It's done! After a week of delay waiting for Swiftech to get me the right backplate for my GPU block, it's all together, tested and running.
Parts used in the project:
Swiftech Apogee CPU block
Swiftech MCW60 GPU block + copper BGA ramsinks
Swiftech MCR220 dual 120mm radiator
Swiftech MCP655 pump
Danger Den acrylic bay reservoir
Danger Den 1/2" x 3/4" vinyl tubing
2 Yate Loon D12SL-12 120x25mm fans
Busted Nidec TA450DC 120x38mm aluminum body fan - organ donor 😀
3 120mm fan guards
10 worm drive hose clamps
Arctic Silver 5
Assorted #4 and #6 machine screws and nuts, a few grommets, plumber's tape, shrink tubing.
Ok, :camera: time.
Pic 1 - Here's the bare case frame after the loop was assembled and leak tested overnight. Everything checked out. Having an old AT power supply with its own on/off switch kicking around was very helpful for filling, bleeding and testing this thing.
Pic 2 - A closer view of the rad mount. There's the gutted Nidec fan body holding the rad to the chassis, and providing just the right amount of clearance for the barbs. The barbs, grommets and tubing are helping with lateral stability, but the real support work is being done with the metal.
Pic 3 - Another view of the rad. Fan wires run through a hole I made in the I/O backplate, not the chassis! One of the project goals was to be as non-destructive as possible to the P160 itself.
By now you're probably wondering, why do I have my rad set up to breathe PSU exhaust? I didn't originally plan it that way. I was going to have the rad body hang down, but when it arrived, I saw that there was a case rail right where I would have had to drill for an inlet. D'oh. I'm not too worried about performance loss, the PSU is big and fat, and it runs amazingly cool under the loads I make it push. It's something I can accept.
Pic 4 - An inside view. There's the X1900XT with block and sinks. Upper left of the mobo is where my rad fan leads come through and attach to controllable fan headers.
Pic 5 - Pump on its shock mount. I drilled 2 extra holes into the mounting plate for extra stability and safety.
Pic 6 - Other side of pump mount. I drilled and tapped the side of the drive cage for #4-40 thread. The 1/2" machine screws are perfectly flush with the metal surface when they are snugged up - making it so the drive trays still travel freely in and out. Non-destructive goodness.
Pic 7 - Cable Management, Pain-style 😀 It ain't the prettiest, but it's outta the way!
Pic 8 - Cables all nice and tidy in the business part of the case, where it matters to me.
Pic 9 - Not much clearance between the rad and the PSU plug... but have no fear, Pic 10 is here. I bought the Swifty rad because it's the narrowest of the bunch. Originally, the narrowness was going to still let me plug my CRT into the dongle on the back of my vid card. Now it's allowing me to still plug my PC into the wall. Whew! Also, if I had used a spacer slimmer than that old fan body, I would have been SOL.
Pic 11 - Front bezel with reservoir of green goo. I like it 😛 I should look for something neat to put into the bottom bay now, but I'm not sure what.
Pic 12 - The fully assembled machine. I've never really beena fan of window cases, but I bought my P160 that way to save a few bucks. I'm glad now, since it's nice for keeping an eye on things in there.
So far so good, the pump isn't the quietest thing in the world, but it has a pot for speed adjustment. I'll see how low I can set it before performance drops off. As for performance, I need to stress test for several hours to make a real evaluation, but here's what I had after 23 or so minutes of CPU and GPU heatup. PWM and NF4 are up, I expected that - and of course, CPU is down. GPU is WAY down.
I'll accept this tradeoff. 😀
UPDATE - 19 Mar 06 - Got a bit more tweaking done. I'm trying out a Storm Rev. 2 waterblock, and I've done some work on PWM cooling for the mobo and X1900. Here's a new Pic for your review.
I've also tapped the fan mounting holes on my rad out to #6/32, so I won't have trouble finding screws for mounting some 38mm fans later for testing - I wanna get my Panaflos on there at some point to see if the extra air pressure improves rad performance. Why Swiftech has the MCR220 fan mounts tapped for metric M1.5 hardware, I just don't know. Finding that stuff can be hell, and they don't give you any screws long enough for mounting 38mm.
I was gonna swap out the fan in my VC-RE chipset cooler for a quieter one, but the replacement embedded fan I bought is just a hair too big 🙁 but it will fit my old Tt G4 VGA cooler perfectly, and it needs a new fan, so that purchase wasn't in vain. I still use my GF4 card in my other box, and it'll be nice to get that big 'ol slab of copper back on it.
Anywho. The fan on the motherboard's PWM is a Delta FFB series 40x28mm, 8 CFM @ 5000 RPM @ 23.9 dBA. It's really a 40x20mm with 8mm of plastic filler/directional vanes on the exhaust side. That extra plastic on the business end actually does me a lot of good, because the way I mounted this little chunker was to put it onto a pair of #6/32 x 2 1/2" screws, and then jam the screw shanks down into the fins of the motherboard's PWM heatsinks 😀 It's only held in by tension, but it's actually pretty secure in its place, due to the way the heatsinks are notched. The extra plastic in the fan body keeps the tension from bending it. This fan is run at ~7 volts via Rheobus, at 7 volts it's pretty much inaudible to me. In case you doubt that this gives enough airflow to cool those FETs down, know that my PWMIC idle temp right now is 35º C, and load is somewhere in the 40s. I'll do some more looping later and report some numbers.
On to the X1900... Yeah, at first glance the fan placement ain't great, half of its flow is going to the wrong place, but the VC-RE cooler is catching a lot of it. NF4 temps are down about 6º C from before right now 🙂 The fan itself is a Delta EFB series 60x10mm, 13.4 CFM @ 3000 RPM @ 24.5 dBA. I think that's overstated. It's running @ full power and I can't say that I hear it. It's rated louder than the PWM fan, which is clearly audible at full power.
My other idea for cooling this area was to strap a pair of 40x10mm together and position them over the VRM, but the mounting options didn't please me, and 40x10mm fans tend to be whiny as hell. Viper John uses a pair of 30mm strapped together and held in with plastic clips for his X1900 mods, but 30mm and smaller fans are rare, and I didn't feel like buying the ones at AllElectronics and praying that they'd be quiet.
This Delta is mounted using #4/40 x 1 1/2" screws, nuts and some plastic spacers to avoid crushing/shorting nearby SMD components. I got the spacers from an unused mount kit for one of the waterblocks. The X1900 PCB has several holes drilled through it (some are for the stock HSF, others are there for no apparent reason), but 2 of them happen to be exactly! the right distance apart for mounting a 60mm using kitty corner screwholes. That was pleasing to me, so 60mm it was.
So, does the fan do anything useful for the X1900? Running the card at full stock 3D clocks and voltage, w/o any PWM cooling, that sensor would register a searing 91º C after an hour of rthdribl. Preliminary testing with the fan in place gave me a PWM load temp of 55º C. A 36º C drop? Yes, that's what it's doing. You can keep your fingers on the card behind the PWM without burning yourself now. In fact, it doesn't even feel hot back there anymore, just warm under load.
I'm going to stick with this solution. In fact, I may even rig up some sort of shroud to direct as much offside airflow as possible down towards the NF4 chip. This is turning out to be 2 for the price of 1.
And that just about wraps up this project. There's one more thing that will be added for sure, I'll be putting micro heatsinks on the X1900's coil blocks. I forgot to order them when I bought the storm and the 60mm Delta, but they'll be here probably Wednesday. Then I'll OC the crap out of that vid card. It'll do 700/800 without much fuss, but I've been hesitant to really blaze that trail due to the PWM temp issues.
The quality Delta fans featured in today's update were purchased at Sidewinder Computers. I'm developing a love/hate relationship with this shop. Their shipping is fast and cheap, and that has made it so much easier for me to nickel and dime myself to death on this project :laugh: but it's turned out well, thanks in part to the cool stuff they sell. I shamelessly plug them because I'm fond of them, and no other reason.
As for the Storm (almost forgot 😀): Well, this has become one hell of a wordy update, but I'll give my first impressions. It's a hell of a piece of work, BUT, I did find some corrosion on the copper plate in and around the cups when I cracked it open for inspection. My guess is that it was coolant/lubricant not properly rinsed away after the the CNC machining process. Swiftech really needs to step up their QC process - they took their lumps with the Apogee and did what they needed to do, but this just ain't cool. For almost twice the price, I expect perfection right out of the box. 10 minutes with some salted lemon juice, a toothbrush and lots of rinse water fixed the problem, but still...
Installation/filling/bleeding: I'm not sure if this paragraph will end up being a tribute to the block, or the power of the Laing D5 pump. They say that this block is very restrictive, but during this process I could see a LOT of water moving through the loop. As I was bleeding, it was obvious that the pump had no trouble shooting coolant through the Storm, the MCW60, the rad and back into the res, with still enough force to generate plenty of turbulence inside the res. This amount of enthusiasm was noted while the pump was at the 3.5 power setting, which I've left it at for most of the time I've had it now. The Rev. 2 block IS tweaked for better flow, but it's still an impingement device, so I'm going to hand a "kickass" award to the D5 here. Nice pump.
Storm performance: I'll get back to ya 😀 I have to let it loop for a while, and it would be more fair to let the AS5 cure for a few days before I do so. It certainly seems to be no worse than the Apogee so far (and it ought not be!), but I won't say it's better in my particular setup at this point. We'll see. 🙂
Parts used in the project:
Swiftech Apogee CPU block
Swiftech MCW60 GPU block + copper BGA ramsinks
Swiftech MCR220 dual 120mm radiator
Swiftech MCP655 pump
Danger Den acrylic bay reservoir
Danger Den 1/2" x 3/4" vinyl tubing
2 Yate Loon D12SL-12 120x25mm fans
Busted Nidec TA450DC 120x38mm aluminum body fan - organ donor 😀
3 120mm fan guards
10 worm drive hose clamps
Arctic Silver 5
Assorted #4 and #6 machine screws and nuts, a few grommets, plumber's tape, shrink tubing.
Ok, :camera: time.
Pic 1 - Here's the bare case frame after the loop was assembled and leak tested overnight. Everything checked out. Having an old AT power supply with its own on/off switch kicking around was very helpful for filling, bleeding and testing this thing.
Pic 2 - A closer view of the rad mount. There's the gutted Nidec fan body holding the rad to the chassis, and providing just the right amount of clearance for the barbs. The barbs, grommets and tubing are helping with lateral stability, but the real support work is being done with the metal.
Pic 3 - Another view of the rad. Fan wires run through a hole I made in the I/O backplate, not the chassis! One of the project goals was to be as non-destructive as possible to the P160 itself.
By now you're probably wondering, why do I have my rad set up to breathe PSU exhaust? I didn't originally plan it that way. I was going to have the rad body hang down, but when it arrived, I saw that there was a case rail right where I would have had to drill for an inlet. D'oh. I'm not too worried about performance loss, the PSU is big and fat, and it runs amazingly cool under the loads I make it push. It's something I can accept.
Pic 4 - An inside view. There's the X1900XT with block and sinks. Upper left of the mobo is where my rad fan leads come through and attach to controllable fan headers.
Pic 5 - Pump on its shock mount. I drilled 2 extra holes into the mounting plate for extra stability and safety.
Pic 6 - Other side of pump mount. I drilled and tapped the side of the drive cage for #4-40 thread. The 1/2" machine screws are perfectly flush with the metal surface when they are snugged up - making it so the drive trays still travel freely in and out. Non-destructive goodness.
Pic 7 - Cable Management, Pain-style 😀 It ain't the prettiest, but it's outta the way!
Pic 8 - Cables all nice and tidy in the business part of the case, where it matters to me.
Pic 9 - Not much clearance between the rad and the PSU plug... but have no fear, Pic 10 is here. I bought the Swifty rad because it's the narrowest of the bunch. Originally, the narrowness was going to still let me plug my CRT into the dongle on the back of my vid card. Now it's allowing me to still plug my PC into the wall. Whew! Also, if I had used a spacer slimmer than that old fan body, I would have been SOL.
Pic 11 - Front bezel with reservoir of green goo. I like it 😛 I should look for something neat to put into the bottom bay now, but I'm not sure what.
Pic 12 - The fully assembled machine. I've never really beena fan of window cases, but I bought my P160 that way to save a few bucks. I'm glad now, since it's nice for keeping an eye on things in there.
So far so good, the pump isn't the quietest thing in the world, but it has a pot for speed adjustment. I'll see how low I can set it before performance drops off. As for performance, I need to stress test for several hours to make a real evaluation, but here's what I had after 23 or so minutes of CPU and GPU heatup. PWM and NF4 are up, I expected that - and of course, CPU is down. GPU is WAY down.
I'll accept this tradeoff. 😀
UPDATE - 19 Mar 06 - Got a bit more tweaking done. I'm trying out a Storm Rev. 2 waterblock, and I've done some work on PWM cooling for the mobo and X1900. Here's a new Pic for your review.
I've also tapped the fan mounting holes on my rad out to #6/32, so I won't have trouble finding screws for mounting some 38mm fans later for testing - I wanna get my Panaflos on there at some point to see if the extra air pressure improves rad performance. Why Swiftech has the MCR220 fan mounts tapped for metric M1.5 hardware, I just don't know. Finding that stuff can be hell, and they don't give you any screws long enough for mounting 38mm.
I was gonna swap out the fan in my VC-RE chipset cooler for a quieter one, but the replacement embedded fan I bought is just a hair too big 🙁 but it will fit my old Tt G4 VGA cooler perfectly, and it needs a new fan, so that purchase wasn't in vain. I still use my GF4 card in my other box, and it'll be nice to get that big 'ol slab of copper back on it.
Anywho. The fan on the motherboard's PWM is a Delta FFB series 40x28mm, 8 CFM @ 5000 RPM @ 23.9 dBA. It's really a 40x20mm with 8mm of plastic filler/directional vanes on the exhaust side. That extra plastic on the business end actually does me a lot of good, because the way I mounted this little chunker was to put it onto a pair of #6/32 x 2 1/2" screws, and then jam the screw shanks down into the fins of the motherboard's PWM heatsinks 😀 It's only held in by tension, but it's actually pretty secure in its place, due to the way the heatsinks are notched. The extra plastic in the fan body keeps the tension from bending it. This fan is run at ~7 volts via Rheobus, at 7 volts it's pretty much inaudible to me. In case you doubt that this gives enough airflow to cool those FETs down, know that my PWMIC idle temp right now is 35º C, and load is somewhere in the 40s. I'll do some more looping later and report some numbers.
On to the X1900... Yeah, at first glance the fan placement ain't great, half of its flow is going to the wrong place, but the VC-RE cooler is catching a lot of it. NF4 temps are down about 6º C from before right now 🙂 The fan itself is a Delta EFB series 60x10mm, 13.4 CFM @ 3000 RPM @ 24.5 dBA. I think that's overstated. It's running @ full power and I can't say that I hear it. It's rated louder than the PWM fan, which is clearly audible at full power.
My other idea for cooling this area was to strap a pair of 40x10mm together and position them over the VRM, but the mounting options didn't please me, and 40x10mm fans tend to be whiny as hell. Viper John uses a pair of 30mm strapped together and held in with plastic clips for his X1900 mods, but 30mm and smaller fans are rare, and I didn't feel like buying the ones at AllElectronics and praying that they'd be quiet.
This Delta is mounted using #4/40 x 1 1/2" screws, nuts and some plastic spacers to avoid crushing/shorting nearby SMD components. I got the spacers from an unused mount kit for one of the waterblocks. The X1900 PCB has several holes drilled through it (some are for the stock HSF, others are there for no apparent reason), but 2 of them happen to be exactly! the right distance apart for mounting a 60mm using kitty corner screwholes. That was pleasing to me, so 60mm it was.
So, does the fan do anything useful for the X1900? Running the card at full stock 3D clocks and voltage, w/o any PWM cooling, that sensor would register a searing 91º C after an hour of rthdribl. Preliminary testing with the fan in place gave me a PWM load temp of 55º C. A 36º C drop? Yes, that's what it's doing. You can keep your fingers on the card behind the PWM without burning yourself now. In fact, it doesn't even feel hot back there anymore, just warm under load.
I'm going to stick with this solution. In fact, I may even rig up some sort of shroud to direct as much offside airflow as possible down towards the NF4 chip. This is turning out to be 2 for the price of 1.
And that just about wraps up this project. There's one more thing that will be added for sure, I'll be putting micro heatsinks on the X1900's coil blocks. I forgot to order them when I bought the storm and the 60mm Delta, but they'll be here probably Wednesday. Then I'll OC the crap out of that vid card. It'll do 700/800 without much fuss, but I've been hesitant to really blaze that trail due to the PWM temp issues.
The quality Delta fans featured in today's update were purchased at Sidewinder Computers. I'm developing a love/hate relationship with this shop. Their shipping is fast and cheap, and that has made it so much easier for me to nickel and dime myself to death on this project :laugh: but it's turned out well, thanks in part to the cool stuff they sell. I shamelessly plug them because I'm fond of them, and no other reason.
As for the Storm (almost forgot 😀): Well, this has become one hell of a wordy update, but I'll give my first impressions. It's a hell of a piece of work, BUT, I did find some corrosion on the copper plate in and around the cups when I cracked it open for inspection. My guess is that it was coolant/lubricant not properly rinsed away after the the CNC machining process. Swiftech really needs to step up their QC process - they took their lumps with the Apogee and did what they needed to do, but this just ain't cool. For almost twice the price, I expect perfection right out of the box. 10 minutes with some salted lemon juice, a toothbrush and lots of rinse water fixed the problem, but still...
Installation/filling/bleeding: I'm not sure if this paragraph will end up being a tribute to the block, or the power of the Laing D5 pump. They say that this block is very restrictive, but during this process I could see a LOT of water moving through the loop. As I was bleeding, it was obvious that the pump had no trouble shooting coolant through the Storm, the MCW60, the rad and back into the res, with still enough force to generate plenty of turbulence inside the res. This amount of enthusiasm was noted while the pump was at the 3.5 power setting, which I've left it at for most of the time I've had it now. The Rev. 2 block IS tweaked for better flow, but it's still an impingement device, so I'm going to hand a "kickass" award to the D5 here. Nice pump.
Storm performance: I'll get back to ya 😀 I have to let it loop for a while, and it would be more fair to let the AS5 cure for a few days before I do so. It certainly seems to be no worse than the Apogee so far (and it ought not be!), but I won't say it's better in my particular setup at this point. We'll see. 🙂
