Watercooling VRs – Nick Poirier
This is my third article for Overclockers.com and it is really an extension of my second article. The second article discussed the installation of a custom water block on top of my motherboard’s CPU voltage regulators. Apparently that article has sparked a lot of interest among the readers of Overclockers.com as I have been receiving a lot of emails from people regarding it.
However, several people let me know that I should have actually tried to cool the bottom of the voltage regulators – and not the top as I had done. This is because it is really the bottom of the VRs which is their hottest side and that it is the bottom of the VRs which is supposed to be cooled. And they’re right!
Consider this example: Anyone who has seen the inside of a PC power supply has seen the large aluminum heatsinks in there. And if you look closely, you will see the heatsinks are attached not to the top side of the VRs but rather to their bottom side. Thus cooling the bottom of the VRs is the correct (and more useful) method.
So, in this third article I will discuss how I set out to cool the correct side of the VRs…by making another VR waterblock and installing on the back-side of the motherboard.The VRs on my motherboard are surface-mount components…thus they do not have legs which go through the motherboardWith that in mind, I took a look at the back of the motherboard to see if there was a flat area in which a water block could be installedAnd luckily the area in question is mostly open and accessibleThere is a strip of open space which is almost 1″ wide right below where the VRs are located on the top-sideThere are many obvious solder points to avoid where the legs of other components stick through the motherboardHowever, there are also many small “bumps” of solder which are all around the areaThese “bumps” of solder stick up off the surface a little but can barely be felt with you fingersNevertheless these bumps had to be considered during the planning stageAs a test to see if cooling the back of the motherboard was even worth doing, I needed to cut an access hole in the motherboard trayThe intended target: my Lian Li PC70’s motherboard trayI cut the hole using my Dremel and those expensive fiberglass reinforced cutting discsI didn’t want to cut away too much of the metal as that would weaken the rigidity and strength of the trayNow to measure how hot the back of the motherboard was gettingI routed a temp probe back behind the motherboard and taped it in place right in the middle of that open areaAnd this is the temperature I recorded:LEFT = coolant temp…RIGHT = back of mobo~75’C !!!At the time I recorded those temps, my CPU was running ~2680 MHz @ ~2.0v and running Prime95 for about 5 minutesThe idle temp for the back of the motherboard was about 60-62’CBut remember…I already have a block installed on the top of the VRsAnd at the time I recorded those temps, I was running my PC with the side panels OFFNot to mention that the hole which I cut in the mobo tray was probably allowing some of the heat to radiate out more easilySo at this point I was convinced cooling the back of the motherboard would help
Nick PoirierTo make the back-side block, I used a somewhat similar method which I used to make my top-side blockFor this one I used some copper flat-bar, 25mm wide and 3mm thick (approx. 1″ and 1/8″)Some thin copper sheet, 1mm thick (approx. 1/16″)And some copper pipe, 10mm OD and 8mm ID (approx. 3/8″ OD, 1/4″ ID)The flat-bar will be used for the base, the thin sheet will be bent to form the water chamber and the pipe will be used for the inlet/outlet connectorsNext I drew up some quick plans to make the water chamberI would end up changing these plans slightly to make the block fit and to make it easier to buildHere is the piece of the copper sheet I cut to make the water chamberI used a hacksaw and it was really easy to cutNext I marked out the cut and fold linesHere is the copper sheet cut to size with the inlet/outlet holes drilledI used my Dremel and those fiberglass reinforced cutting discs to make the cuts this timeAgain the metal cut very, very easilyHere is the base of the block cut to length (75mm)I had to make some notches in the side and trim the lower right corner to avoid contact with some solder points on the underside of the motherboardIt was these notches which forced me to adjust my original dimensions for the water chamber (the water chamber had to be reduced from 15mm to 12mm wide)I also planned on bending out the bottom 2mm of the edges of the water chamber so that it would have a flat surface to meet the base of the blockHowever, I found it next to impossible for me to bend the thin copper sheet in this way by handI would have needed a nice bending machine to get the 2mm lip I wanted at the bottomHere you can see the inside of the water chamber after folding up the thin sheet copperFolding the copper was somewhat difficult – time consuming mostlyI used my hands and some regular pliers to do the bendingThe inlet/outlet pipes were cut to no particular lengthThe ends of the pipes which get attached to the block were cut at an angle to correspond to the shape of the blockAlso, I drilled lots of “divots” or “starter” holes in the base where the water will flowThe block is basically open on the inside with little flow restriction, but it was my hope that the holes will help increase the surface area in contact with the coolant and to maybe disrupt the flow a little
CONTINUED page 3…
Nick PoirierHere is the block ready to be solderedAnd here is the block all solderedI tested the block for leaks and there were noneI spent a little time cleaning up the blockI tried to file away most of the excess solder but I really didn’t put a great deal of effort into it since I will never see the blockI used some 400 grit sandpaper and some metal polish/cleaner to finish upThe block is now ready for installationThis is where the block is goingMy plan is to use thermal tape to act as the separating layer between the copper block and electrical traces on the motherboardI didn’t plan on the using thermal tape alone to secure the blockSo I had to design some kind of mechanism to hold the block in placeI made a hold-down bracket out of some aluminum square tubing I had lying aroundDue to the tight space limits behind the motherboard, I had to cut the aluminum tube in half lengthwise and file down the edges a littleI also had to worry about the heads of the mounting screws possibly touching some solder points on the back of the motherboardThe original hole I made in the motherboard tray had to be increased somewhat to allow for the hoses to get connected more easilyAnd here is something I had not considered at all during the planning/construction of the blockThe side panel of my Lian Li PC70 has a handle which is “punched” into the metal surfaceThis area, of course, is right where the block is getting installedThis handle reduces the depth of the overall space I have to work with between the back of the mobo and the inside surface of the installed side panelThankfully, just by pure luck, the handle ends up right between the two mounting screws (where there is some free space)I remember saying some choice words when I realized I forgot about the handle on the side panel 🙂
CONTINUED page 4…
Nick PoirierThe hold-down bracket I made works greatIt can apply plenty of down-force on the block and it doesn’t flex the motherboard tray out of shape (even in its cut-up state)I didn’t tighten the nuts too much…just as much as I could with my fingersI was worried about those small solder “bumps” I mentioned before working their way through the thermal tape if I put too much pressure on the blockHere are some pictures of the block installed with the bracket holding it in place:This picture is from the side and shows the tight space constraints behind the motherboardMy camera was half inside the case looking down the length of the caseAt this point, I knew that the block fit onto the motherboard and it was ready to be fully installedSo that meant I needed to apply the thermal tape to the baseThe thermal tape I used came from some Thermaltake heatsinks I came across during my job as a PC repair tech
CONTINUED page 5…
Nick PoirierHere the block is installed on the motherboard with the thermal tape in placeThe thermal tape actually holds the block in place VERY well – I am quite surprised!After seeing how well that thermal tape worked I am not worried in the slightest that the block will shift, slide over, and make contact with any of the surrounding solder points sticking up around itHere the hoses are hooked up to the blockI used plain regular nylon hoseThe top of the block is where the coolant comes in and the bottom is where it exitsIn this particular circuit, the coolant comes up from the pump, goes thru the top-side block, loops back and down thru the new back-side block, then returns back to the reservoir (ie no rad in this loop)
And how well does it work, you ask?Here is a temp readout with the back-block installed (again…LEFT = coolant temp, RIGHT = back of mobo)That picture was taken with my CPU running @ ~2600 MHz, 1.95v, 100% load running Prime95Remember that the motherboard temp before my new back-side block was ~75’C during 100% load times!!
Some ResultsThe biggest change I have noticed since the addition of the back-side block is that my available CPU voltage has been increased and that the CPU voltage has stabilizedMy NF7-S Rev.2 motherboard allows you to select a CPU voltage as high as 2.3v but anything over 2.125v would make the system hang during POSTI was never able to use a voltage above 2.125v (no matter what my CPU speed, multiplier, or FSB was)Now with the increased cooling, I can use a CPU voltage as high as 2.22v (2.25v or higher still locks up the system)Also, the CPU voltage does not vary as much anymoreBefore you could see the CPU voltage changing by as much as 0.03v regularly by watching MBMNow the CPU voltage only changes by 0.01v occasionallyUnfortunately this has not helped my CPU overclock any further really (maybe another 50 MHz)I think I have simply maxxed out my 1700+ at ~2700 MHz with a voltage of ~2.0vOr my 431w Enermax P/S has hit it’s limitOr my cooling system has hit its limit for cooling a CPU at this MHz/voltageThe higher CPU voltage would be useful if I was super-cooling the CPU I assumeI am just running plain water-cooling here…pump, blocks, radsInterestingly, my reported CPU temperature did not get affected by reducing the excess heat around the CPU socketI thought for sure I would see the CPU temp drop a few degrees, but it hasn’t happenedMBM still reports the same CPU temps I am used to seeingHowever, if the reported CPU temp dropped, say 10’C, I realize that my CPU would probably still be running as hot as it always wasI was hoping for a few ‘C temp drop thoughMaybe the latest BIOS for the NF7 reports the internal CPU diode temp instead of the in-socket probe temp??My FSB still maxes out at ~223 MHz, but I think that’s because I am just running the chipset at 1.7v (the max in the BIOS)
ConclusionsMaking the back-side block was a fun experience – I am glad that I did it!With my particular water-cooling setup, it was not hard to add another block to the systemHowever, it was a lot of work to plan, build and install the blockI still felt it was certainly something worth trying!And there’s always the bragging rights 🙂However, I am sure that I have not fully discovered all the potential pro’s and con’s of using a back-side blockThere may be problems or issues that have yet to surface from thisSo at this time I cannot recommend that everyone try this in their systemsInstalling a copper block on the surface of the motherboard might not work for everyoneThe obvious benefits which I have discovered so far is that the max available CPU voltage went up and that the CPU voltage did stabilize a lotIf your system suffers from a low and/or erratic CPU voltage, then maybe this is something worth looking intoAlso the excess heat around the CPU has been significantly reducedAnd the cost of the materials to make the block was very low (maybe $5 MAX) which is always good
Thanks for reading! 🙂