Bought a HR-03 for my X1950 Pro

Matija · Post by **Matija** » Wed May 23, 2007 9:49 am

As the subject says, I bought a HR-03 for my X1950 Pro

It's still in the box. I'm gathering the courage to try to install it... It looks kind of horrifying with the backplate and the silicon spacers and everything. Has anyone mounted that beast on their X19yy? What do I have to pay special attention to? I've heard that some haven't used the silicon spacers, and the Thermalright website shows a picture of a SMD element that gets in the backplate's way (luckily, there's a washer extra in the box). Anything else? There are 7 RAM chips and 3 VRM chips on the card; there are 11 small heatsinks in the box. What do I use the extra one for?

My intention is to mount the thing in the primary position, so that the heatsink is on top of the card. Is that a good idea, or should I mount it so that it hangs below the card? If I mount it on the top, will there be room for a HR-05 on my P5B Deluxe? Edit: Or a Noctua NC-U6? I'll also probably buy a Noctua NH-U12F for the CPU, so it might get a bit crowded in the top area...

BTW, I have a SLK3000B, and I'll probably be closing the side vents. Too much dust goes in.

LuckyNV · Post by **LuckyNV** » Wed May 23, 2007 4:25 pm

thermalright included extra sinks because different cards use a different number of them. So having some left over is np.

If you use tall aftermarket heatsinks for the northbridge like the HR-05 or similar then rear mounted method will get in the way.

Rear mounted will also interfere with large CPU heatsinks like the Noctua, Ultra120, Tuniq Tower etc...

so if you choose rear mounted you are most certainnly not able to use tall custom chipset heatsinks and limited to certain orientations for large CPU heatsinks.

regal196 · Post by **regal196** » Wed May 23, 2007 6:31 pm

I have the HR-03 installed on my X1900GT. It works beautifully. I am running it passive. see my sig link to see my set up.

Installation was easier than the Ninja. It was very straight forward.

-Pennell

Matija · Post by **Matija** » Wed May 23, 2007 9:16 pm

Is there a possibility to remount the HR-03 the other way, or will the base be scratched by the cage thing holding it? I'm not sure if the combination of HR-03 + CPU heatsink + NB heatsink will fit until I have all three, and that's going to take a while.

regal196 · Post by **regal196** » Wed May 23, 2007 9:23 pm

The HR-03 can be mount wrap over way (the way my is installed). or it can be install non-wrap over.

The contact surface should be fine, the bracket only applies pressure to the corners of the block.

Matija · Post by **Matija** » Wed May 23, 2007 9:47 pm

Ah, thanks. I might finally try it today. I'll mount it on the top to get an idea of the space left in the CPU/NB area. It's really hard to tell.

Matija · Post by **Matija** » Sun May 27, 2007 5:15 am

Wow. This was one of the worst hardware installation experiences I have ever had.

First of all, when I removed the stock cooler, I was greeted by the GPU swimming in a sea of silver paste, and RAM chips that were covered with sticky thermal pad residue. It took me two bloody hours to clean that mess.

Then I found out that the backplate needs not one, but TWO washers to mount. As there were five in the box, I needed to find one extra... So I cut a rubber HDD grommet in half.

The tiny heatsinks that came in the box were in fact too big for the VRMs. I managed to squeeze them in, but I have no idea if they're seated properly.

The heatpipes have contact with two short RAM heatsinks. I've bent them slightly, but I'm afraid to bend them some more. I have no idea what kind of contact the HR-03 base has with the GPU. Probably not a great one. I opted for the "regular" mounting method, the HR-03 below the CPU, by the way.

Anyway, the temperatures... With the stock cooler, and according to ATT, they were around 60 for the GPU and 50 for ENV in idle. Now they are 10 degrees lower.

But (there's always a but)... The computer shut down while I was typing this for the first time, with ATT showing the 3D renderer in the background... So it's not stable. Aaaaargh!!!!!

I'll send this and run it again, with logging. I want to see where and why it reboots. Maybe I need to mount a fan on it, in which case I have to remount the HR-03 upside-down... And buy a fan.

Edit: The shutdown happened soon after the GPU temperature hit 75. That's weird, since I've seen it go over 80 under stock cooling.

What now?

Another edit: It looks like VRMs might be the problem. I just tried touching the two ones close to each other (see this image, and they are scorching hot in IDLE! If I kept my finger on them for more than two seconds, it'd get burned. I'm using the short heatsinks on them. The third VRM chip (the larger one) has a tall heatsink on it, and doesn't seem nearly as bad.

So, what the frack should I do about those two burning little cockroaches? Try to fit the taller heatsinks on them, and put the short ones on RAM? I have a feeling I'd have to end up buying thermal glue, as they wouldn't hold after removing and refitting them...

mr. poopyhead · Post by **mr. poopyhead** » Sun May 27, 2007 6:18 am

if the ramsinks are interfering with the way the cooler sits, i think you'd be better off not using them for those chips... i don't think they actually need extra cooling, but i'd ask around if i were you... there are some threads about it....

if your GPU is shutting down around 80, then something is wrong... i've been told that the R580+ (XT and XTX... close enough) can handle over 100 degrees...

Matija · Post by **Matija** » Sun May 27, 2007 6:24 am

Heh, you wrote this while I was editing

Please see the last edit. That might be the problem. What do you think?

I don't think it's the GPU, as it survived over 80 with stock cooling without shutting down, and now it's gone to hell with 75.

The HR-03 seems to be mounted okay, as it's warm to the touch.

Edit: Placed the larger heatsinks on the VRMs. No good. Put the card under load, got burned almost immediately by touching the little buggers. The PCB around them is also scorching hot. I wouldn't be surprised if their temperature was 120-130 degrees. This time, with an open case, the GPU reached 73 degrees before the machine shut down.

I think I need a fan (somewhere, somehow), or a really massive heatsink. The fun part is that Sapphire left a bunch of holes on that part of the PCB. Lo and behold, their X1950 Ultimate features a huge golden heatsink... Mounted using those holes.

I'm wondering whether mounting the HR-03 in reverse (hanging on the bottom of the card), along with a fan, would help things. It really needs the airflow.

Argh.

New edit: I took a 120mm fan and let it blow over the card, sideways. Didn't crash. Then I placed the fan on the HR-03. The VRMs did get very hot (slightly less than previously, though - even airflow over the back side of the PCB helps!), didn't crash. I think the best thing is to remount the HR-03 in the other direction and mount a fan on it, everything be damned. This setup simply cannot run with passive cooling.

.

z3r0 · Post by **z3r0** » Sun May 27, 2007 10:01 am

You could try removing the pci cover on the slot below your card, it might help with VRM cooling as a negative pressure case should end up using it as a intake of sorts. And that is also how the Accelero S1 that ships as a stock cooler on a x1950pro works, thats gotta count for something.

Matija · Post by **Matija** » Sun May 27, 2007 10:50 am

It's open since I got the card...

I've just remounted the damn thing, and it seems to have a bit better contact now. I've bent the heatpipes the other way (it's a wonder they still work

), and now they don't touch the RAM heatsinks. As long as there's a fan blowing on the card, everything is okay... But without the fan, it's just a matter of time until the VRMs overheat. After the remounting, it crashed in ATT's renderer after about 20 seconds, and the GPU barely hit 50 degrees.

I'm far from happy

I mean, I've read about the VRM issue, but damn, I didn't think those bastards were THAT hot. I'll try getting a big heatsink for them and perhaps mounting them with thermal adhesive. So much for the warranty...

Matija · Post by **Matija** » Mon May 28, 2007 10:37 am

Guys, help me. I don't know what to do... I need to attach some big heatsinks to the VRMs. I could use thermal adhesive, but then it's game over, they can't be removed (okay, maybe with the freezing or heating trick, but it's too risky). I'd like a removable solution.

Here's a couple of pictures of the X1950 PCB:

http://www.3dnews.ru/_imgdata/img/2006/10/17/29860.jpg
http://img.techpowerup.org/070504/untitled.jpg

As you can see, there are holes on the right side of the PCB, Sapphire uses them for its VRM heatsink on the Ultimate.

The tiny ceramic heatsinks on the first picture above aren't something I'd trust, based on the temperatures I've personally witnessed (and my left index finger can attest to it).

I have a northbridge heatsink sawed in half and the two parts filed and sandpapered, so the leftover contact surface is 9mm, which would fit on the VRMs. But I have no bloody idea how to mount the HS leftovers. The holes next to the bottom VRMs are roughly 30mm apart. The HS parts are 37mm wide. If I have to spend another 4 hours with the saw and the file, I'm going to go insane

Sooooo... I was thinking about tin-soldering some cut screws (fine threaded) on the HS bases and effectively making a bolt-through. Would that work? Have you ever tin-soldered something to aluminum? Is it even possible? Do I need special solder? What about oxidation? Am I grasping at straws? What do I do???!

WR304 · Post by **WR304** » Mon May 28, 2007 1:18 pm

If you're using thermal adhesive and want it to come apart in future dilute the adhesive with thermal paste. You'd have roughly 1 part adhesive to 3 parts paste.

The heatsink will still stick but the bond is weaker and can be broken without destroying the components.

mr. poopyhead · Post by **mr. poopyhead** » Mon May 28, 2007 9:44 pm

have you checked your VRM temps? i run mine up to 100 degrees on a hot day with no problems...

i'd really be more worried about the GPU not being able to handle over 80 degrees... contact their tech support and ask them how much the GPU can handle... if they tell you around 100 degrees, then you can tell them that yours isn't performing to that level and RMA it...

Matija · Post by **Matija** » Mon May 28, 2007 9:57 pm

Can't check them with anything but my finger (which barely stopped hurting yesterday morning). They are HOT. As in, really really hot.

The GPU (if the sensor is inside the GPU) could handle 80 degrees with stock cooling. The last time I tried to put the card under load, it shut down when the indicated GPU temperature was around 50 degrees (haven't had it logged). As I'm writing this, the card idles at 36 for GPU, 34 for ENV, whatever and whereever those temperatures are. The HR-03 is hanging below the card and being cooled passively.

I don't think the GPU is the problem...

I can try cutting the RAM heatsinks currently on the VRMs to make them fit better (currently they barely fit between the Pulse chips and the SMDs), but that's it as far as simple mods go.

mLy! · Post by **mLy!** » Tue May 29, 2007 12:34 am

you could try the VRM heatsink that comes with the accelero S1, they are beeing sold from arctic's ebay store.

Matija · Post by **Matija** » Tue May 29, 2007 12:44 am

The item would probably be confiscated by the police and I'd be brought in. Or I'd have to pay a billion $$$ in the end. There's a ton of horror stories I've heard about customs and taxes.

Maybe I should have a metal worker make me a copy of that thing...

I don't know. Everything I think of has SOME kind of a flaw that might bite me in the ass eventually.

mr. poopyhead · Post by **mr. poopyhead** » Tue May 29, 2007 7:35 am

Matija wrote:Can't check them with anything but my finger (which barely stopped hurting yesterday morning). They are HOT. As in, really really hot.

you can check the VRM temps in rivatuner... it should be one of the default readings. if your VRMs and GPU are crashing under 100 degrees, you may want to look into an RMA...

Matija · Post by **Matija** » Tue May 29, 2007 8:10 am

Ati Tray Tools lists two temperatures: GPU and ENV. I'm assuming that GPU is really the GPU temperature, and ENV is... Well, somewhere. But it's not the VRMs. I installed RivaTuner, and it lists GPU temperature and Ambient temperature, as expected.

Once again, they are both much lower with the HR-03 than with the stock cooler, and the card worked perfectly with stock cooling. I can't RMA something that works

I really do need a big chunk of metal on the VRMs, as tiny heatsinks aren't enough, but I don't have a single idea on how to mount something bigger.

mr. poopyhead · Post by **mr. poopyhead** » Tue May 29, 2007 9:09 pm

Matija wrote:Ati Tray Tools lists two temperatures: GPU and ENV. I'm assuming that GPU is really the GPU temperature, and ENV is... Well, somewhere. But it's not the VRMs. I installed RivaTuner, and it lists GPU temperature and Ambient temperature, as expected.

Once again, they are both much lower with the HR-03 than with the stock cooler, and the card worked perfectly with stock cooling. I can't RMA something that works I really do need a big chunk of metal on the VRMs, as tiny heatsinks aren't enough, but I don't have a single idea on how to mount something bigger.

go to the hardware monitor in rivatuner and click setup... there should be a list of values RT can report. voltage regulator temp should be one of them.

i have a ram sink on my VRM, it feels hot as hell. rivatuner reports that it's around 80-90 degrees. yours should be able to run safely around those values as well. again, check the VRM and GPU temps in rivatuner. if it consistently crashes around the same temperature, then overheating might be the problem. otherwise, i have no idea. maybe RAM overheating? doubt it though... i used to run GDDR3 on my old 6800 ultra without any cooling.

Matija · Post by **Matija** » Tue May 29, 2007 9:55 pm

Nope, no VRM temps available.

I did find out that GPU is indeed the GPU, and ENV is the temperature of the monitoring chip itself (located on the back of the card; it was the GPU that was previously heating it to obscene values with stock cooling, as with the rest of the nearby PCB).

I'll try fitting the heatsinks on the VRM better today.

Matija · Post by **Matija** » Thu May 31, 2007 9:35 am

Small update: I removed the heatsinks from the VRMs and cut off a rib from each, so they'd fit better.

I also uncovered the bottom vent on the SLK3000B side panel to improve airflow (and increase noise).

After 6 minutes of the 3D renderer in ATT, the GPU hit a maximum of 58 degrees. No shutdown. So I guess it's a bit more stable now. Touched the heatsinks and the PCB, almost got my finger burnt again...

I'm going to try and load the card for more than 6 minutes when I find the courage

But the VRM cooling is still woefully inadequate, a disaster waiting to happen.

I'm planning to add an intake vent to be on the safe side with HDDs (whose temperature I cannot read as they are in RAID). That should hopefully lower the VRM temps a bit further, as there'll be more airflow. But I still need to properly take care of this stupid problem.

Edit: The card survived almost 30 minutes of ATT, and the GPU eventually hit 60.88 degrees. Good - I can now officially call the card stable. I have to mod the case slightly on Saturday, so I'll update you on what happens when there's an intake vent to improve airflow, but I think I'll eventually have to resort to a proper VRM heatsink to be 100% safe.

Moogles · Post by **Moogles** » Fri Jun 01, 2007 11:46 am

When I assembled my computer (early January) I looked into getting an X1950Pro with a HR-03, but the VRM issue kept me from going through with it. I ended up with XFX's passively cooled 7950GT. Sorry your experience was bad, but it does validate my decision... so thank you for that.

I think passive cooling could be done so much more efficiently. The problem with these aftermarket solutions is that they have to fit on a wide range of different PCB designs, and it's impossible to cool the VRMs properly. Even though the price would go up tremendously, they should make copper blocks (i.e. the watercooling block EVGA uses on their "black pearl" 8800GTX) that are card specific and cool every vital part of a video card.

Matija · Post by **Matija** » Fri Jun 01, 2007 12:49 pm

I was also thinking about an Nvidia card due to lower power consumption and easier passive cooling. But since I refuse to play games without forced 16AA, and Nvidia's 7xxx method is truly awful, I decided to get the 1950, which I still think it's the best bang for the buck as far as speed and image quality are concerned.

Sadly, yes, it's plagued by horrible stock cooling (though the reference Ati design, full copper, might have been better) and the horror that are VRMs (which were dumped by card manufacturers lately). Powercolor did the right thing by including the VRM plate with the S1.

Matija · Post by **Matija** » Sat Jun 02, 2007 3:46 am

Well, so much for that.

I was happily playing a game, when all of a sudden, "klack-klack".

The VRM heatsinks fell off.

wutang · Post by **wutang** » Sun Jun 03, 2007 9:37 am

cooling the vrm on the x1950pro was a nitemare for me when i changed the stock cooling, good luck!

Matija · Post by **Matija** » Sun Jun 03, 2007 9:55 am

So how did you do it? Or is that why you have an 8800 now?

I'd have no problems thinking of a solution if I didn't want to keep the warranty valid. Thermal adhesive would have done wonders, but then, as I said, it's game over if something happens to the card.

My latest idea is to take one half of the sawed NB heatsink and drill holes through it. Since I have no screws that would fit through the PCB holes (they have a really small diameter), I was thinking of using a fishing line to hold it in place.

God, it's been a week, and I still haven't managed to do anything with the damn VRMs...

ryboto · Post by **ryboto** » Sun Jun 03, 2007 5:02 pm

i don't have anything on the vrms on my 1950 pro. Never had an issue, not even while folding with it.

Matija · Post by **Matija** » Sun Jun 03, 2007 9:28 pm

i don't have anything on the vrms on my 1950 pro. Never had an issue, not even while folding with it.

You lucky bastard

Slumberland · Post by **Slumberland** » Wed Jul 18, 2007 12:43 am

These may already be old ideas but I thought I'd mention them, just in case:

1) Arctic silver makes a two-compound thermal glue: you get two tubes, mix a bit of each on top of the chip, hold the heatsink in place, and it all fuses together.
I'm not suggesting you're unfamiliar with the method-- but Arctic Sliver's solution has *excellent* thermal properties compared with other, generic thermal fixatives.

2) Is it possible that the vido card tools you are using have disabled ATi's polling, and/or are not able to change the voltage?
I ask, because:

I have an x1950xt, and I had some weird auto-shutdown problems after loading AtiTool and, later ATi Tray Tool.
Under Windows XP x64, neither program could set the voltage on my card. One of the utilities needs ATi's background poller shut off, so the only way to get my 3D clocks back was to "overclock" in the utility..... but the voltages were stuck at 2D settings.
I had bizarre shutdown issues for several days, at temperatures that weren't high enough to be the cause. All of a sudden, I'd have a blank screen and the computer would reboot.
Uninstalling all tools and the ATI drivers, then re-installing the drivers finally got me back to square one. But until I had uninstalled EVERYTHING, the problem persisted. I assume that registry settings never got purged. (?)

In any case, you may want to try uninstalling all tools (and manually purge their folders & .ini profiles), uninstall your drivers, install the CCC, and *make sure the two ATI services are not disabled* (Start>Run; type "services.msc", without quotes, at the command prompt. There should be two services registered to ATI. They don't need to be running, as they will shut themselves off when not needed, but they need to be enabled for CCC to control your card. You can always disable them later and go back to ATi Tray Tool).

Bought a HR-03 for my X1950 Pro

Bought a HR-03 for my X1950 Pro

Just making sure...