Project QBG
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
Project QBG
Edit: Because this post is looooong, click here to jump to the latest information and pics.
Intro
In addition to reading about quiet components and noise reduction, I’ve spent considerable time learning about watercooling, knowing that I would try it eventually.
Now is the time.
With watercooling, my PC finally gets a name: Project QBG - Quiet But Ghetto. The pictures should explain that well enough. If a case by MNPCTech or Mashie is the polished CAD rendering that Boeing shows to the Pentagon, my system is the cardboard and duct tape contraption that the engineers actually used to design the thing.
This thread will document the system’s current state and include specific information about my silencing and watercooling techniques and experiences. While this is neither a worklog nor a how-to, I hope most readers will find something of interest here.
Starting Point
I haven’t built myself a completely new system since September 2000, though no parts of that computer are still in use here. This PC’s configuration is the result of several years of steady upgrades. About a year ago, I got serious about quieting the machine and moved everything into a new case which was modified for lower noise. More recently, the PSU and HD have been changed for quieter components, leaving the following setup pre-watercooling:
• Chenbro Genie SR10405, modified
• Seasonic Super Silencer 350 A3
• Zalman CNPS-7000AlCu
• Intel P4 2.4C (Northwood)
• Abit IC7
• PQI PC3200 Turbo 2x512MB (Samsung TCCD)
• Visiontek Geforce 3
• Samsung SP1614C (JVC motor)
• PDC 52x CD-RW
• 3.5” floppy
• Chaintech AV710
• Linksys NIC
The Zalman HSF and stock NB cooler were both controlled by Speedfan. After much experimentation, I set the CPU fan to 28% idle (~930rpm) and 40% (~1400rpm) above 55C. The NB ran from 40% to 50%, or roughly 2300rpm to 3000rpm. Other cooling included:
• Sunbeam Rheobus, controlling the case fans
• Evercool aluminum 120 exhaust, lower rear, <4V
• Panaflo 92M1A exhaust, upper rear, ~5V
• Enermax 80, ziptied to GF3, ~5V
This was a very quiet PC but remained audible in my even quieter room. At idle, the PSU produced the most noise, though the Panaflo 92 ran a close second. Under load, the Zalman HSF became audible. The SP1614C is isolated with a combination of Sorbothane and elastic, and I’ve never noticed any drive noise.
Case Modifications
I bought the Genie from Coolcases as part of a silencing upgrade a year ago. I removed the fan grilles, damped the side panels with McMaster 54665T32, and lined most of the case with the foam from McMaster’s melamine ceiling tiles. For convenience, I added casters and handles, but have detected no noise benefits from those changes.
This case proved an excellent foundation for a quiet PC. And that’s the right word, since the PC is extremely heavy. The damping works wonders against chassis resonance, and rapping a panel gives a dull thunk with no ringing. The combination of the Genie’s rear-mounted HD cage and the damping materials prevents drive noise from escaping the front of the case. Airflow in this case is well designed and effective, as is the large intake filter. I could nitpick a few areas, but this case is an excellent choice for anyone who can accommodate its size and price.
I selected this case partly for its suitability for watercooling. Thanks to the Genie’s large open space in the lower compartment, a large radiator fits easily with minimal mods. Knowing that I would run my fans as slowly as possible, I decided that the open 120mm intake would be sufficient and did no more cutting as part of the WC conversion.
WC Components Used
All of my reading indicated a few simple rules for quiet watercooling. Fundamentally, WC is aircooling with a complicated ducting arrangement, so most of our usual airflow goals apply.
To flow air through the radiator quietly, you want a radiator with a lot of frontal area, reasonable thickness, and open fin spacing. I elected to use a heater core for its good price/performance ratio and went for the biggest frontal area I could fit. The core is for a 1995 Chevrolet G Van (Fedco 2-333 or GDI 399095) and has rated core dimensions of 9-1/2” x 7-3/8”. Mine is slightly smaller than rated and an excellent fit in the Genie. While the core’s 2” thickness is greater than ideal, I believe the large frontal area compensates. This core offers almost twice the area of the popular Chevette/Pro-120 cores, and about 50% more than a Black Ice Pro 2.
For the CPU waterblock, I was lucky to get a Storm G4 from Little River Waterblocks.
Any temperature advantage available in the block should allow a slower fan or pump speed, so a quiet system needs a good block. The G4 is a good block. Mine came configured with ½” barbs.
Getting a quiet pump was essential. In addition, I wanted a DC-powered pump, so the current choices were the C-Systems CSP-750 or the Innovatek/Eheim HPPS. Edward Ng’s used CSPs were available for a fraction of the HPPS’ price, so the decision was made.
Other Parts & Supplies
McMaster-Carr, the local hardware store, and the auto parts store were raided for additional parts. These included 7/16”-ID Masterkleer tubing (only $0.36 per foot!), various hose fittings, hose clamps, weather stripping, water-pump lubricant/corrosion inhibitor, etc. A project like this can go a lot faster when you remember to ignore Home Depot and Lowe’s and find a REAL hardware store.
Installation Steps
I will just touch on the highlights here, with the understanding that there’s more to installation than detailed here.
The radiator’s hose fittings required modification. I removed the ends of both tubes and epoxied ½” brass barbs onto the remaining pieces. Because the core’s tubes run at an angle, I couldn’t get as tight a mechanical joint as I’d hoped, but the connections passed a pressure test after letting the JB Weld set up for 24 hours. This qualified as Good Enough.
I had decided to use the case as my fan shroud, so the HC was mounted to the chassis’ front panel with zip ties and 1” nylon spacers. These secure the core very well while requiring almost no work. The core is sealed to the case with weather stripping and scrap foam so that all intake air must pass through it. The case’s side panels must be installed for proper cooling.
Mounting the waterblock was extremely simple, though it did require removing the motherboard and the stock HSF retention frame. No real secrets or tricks here.
The real art of watercooling lies in hose routing. I wanted to use the least amount of hose possible, so the pump is suspended at the same level as the waterblock and radiator fittings. The whole loop has about 2 feet of tubing.
I mixed distilled water with the recommended dose of Prestone pump lubricant and 1 drop per gallon of dishwashing soap. This is probably the cheapest acceptable blend available.
Installation Issues & Problems
While I suffered no serious problems, various smaller issues arose. These affect convenience more than performance, but will lead to changes at some point.
The HC’s connections are the system’s high point, and this makes bleeding a giant pain. I’d thought that tilting the PC during bleeding would solve this problem (It works fine on a shifter kart…), but it has not been as fast as I’d hoped. When I redo the loop, I will happily add more hose if it allows better bleeding. With my modest water flow, I don’t believe that an extra foot of tubing will hurt my cooling significantly.
The HC’s angled connections limit my mounting options. This was no problem for my initial setup, but it will prevent inverting the radiator. To solve this, I’ll need to break out a torch and replace the curved tubes with straight copper pipe. And JB Weld seemed like such a time saver. Sigh…
The 1” spacers I used between the HC and the case are a little too short. The Genie’s bezel hinges force the HC to a small angle. Easily fixed next time.
I ordered 5/8” spring hose clamps from McMaster for my 5/8” OD tubing. They are great clamps, but absolutely will not fit over the tubing when it’s mounted on a ½” barb. I am incredibly foolish and running without hose clamps at the moment. For this reason alone, my redo will probably be soon. To be honest, the tubing is very secure on ½” barbs, and I’ll probably need to cut it for removal. Careful eyes will notice that there is one clamp in the system. I put my T-line too near the pump’s intake, and the tubing didn’t quite seal on the pump. Using a worm-drive clamp stopped the slight leak, though I should have seen this problem in advance – there’s almost no tube between the 2 barbs.
Results
Surprisingly good, but also disappointing.
The CPU temperature is great – something around 10C less than with the Zalman. And no fan ramping up under load! FWIW, the IC7 reports roughly +17C over ambient when idling and +35C under CPUburn. Considering the IC7’s reputation, those numbers satisfy me. The 120 exhaust still runs at <4V (below startup voltage), which gives extremely low airflow. Regardless of fan speed, the 2.4C refuses to clock ANY higher with WC. EDIT: I've been running at 260FSB, or 3.12GHz, with 1.625V set in the BIOS. Not too surprising for a P4, but I’d hoped for a little something extra. However, this does mean my heat production hasn’t changed, and the temp comparisons are as meaningful as mobo readings ever get.
Initially, I had a problem with the PSU fan ramping up. Not too surprisingly, the Seasonic didn’t want its air preheated by the radiator. Because my Genie has a door over the external bays, a traditional PSU duct wasn’t possible. So I tried to come up with something more ghetto for the truly lazy WCer. With painter’s tape and a scrap of foam, I partitioned the Genie’s top compartment into a separate zone. Next, I used the Rheobus to turn off the 92mm exhaust. With 3 minutes’ work, the PSU pulls ambient air in over the hard drive, and I eliminated a fan. Very satisfactory engineering! Under long-term load, the PSU fan ramps up to ~1350rpm, only 100 above idle.
With 2 fans removed, pump noise became an issue late at night, when a faint HD-like whine came through the case. Running the pump at 8-9V solves this problem completely, but costs ~2C on the CPU. However, the SS becomes the loudest component, even at idle. Having a Super Silencer A3 as the “noise problem” in my system, I am perfectly satisfied with its noise signature. I’ve been living with its current configuration for about a month and have noticed no irritating noises.
Conclusion & Thoughts
For a mechanically minded gadget freak, watercooling is cool! I have significantly improved temperatures and noise compared to a quality HSF, and I blame this system’s inability to OC higher on my P4. However, I suspect that ducting an XP120 to the 120mm exhaust would have given better results by a price/performance standard. Such an approach would have required less work as well.
My results reflect a WC system that prioritizes low noise in a self-contained system, and I’m delighted with the outcome. I firmly believe that a good watercooling implementation requires significant costs in time, space, case modding, and money. While I accepted all that, newbies should approach watercooling with open eyes rather than expecting water to perform miracles by itself.
There are a couple of areas that I still wish to improve. The hose routing needs a better layout for filling and bleeding. In addition, I’d prefer to replace the aluminum pump to avoid any galvanic corrosion issues. I will inspect the components when I replumb the loop and report any problems. Finally, Zerex Racing Super Coolant should be a better additive than the Prestone lubricant, so I’ll make that change too.
Pictures
Clicking the thumbnail will give a somewhat bigger picture. Links also included to the official, Monster-Sized shots (in case you want to count the dust motes on my soundcard). The biggies are approximately 2-3MB, so consider yourselves warned. BTW, the Image Zoom extension for Firefox is very nice indeed.
Monster
Monster
Monster
Monster
Intro
In addition to reading about quiet components and noise reduction, I’ve spent considerable time learning about watercooling, knowing that I would try it eventually.
Now is the time.
With watercooling, my PC finally gets a name: Project QBG - Quiet But Ghetto. The pictures should explain that well enough. If a case by MNPCTech or Mashie is the polished CAD rendering that Boeing shows to the Pentagon, my system is the cardboard and duct tape contraption that the engineers actually used to design the thing.
This thread will document the system’s current state and include specific information about my silencing and watercooling techniques and experiences. While this is neither a worklog nor a how-to, I hope most readers will find something of interest here.
Starting Point
I haven’t built myself a completely new system since September 2000, though no parts of that computer are still in use here. This PC’s configuration is the result of several years of steady upgrades. About a year ago, I got serious about quieting the machine and moved everything into a new case which was modified for lower noise. More recently, the PSU and HD have been changed for quieter components, leaving the following setup pre-watercooling:
• Chenbro Genie SR10405, modified
• Seasonic Super Silencer 350 A3
• Zalman CNPS-7000AlCu
• Intel P4 2.4C (Northwood)
• Abit IC7
• PQI PC3200 Turbo 2x512MB (Samsung TCCD)
• Visiontek Geforce 3
• Samsung SP1614C (JVC motor)
• PDC 52x CD-RW
• 3.5” floppy
• Chaintech AV710
• Linksys NIC
The Zalman HSF and stock NB cooler were both controlled by Speedfan. After much experimentation, I set the CPU fan to 28% idle (~930rpm) and 40% (~1400rpm) above 55C. The NB ran from 40% to 50%, or roughly 2300rpm to 3000rpm. Other cooling included:
• Sunbeam Rheobus, controlling the case fans
• Evercool aluminum 120 exhaust, lower rear, <4V
• Panaflo 92M1A exhaust, upper rear, ~5V
• Enermax 80, ziptied to GF3, ~5V
This was a very quiet PC but remained audible in my even quieter room. At idle, the PSU produced the most noise, though the Panaflo 92 ran a close second. Under load, the Zalman HSF became audible. The SP1614C is isolated with a combination of Sorbothane and elastic, and I’ve never noticed any drive noise.
Case Modifications
I bought the Genie from Coolcases as part of a silencing upgrade a year ago. I removed the fan grilles, damped the side panels with McMaster 54665T32, and lined most of the case with the foam from McMaster’s melamine ceiling tiles. For convenience, I added casters and handles, but have detected no noise benefits from those changes.
This case proved an excellent foundation for a quiet PC. And that’s the right word, since the PC is extremely heavy. The damping works wonders against chassis resonance, and rapping a panel gives a dull thunk with no ringing. The combination of the Genie’s rear-mounted HD cage and the damping materials prevents drive noise from escaping the front of the case. Airflow in this case is well designed and effective, as is the large intake filter. I could nitpick a few areas, but this case is an excellent choice for anyone who can accommodate its size and price.
I selected this case partly for its suitability for watercooling. Thanks to the Genie’s large open space in the lower compartment, a large radiator fits easily with minimal mods. Knowing that I would run my fans as slowly as possible, I decided that the open 120mm intake would be sufficient and did no more cutting as part of the WC conversion.
WC Components Used
All of my reading indicated a few simple rules for quiet watercooling. Fundamentally, WC is aircooling with a complicated ducting arrangement, so most of our usual airflow goals apply.
To flow air through the radiator quietly, you want a radiator with a lot of frontal area, reasonable thickness, and open fin spacing. I elected to use a heater core for its good price/performance ratio and went for the biggest frontal area I could fit. The core is for a 1995 Chevrolet G Van (Fedco 2-333 or GDI 399095) and has rated core dimensions of 9-1/2” x 7-3/8”. Mine is slightly smaller than rated and an excellent fit in the Genie. While the core’s 2” thickness is greater than ideal, I believe the large frontal area compensates. This core offers almost twice the area of the popular Chevette/Pro-120 cores, and about 50% more than a Black Ice Pro 2.
For the CPU waterblock, I was lucky to get a Storm G4 from Little River Waterblocks.
Any temperature advantage available in the block should allow a slower fan or pump speed, so a quiet system needs a good block. The G4 is a good block. Mine came configured with ½” barbs.
Getting a quiet pump was essential. In addition, I wanted a DC-powered pump, so the current choices were the C-Systems CSP-750 or the Innovatek/Eheim HPPS. Edward Ng’s used CSPs were available for a fraction of the HPPS’ price, so the decision was made.
Other Parts & Supplies
McMaster-Carr, the local hardware store, and the auto parts store were raided for additional parts. These included 7/16”-ID Masterkleer tubing (only $0.36 per foot!), various hose fittings, hose clamps, weather stripping, water-pump lubricant/corrosion inhibitor, etc. A project like this can go a lot faster when you remember to ignore Home Depot and Lowe’s and find a REAL hardware store.
Installation Steps
I will just touch on the highlights here, with the understanding that there’s more to installation than detailed here.
The radiator’s hose fittings required modification. I removed the ends of both tubes and epoxied ½” brass barbs onto the remaining pieces. Because the core’s tubes run at an angle, I couldn’t get as tight a mechanical joint as I’d hoped, but the connections passed a pressure test after letting the JB Weld set up for 24 hours. This qualified as Good Enough.
I had decided to use the case as my fan shroud, so the HC was mounted to the chassis’ front panel with zip ties and 1” nylon spacers. These secure the core very well while requiring almost no work. The core is sealed to the case with weather stripping and scrap foam so that all intake air must pass through it. The case’s side panels must be installed for proper cooling.
Mounting the waterblock was extremely simple, though it did require removing the motherboard and the stock HSF retention frame. No real secrets or tricks here.
The real art of watercooling lies in hose routing. I wanted to use the least amount of hose possible, so the pump is suspended at the same level as the waterblock and radiator fittings. The whole loop has about 2 feet of tubing.
I mixed distilled water with the recommended dose of Prestone pump lubricant and 1 drop per gallon of dishwashing soap. This is probably the cheapest acceptable blend available.
Installation Issues & Problems
While I suffered no serious problems, various smaller issues arose. These affect convenience more than performance, but will lead to changes at some point.
The HC’s connections are the system’s high point, and this makes bleeding a giant pain. I’d thought that tilting the PC during bleeding would solve this problem (It works fine on a shifter kart…), but it has not been as fast as I’d hoped. When I redo the loop, I will happily add more hose if it allows better bleeding. With my modest water flow, I don’t believe that an extra foot of tubing will hurt my cooling significantly.
The HC’s angled connections limit my mounting options. This was no problem for my initial setup, but it will prevent inverting the radiator. To solve this, I’ll need to break out a torch and replace the curved tubes with straight copper pipe. And JB Weld seemed like such a time saver. Sigh…
The 1” spacers I used between the HC and the case are a little too short. The Genie’s bezel hinges force the HC to a small angle. Easily fixed next time.
I ordered 5/8” spring hose clamps from McMaster for my 5/8” OD tubing. They are great clamps, but absolutely will not fit over the tubing when it’s mounted on a ½” barb. I am incredibly foolish and running without hose clamps at the moment. For this reason alone, my redo will probably be soon. To be honest, the tubing is very secure on ½” barbs, and I’ll probably need to cut it for removal. Careful eyes will notice that there is one clamp in the system. I put my T-line too near the pump’s intake, and the tubing didn’t quite seal on the pump. Using a worm-drive clamp stopped the slight leak, though I should have seen this problem in advance – there’s almost no tube between the 2 barbs.
Results
Surprisingly good, but also disappointing.
The CPU temperature is great – something around 10C less than with the Zalman. And no fan ramping up under load! FWIW, the IC7 reports roughly +17C over ambient when idling and +35C under CPUburn. Considering the IC7’s reputation, those numbers satisfy me. The 120 exhaust still runs at <4V (below startup voltage), which gives extremely low airflow. Regardless of fan speed, the 2.4C refuses to clock ANY higher with WC. EDIT: I've been running at 260FSB, or 3.12GHz, with 1.625V set in the BIOS. Not too surprising for a P4, but I’d hoped for a little something extra. However, this does mean my heat production hasn’t changed, and the temp comparisons are as meaningful as mobo readings ever get.
Initially, I had a problem with the PSU fan ramping up. Not too surprisingly, the Seasonic didn’t want its air preheated by the radiator. Because my Genie has a door over the external bays, a traditional PSU duct wasn’t possible. So I tried to come up with something more ghetto for the truly lazy WCer. With painter’s tape and a scrap of foam, I partitioned the Genie’s top compartment into a separate zone. Next, I used the Rheobus to turn off the 92mm exhaust. With 3 minutes’ work, the PSU pulls ambient air in over the hard drive, and I eliminated a fan. Very satisfactory engineering! Under long-term load, the PSU fan ramps up to ~1350rpm, only 100 above idle.
With 2 fans removed, pump noise became an issue late at night, when a faint HD-like whine came through the case. Running the pump at 8-9V solves this problem completely, but costs ~2C on the CPU. However, the SS becomes the loudest component, even at idle. Having a Super Silencer A3 as the “noise problem” in my system, I am perfectly satisfied with its noise signature. I’ve been living with its current configuration for about a month and have noticed no irritating noises.
Conclusion & Thoughts
For a mechanically minded gadget freak, watercooling is cool! I have significantly improved temperatures and noise compared to a quality HSF, and I blame this system’s inability to OC higher on my P4. However, I suspect that ducting an XP120 to the 120mm exhaust would have given better results by a price/performance standard. Such an approach would have required less work as well.
My results reflect a WC system that prioritizes low noise in a self-contained system, and I’m delighted with the outcome. I firmly believe that a good watercooling implementation requires significant costs in time, space, case modding, and money. While I accepted all that, newbies should approach watercooling with open eyes rather than expecting water to perform miracles by itself.
There are a couple of areas that I still wish to improve. The hose routing needs a better layout for filling and bleeding. In addition, I’d prefer to replace the aluminum pump to avoid any galvanic corrosion issues. I will inspect the components when I replumb the loop and report any problems. Finally, Zerex Racing Super Coolant should be a better additive than the Prestone lubricant, so I’ll make that change too.
Pictures
Clicking the thumbnail will give a somewhat bigger picture. Links also included to the official, Monster-Sized shots (in case you want to count the dust motes on my soundcard). The biggies are approximately 2-3MB, so consider yourselves warned. BTW, the Image Zoom extension for Firefox is very nice indeed.
Monster
Monster
Monster
Monster
Last edited by HammerSandwich on Tue May 03, 2005 11:00 pm, edited 2 times in total.
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
Revision 1
I spent considerable time rebuilding the loop today. No pictures yet.
A friend also using the CSP-750 pump (related thread) told me that his pump had failed with water in the motor housing. Examining my T-line, I saw that the water level had dropped, and I started sweating. Happily, this occurred on the day that the Laing DDC pump became available. The DDC is the pump which Apple uses in the dual-2.5GHz Mac and was engineered for serious reliability. And it has no shaft seal to fail...
I ordered a pump from Dangerden, and it arrived today. My plans included replacing the pump, rerouting tubing for quicker bleeding, and inspecting my components.
My most important finding was water in the CSP-750's motor assembly. It was not enough to cause any corrosion or electrical problems - YET.
I also found faint traces of slime in the loop. This may just be from the pump lubricant, but I was concerned about biological growth and boiled the distilled water for the new coolant. The waterblock's internals were clean, so no real problems at this point anyway.
I did not enjoy incorporating the DDC into my loop. The pump has integrated barbs for 3/8" tubing, so I had to handle different barb sizes throughout. I found that I could get a spring clamp for 1/2"-OD tubing over my 5/8"-OD. This wasn't easy - particularly with the DDC's parallel barbs - but it safely secures my larger tubing onto the pump. Using pumps that provide ~1gpm, I should probably just use 3/8" tubing and will consider converting with the next redo.
The loop now has the DDC on the case's floor by the PCI slots. The pump's barbs point up, with the output going straight to the waterblock. The block's output runs to a T-line and into the radiator. From the rad, the loop returns to the pump, and this run has a second T-line. With two tees, I can fill and bleed in just a few minutes.
Because the thermal grease is still curing, I'm not certain if temperatures improved. The new pump should give more flow, but I suspect the difference can't be seen with motherboard reports.
The good news is that the DDC is significantly quieter than the CSP-750. The DDC does not have an annoying high-frequency tone, and I am running it at 12V. (And it will not undervolt very far at all, certainly not to the rated 6V.) Vibration is an issue, as with most (all?) pumps. My DDC sits on ~1/3" of melamine foam, and this seems okay so far.
EDIT 1: The pump has been running for about 18 hours now, and I just measured the voltages. I believe the pump's current draw (not measured!) stressed my Sunbeam controller - that channel's heatsink is hotter than the others', and the maximum voltage is only 10.8v or so. The pump will start only slightly below maximum and stabilizes at 10.1v at that knob setting. It will then run down to ~7.5v. Slower speeds reduce vibration considerably. I haven't looked into the impact on cooling.
EDIT 2: I hooked up the RPM wire today. Assuming I have configured Speedfan properly, the DDC seems to run around 1800rpm at 10.1v. The readings are slightly unstable, however, so I'm not sure they're trustworthy. With the new wiring setup, I'm not recording voltage, but I assume that maximum undervolt is still about 7.5v. This gives approximately 1150rpm. This picture should give a better, uh, picture of both speeds with a stall when I tried to go even lower:
If this pump lives up to its reliability hype, I'll be happy enough. I can definitely recommend the DDC for quiet watercooling.
I spent considerable time rebuilding the loop today. No pictures yet.
A friend also using the CSP-750 pump (related thread) told me that his pump had failed with water in the motor housing. Examining my T-line, I saw that the water level had dropped, and I started sweating. Happily, this occurred on the day that the Laing DDC pump became available. The DDC is the pump which Apple uses in the dual-2.5GHz Mac and was engineered for serious reliability. And it has no shaft seal to fail...
I ordered a pump from Dangerden, and it arrived today. My plans included replacing the pump, rerouting tubing for quicker bleeding, and inspecting my components.
My most important finding was water in the CSP-750's motor assembly. It was not enough to cause any corrosion or electrical problems - YET.
I also found faint traces of slime in the loop. This may just be from the pump lubricant, but I was concerned about biological growth and boiled the distilled water for the new coolant. The waterblock's internals were clean, so no real problems at this point anyway.
I did not enjoy incorporating the DDC into my loop. The pump has integrated barbs for 3/8" tubing, so I had to handle different barb sizes throughout. I found that I could get a spring clamp for 1/2"-OD tubing over my 5/8"-OD. This wasn't easy - particularly with the DDC's parallel barbs - but it safely secures my larger tubing onto the pump. Using pumps that provide ~1gpm, I should probably just use 3/8" tubing and will consider converting with the next redo.
The loop now has the DDC on the case's floor by the PCI slots. The pump's barbs point up, with the output going straight to the waterblock. The block's output runs to a T-line and into the radiator. From the rad, the loop returns to the pump, and this run has a second T-line. With two tees, I can fill and bleed in just a few minutes.
Because the thermal grease is still curing, I'm not certain if temperatures improved. The new pump should give more flow, but I suspect the difference can't be seen with motherboard reports.
The good news is that the DDC is significantly quieter than the CSP-750. The DDC does not have an annoying high-frequency tone, and I am running it at 12V. (And it will not undervolt very far at all, certainly not to the rated 6V.) Vibration is an issue, as with most (all?) pumps. My DDC sits on ~1/3" of melamine foam, and this seems okay so far.
EDIT 1: The pump has been running for about 18 hours now, and I just measured the voltages. I believe the pump's current draw (not measured!) stressed my Sunbeam controller - that channel's heatsink is hotter than the others', and the maximum voltage is only 10.8v or so. The pump will start only slightly below maximum and stabilizes at 10.1v at that knob setting. It will then run down to ~7.5v. Slower speeds reduce vibration considerably. I haven't looked into the impact on cooling.
EDIT 2: I hooked up the RPM wire today. Assuming I have configured Speedfan properly, the DDC seems to run around 1800rpm at 10.1v. The readings are slightly unstable, however, so I'm not sure they're trustworthy. With the new wiring setup, I'm not recording voltage, but I assume that maximum undervolt is still about 7.5v. This gives approximately 1150rpm. This picture should give a better, uh, picture of both speeds with a stall when I tried to go even lower:
If this pump lives up to its reliability hype, I'll be happy enough. I can definitely recommend the DDC for quiet watercooling.
Last edited by HammerSandwich on Mon Dec 06, 2004 6:22 pm, edited 3 times in total.
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
-
- Posts: 86
- Joined: Sat Aug 28, 2004 5:29 pm
lol, I've never used CAD. Just plop it on the workbench and wait for whatever inspires me. The journey is more exicting that way.If a case by MNPCTech or Mashie is the polished CAD rendering that Boeing shows to the Pentagon
You've done a great job with the Genie. Very clean execution. I've never thought of taping ventilation holes. Nice find on the Chev raditor
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
Described a complete rebuild with a Laing DDC pump. Click to jump up.
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
Edited revision with DDC undervolting info.
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
Edited revision with DDC RPM info.
-
- SPCR Reviewer
- Posts: 2696
- Joined: Thu Dec 11, 2003 9:53 pm
- Location: Scarsdale, NY
- Contact:
Great; definitely what I was looking for...
I got to take a closer look at it this evening and noticed that they actually use those starburst pattern screws (I wish I knew what they're called!) to hold the pump together; it's not really worth it for me right now to go out and buy yet another tool just to get into the pump. I'll just use it without worrying about what's inside.
Still trying to figure out exactly how I want to implement mine. I'm considering downvolting and downclocking stuff and running fans slower to really enjoy the quieter pump, but suddenly my OCZ PSU is looking more and more likely to become a bigger noise issue...
Extracting the PSU for modification or for swapping in my case is not easy...not easy at all.
-Ed
I got to take a closer look at it this evening and noticed that they actually use those starburst pattern screws (I wish I knew what they're called!) to hold the pump together; it's not really worth it for me right now to go out and buy yet another tool just to get into the pump. I'll just use it without worrying about what's inside.
Still trying to figure out exactly how I want to implement mine. I'm considering downvolting and downclocking stuff and running fans slower to really enjoy the quieter pump, but suddenly my OCZ PSU is looking more and more likely to become a bigger noise issue...
Extracting the PSU for modification or for swapping in my case is not easy...not easy at all.
-Ed
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
Torx. If you look again, you'll see the screws are also slotted for a standard screwdriver. I didn't have any problems opening mine that way, and I do have Torx drivers in the toolbox.Edward Ng wrote:...they actually use those starburst pattern screws (I wish I knew what they're called!) to hold the pump together...
Didn't you say it was okay if this PC turned out noisy?Edward Ng wrote:I'm considering downvolting and downclocking stuff and running fans slower to really enjoy the quieter pump...
-
- SPCR Reviewer
- Posts: 2696
- Joined: Thu Dec 11, 2003 9:53 pm
- Location: Scarsdale, NY
- Contact:
Depends; I don't think 400/1100 and 2400 are that much slower on GPU/memory and CPU than 425/1180 and 2600, particularly if I can keep the memory speed up and memory timings tight. There's a gigantic voltage difference on the CPU between 2400 and 2600, and I can bring the voltage on the GPU back down as well.
But yeah, it won't make a significant difference if I don't cut down on the PSU noise...
-Ed
But yeah, it won't make a significant difference if I don't cut down on the PSU noise...
-Ed
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
I took some air temperatures last night. I still haven't calibrated my CPU/mobo temps and used my GOAT (Good Ole Analog Thermometer...) for the others.
After 1 hour of Prime95 Large FFTs:
- 120 exhaust (rad & mobo) was +9C over ambient.
- PSU exhaust was +13-14C.
- CPU temp (reported) was +30C.
- PSU fan was 1400rpm.
The rad exhaust temp does include heat from the mobo, vidcard, etc. However, I think it's safe to say that my low airflow gives much higher water temps than most WCers like to see. I'll try to get those in the next day or two.
Water Temps
- Idles at +5-6C over ambient.
- Under same load as above, it is about +11C.
Looks like a big rad is a valid substitute for airflow.
After 1 hour of Prime95 Large FFTs:
- 120 exhaust (rad & mobo) was +9C over ambient.
- PSU exhaust was +13-14C.
- CPU temp (reported) was +30C.
- PSU fan was 1400rpm.
The rad exhaust temp does include heat from the mobo, vidcard, etc. However, I think it's safe to say that my low airflow gives much higher water temps than most WCers like to see. I'll try to get those in the next day or two.
Water Temps
- Idles at +5-6C over ambient.
- Under same load as above, it is about +11C.
Looks like a big rad is a valid substitute for airflow.
Last edited by HammerSandwich on Fri Dec 17, 2004 4:59 pm, edited 1 time in total.
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
Frankly, I think my configuration is wrong, since others have reported ~3900rpm. I trust Roscal to be more thorough than I am.Bat wrote:When monitoring the pump speed, are you taking into account that the pump produces ten pulses per revolution? (I'm guessing you are, since you report 1800rpm or 30Hz, but it's as well to check.)
I have Speedfan set with:
- Fan1 Divisor = 1
- Fan1 Mult = 1
- Fan1 Div = 10
I should check on this with Bill Adams...
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
Bump for water temp edit.
-
- Posts: 271
- Joined: Thu Jun 24, 2004 3:19 pm
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
Thanks for the praise, James. And I've got a special treat in store for zoob the patient. Just sneaking in under the 6-month wire since his initial request for pics, I am reminded of how quickly everything gets posted on the net.
Okay, let's really get to work. There have been a couple small changes since the last update.
In reverse order, I finally broke down and put a good Japanese L1A into my Super Silencer. This reduced noise noticeably, which implies good things about the system's overall loudness. Haven't seen any troubles with startup (though this box runs nearly 24/7), and the exhaust stays cool enough that I haven't bothered measuring it. You guys will notice that I forgot to bend the intake slats while I had it apart, but I can't be bothered now. Either way, I highly recommended the swap unless you have high ambient temps or some such.
A Yate Loon D12SL replaced the Evercool a little while ago. I skipped intensive testing, but a quick Prime95 run showed that temps were maybe 1-2C better with a lower noise level. The PSU's subsequent fan swap allowed a lower speed on the YL, and I've been running it at <5V. Airflow is damned near non-existent, but temps and stability show no problems.
Here's a teaser pic (that fits forum rules):
Click for the other, bigger pics.
Same as above. A shot towards the rear. Towards the front. HD mounting details. It uses Stretch Magic tied to the cage to compress the drive down into Sorbothane. Front view of the top section. Check out the awesome hose plugs I found for $.47 each! The wire running out the front goes to the Sunbeam's 4th channel and makes fan testing a breeze. Top down. Excess wiring detail. The Genie has a convenient ledge next to the 5" bays that gives good cable routing after a little Dremel action.
The biggest improvement I can imagine would be a thinner, single-pass rad with the DDC intake mod and somewhat cleaner hose routing. And the next time the mobo comes out, I'll drill a couple holes in the shelf to hold the T-lines. (Near the cable tie on the right of the "top down" pic.)
Frankly, I'm in no hurry on that wishlist. The machine has been working very nicely and quietly. Apart from the changes described above, I've topped up the water once - maybe twice - in the last 6 months, cleaned the air filter several times, and simply used the PC. It may be time for a new project...
Okay, let's really get to work. There have been a couple small changes since the last update.
In reverse order, I finally broke down and put a good Japanese L1A into my Super Silencer. This reduced noise noticeably, which implies good things about the system's overall loudness. Haven't seen any troubles with startup (though this box runs nearly 24/7), and the exhaust stays cool enough that I haven't bothered measuring it. You guys will notice that I forgot to bend the intake slats while I had it apart, but I can't be bothered now. Either way, I highly recommended the swap unless you have high ambient temps or some such.
A Yate Loon D12SL replaced the Evercool a little while ago. I skipped intensive testing, but a quick Prime95 run showed that temps were maybe 1-2C better with a lower noise level. The PSU's subsequent fan swap allowed a lower speed on the YL, and I've been running it at <5V. Airflow is damned near non-existent, but temps and stability show no problems.
Here's a teaser pic (that fits forum rules):
Click for the other, bigger pics.
Same as above. A shot towards the rear. Towards the front. HD mounting details. It uses Stretch Magic tied to the cage to compress the drive down into Sorbothane. Front view of the top section. Check out the awesome hose plugs I found for $.47 each! The wire running out the front goes to the Sunbeam's 4th channel and makes fan testing a breeze. Top down. Excess wiring detail. The Genie has a convenient ledge next to the 5" bays that gives good cable routing after a little Dremel action.
The biggest improvement I can imagine would be a thinner, single-pass rad with the DDC intake mod and somewhat cleaner hose routing. And the next time the mobo comes out, I'll drill a couple holes in the shelf to hold the T-lines. (Near the cable tie on the right of the "top down" pic.)
Frankly, I'm in no hurry on that wishlist. The machine has been working very nicely and quietly. Apart from the changes described above, I've topped up the water once - maybe twice - in the last 6 months, cleaned the air filter several times, and simply used the PC. It may be time for a new project...
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
If you see a way to make the shroud bigger, let me know!
HammerSandwich wrote:I had decided to use the case as my fan shroud, so the HC was mounted to the chassis� front panel with zip ties and 1� nylon spacers. These secure the core very well while requiring almost no work. The core is sealed to the case with weather stripping and scrap foam so that all intake air must pass through it. The case�s side panels must be installed for proper cooling.
Have you considered flipping the case fan so that it blows air into the case rather than out of it? As it is, you've got warm CPU heated air entering the main chamber--not the best thing for GPU and chipset cooling. If the rear case fan is flipped, then CPU heated air will immediately exit the case out the front.
-
- *Lifetime Patron*
- Posts: 1288
- Joined: Sat Oct 25, 2003 3:21 pm
- Location: 15143, USA
- Contact:
I haven't tried that since I went WCed, but the idea has merit with the VGA Silencer in place. But I'm not going to play with it, because the temps are more than fine, noise is truly low, and the Genie's intake filter works as is. The PC runs so well that I'd rather just use it for a while.
There may be rough plans for a custom case forming here, though. Nothing will happen there for a little while, thanks to unfortunately high material cost estimates.
There may be rough plans for a custom case forming here, though. Nothing will happen there for a little while, thanks to unfortunately high material cost estimates.