PSU on top: why?
Posted: Mon Nov 21, 2011 9:59 am
Many modern cases still have a top-mounted PSU. I was wondering: why? What's the disadvantage of bottom-mounted PSUs?
Discussions about Silent Computing
https://silentpcreview.com/forums/
https://silentpcreview.com/forums/viewtopic.php?f=15&t=63459
The original ATX Chassis was designed so that the PS fan would cool the CPU heat sink via a duct and many systems with low wattage CPU's are still built this way.Olaf van der Spek wrote:Many modern cases still have a top-mounted PSU. I was wondering: why?
You will have to select a more expensive PS with longer cables.Olaf van der Spek wrote:What's the disadvantage of bottom-mounted PSUs?
I know, but even cases like the new Solo II have a PSU on top.Arbutus wrote: The original ATX Chassis was designed so that the PS fan would cool the CPU heat sink via a duct and many systems with low wattage CPU's are still built this way.
The solo II has a vent above the PSU, so you can have it draw air from the outside - as such, it should be functionally equivalent to most cases with bottom-mounted PSUs (and it should be a particularly good choice for a fanless PSU). It's definitely not the same as traditional ATX cases where the PSU draws air from the inside of the case.Olaf van der Spek wrote:I know, but even cases like the new Solo II have a CPU on top.Arbutus wrote: The original ATX Chassis was designed so that the PS fan would cool the CPU heat sink via a duct and many systems with low wattage CPU's are still built this way.
Ah. But then PSU fan noise has a direct exit path.Cyäegha wrote:The solo II has a vent above the PSU, so you can have it draw air from the outside - as such, it should be functionally equivalent to most cases with bottom-mounted PSUs (and it should be a particularly good choice for a fanless PSU). It's definitely not the same as traditional ATX cases where the PSU draws air from the inside of the case.Olaf van der Spek wrote:I know, but even cases like the new Solo II have a CPU on top.Arbutus wrote: The original ATX Chassis was designed so that the PS fan would cool the CPU heat sink via a duct and many systems with low wattage CPU's are still built this way.
Not only that but any spills, etc go straight into the PSUOlaf van der Spek wrote:Ah. But then PSU fan noise has a direct exit path.Cyäegha wrote:The solo II has a vent above the PSU, so you can have it draw air from the outside - as such, it should be functionally equivalent to most cases with bottom-mounted PSUs (and it should be a particularly good choice for a fanless PSU). It's definitely not the same as traditional ATX cases where the PSU draws air from the inside of the case.
IMO they`re far from negligible, especially when dealing with higher ambient temps and slower psu fans. Moving the psu at the bottom noticeably reduced exhaust temperature for me and allowed me to use a slower fan. Another drawback with the typical placement at the top is that the psu fan is often competing for airflow against the rear case exhaust. Having the psu at the bottom with it`s own intake you don`t have to worry about it messing with case airflow.IMO, I don't think there's any reason to move the PSU to the bottom location apart from marketing. It has just become a hype. Thermal gains are negligible. The "convection theory" falls apart if there are any fans in the case.
I never liked top vents, especially after a friend of mine brought in his pc a few days ago, after having spilled coke into it..I very much dislike top fan vents for this reason and wish it was easier to find a case without them, as I'm only going to have to try and block them up to protect my valuable components from damage and it's not easy to make them water-tight.
I have a different experience. My previous system (C2D 4300 OC'ed to 3GHz) has been delegated to serve as my parents' general computer. I've moved the system to a new case with the PSU at the bottom. Temps are exactly the same as they were while the PC was housed in my Antec Solo. The heatsink is a passive Scythe Ninja 2.IMO they`re far from negligible, especially when dealing with higher ambient temps and slower psu fans. Moving the psu at the bottom noticeably reduced exhaust temperature for me and allowed me to use a slower fan. Another drawback with the typical placement at the top is that the psu fan is often competing for airflow against the rear case exhaust. Having the psu at the bottom with it`s own intake you don`t have to worry about it messing with case airflow.
A C2D isn't exactly a high-end CPU that produces a lot of heat. I assume you didn't have a dedicated GPU either.KadazanPL wrote: I have a different experience. My previous system (C2D 4300 OC'ed to 3GHz) has been delegated to serve as my parents' general computer. I've moved the system to a new case with the PSU at the bottom. Temps are exactly the same as they were while the PC was housed in my Antec Solo. The heatsink is a passive Scythe Ninja 2.
Again, a 5750 isn't a high-end GPU...Current system (see my sig), despite a substantial overclock, remains cool at all times, thanks to the Solo's design, i.e. optimal front-to-back airflow.
I don't think fans running below 1000RPM will ever compete for airflow. There's simply not enough suction.
How about the temps inside the psu though? Even if the psu fan stays at a low speed, it doesn`t mean that all the components inside are adequately cooled, since there are no sensors on each one of them. While the main heatsink could stay cool enough, components receiving less airflow like the secondary capacitors could suffer. At best you`d get reduced efficiency because of the increased temperatures.I have a different experience. My previous system (C2D 4300 OC'ed to 3GHz) has been delegated to serve as my parents' general computer. I've moved the system to a new case with the PSU at the bottom. Temps are exactly the same as they were while the PC was housed in my Antec Solo. The heatsink is a passive Scythe Ninja 2.
Well, it isn't a high-end CPU but it does produce a lot of heat. Please consider that there's a heavy overclock involved - from 1.8 to 3.0GHz and the production process is 65nm. All energy-saving features had to be turned off and voltage has been increased for stability reasons. All in all, it is a hot CPU, but nothing Ninja 2 couldn't handle passively.Olaf van der Spek wrote: A C2D isn't exactly a high-end CPU that produces a lot of heat. I assume you didn't have a dedicated GPU either.
Granted, it isn't high-end, but it's passive and all the heat is dumped inside the case. Temps are 40 at idle and 80 under load. I'm quite certain that a hotter card would do just fine, I was hoping to swap it for a passive HD6770, but will probably wait for the 7xxx generation to arrive (5750 is adequate for my gaming needs just now).Again, a 5750 isn't a high-end GPU...
You can't just generalize from your system.
The savings on the increased efficiency of a cool PSU will take ages to equal (or justify) the cost of a new case (that basically was my point). A test should be conducted - same components, same fans&flow - different cases (one top-mounted PSU, one bottom-mounted PSU). And then measure the temperature of the air coming out from the PSU.How about the temps inside the psu though? Even if the psu fan stays at a low speed, it doesn`t mean that all the components inside are adequately cooled, since there are no sensors on each one of them. While the main heatsink could stay cool enough, components receiving less airflow like the secondary capacitors could suffer. At best you`d get reduced efficiency because of the increased temperatures.
What's a lot of heat?KadazanPL wrote:Well, it isn't a high-end CPU but it does produce a lot of heat. Please consider that there's a heavy overclock involved - from 1.8 to 3.0GHz and the production process is 65nm. All energy-saving features had to be turned off and voltage has been increased for stability reasons. All in all, it is a hot CPU, but nothing Ninja 2 couldn't handle passively.
This CPU produces 135W of heat. (when overclocked, it's TDP is 135W to be precise) That's a lotOlaf van der Spek wrote: What's a lot of heat?
It's still just a dual core. Others might have quad (or more) cores, producing even more heat. A friend of mine has a Q6600 overclocked to 3+ ghz.
It consumed 108W of energy (peak 3D) but produced quite a lot of heat due to its inefficiency. It also idled quite high. On the other hand, the heatsink was designed in such a way that it blew hot air outside the case. A passive card can't do that.A 8800 GTS uses less energy and produces less heat than more modern cards (AFAIK).
I'm not going ot deny thatA PSU that's fed cold air takes less to cool than a PSU that's fed hot(ter) air.
I didn't say that.AFAIK there's no reason why mounting a PSU at the bottom would require a larger case.
I totally agree.KadazanPL wrote:IMO, I don't think there's any reason to move the PSU to the bottom location apart from marketing. It has just become a hype. Thermal gains are negligible. The "convection theory" falls apart if there are any fans in the case.
In 1998 you had generic beige boxes with moulded fascia.
In 2004 you had "Prescot ready cases" and windows.
In 2006 you had aluminum cases and kitsch cold-cathodes.
In 2010 you had bottom-mounted PSUs, black painted interiors and motherboard cutouts.
All of the above don't improve much performance or thermal-wise, but are simple selling-points to convince you that you need to change a perfectly good case to a newer, allegedly "improved" one.
I have seen builds with original design cases where the case fan was so much more powerful than the PS fan the there was no air flowing through the PS.ntavlas wrote:...with the typical placement at the top is that the psu fan is often competing for airflow against the rear case exhaust.
I know. Where did you get that 135 W value from?KadazanPL wrote:This CPU produces 135W of heat. (when overclocked, it's TDP is 135W to be precise) That's a lotOlaf van der Spek wrote: What's a lot of heat?
It's still just a dual core. Others might have quad (or more) cores, producing even more heat. A friend of mine has a Q6600 overclocked to 3+ ghz.
It doesn't matter how many cores you have - only the amount of heat generated (or, speaking precisely, the TDP). A single core can produce that much (remember Pentium4?), as well as a hexacore. The efficiency is the key and it varies.
You do know that every W that enters your computer as power is transformed into a W that exits your computer as heat, right?It consumed 108W of energy (peak 3D) but produced quite a lot of heat due to its inefficiency.
You know, every time you dismiss something fairly high-powered of his, you're further restricting the cases where a bottom-mounted PSU is relevant. Most machines now are in the vicinity of 100w total, being Intel systems with integrated graphics and a couple drives. Many will be more realistically in the vicinity of 50w. If his ~250w machine "isn't very much", then what does need a bottom-mounted PSU: SLI rigs with Extreme or Bulldozer CPUs? That makes the orientation irrelevant to the vast majority of people.Olaf van der Spek wrote:Still, 108 W isn't much for a videocard.
The vast majority of people don't care about noise that much, so it's already irrelevant for them.Scrooge wrote:That makes the orientation irrelevant to the vast majority of people.
I'm glad it's not just me then. If you know of any decent budget (<£40) cases without them (I need four 5.25" bays (in fact preferably all the way from top to bottom as then I can fit my HDs with adapters, but if not either a removable drive cage or one with half-decent vibration-dampening rails) and external removable intake fan filter(s) or space to fit magnetic ones) please let me know, as everything I've found seems to have top vents. I was looking for decent cable management, but I'll sacrifice that now in desperation, and I couldn't really care where the PSU is (providing if it's bottom mounted it has an external, removable filter).ntavlas wrote: I never liked top vents, especially after a friend of mine brought in his pc a few days ago, after having spilled coke into it..
The point I wanted to make was that when building a new machine, of from the manufacturer`s/designer's POV, when developing a new case, moving the psu to the bottom is a sound choice and not just a marketing gimmick. It adds no extra cost or weight and at worst offers a small improvement in temperatures and noise. The main drawback is that a longer 4pin power cable is needed something that could be a bit of a PITA before psu makers adapted, but nowdays, it`s much less of an issue.The savings on the increased efficiency of a cool PSU will take ages to equal (or justify) the cost of a new case (that basically was my point). A test should be conducted - same components, same fans&flow - different cases (one top-mounted PSU, one bottom-mounted PSU). And then measure the temperature of the air coming out from the PSU.
Perhaps this has been done somewhere? Does anyone have a link?
You'd put the PSU intake at the bottom, not at the top.PartEleven wrote:His third graphics card essentially sits on top of the PSU intake fan.
Your first point is a cop-out: I don't believe the bottom-mounted PSU makes any difference to noise for the vast majority of builds, so people's preferences don't matter. Anyone building even a typical gaming machine in a mid-tower should be able to have it quiet and ventilated with either orientation, and only people who care enough to build their own machines will even be aware of the choice of PSU location, so we're starting with a more aware group of users.Olaf van der Spek wrote:The vast majority of people don't care about noise that much, so it's already irrelevant for them.
BD/CF/SLI are exaggerated, I was thinking about R6950 or GTX 560 Ti...
That said, I'm still not aware of any disadvantages of PSU on bottom.
It does influence me somewhat, as I don't want my PSU sucking unfiltered dust from outside the case so if it doesn't have an external removable filter it puts me off. I don't particularly want to put the PSU fan side up either, as that not only moves the cables further from where they need to go (which might mean having to buy/use extension cables) but also means that screws, etc can be dropped in it when working on it!Scrooge wrote:That said, I wouldn't base my choice of case on the position of the PSU. It does make cable management harder/ more annoying, but many cases do that on their own. It's neutral, to me, I just don't believe it's the big deal some are making it out to be for the vast majority of builds.
Then the PSU's thermal effect on the rest of the system is really no different from having the PSU on top with the intake at the top of the case. Then PSU placement really comes down to preference, as there's no objective advantage of either position over the other.Olaf van der Spek wrote:You'd put the PSU intake at the bottom, not at the top.PartEleven wrote:His third graphics card essentially sits on top of the PSU intake fan.