissipate-as-heat up to 100W. Correct?

Thus when moving the PSU inside and onto the same cooler, the cooler will have to cope with 25% more heat (no matter what the absolute figure turns out to be). It's less of an increase for higher PSU efficiencies, but anyhow, making a cooling system 25% better isn't that much of a huge leap IMO, and also IMO it could even be a net cost reduction if it means you can get rid of a secondary cooling system altogether, not to mention plugs and cabling and extra packaging costs.

Otherwise you'd be right back to where you have your complaint, which is that the power supply is overheating by itself. Now add the atmosphere of the inside of a console and its going to be worse, not better. Seriously imagine it like a computer for a moment. Whats cooler for the PSU, the temp of the air outside of the case or the temp of the air inside the case? Why do you think all modern computer power supplies are equiped with their own exhaust fans?

In a lower-end ATX system the fan inside the PSU is the only fan that moves any air through the system at all. Also the PSU sits at the bottom of the airflow food-chain in an ATX case, it's cooled effectively with "exhaust" air and still works.
An ATX PSU isn't the right example to convince me that integrated PSUs are problematic I'm afraid.

The point is not ambient temperature. If ambient temperature was the deciding factor in cooling, we wouldn't use fans. Those bricks have failed because of insufficient airflow, not because of warm ambient air.

Atmosphere inside a console is going to be worse to the PSU then the atmosphere on the floor behind your entertainment center.

I believe the opposite to be true, due to "natural" airflow inside the console case.

 

[Bigus Dickus]

The environment inside the 360 is much warmer than ambient temp. Had the PSU components been included inside the main case, chances are that natural convection would not have sufficed due to the high case environment temp. Thus, another HS and fan for the PSU or a larger HS and fan that covered CPU/GPU/PSU would have been required.

With an external brick, it is possible that natural convection provides adequate cooling so long as ambient temp remains within the expected range (i.e., not closed up in a little box without any ventilation (not forced) to room air).

So the trade study in cost is the extra price of the PSU housing and extra cabling and connectors, vs. the extra heatsinking and fan capacity, as well as case size had it been included internally. That, combined with possibly the wish to keep the main case smallish for marketing reasons.

Now, I don't know if the PSU brick has natural or forced convection... I'm just making a "what if" analysis, but it wouldn't surprise me at all to find the brick fanless. And it wouldn't surprise me at all if that actually turned out less expensive, or perhapse with such a marginally higher cost that it was a wash economically.

 

[Squeak]

Imagine you had a heatpipe with a dinky little radiator (heatsink) at the other end.

That's my whole point. You can put all the copper in the world on a processor, at some temperature the copper won't be able to get the heat away fast enough from the small point. In that case a heatpipe will be able to spread the heat dissipation out over a much larger surface, allowing the design to have far greater thermal output before a fan is needed. So of course you will need a big radiator, but with a heatpipe you can actually use it.

 

[SugarCoat]

Zecken i'm not talking at all about the heat output of a small psu, i am telling you that its fact that you cant just put the power unit into a console without making the cooling better. You're saying its bad that the Xbox360 power unit is in a plastic case outside the console but its somehow healthier inside a stuffy console and i'm telling you that makes zero sense unless you have improved cooling to compensate and keep those components cooler then what they were malfunctioning at outside of the case.

Basically from what you're saying, is that a 50C range of power supply components outside of a computer or console is worse then a 60C range of those same components if it were inside but have that hot air blowing on them. That doesnt make sense. Just because the air is moving doesnt magically make it any better on the components themselves. That defies the laws of thermal physics.

If the failure temp is 50C, and you internalize the component, you need to make absolutly sure that the air blowing through those components is cooler then 50C. The temperature at which the components fail doesnt magically go up if the air is moving.

This is all assuming that temperature is indeed a main cause of failure.

 

[OtakingGX]

The environment inside the 360 is much warmer than ambient temp. Had the PSU components been included inside the main case, chances are that natural convection would not have sufficed due to the high case environment temp. Thus, another HS and fan for the PSU or a larger HS and fan that covered CPU/GPU/PSU would have been required.

The PSU would not be subjected to natural convection inside the case though. Air is moving around. This increases the convective heat transfer greatly over free convection. And while noisy, turbulent air convects heat much better.

That's my whole point. You can put all the copper in the world on a processor, at some temperature the copper won't be able to get the heat away fast enough from the small point. In that case a heatpipe will be able to spread the heat dissipation out over a much larger surface, allowing the design to have far greater thermal output before a fan is needed. So of course you will need a big radiator, but with a heatpipe you can actually use it.

Given the same heat sink, a heat pipe and a standard heat sink will still need to remove the same amount of heat. The advantage the heat pipe has is that it can give you a heat sink with a more evenly distributed temperature than the non heat pipe. This means less air flow for the same cooling, but not no air flow. Given the amount of heat we're estimating here, it will most likely need to move a lot of air. The hardware can't be designed for best cases, it needs to still work in Arizona when it's 110 F outside. If you want a maximum temperature of 150 F on your chips, that doesn't give you a big theta to work with when removing all that heat.

Basically from what you're saying, is that a 50C range of power supply components outside of a computer or console is worse then a 60C range of those same components if it were inside but have that hot air blowing on them. That doesnt make sense. Just because the air is moving doesnt magically make it any better on the components themselves. That defies the laws of thermal physics.

Amazingly, it does! That's what forced convection is all about. The equation to find heat transfer rate is Q = h * theta (where theta is the difference between surface and ambient temperature). h gets big quickly when air is moving. If you can triple h then having a theta that is only half as large is worth it.

 

[SugarCoat]

Really. So you agree then that if the Xbox360s power unit was internalized without any improvment to the cooling of the box as it stands now the failure rate would be lower?

 

[macabre]

I think you're overestimating the significance of PSU heat output. It's important to note that PSU (peak) heat dissipation is directly proportional to its rated power output*1/efficiency.
99%+ of the electric energy going into the console will have to leave it as heat as well.

A PSU for a console that draws 100W (reasonable guess) max with 80% efficiency (conservative guess) will draw 125W from the outlet and dissipate up to 25W of heat, while the console it powers will consume/dissipate-as-heat up to 100W. Correct?

As PS3 is supposed to use Vicor`s tech, they have some numbers on the thermal management here:

http://www.vicorpower.com/products/vichip/bcm/thermal_management/
http://www.vicorpower.com/products/vichip/vtm/thermal_management/

 

[OtakingGX]

Really. So you agree then that if the Xbox360s power unit was internalized without any improvment to the cooling of the box as it stands now the failure rate would be lower?

Yes. Get some of that forced airflow over the PSU components. As far as I'm aware, the overheating problems all have to do with the PSU. The cooling of the chips in the Xbox 360 is fine. Even if the air is warmer from waste heat of the CPU and GPU it's better that it's moving.

 

[Fox5]

Yes. Get some of that forced airflow over the PSU components. As far as I'm aware, the overheating problems all have to do with the PSU. The cooling of the chips in the Xbox 360 is fine. Even if the air is warmer from waste heat of the CPU and GPU it's better that it's moving.

Doesn't the air still have to be cooler than the psu in order to cool the PSU instead of warming it?

 

[OtakingGX]

Doesn't the air still have to be cooler than the psu in order to cool the PSU instead of warming it?

Yes. But the air coming out the back of the 360 is far cooler than the electronics will be. Some of the hottest components, the switching regulators, have operational temperatures from -40C to 125C for better components. I don't own an Xbox 360, but I doubt it belches flames out the back. The exhaust should be well below 125C.

Good 60mm fans can move about 25 cfm of air. I'll err on the side of safety and say that Xbox 360's two fans can move 30 cfm combined. Also I'm assuming the PSU is 80% efficient, so the recorded 160 watts at the wall means that the internals are turning 130 watts into heat. At 30 cfm 130 watts will increase the temperature of air by 7.3 C. If your living room is 25 C, now you're cooling with 32.3 C air. With an internal power supply you need to cool those remaining 30W of power that the PSU is turning into heat. That's another 1.7 C. So the exhaust on the back should be a nice toasty 34 C.

 

[3dcgi]

As far as I'm aware, the overheating problems all have to do with the PSU. The cooling of the chips in the Xbox 360 is fine.

I think you are mistaken. I've talked to more than 20 Xbox 360 owners and not one has had a problem with the power supply. I've personally had to return an Xbox 360 due to overheating and it was the main box. I didn't even send the PSU to Microsoft when I returned the main unit.