Silenced TJ07 (D5 and DDC 2 in same loop)

[utc_pyro]

This is extremely quiet TJ07 I’ve been working on for the last month or so. Half way through is kind of a bad place to start a worklog, but I just borrowed my GF’s DSLR today to take pics.

The actually electronics have not been purchased yet, but here is the case and cooling components so far.

Case components:
Silverstone TJ-07 Black, non windowed side panel.
eDead 80 dampening on sides and frame of case
eDead 120 dampening on MB back tray
eDead v4/TekLite ontop of all eDead 80 to enhance high frequency absorption
Silverstone CMD01-EAS software based fan controller
Silverstone CFP51 drive bay W/Scythe Slip stream fan
Silverstone SDP08 2.5mm drive adapter (for SSD)
Silverstone AreoSlots
Noiseblocker NB XE1 92mm exhaust fans
Corsare HX650 Modular PSU
Every-Card-Known-to-Man Card Reader (not really, but it has SD, XD, CF, and Micro SD so I’m happy) – Mounted in old I/O port slot

Major Cooling Components (no fittings listed):
Pump 1 = Laing D5 Vareo with Koolance PMP450/S top and 120mm long res
Rad 1 = XSPC RX360 W/ Gentle Typhoon 1850RPM fans
Pump 2 = Laing DDC 1+ (aka DDC2) w/SXPC top on custom noise dampened mount
Rad 2 = Hardware Labs Black Ices Stealth GT 240mm W/ Gentle Typhoon 1450rpm fans
CPU = Heatkiller 3.0 (Sorry Swiftech, not this time)
GPU1 = Swiftech MCW60 W/GTS 275 full coverage heat sync
GPU2 = Koolance GPU-180 pulled from my first WC rig

Tubing will be a mix of ½”, 3/8”, and ¼” at different portions of the loop, and I’ve spent over $200 on just fittings so far to keep everything clean. There is also a coolent temp probe right before Pump 2 and one on the GPU1/2’s output to give the temperature rise through the blocks.

Two miss-matched Pumps:
I’ve seen it said here numerous times that one should not put two miss matched pumps in the same loop because you’ll overspend the second pump. From an engineering perspective (mind you I’ll and EE, not an ME) that sounds like a load of bull.

Each pump will find its own point on its PQ curve depending on the back pressure it “sees” for a given flow rate and stay there regardless if anther pump is in the same loop. If anther pump is in the loop, it just changes the back pressure it “sees” to the pressure drop minus the pressure from the other pump. Thus as long as the pressure drop between the first pump in the second pump isn’t more then the first pump is providing, and the equivalent flow isn’t more then the 0psi head max flow of ether pump, it will work just fine.

I’m going to install a test line off of the 5-way Koolance splitter after Rad1 and before Pump 2. On this test line there will be a compound pressure/vacume gauge that I’ll use to set speed of Pump 1 so that Pimp 2 sees about 0.2PSI on the input, and thus keeping the system as quiet as possible. Looking at the pressure drop charts for each of my components and both pumps, I’ll need to run the D5 at the equivalent of 3 if flow is 2gpm to get the pressure were I want it.

Also on the pumps, I’ve put in a pump control panel to help with priming, bleeding, and adjusting things. There is one switch for each pump, and a 10-turn 10K pot that replaced the built in one in the D5. To prime/bleed the system, I’ll probably just use the D5 at setting 1.5 to get it mostly filled, and then use just the DDC to get the bubbles out.

Things being cooled:
This is the plan as of now. Things may change depending on what I get for Graduation (Woohoo for a B.S.E.E.) or Christmas.

Intel 860 Core i7 CPU. Plan to OC with turbo mode and speed-step on
ASUS Maximus III Formula
EVGA GTX 275 Superclocked (verified to run about 740mhz core on air)
4GB DDR2 2000, probably Corsair.
Intel X25-M G2 80GB Boot/App drive
HITACHI Deskstar 7K1000.C Storage Drive
Cable Card Tuner (next year)

Pictures:
Below are the pics I took today of the work in progress.


Overview of the case




I mounted this multi-card reader in place of the IO panel. To make it were one could easily insert and remove the cards, I modified the tray to move the whole reader forward 1cm were it’s almost flush with the door.




This is the fan controller out of the CMD-01 mounted to the mid-plate. I moved it here so I had three front slots open for future projects. It’s on standard MB standoffs and has a hole under it to allow the wires to be passed through.




This is my pump-control panel inform of the fan controller. It’s just two switches for pump 1 and 2 and a 10K wire wound variable resistor to control pump 1. I wired that up on a spare fan extender I had, and if unplugged it sounds like the pump reverts to setting 4 (just like the MCP 655B).




This is the top mounted Stealth GT radiator, using the stock fan mesh for the mount. I pulled it out and re-formed it around this radiator to serve as an air-intake in the top chamber. The Stealth GT is an extremely restrictive rad, so at some point it will be replaced with a custom top plate and Swiftech MCR240. The good thing about it is that you cannot hear the fans from the top at all.




Top mounted radiator from the top. One can see the Koolance fill port on the right side




This is the back view of the XSPC RX360 radiator in the bottom chamber. I mounted it on the opposite side of most people do with the TJ-07 due to were my desk is relative to the computer. It will get cooler air from the right side of the case then the left. One can also see the 1/8” angle aluminum used as a mount for this radiator, and the eDead 80 on the back dampening the transferred vibrations.




Here is the other side of the RX360 with the Gentle Typhoons. Each is on a vibration mount to keep noise down, but I don’t expect to run them over 700 RPM at idle due to noise. The entire lower chamber (top and bottom) is lined with eDead as this area will have the most sound being generated. One can also see the DDC 2 on its XSPC based mount.




The DDC2 has a XSPC top, ½” barbed intake and 3/8” output. The back of the rad is against some 1/16” aluminum sheet going out toward the fans to act as a heat sync. That is all mounted on a 1/8” thick pad of sorbothane before the XSPC mount. The back of the mount is covered in eDead again and the whole thing is glued to a ¼” thick pad of TPE (same as Petra’s gel stuff) on an aluminum mount. When running, very little to no vibration from the DDC makes it into the case.




This is a Koolance 5-way splitter on the output of the RX360. It has one of my two temperature probes, the drain port (silver tubing going to the QDC), and once the parts get in have a ¼” hose going to the mid-panel pressure port. As one could guess, it’s really a pain to get every thing screwed into this one.




This fill port will be used to install a compound vacuum/pressure gauge to monitor loop pressure between the pumps and pressure during leak testing.




This is the Koolance Pump/res combo attached to the MB tray. So far in test running it I’ve noticed it does not do the best job at removing bubbles, but its compact and looks good in the loop.




Side view of the pump mount. The pump base is on a ¼” thick pad of 40OO sorbothane ad the top bracket is on a set of ultra-soft vibrating isolation standoffs. I had to trim the base plate a bit to make it fit on the tray, and add a 1/16” thick bar of AL to support the top. It’s not the best looking, but it’s functional at keeping the noise levels down.




Overview of the MB tray as it is now. On it right now is an OLD nforce based socket 939 motherboard that I use for mockup and my GTX 275 for making sure everything fits. The GTX 275 will be watercooled as well, but I need to install my v-mod tap points before installing the new cooler.




Two ultra quiet Noise Blocker exhaust fans. Silent PC Review gave there their top rating for 92mm fans in low restriction environments, so hopefully they’ll get the job done here. Wish they weren’t purple though.




Both side panels were covered in eDead 80 and eDead V4 to dampen vibration and reflected sound. This wont be the most effective thing I did to the case (by FAR), but it should help.




This is an overall shot of the case with the components installed. I’ll be re-installing the drive rails next and getting those dampened to reduce HDD noise.

[Antioch]

How's that SilverStone fan controller going to work? Isn't it some ESA thing where you need the motherboard to be ESA certified as well? Not like an Aquaero or anything where you can just program it and let it do it's thing without any special "Certified" parts?

[utc_pyro]

As long as you can install the Nvidia Control Panel (as in your have an Nvidia video card or MB) it will work. Also Speedfan and I think Everist can control it, but you cant set up the rules. The only thing that you miss by not having an EAS motherboard is the more detailed readout that you could get, but for managing basic cooling it still gets the job done.

I have one in my old desktop that is now transplanted into a U4 server case. It has a 8800gts that is used just for folding, so it let me install the drivers. I can pick up the video card temps, CPU temps, HDD temps, thermal probe temps, and CPU load and use all of these to set the fan speed rules. It has an Asus P5B somethign another (Intel P45 chipset), so it's not in the least EAS compatible.

[utc_pyro]

Quick update. Time to test fit the VGA sync on GPU1. GPU2 is still in use, so it will be the last component installed.



This is my EVGA GTX 275 Superclocked with the heat sync removed. So far it looks like a standard reference design (P897)



So moment of trough with the Swiftech GTX275 full coverage heatsync. Getting this was simplier/cheaper then buying a bunch of individual syncs.



Fits like a glove! Props to Swiftech for making something that fits this well with all af the components on the board. The clearances on the caps is pretty tight, so it’s amazing they were able to pull this off.



My good old first gen MCW60. This is about the only component I have that’s been in all my rigs sence I graduated from a Koolance PC2 back in 2005.



Bit warn down from all the mounts and odd HS grease, but it should clean up quite well before I hard mount it



Fits fairly well, and the blocks at the top clear the card like I wanted. It still has a G80 mounting plate on it (can’t find my d*** allen wrenches) , but it shows that everything should fit.



Huge mass of fittings…. Intake (on left) comes in on a Koolance 3/8” compression fitting. That goes into the Danger Den “T” block. There is Koolance ¼”compression fitting that will feed the GPU-180-H06 on the 9500GT, and a Koolance CNT-VD3 feeds down into the MCW-60. Another Koolance CNT-VD3 feeds back into the Bitspower Black Sparkle “Q” plus fitting. This also has a ¼” Koolance compression fitting for the return of GPU2, a Koolance 3/8” compression fitting outlet to the res, and another Bitspower temperature sensor.

God this is going to be a B**** to leak test.
Sorry for the crappy pics, my Olympus is still dead and the old Cannon is at work from my last site visit looking at the comm equipment at a hydro plant. All pics were from my Moto Droid, and low light is not the strong point of its tiny cam module.

[utc_pyro]

Leak test time!

I have everything plumed and filled with distilled for leak testing as this is posted. It’s been running for about two days now, and I’ve been able to fix any issues except for the res leaking. The Koolance res’s don NOT like to seal very well, so I’m unable to test just by pressure with my gauge.



Also the Koolance Res-Top does not bleed easily as the flow does not pass through the top chamber. After running over night most of the bubbles are out, but an odd number stick to the side of the res like in the pic below.



__________________________________________________ _______________

Performance Testing:

With the two unlike pumps in this system, I want to show that it’s a load of bull that one cannot do such. Normally it’s stated that one will “over speed” the second pump or starve the second one of flow depending on how they are matched.

That’s not quite how it works in most cases. As long as you are not pushing more flow thought he second pump then it’s 0 head flow rate (max flow), you cannot over speed it from doing this. Also as long as you don’t have a negative relative to atmosphere pressure going into the second pump, then you are adding to “flow” by increasing total pressure, not restricting it.

Right now I have a Compound Pressure gauge with a range of +10 to -15PSI in the system at the inlet to the second pump. With this gauge, one can judge how the two pumps are working together. System flow wise it’s like this:

Res -> D5 W/Koolance top -> RX360 rad -> Koolance 5-way port (pressure tap here) -> DDC-2 with XSPC top -> extremely restrictive loop

I’ll be comparing this pressure, minus the presumed pressure drop of the rad, to the following graph that Martinm210 produced in the thread ” Some Laing D5 Testing, 13.1-24V is the same” back in 2007.


Graph of D5 flow vs pressure at various settings by Martinm210.

First, a zero PSIG point must be put into the loop to ensure accuracy of the gauge. Thus, I put a brass ¼” BSPP breather port into my fill port to bring the res pressure (and pump 1 inlet pressure) to 0 PSIG.



Next we need to see what the gauge thinks 0 PSIG looks like


So it’s at about 0 PSIG = .25 PSI on this gauge. The adjustment screw is not accessible, so I’ll just leave it like this. What can you really expect form a $10 harbor freight kit?

Now with just the D5 on, and at setting 5.


This gives a little under 5PSIG, and if you subtract out the offset and presumed loss in the rad you can presume this to be 4.5PSI as the outlet pressure of the D5. Looking on the graph, this comes out to be some were around 1GPM on the “5” setting.

Now let’s try with just the DDC2 on.


Here we can see a slight vacuum at the pump inlet. Taking out the offset, it’s about 2 in Hg, or about -1 PSIG at the pump inlet here. It’s hard to tell, but looking at the pressure drop graphs for the RX360, this is as least 1GPM of flow.

Now let’s try both on full blast.


This is about 3.60 PSIG on the gauge, or right around 4 PSI after you take into account offset and pressure loss through the rad (a little less then .5PSI @ 2GPM). 4PSI at setting 5 on Matrianm’s graph above would put the total flow output just under 2GPM, so that’s quite good with the components I have. The GT 220 it’s dropping over 2PSI at 2GPM, the MCP-60 1PSI, and I cant find graphs on the Heatkiller 3.0.

Now let’s tune this thing for silence by adjusting the D5’s speed. I have the 10-turn port that I can finely adjust the speed on until the pressure and noise levels are were I want them.


This is at about setting 2 on the DDC, and both the D5 and the DDC are much quieter at this operating point. They aren’t that loud at full blast, but I’m gunning for silence here. Pressure at the DDC is a little above zero (.25 to .5 PSIG), so I’m still getting a performance boost and it’s still high enough to give cool the system if the DDC2 decides to bite the dust.

__________________________________________________ ______________

Due to a request by another user, i made a few addition pics of the top rad. I'll post them here as well: