New Maze-4 AMD Unseating CPU

[Warmart]

Having a problem with a Maze-4 install on my Epox 8RDA+. I'm using the 4 motherboard mounting holes and the nylon hardware Danger Den sent along. I used small washers and nuts on each side of the board, then carefully put the block on, then 4 more small washers, then 4 springs, 4 thick washers, and finally thumb screws.

I'm getting an immediate FF post code on the motherboard led's with the Maze 4 installed. If I remove it, and put on one of my regular heatsink/fans that clip to the socket, I boot fine.

Wondered at first if my board was flexing too much from the weight of the block but then I noticed that after removing the Maze 4 my cpu is becoming slightly unseated in the socket! Well that's not a good thing obviously. Am getting a perfect imprint of Artic Silver on the block but it's not gonna work well if the cpu is not making solid contact with the socket pins.

I had cut a rectangle out of my motherboard tray to make it easier to install the block and am wondering now if that was a good idea. Maybe that's hurt the rigidity of my board causing the slot to flex just enough to remove the cam pressure? Thought about excessive hose tension (1/2 ID Clearflex) but my install doesn't look as pressure inducing as some others I've seen.

I guess I should drain the system and try a quick boot with the block on and no hoses. Or an insulated piece of metal on the back of the socket with holes drilled for the 4 mounting bolts maybe? Or is the nylon hardware not the best way to fly?

Or is my socket weak on my 8RDA+?

Kind of a bad initial indoctrination to water cooling for me :( Well I have no leaks yet, knock on wood.

Thanks

[Voodoo Rufus]

Hmm, how about backing off the pressure on the CPU slightly? If there's too much pressure, I guess that problem could happen, never heard of that before.

I wouldn't boot without water in the system EVER. Good way to destroy the CPU in short order.

[mata2974]

I know exactly what you a talking about, I had the same problem with my old Maze 4, siren or no boot.

Yes its a good idea to cut out that rectangle on the back of case, otherwise you gonna have a lot of trouble fastening those nuts, sometimes you need to hold them from the bottom whislt fastening the front. I dont think there is any pressure lost due to the cut.
What I finnally figured out was to really tighten those thumb screws tight, otherwise it will not post or siren on you. Just be careful that the pressure is even and you have a super thin layer of AS5 or whatever grease you use. Just be patient

[Warmart]

I had eventually tried them pretty tight, but of course by then the cpu was probably already popped out, so I was wasting my time.

I think I'm going to need to determine exactly what is causing the cpu to pop ever-so-slightly loose. So I'm probably gonna drain and put just the block on (no hoses). Then remove it, and see if the cpu remained seated. Then I guess I'd know it was due to hose pressure. Still don't think I have them installed badly though.

Thanks for the ideas.

[Edward2]

I am not familiar with water cooling on an AMD system, but I recently installed a WC setup on my P4 system and initially had problems booting. It turned out that my waterblock/mounting plate was grounding something on my mobo. I had to use some electrical tape to insulate the waterblock/mounting plate from the components on the mobo. Once I did this, it booted up just fine.

Just thought I would mention it. You might take a close look at your setup, and make sure that your WC setup is not touching any of the components on your mobo.

Also, the CPU may not be loose in the socket while the waterblock is installed (like you have described), since there is pressure from the mounting bolts holding the waterblock against the CPU against the socket. The CPU may be getting loose while you remove the waterblock (the thermal grease under pressure kinda acts like a suction cup holding the CPU to the heatsink). I have seen CPU's get stuck to the backside of the heatsink and be completely removed from the socket when the heatsink is removed.

[Warmart]

Originally posted by Edward2
Also, the CPU may not be loose in the socket while the waterblock is installed <snip> The CPU may be getting loose while you remove the waterblock


Well that got me thinking Ed2. I bought a telescoping inspection mirror on the way home from work and then peeked around my socket the best I could, and it sure didn't look as if the cpu was loose at all.

I also bought some 3" metal threaded rods to replace the plastic ones DD sent me. I was not at all happy with their extreme flexibility. Well actually screwed up and didn't read each label from the bin at Lowes and ended up only being able to use 3 of the metal, and 1 of the DD nylon rods as one was 6-32 and I needed 8-32 threads. Stupid me hehe. Well after remounting, the block is much more stable. After putting the board back in I just hooked up the power switch lead from case to moboard and I posted fine. w00t I think.

Then I power down and hook up remaining case leads to moboard, try to boot up, no dice! WTF I think? Hmm. Unplug reset switch header, boot fine. Bingo. Pop off front casel bezel and find my weatherstripping was pushing on my reset switch, lmfao. I felt pretty damn stupid to say the least. Well everything works now so I don't care. Just two days of misery for a laughable (now anyway) mistake.

The first day I had this FF boot error I had went and found the cheapest 1/2 decent cpu I could find locally (a 2500+ Barton core), and that was the chip I left in since I knew it worked when I had pulled the board and installed a normal hs/fan (of course the case reset sw wasn't hooked up then). It's a stock 166/333 chip and I got it to 220x11 @ 1.8v last night in about an hour of testing (stable prime95 torture). Need 1.8vcore for that. Tried 225 the whole way to 1.95vcore but no joy. Can't raise vdimm any higher then 2.9v without a mod. Course I could adjust my already modded NF2 chipset (vdd) voltage tonight perhaps. Since this is one of the new multi locked Bartons it's kinda hard to tell if it's RAM or CPU limitations holding me back. My guess is either RAM or Northbridge, since it took small increases in vcore for about every 10Mhz FSB, and temps didn't rise much. Not usual for a cpu to just stop like a rock and still be cool (and not respond to plenty of vcore).

Though I can always put my unlocked 2700+ back in and tell for sure. Can't really complain with 220/440 DDR FSB and I guess I'm getting some benefit from the 512M cache on this 2500 Barton.

Sorry for the book. I'm very happy now and wanted to let everyone know what the REAL problem was, it was ME. :)