Soldato
- Joined
- 18 Mar 2008
- Posts
- 12,751
I think that the final thing will be really amazing mate 
Project: Desk Rad
- Thread starter JohnTedd
- Start date
Here's the room in its "finished" state (as if anything ever is!):
And here's the inside of the finished case (apart, perhaps from some more cable tidying):
And one more close up:
And here's the inside of the finished case (apart, perhaps from some more cable tidying):
And one more close up:
Too many cables!!
I think I have the minimum number of cables for this set up, which includes 4 hard disks + one blu-ray burner (5 SATA cables), 6 temperature sensors, 4 fans, one pump and radiator fans, 4 PCI power cables, CPU cables (two), power cable for disks + power cable for bigNG, fans + blu-ray burner.
However I do agree that they could be routed a lot more neatly, although that's never been my strong point. In fact my wife says that "Mess" is my middle name...
Associate
Thought about making a little backing for the desk that goes all the way down to the floor with holes to route cables? Would cover up the mass of cables under your desk and you could remove it to get to what's there if you need to.
It seems to be working out quite well now but I am in the process of looking at the overclocking and the configuration of the bigNG. I have decided that the radiator fans have to be run at 100% all the time. They are very low speed anyway (750 RPM) and are virtually silent, especially behind the desk and dropping the speed down too much does appear to make the water temperature rise and never fall. At this stage I don't really know how far the water temperature would go (nor how disastrous that would be) so they are fixed at 100% for now. I have a few spare 12cm fans that I might additionally point at the radiator to see if they give extra cooling. That would just be an experiment rather than rushing out and buying higher RPM fans, but the aim would be to decide whether or not it's worth buying higher RPM fans. Also, I may experiment with making the fans push instead of pull.
Hours of fun ahead
That is certainly a possibility I hadn't considered although I am not sure if I would bang my feet on it. I certainly agree that the cables don't look great - I hadn't really noticed them until I looked at the photo of the final room.
My current idea for the cables is to staple some cable ties to the wooden desk back panel (which I have already done to some extent) and route the cables more tidily along there. Although it would have been neater to do this on the other side of the wooden back panel, I decided to keep water routing and cable routing on separate sides. Of course there shouldn't be a problem with them being together (they are inside the case), but I felt it was safer.
Anyway thanks for the option, which I will certainly consider.
John
So I spent the last few days doing a bit of bigNG configuration and overclocking. I will post the results later. However I did finally get round to looking into an issue that I'd rather ignored up until now... Like a number of people on various forums I had an issue with the BIOS seeing all 3 of my DDR3 RAM. The BIOS could detect that 3 sticks were installed, but only "Enabled" 2 of them. I had originally thought this was a BIOS issue and that I'd have to wait for Gigabyte to resume their support (yesterday) before getting it resolved. However I decided to do some googling and I discovered that:
1. It could be voltages set by the BIOS for the RAM and the vtt core (so try a new BIOS or manual settings).
2. It could be duff RAM (so inspect it and the slot and move the sticks around).
3. It could be that the processor/socket has bent pins or needs re-seating (so take a look and re-seat it).
I approached these options in the order above. I'm not an expert on voltages and so I only tinkered with this a bit. I had previously tried all the BIOSes available on the Gigabyte website, but I gave them another go. The newer BIOSes recognised that not all of the memory had been found and attempted to adjust itself, showing the message:
"Recovering lost DRAM size..."
and then rebooting. It did this a number of times before giving up... Made the boot time very long, so during the initial build I went back to the original BIOS that didn't do this. That's what made me think it was a BIOS issue. Anyway this time I tried the new BIOSes and fiddled with the voltages, but no luck.
I then tried all the sticks in all of the slots. This made no difference either, so I was left with the CPU. Frankly I was dubious that the only thing affected could be the number of DRAM slots enabled, but this had been the eventual cure for some people (including on this forum) so I gave it a go. Imagine my surprise when it immediately detected all 3 sticks. I was chuffed, but that soon turned to dismay when the dreaded blue-screen hit... I am not precisely sure whether the RAM was implicated in this, or whether the various other changes (e.g. to the BIOS) caused the blue-screen, but the upshot was the disk was bad and windows couldn't recover it for some reason....
My OS disks are in a Mirrored RAID configuration, but that didn't seem to help me get out of this situation. In fact I think it might be partially responsible for it, as one time doing the BIOS flashing I forgot to set the Intel controller to RAID configuration. So the long and the short of it is that I had to restore an older backup, and I am still getting things back to where they were before. The good news is that I managed to retrieve my bigNG settings from the corrupted disk before wiping it. They are in a database stored in:
\AppData\Local\VirtualStore\Program Files (x86)\mCubed\T-Balancer Network Server (alpha)\Navigator 2.12
Anyway more to follow in overclocking and bigNG configuration....
1. It could be voltages set by the BIOS for the RAM and the vtt core (so try a new BIOS or manual settings).
2. It could be duff RAM (so inspect it and the slot and move the sticks around).
3. It could be that the processor/socket has bent pins or needs re-seating (so take a look and re-seat it).
I approached these options in the order above. I'm not an expert on voltages and so I only tinkered with this a bit. I had previously tried all the BIOSes available on the Gigabyte website, but I gave them another go. The newer BIOSes recognised that not all of the memory had been found and attempted to adjust itself, showing the message:
"Recovering lost DRAM size..."
and then rebooting. It did this a number of times before giving up... Made the boot time very long, so during the initial build I went back to the original BIOS that didn't do this. That's what made me think it was a BIOS issue. Anyway this time I tried the new BIOSes and fiddled with the voltages, but no luck.
I then tried all the sticks in all of the slots. This made no difference either, so I was left with the CPU. Frankly I was dubious that the only thing affected could be the number of DRAM slots enabled, but this had been the eventual cure for some people (including on this forum) so I gave it a go. Imagine my surprise when it immediately detected all 3 sticks. I was chuffed, but that soon turned to dismay when the dreaded blue-screen hit... I am not precisely sure whether the RAM was implicated in this, or whether the various other changes (e.g. to the BIOS) caused the blue-screen, but the upshot was the disk was bad and windows couldn't recover it for some reason....
My OS disks are in a Mirrored RAID configuration, but that didn't seem to help me get out of this situation. In fact I think it might be partially responsible for it, as one time doing the BIOS flashing I forgot to set the Intel controller to RAID configuration. So the long and the short of it is that I had to restore an older backup, and I am still getting things back to where they were before. The good news is that I managed to retrieve my bigNG settings from the corrupted disk before wiping it. They are in a database stored in:
\AppData\Local\VirtualStore\Program Files (x86)\mCubed\T-Balancer Network Server (alpha)\Navigator 2.12
Anyway more to follow in overclocking and bigNG configuration....
Permabanned
- Joined
- 9 Dec 2010
- Posts
- 18
Great kit, would love to see more pictures of the PC!
I've finally got round to taking some more photos. However the thing you really notice when you look at the photos is that I need to do some cable management and/or get a more modern case that provides a) more space and b) a motherboard tray behind which cables may be routed.
I've also done some overclocking and have a stable system operating at 4.29 GHz with FSB 165 MHz. None of the temps (of both GPUs and CPU + NB) gets to 60 deg C doing benchmarking or baking (3D/PCMark Vantage, Hot CPU or converting lots of videos). Turns out that the sternest test is converting videos and the conversion speed of this new PC is 4-5 times faster than my old one, which is great as this is why I built it
I've also done some overclocking and have a stable system operating at 4.29 GHz with FSB 165 MHz. None of the temps (of both GPUs and CPU + NB) gets to 60 deg C doing benchmarking or baking (3D/PCMark Vantage, Hot CPU or converting lots of videos). Turns out that the sternest test is converting videos and the conversion speed of this new PC is 4-5 times faster than my old one, which is great as this is why I built it
Here's a general photo of the main case:
And here are some close ups.
CPU
Northbridge
Both...
Looking along the GTX480s with their shiny Nickel backs:
My case mods for the tubing to get in... The holes are closer to square, but the grommets make them look somewhat circular...
RAM + CPU + NB
OK, so I went a bit mad there...
Finally the bigNG (OK you can hardly see it)...
CPU
Northbridge
Both...
Looking along the GTX480s with their shiny Nickel backs:
My case mods for the tubing to get in... The holes are closer to square, but the grommets make them look somewhat circular...
RAM + CPU + NB
OK, so I went a bit mad there...
Finally the bigNG (OK you can hardly see it)...
Last edited:
Overclocking
I have overclocked the FSB from its factory default of 133 MHz to 170 MHz. This gives a processor lift from 3.33 GHz to 4.29 GHz. I did find that I could go to 180 MHz (4.5GHz) and it worked fine under load. However I've had a few strange blue screens on system boot so I decided to wind it back down for now. I did also try changing the CPU multiplier from 25 upwards but didn't get any better results.
As a result of this the CPU idle temp has risen from 23-24 to 33-35 deg C. Under load they can reach ~60 deg C (I think once some of the cores may have been near 70).
I have also overclocked the ASUS GTX 480s from 700 MHz to 804 MHz (and DDR from 3696 to 4004). This increases their idle temps from 30-31 to 42-45 deg C. Under load they get to 55-60 deg C.
I have overclocked the FSB from its factory default of 133 MHz to 170 MHz. This gives a processor lift from 3.33 GHz to 4.29 GHz. I did find that I could go to 180 MHz (4.5GHz) and it worked fine under load. However I've had a few strange blue screens on system boot so I decided to wind it back down for now. I did also try changing the CPU multiplier from 25 upwards but didn't get any better results.
As a result of this the CPU idle temp has risen from 23-24 to 33-35 deg C. Under load they can reach ~60 deg C (I think once some of the cores may have been near 70).
I have also overclocked the ASUS GTX 480s from 700 MHz to 804 MHz (and DDR from 3696 to 4004). This increases their idle temps from 30-31 to 42-45 deg C. Under load they get to 55-60 deg C.
T-Balancer BigNG
All of the above overclocking was done with the pump being controlled by the T-Balancer BigNG and operating at around 50% (at the moment 54% for water temp of 28.5). Also the case fans are controlled by the BigNG and are also at 50%. The pump and rad fans are inaudible and the system itself is not noisy although you can hear the fans. The main culprit is the Dominator RAM Airflow fan which I plan to replace with the EK-RAM waterblock when it comes out.
I am currently using all of the temperature sensors in my BigNG setup rather than trying to interface it to the information available from the various temperature sensors in the system itself, such as the CPU core temps. I have the digital sensors set to measure the difference in water temp between the input to the CPU WB and the output of the graphics cards. I have the analogue sensors measuring HD temp, RAM temp, NB temp and case temp.
I would like to use the temp difference in the water but the BigNG doesn't seem to support that - it only appears to work on absolute temps. I therefore run the pump according to the water temp.
I run the case fans according to the case temp and I try to run the Airflow fans according to the RAM temp, but it seems to ignore my settings.
I also run the bottom bay fan from the HD temp (although that now seems to be mis-measuring and I have not yet investigated why).
I have not finished tinkering with this yet. Not by a long way, but it does seem to largely be operating as I want, making the system quiet without manual intervention.
I did upgrade to the 2.34 firmware which seemed to help some of my issues (although initially it kept beeping at me which was very annoying) and the alpha 2.12 Navigator. I have yet to get on with the "server" version that allows networked access and, it claims, a cleaner API.
All of the above overclocking was done with the pump being controlled by the T-Balancer BigNG and operating at around 50% (at the moment 54% for water temp of 28.5). Also the case fans are controlled by the BigNG and are also at 50%. The pump and rad fans are inaudible and the system itself is not noisy although you can hear the fans. The main culprit is the Dominator RAM Airflow fan which I plan to replace with the EK-RAM waterblock when it comes out.
I am currently using all of the temperature sensors in my BigNG setup rather than trying to interface it to the information available from the various temperature sensors in the system itself, such as the CPU core temps. I have the digital sensors set to measure the difference in water temp between the input to the CPU WB and the output of the graphics cards. I have the analogue sensors measuring HD temp, RAM temp, NB temp and case temp.
I would like to use the temp difference in the water but the BigNG doesn't seem to support that - it only appears to work on absolute temps. I therefore run the pump according to the water temp.
I run the case fans according to the case temp and I try to run the Airflow fans according to the RAM temp, but it seems to ignore my settings.
I also run the bottom bay fan from the HD temp (although that now seems to be mis-measuring and I have not yet investigated why).
I have not finished tinkering with this yet. Not by a long way, but it does seem to largely be operating as I want, making the system quiet without manual intervention.
I did upgrade to the 2.34 firmware which seemed to help some of my issues (although initially it kept beeping at me which was very annoying) and the alpha 2.12 Navigator. I have yet to get on with the "server" version that allows networked access and, it claims, a cleaner API.
Last edited:
Broken radiator update
As you may have seen I started another thread off about fixing my leaking radiator. In the end I used Plastic Padding's Leak-Fix and it seems to have worked like a charm. So I am planning on adding this as a second radiator, also behind the desk to add some more cooling. Hopefully I'll be able to get the water temps down to, say, 25 deg C.
As you may have seen I started another thread off about fixing my leaking radiator. In the end I used Plastic Padding's Leak-Fix and it seems to have worked like a charm. So I am planning on adding this as a second radiator, also behind the desk to add some more cooling. Hopefully I'll be able to get the water temps down to, say, 25 deg C.
Final update for now... Having seen everyone else's project I have been inspired to do some braiding and have purchased a variety of sizes. They should come tomorrow and I hope to be able to have a go at the weekend.
Associate
Yeah AMEX do purchase protection like the other cards plus a user damage feature, much like some component suppliers do whereby they basically insure your items for a couple of weeks after purchase in case you damage them during fitmentmight be reading it wrong but you saying you punctured the rad, and then got a refund because you broke it?
YUP, sounds like you explained it best way there and they accepted it.
LOL ... those P18x cases soon fil;l up with cables don't they !
and that PSU dwarfs the space available, good to see yo uwent for the front fan mod, that was one thing i never understood with antec, they overlooked some of the simplest of things, and that case was supposed to have been designed by expert users !!
I'm loving the overall "different" approach to watercooling arrangements, and i'm liking your sideways thinking
I too tend to agree that it makes more sense to have the RAD etc outside the case for cleaning purposes !
Think i'll keep an eye on this thread and your progress
BTW ...... impressive "first shopping list" !!
EDIT:
have you looked at the MIPS RAM blocks ? it's what i'm going to give a whirl to cool the 8GB (4x2GB) of Corsair Dominator GT RAM in my setup rebuild.
Last edited:
I totally agree that the fan in the middle of the bottom compartment has always been a pain even when I had a much smaller PSU. Cables were always in danger of hitting the blades too (due, of course, to my lack of cable management
).I'm loving the overall "different" approach to watercooling arrangements, and i'm liking your sideways thinking
I too tend to agree that it makes more sense to have the RAD etc outside the case for cleaning purposes !
Think i'll keep an eye on this thread and your progress
BTW ...... impressive "first shopping list" !!
Thank you for the +ve comments. The main reason I like the external RAD idea is that the idea is to get the heat away from all the computer components, so why not go the whole hog and take them out of the case to a frankly huge heat sink (the room). Following my re-org to having the reservoir at the corner of the desk, bleeding the system is a piece of cake
.My original shopping list included Dominator GT RAM and I wasn't sure whether the MIPS was suitable (yes I was aware of it). However the Dominator GTs seemed to be out of stock everywhere and no one seemed to know when they would be in stock and since I was doing this build over Xmas I downgraded to the Dominator RAM. My plan was to see how bad the AirFlow fans were and to take it from there. I also thought (wrongly for some reason) that I could slow them down with my BigNG. Furthermore I must confess to being quite a big EK fan and all their products I've used to date are top notch, so I'm still leaning towards the EK ones.
Project Update
After having a go at overclocking and seeing the water temp rise to 29 deg C I decided that I needed a bit more cooling in my loop. Since I still had the punctured RAD I decided to try to fix it. I started a thread on the subject (http://forums.overclockers.co.uk/showthread.php?t=18227922). The summary is that the options are:
1. Don't risk it - buy a new one.
2. Solder it.
3. Repair it with car radiator repair kits - Liquid Metal or Leak-Fix.
I chose Leak-Fix mainly because it was in stock.
I immersed the rad in water and located the puncture (luckily there was only one despite me using the wrong screws in 6 holes). The location of the leak:
It took two coats of Leak-Fix to cover the hole to my satisfaction. It then looked like this:
I then immersed it in water again and blew into it, confirming the fix was water-tight. I then did the same thing but with a compressor (thanks to my father-in-law) and put 2 bar into it (that's equivalent to 20m of head pressure which is far more than my pump - 4m). With no leak detected I am pretty confident about the fix and I am now in the process of adding it behind the desk as a second rad.
More later...
After having a go at overclocking and seeing the water temp rise to 29 deg C I decided that I needed a bit more cooling in my loop. Since I still had the punctured RAD I decided to try to fix it. I started a thread on the subject (http://forums.overclockers.co.uk/showthread.php?t=18227922). The summary is that the options are:
1. Don't risk it - buy a new one.
2. Solder it.
3. Repair it with car radiator repair kits - Liquid Metal or Leak-Fix.
I chose Leak-Fix mainly because it was in stock.
I immersed the rad in water and located the puncture (luckily there was only one despite me using the wrong screws in 6 holes). The location of the leak:
It took two coats of Leak-Fix to cover the hole to my satisfaction. It then looked like this:
I then immersed it in water again and blew into it, confirming the fix was water-tight. I then did the same thing but with a compressor (thanks to my father-in-law) and put 2 bar into it (that's equivalent to 20m of head pressure which is far more than my pump - 4m). With no leak detected I am pretty confident about the fix and I am now in the process of adding it behind the desk as a second rad.
More later...
First Braiding
As I was re-organising my radiators I decided it was time to try out braiding for the first time ever (on an area that would be behind the desk in case it didn't go too well) and braid the fan cables.
I had trouble getting the first fan connector off and needed my USB microscope to see why it was proving so difficult (cos my eyes aren't as good as the used to be). Turned out that the securing pin had curled up rather than being pushed down below the plastic. With the help of the microscope I was able to push them down and get the pins out. Each subsequent fan connector proved easier to remove than the last, so I feel a lot more confident about doing that than when I started.
I also didn't really know what sized braid I needed. I bought short lengths of a variety of sizes - 3mm, 6mm and 10mm. When I received them the 3mm looked tiny, although it does expand. I gave it a try on my first fan cable and it certainly expanded OK, but the result is that you can see the wire through the braid and the braid itself appears slightly weaker and has lost its shape.
The second one I used the 6mm, which was easy to thread with the fan cable and looks much better.
I also wanted the 3 fan cables on the rad to end up roughly the same length, so I wanted to bunch the cable nearest to the side of the RAD. I tried using the largest braid but it's such a short length it kept disintegrating. In the end I gave up and braided it with another bit of partly knackered 3 mm and then coiled it and sealed it along with the other two braided wires within a large piece of heatshrink. Perhaps you'd call it a bodge, but I think it looks OK.
As I was re-organising my radiators I decided it was time to try out braiding for the first time ever (on an area that would be behind the desk in case it didn't go too well) and braid the fan cables.
I had trouble getting the first fan connector off and needed my USB microscope to see why it was proving so difficult (cos my eyes aren't as good as the used to be). Turned out that the securing pin had curled up rather than being pushed down below the plastic. With the help of the microscope I was able to push them down and get the pins out. Each subsequent fan connector proved easier to remove than the last, so I feel a lot more confident about doing that than when I started.
I also didn't really know what sized braid I needed. I bought short lengths of a variety of sizes - 3mm, 6mm and 10mm. When I received them the 3mm looked tiny, although it does expand. I gave it a try on my first fan cable and it certainly expanded OK, but the result is that you can see the wire through the braid and the braid itself appears slightly weaker and has lost its shape.
The second one I used the 6mm, which was easy to thread with the fan cable and looks much better.
I also wanted the 3 fan cables on the rad to end up roughly the same length, so I wanted to bunch the cable nearest to the side of the RAD. I tried using the largest braid but it's such a short length it kept disintegrating. In the end I gave up and braided it with another bit of partly knackered 3 mm and then coiled it and sealed it along with the other two braided wires within a large piece of heatshrink. Perhaps you'd call it a bodge, but I think it looks OK.