Single vs. two-pass heater core

[fvr]

As I understand the number of passes refers to the number of times fluid reverses its direction of flow. For example, two-pass heater core will have a U turn internally. Correct? If so, then what is the point of this design and why not just make all heater cores single-pass? The answer I’m not looking for, because it’s obvious, is that it is easier to have the inlet and an outlet close together.

Here is my take on this:
1. The two-pass construction is like connecting two smaller HC in series, thus increasing fluid velocity. Increase in velocity may result in a higher convection coefficient if the flow rate stays the same.

2. However, dT entering the "second" radiator (or second half if you will) is lower. This makes the second half less efficient at displacing heat, which is a common argument.

I might consider modifying my existing two-pass heater core into a single-pass type, but only if there is going to an improvement.

Thanks.

[Since87]

The one issue you didn't mention is how the lowered flow resistance of the heatercore affects the overall flowrate. The increased flowrate with a single pass heatercore will reduce the thermal resistance of the waterblock. The thermal resistance of the waterblock is much more dependent on flowrate than the heatercore. (over the range of flowrates typically seen in watercooling systems.)

It's worth considering the relative flow resistances of the heatercore and waterblock though. If the waterblock is much more restrictive than the heatercore, then reducing the flow resistance of the heatercore by a factor of four won't have much impact on the flowrate through the system.

Aside from that, it looks like you've considered all the relevant issues.

[Gooserider]

IMHO if all else is equal, a single pass is better than a double pass... Probably one of the biggest reasons that most people use double pass cores is that they are FAR more common! I've gone through the entire Leakycar photo set, and judging by the pictures, there are only about 20 single pass cores, vs. several hundred dual pass. The single pass cores are also very similar in size, and most have I/O fittings that stick out the sides or ends of the tanks rather than the face, which makes them harder to fit for many applications. Lastly, it appears that several of that small number are for 'wierd' car apps, which makes them more expensive / hard to get.

I'm using a 2-342 (6"W x 9.5"L x 2"T)in my rig, which is from one of several Chevy / GM trucks made in the mid 60's. It was only about $40.00 USD, which is fairly cheap, but I had to special order it which took over a week.

2. However, dT entering the "second" radiator (or second half if you will) is lower. This makes the second half less efficient at displacing heat, which is a common argument.

True, but this really applies to a single pass core as well - the dT of the coolant entering will be higher than the dT of the water exiting, with a temp gradient accross the face of the core, running from inlet to outlet. In a DP core, it's more complex, as each segment will have it's own gradient, with each gradient running in the opposite direction. The two segments are also closer and narrower, so it's harder to take advantage of the gradient.
In my system, I'm going to take advantage of the gradient. Hot coolant will enter the bottom of my rad, and exit out the top. I have two 120 mm fan zones side by side sucking through the rad via a divided duct. The bottom fan which sees the hottest coolant will blow out a hole in the bottom of the case (which is mounted on wheels) getting most of the heat out of the case ASAP. The upper fan blows into the case, and is the sole air intake for the system. In effect I have made the single core into two 'virtual rads', each 6"W x 4.5"H. I suppose it would be possible to do something similiar with a DP core, but it would be far more complex and difficult.

Gooserider

[fvr]

Gooserider, is you air intake right next to your hot air outlet? I didn't quite understand that part. Anyway, I was wondering why more car manufacturers don't use SP heater cores.

I'm thinking about getting two 9.5"x6"x2" heater cores and constructing a 9.5"x6"x4" single pass radiator. Two 120mm fans will pull the cold air through it into the case (26" Dragon). I know it's not a good idea to increase case temperatures, but I'd rather have my water cold the my case cool.

Also, I think it would be better to connect the two heater cores in parallel and set them apart or run high CFM fans for the setup I described above. Unfortunately, I just don't have the room for two huge heater cores to be placed separately. Basically speaking, I'm loosing efficiency due to lower dT(air) in a 4" thick heater core. (Part of the HC where the air exits or is about to exit will be less efficient.) Moreover, efficiency will further be lost since two HCs have higher airflow resistance. What I’m saying is that there are many factors that have to be considered.

[Gooserider]

Gooserider, is you air intake right next to your hot air outlet? I didn't quite understand that part.

Sort of... The radiator is mounted in the lower left front face of the case (An AMS CK1100B server cube). The rest of the case will be sealed so that the rad is the only significant air intake.

The rad face is approx 9.5"H x 6"W, and I have made a duct with two chambers that mounts inside the case and seals around the inside face of the rad. The two chambers are created by a partition that almost touches the face of the rad and divides it into an upper and lower section.

The upper section holds a 120mm x 38mm fan mounted parallel to the rad face, about 1.5" away from it. This fan sucks air through the rad, and blows it into the rest of the case.

The lower chamber makes a 90* turn down to the floor of the case where I have a second 120mm x 38mm fan mounted on the bottom of the case going out a blowhole. The case is on casters that hold it about 1.5" off the floor. Thus air gets sucked in through the rad, turned 90* and blown right back out the case. The inlet and outlet are close to each other but on different surfaces. If I find that I have problems with either air recirculation or exhaust noise, it should be no problem to add a 'mudflap' to the bottom front edge of the case to force the airflow and noise to go the other way.

Anyway, I was wondering why more car manufacturers don't use SP heater cores.

Well they are doing a different application, with much higher dT's, less concern about flow restrictions, and more of an issue with ease of hookup. It is far easier to hook up the core if the I/O's are next to each other, and only having to reach one end of the core makes designing the heater box easier. (Indeed if you look at the 2-342 that I'm using you will see that it has some really funky extra pipes to put the I/O's next to each other - the first mod needed for the core is to shorten them. IOW, they don't gain anything useful from using an SP core as opposed to a DP core.

I'm thinking about getting two 9.5"x6"x2" heater cores and constructing a 9.5"x6"x4" single pass radiator. Two 120mm fans will pull the cold air through it into the case (26" Dragon).

IMNSHO this is a *BAD* idea! This is the same size as the core that I'm using, and is a bigger core than the vast majority of WC systems use. I chose it because I'm building a dual processor rig, with a bunch of hot SCSI drives. The 2-342 has far greater heat dissipation ability than any reasonable load you are likely to get in a single case machine from a waterflow standpoint.

If you read the articles on applied WC thermodynamics, particularly BillA's work, both here and on OverClockers, you will note that radiator design is a compromise. The better you get with waterflow, the worse you get with airflow.

Your heat dissipation is limited to the capacity of the most limited flow - infinite water flow won't cool anything unless there is air flow to cool the water, and infinite airflow won't cool unless there is enough water flow to carry the heat to it. In almost all cases, the 'bottle neck' parameter is airflow, especially if one wants to keep the amount of fan noise down. (The reason I'm using 38mm thick fans in suction mode, they have the lowest volume losses going through the rad...) This is because air has much lower heat capacity than water does. At the same time, you will loose about 50% of the fans rated free air volume due to the restriction of a 2" thick rad core (i.e. an 80CFM fan will only actually move 40CFM through the rad).

Doubling the core thickness will reduce your airflow by half (or worse)! I would expect your temps to go up considerably if you do this. Your water flow is already fine with a single core, do more work on increasing your airflow.

I know it's not a good idea to increase case temperatures, but I'd rather have my water cold the my case cool.

I agree. However the heat budget for my system called for 80CFM worth of airflow through the rad, and I realized that there wasn't much advantage to blowing that much air through the rest of the case. By splitting the airflows I'm dumping the hottest air outside the case so I get the best of both worlds.

Gooserider

[Since87]

Originally posted by Gooserider
Doubling the core thickness will reduce your airflow by half (or worse)! I would expect your temps to go up considerably if you do this.

I would expect two rads in series (airflow wise) would get roughly 85% of the the single rad air flowrate. Substantially more than 50% in any case.

Got backup for your statement?

[Les]

Originally posted by Since87
I would expect two rads in series (airflow wise) would get roughly 85% of the the single rad air flowrate. Substantially more than 50% in any case.
.....

Yeah, thats about what I (Kryotherm guided) make it.
For one example increasing from ~77% to ~89% for an increase in "clearance between fins" from ~0.1mm to ~5mm

[fvr]

I agree. There is no point in having twice the surface area if one doesn't run more powerful fans to compensate for higher resistance of the 4" setup. My original I deal was to get as close to ambient temperature as possible, but now I think it's not worth the extra noise.

[Les]

Originally posted by Les
Yeah, thats about what I (Kryotherm guided) make it.
For one example increasing from ~77% to ~89% for an increase in "clearance between fins" from ~0.1mm to ~5mm

Crap.
Well not entirely, but only applied to the specific case I looked at( unrealistic back pressure of ~57Pa)
Investigating another case with a more realistic back pressure of ~13.5Pa get the flow rate reduced to 50% by doubling the thickness from 2" to 4".
Will keep my mouth shut until have a better understanding of the full picture.

[Blackeagle]

Considering the problem at hand, would not a single pass large rad (1) with 4 fans of a high pressure design (Delta 38mm?) be the best performing?

[Gooserider]

Since87: I would expect two rads in series (airflow wise) would get roughly 85% of the the single rad air flowrate. Substantially more than 50% in any case. Got backup for your statement?

No math numbers, sorry. I was working on the basis of one of BillA's articles that indicated an approximate average airflow loss of 50% on a 2" thick core.

(exact loss varies depending on the fan, how it's mounted, the static pressure, and the PQ curve of the fan - some fans seem to push pretty well as the load increases, some drop pretty fast, on average thicker fans handle flow resistance better than thin ones)

I then made the assumption that doubling the thickness doubles the flow loss, giving an effective flow of 25%.
There may well be problems with my methodology, but I do feel comfortable with my bottom line assertion that doubling the rad thickness is not a good way to go for better cooling.

Gooserider

[Since87]

I suspect that what Bill was saying was along the lines of 'Don't expect to get more than half an axial fan's free air flowrate spec when using it with a 2" heatercore.' (I don't really know though since I don't know what article you are referring to.)

I agree that stacking heatercores is not a good way to go, mainly because the fan noise required to get good performance is too high. Even if the air flowrate only drops to 85% (of single HC flowrate) with two heatercores stacked, (and reasonably quiet fans) the air will be heated so much by the first HC that the second HC won't be able to dump much heat.

It's certainly possible to get improved performance with stacked HC's, just not worth the noise to me.

[Gooserider]

Since87 I suspect that what Bill was saying was along the lines of 'Don't expect to get more than half an axial fan's free air flowrate spec when using it with a 2" heatercore.' (I don't really know though since I don't know what article you are referring to.)

More or less, mostly less. Apparently some of the popular 'silent' fans really fall off badly. The 38mm 120's were the best of the bunch, but even they weren't that good. Alot of what I got was from reading his graphs that showed airflow at certain static pressures, and attempting to map that back onto the fan PQ charts.

I agree that stacking heatercores is not a good way to go, mainly because the fan noise required to get good performance is too high. Even if the air flowrate only drops to 85% (of single HC flowrate) with two heatercores stacked, (and reasonably quiet fans) the air will be heated so much by the first HC that the second HC won't be able to dump much heat.

Exactly, the second core won't dump much heat, and I'm not even convinced that the amount of heat dumped by the second core would make up for the overall loss of cooling from reduced airflow. It would depend on just what the exact setup was.

The other issue is that the goal of cooling to ambient is an asymptotic curve, and increasing cooling comes at astronomically increasing cost. A 2-342 sized core has so much cooling capacity potential to start with that I would not expect much additional cooling from any further reasonable increase in rad size. I decided to go with a heater core very early on in my planning, so I haven't paid a great deal of attention to the commercial WC rad options. However, I've gathered the impression that a 2-342 has 2-4x the surface of most popular commercial rads, lower flow resistance, and better thermal transfer ability, so it seems like I already have far more than average performance potential.

It's certainly possible to get improved performance with stacked HC's, just not worth the noise to me.

I agree. I also suspect that the space penalty associated with stacking and getting improved performance is more than most people would anticipate.

I believe that it is possible to get reasonable flow through a 2" core by mounting fans on just the suction side. I also think that if one wants acceptable performance and tolerable noise, a duct is vital. The duct specs I've seen say that a fan needs at least 1.5" of free air space on each side to get optimal flow and reduced noise. This means that a 38mm (or 1.5") thick fan will require a minimum 4.5" of clear space for good airflow, 6.5" if one includes the core.

IMHO, to get reasonable flow through a 4" core a push-pull fan setup is indicated. If one assumed 4.5" on each side, plus the core, then it would take 13" of free space to fit the rad! That's more space than most cases have... Even with my monster sized cube case, I had a real squeeze to get just my 2" core mounted with the fans on ducts and preserving the space it needs.

Gooserider

[sevisehda]

The other day I had the idea of getting 2 heatcores, cutting off the bottoms then soldering them together end-to-end. Then place the now 16" x 8" rad at the bottom of my case. Theoretically it should have the resistance equal to that of the orginal setup just with better cooling. With the 4 fittings it would be very easy to do a 2 pump setup.

The 2 catches would be finding a case with enough clearence at the bottom, and having the need for that much ccoling.

[BrianW]

Yup that would be nice..... Long ass heatercore.

BrianW

[Gooserider]

sevisehda The other day I had the idea of getting 2 heatcores, cutting off the bottoms then soldering them together end-to-end. Then place the now 16" x 8" rad at the bottom of my case. Theoretically it should have the resistance equal to that of the orginal setup just with better cooling. With the 4 fittings it would be very easy to do a 2 pump setup.

Interesting idea, certainly I think it would work better than trying to do a double thickness core.

However depending on how you plumbed it, you might or might not have the same resistance as a single core.

(Note, all flow resistance discussion relates to water flow - airflow resistance would be unchanged for any given part of the core, but the total airflow resistance would drop simply because of the increased rad surface area, just as if two cores were used next to each other)

(Second note - DP = dual pass core, SP = Single pass core, and all comparisons are based on using the same base core number (doesn't matter what it is) for all scenarios)

There is considerable flow resistance from the tubes themselves, so unless you plumbed the thing in such a way that the effective tube length and number of tubes remained the same, then the resistance would go up.

Let me think about the picture for a moment I am assuming that you are using DP cores since you mention ending up with 4 I/O connections. In effect you would have created two parallel SP rads in the same body, each having a number of tubes equal to one pass in the DP core, and twice the DP core length. Assuming that both passes had the same # of tubes (not always the case) then:

1. If you ran two seperate loops, the flow resistance would be about the same as if you had a seperate DP core on each loop.

2. If you had a single loop, and 'Y'ed the flow so that each pass got about half the coolant then the flow resistance would be lower, approx what you would get by running two DP cores in parallel. (or one SP core of the same size as the combined unit)

3. If you used a loop of tube to tie the output of the first pass to the input of the second (effectively creating a bigger DP core) then the coolant flow resistance would go UP due to the increased tube length, approximating what you would get with two DP cores run in series.

I can't think of any other plumbing setup scenarios.

Aside from the 'coolness' factor from doing something unusual and possibly reducing the total size by the length of two tanks (approx 1.5" on average) I don't see ANY advantage that your proposed setup would offer over doing the same thing with two cores not soldered together.
----------
From a structural standpoint, I think you would also have problems. I will admit my exposure to the innards of heater cores was brief and many years ago, but my recollection is that core construction would make your idea difficult to do.

A typical core is made from a bunch of ribbons of brass that are formed, pressed together, and soldered along their edges to form the tubes. (more ribbons go in between the tubes to make the cross fins) The ribbons are folded back and forth such that the folds define the tank-tank length of the core. The tanks are soldered around the perimeter of this folded area, and don't actually touch the tubes themselves. While the core-tank mating surface of the folded area is kept very flat, the tube openings are irregular and not consistent in their relationship to the mating surface.

I suspect rather strongly that you would find it very difficult to fasten the two core lengths together unless you created some sort of 'adapter tank' with a mating surface on each end to match the two oore mating surfaces. There are a couple ways I could see of doing this, but I don't see an easy approach that would be worth the effort. I also suspect that the soldering would be a bear as well since you would have to figure out how to make two joints in close proximity without having making the second joint melting the first joint (or the core)

I don't know just what such an 'adapter tank' would do to the flow resistance, I suspect nothing good. As a first guess, I would expect it to be approximately equal to the impact of the bottom crossover tank on a DP core, which again would suggest that there is little gain from what you propose.

I suspect that if you actually needed that much cooling, and wanted low flow resistance the best way to achieve it would be to run two SP cores with no attempt to connect them except with plumbing. (If both would be used in the same loop, run them in parallel rather than series.)

Gooserider

[sevisehda]

I hadn't really thought out the idea much. I would basically shave off the bottom of the 'tank' of 2 cores. Then solder 2 of them together. After 2 days of searching I have yet to find a case with enough clearence between the MB and bottom to fit a radiator.

This problem was solved with another crazy idea after reading the artcle on BTX.

I just came into some unexpected cashso, maybe its time I constructed a 'concept' case.

[Gooserider]

sevisehda: After 2 days of searching I have yet to find a case with enough clearence between the MB and bottom to fit a radiator.

It would seem to me like you aren't likely to - I can't think of any advantage that a 'normal' case design would get from pushing the mobo up in the case. OTOH, pushing the board up would mean needing to either make the case taller or putting less stuff in the top of it.

What sort of air flow pattern were you thinking of anyways? I can't think of anything offhand that puts the rads in the bottom of the case without either blowing rad heated air into the case, using some really complex ducting, or having to force the air to flow in strange directions.

OTOH, It would seem like a fairly neat thing to put the rad on the top of the case, with an enclosure that acted as a duct to send all the hot air out the back of the case. Then pull air in the bottom of the case and blow it out the top...

You might get some extra room from a case w/ a sliding mobo tray. I know my case has about an inch of clearance between the case bottom and the bottom edge of the tray; but that space is used by the guides the tray slides on.

A few other random thoughts -

1. Mount the rad through the case, so that it only sticks in part of the way.

2. Get a wider than normal case so that you can put the rad in front of the mobo. This might block one or more slots, but might be OK.

3. What is the relationship between the dimensions of the mobo and the I/O plate position? If you got a case sized for an eATX size board, and put a smaller size board in it, might the smaller board not hang down as far?

4. Make a mini box under the case, and put the rad in that, then the only thing in the main case would be a hole in the floor for the air flow.

Gooserider

[sevisehda]

If I had to put a dual-core rad in a case(with enough clearence) right now...

I'd cut openings in the bottom of the case, then put the rads on top. I'd then use the PS and system exaust fans to draw air out of the case. Air naturally would be sucked through the rad, so in essence the case itself would the shroud. The downside would be slighter warmer air inside the case but with efferything watercooled it wouldn't be that bad.

[Blackeagle]

Interesting disscussion.

What about a bit differant approach.

Set up the pump, res and rad in a external case of their own. This would allow you to design for a Xtreme sized rad. I'd suggest looking at rad #2-192 in the heater core data base.

10.75" X 10.25" X 2" for the finned area. A huge core that can handle about anything I can imagine a single system having, even TEC's on CPU & GPU. And no messy cutting of tanks, just shorten the odd inlet & outlet tubes.

And this large of a rad would greatly reduce air flow resistence form your dual 120mm fans. Or perhaps some really large fans, 172mm push=>pull anyone? Small squirrel cage blower of 8-10" tall x 6-8" wide?

And as long as you are going with a external case for max attempt at reaching ambient temps you will have room for a pump of any size you desire (Little Giant or Iwaki?). And the above rad has 5/8" inlet & outlets so you could also increase your line size to 5/8" as well for lower flow resistence in the lines.

This could become a never ending spiral, eh? But if anything is going to come very close to ambient, this monster sure should!

I intend to build something along these lines for my new system. The exterior cooling case will be elongated and open at both ends as a sort of wind tunnel. And with the cooling case elongated the puter case will sit on top of it. My water lines will go up through the puter case's floor. Rad will be 2-342 however and pump a Iwaki MD-15 & DIY custom res. IN my build one goal will be to maintain the lowest noise level I can.

[Gooserider]

sevisehda:
If I had to put a dual-core rad in a case(with enough clearence) right now... I'd cut openings in the bottom of the case, then put the rads on top. I'd then use the PS and system exaust fans to draw air out of the case. Air naturally would be sucked through the rad, so in essence the case itself would the shroud. The downside would be slighter warmer air inside the case but with efferything watercooled it wouldn't be that bad.

Either I'm misunderstanding you, or your idea has problems (possibly both ) If your rads are on the top of the case, and you have holes in the bottom, then convection is going to pull air in through the bottom. Your fans are also going to pull air through the bottom in preference to the more restrictive rad core - I would expect little or no airflow through the rad. The only way having a rad at the intake and fans at the exhaust will get airflow through the rad is if the case is otherwise sealed.

Looking at your post again, perhaps you were meaning to put the rads directly on top of the holes, in which case it would work if the case were otherwise sealed. However the previous posts were suggesting that you couldn't find a case with the clearances, which is why I was thinking about alternative designs.

Blackeagle:
Interesting disscussion. What about a bit differant approach. Set up the pump, res and rad in a external case of their own. This would allow you to design for a Xtreme sized rad.

Not a bad idea, and one that I've seen many times in different forms. As I see it the seperate box approach has advantages and disadvantages. It simplifies system design since you no longer have to figure out how to squeeze all the plumbing into a standard PC case (and possibly be forced to swapp cooling for compactness) It is also possible to locate the cooling box (and it's noise) away from the system. OTOH, it makes moving the system more of a headache, and possibly causes other problems with more complex plumbing.

I'd suggest looking at rad #2-192 in the heater core data base. 10.75" X 10.25" X 2" for the finned area.

Not bad, but if you want big, why not just go whole hog and get an actual car RADIATOR - say something from an old V-8 smogmobile...

A huge core that can handle about anything I can imagine a single system having, even TEC's on CPU & GPU.

I don't claim expertise on extreme / subambient cooling, but my understanding is that if one is using a TEC, etc. it is best if you use multiple independent circuits. The reason is that the different parts are operating at different temeprature ranges and you defeat the purpose if you mix the coolants.

And no messy cutting of tanks, just shorten the odd inlet & outlet tubes.

Good, but note that often the size of the intended hose connection is different from the size of the tubes coming out of the core. This isn't a big deal, but you need to be aware of it. Also if necessary, it is not that bad to add adapters to change the size or type of fittings coming out of the rad.

And this large of a rad would greatly reduce air flow resistence form your dual 120mm fans. Or perhaps some really large fans, 172mm push=>pull anyone? Small squirrel cage blower of 8-10" tall x 6-8" wide? And as long as you are going with a external case for max attempt at reaching ambient temps you will have room for a pump of any size you desire (Little Giant or Iwaki?).

Note that getting to ambient (w/o using pelts, or other radical cooling tech) is nearly an impossible goal, that you approach on an assymptotic curve - the closer you get, the harder you have to work, and you rapidly get into the diminishing returns zone. At the same time much of what you're mentioning is going to come with a major noise cost - IMHO this isn't acceptable, my objective is 'good enough' cooling with the lowest possible noise level. Going to larger pumps can also be self defeating as larger pumps will put more heat into the system. Get a big enough pump and your system will be doing more to cool the pump than it is to cool the CPU!

And the above rad has 5/8" inlet & outlets so you could also increase your line size to 5/8" as well for lower flow resistence in the lines. This could become a never ending spiral, eh? But if anything is going to come very close to ambient, this monster sure should! I intend to build something along these lines for my new system. The exterior cooling case will be elongated and open at both ends as a sort of wind tunnel. And with the cooling case elongated the puter case will sit on top of it. My water lines will go up through the puter case's floor. Rad will be 2-342 however and pump a Iwaki MD-15 & DIY custom res. IN my build one goal will be to maintain the lowest noise level I can.

This last proposal sounds more reasonable, and has a goal I'd agree with. (It also isn't far from my own system in terms of components.) The only thing I would be concerned about in terms of noise is your mention of having an opening on the front of your box. Openings let noise out, so you might consider how to muffle the front opening. Alternatively, consider doing a 'U' shaped box with both input and output at the back, or some other alternative. You might want to look at the Silent PC Review website for other ideas on building a quiet cooling box. (I like to go where the experts are, the folks here are great on cooling, but tend to be noisy. They are silent, but tend to be hot. I go here to learn about cooling, there to learn about silence, and get the best of both worlds....)

Gooserider

[Blackeagle]

Gooserider,

The "wind tunnel" will be approx. 26" X 16" X 12" and will have baffles that will create a S shaped flow route for the air coming into the rad & fans area and again exiting after passing over my pump.

Whole tunnel will be lined with two layers of sound deadening materials. One dense 1/2" foam layer overlayed with a 1" lower density foam of a egg flat type. These foam layers are, in large part, the reason for the size of the tunnel. The tunnel will be made from wood for it's natural sound reducing qualitys.

With 4 120mm fans set up with 2 pushing and 2 pulling hooked up to a reobus that will allow me to run them from 5v-12v it should offer a good noise value as well as extreme performance when desired. Although with that many 120mm fans at 12v I don't think even the sound foam would keep it from being a bit much.

Case is a PC76 Lian-Li server tower, which will also be lined with the same foam(s). I'll close off most of the slot vents on the case with the foam as well as part of the rear fan openings. I'll install a small window in the case top and replace the 120mm fan there with a 80mm, and the window material will be 1/4" to weight the panel a bit as well as offer lower transmission of sound due to the greater thickness of the plexi. The 4 front 80mm mounts will be removed and a single 80 or 92mm will replace them centered on a single hhd bay with the other drive bay removed.

PSU will be a 510 PC P&C, not the most quiet, but a very good PSU other wise with high 12v output. PSU and all case fans will be mounted with the new silicone vibration dampeners or a layer of the 1/2" foam between the fan and the case.

Over all I'm hopefull it will offer quiet combined with a pretty high O/C. This system will be moved seldom, if at all, so I'm not really concerned regarding that point.

Now that I've detailed a bit better my design ideas I'd love to hear your input on further improvements to it. One question I have no answer to it this: What amount of air flow through the case will be needed to keep a low enough case temp to keep the PSU fan from wanting to increase it's RPM's? I'm guessing 40+CFM should do it, but not sure. The water cooling of the CPU & GPU should help this some.

Thanks

[fvr]

Nice discussion guys. Keep those ideas coming.

[sevisehda]

Originally posted by sevisehda
If I had to put a dual-core rad in a case(with enough clearence) right now...

I'd cut openings in the bottom of the case, then put the rads on top. I'd then use the PS and system exaust fans to draw air out of the case. Air naturally would be sucked through the rad, so in essence the case itself would the shroud. The downside would be slighter warmer air inside the case but with efferything watercooled it wouldn't be that bad.

I would place the rads on top of the openings at the bottom of the case.

[Gooserider]

Most of your ideas sound really good on what you are planning Blackeagle. A few misc. thoughts...

1. Another element of sound reduction that you might consider for all the metal surfaces is to line the metal with 'asphalt roofers tape' - The folks at SPCR think very highly of it for budget sound damping, it really kills resonance. I've tried it on some of my case side panels; before if I hung a panel on a bungie and whacked it with a screwdriver handle it made a pretty decent gong. After lining it, I would get a dull thud with no resonance at all. Since alot of PC noise comes from resonant amplification by the sheet metal, the SPCR folks generally recomend a layer of roofers tape or equivalent as the first layer of soundproofing on metal cases. You can get a roll big enough to do a couple normal cases for around $12 at Lowes or Home Despot type stores. It's basically a thin layer (~1/16") of asphalt adhesive backed with a thin layer of soft aluminum.

2.

PSU will be a 510 PC P&C, not the most quiet, but a very good PSU other wise with high 12v output. PSU and all case fans will be mounted with the new silicone vibration dampeners or a layer of the 1/2" foam between the fan and the case.

I'm using the same PSU. I don't know if you've already done this or not, but PCP&C offers an option to replace the stock fan with one of their thermal controlled lower noise fans - I think it's like $15 extra...

3.

One question I have no answer to it this: What amount of air flow through the case will be needed to keep a low enough case temp to keep the PSU fan from wanting to increase it's RPM's? I'm guessing 40+CFM should do it, but not sure. The water cooling of the CPU & GPU should help this some.

Tough guess, but I would tend to agree given that you are doing the WC stuff. Probably experimentation is the only easy way to tell. You might find that it is also worth cooling your hard drives. If I'm understanding your description properly, you'll end up with an 80 or 92 on the intake, and another 80 plus the PSU on the exhaust? That should be enough given the WC system.

I would advise setting your fans to a silent level, and see how the PSU does, if it gets loud then take the case off and see what else is hot that might be worth giving the WC treatment to. Repeat as needed.

4. I notice you're planning to use a rheobus for fan control. This is sort of a pet peeve of mine, I *HATE* the idea of having to manually control my fan speeds, it seems really stone age. I am searching for an automatic thermal control that will constantly adjust my fan speeds automatically to the minimum needed to maintain my desired level of cooling.

I insist that it be 100% functionally independent of the mobo, other than possibly reporting and configuration software (if any), as I don't want to waste CPU cycles on temperature control, or risk problems because of an O/S hang (however unlikely that is with Linux) I also require Linux compatibility. (I don't allow Windows on my systems)

My ideal controller would also offer both a Digi-Doc style LCD status screen w/ temp and fan RPM readouts, and the ability to report back to the PC for logging or other functions. This implies a microcontroller based solution.

Unfortuneately, I haven't been able to find anything that would give me that sort of function, though a couple of folks on SPCR have been working on custom controllers that are close. I'm probably going to end up doing an op-amp solution that won't have the display functions, but will hopefully do what I want.

Gooserider