Mach I cooling dilema

[killingdjef]

Yesterday I prepared my rig for some prometeia action however I have a few problems I can't seem to solve.

1) Whenever I turn on my prometeia rig, it doesnt wait for the prometeia to drop to -33, instead it just switches on. My CPU doesnt reach extreme temperatures by this but it makes me uncomfortable namelijk because of problem #2.

2) When I've got everything attached and turned on, the lowest temperature I get is 21 degrees celcius (while the display says -35 / -38 ). I've checked if I had used too much seal kit but the evaporator and the processor make contact (cooling paste gets smeared when I put a drop in the center). I've also checked if the evaporator actually cools, so I ran it detached from the board and after a while frost appeared on the evaporator.

The prometeia did cool to -20 on my previous mother board but after I had used too much sealing kit (it even covered my CPU a bit) temperatures have always be above sub-zero. I've removed most (if not all) excessive sealing kit and started over, with no result though. The readout of above sub-zero temperatures were on my previous as well as current motherboard.

Finally, I seem to have forgotten to add the small hard-plastic layer that goes on top of the evaporator, I doubt thats the problem but I thought it would be worth mentioning.

If anyone has a tip or idea what I can do please let me know.

Thank in advance

Setup:

Epox 9npa+ SLi

Operton 165

AOpen 700 watt PSU

2x Geforce 7800 GT

OCZ PC 4800 Platinum EE

 

[stocky]

Is the Prometeia properly installed and recognised in the control panel and control centre?

My first thought for the high temp would be contact, and second thought would be that the new motherboard doesn't read negative temps correctly - but you seem to have covered these

 

[Gray Mole]

Right...

Sounds like your contact 'should' be good, but have you checked that the evap and cpu surface is flat? That the head seems fully straight on the socket? Just in case the evap head was somehow damaged, or your hold down kit is twisted giving you the impression of good contact.

I only suggest it as the temps seem a little too cold on the Mach display. Could indicate partial contact.

Also, you can load the cpu and watch both the cpu core and display temps to get an idea of how good your contact is, and whether the mobo readings are changing at all.

Is your system stable at high overclocks and vcore? Does the temp from the mobo 'shoot up' when you go to high vcore or put it under load?

All good things to check and be sure of so you can be sure of the problem, and IF there actually IS a problem.

Otherwise I know of a shop that can modify your unit to 'Chilly1' specification, and you'd never have to worry about your temps again

*hint, it's EPUK*

Cheers

Gray

 

[killingdjef]

Right...

Sounds like your contact 'should' be good, but have you checked that the evap and cpu surface is flat? That the head seems fully straight on the socket? Just in case the evap head was somehow damaged, or your hold down kit is twisted giving you the impression of good contact.

I only suggest it as the temps seem a little too cold on the Mach display. Could indicate partial contact.

The temperature is always arround -33 to -40 when im idle on the display. When stressed it usually drops to about -25. I made sure I screwed every thing as tight as possible without breaking anything, the kit even came underneith it.

Also, you can load the cpu and watch both the cpu core and display temps to get an idea of how good your contact is, and whether the mobo readings are changing at all.

Is your system stable at high overclocks and vcore? Does the temp from the mobo 'shoot up' when you go to high vcore or put it under load?

I've overclocked from 1,4v to 1,5v and from 200 fsb to about 280. Idle temperature is about 24 degrees. I did that overclock on an air cooler. Whenever I fully stress the system (2x SuperPi2004) the temperature doesnt shoot up, it gradually goes up to about 32-33 degrees. Whereas when I did that on an air cooler it would go from 35 idle to mid 40's

All good things to check and be sure of so you can be sure of the problem, and IF there actually IS a problem.

Otherwise I know of a shop that can modify your unit to 'Chilly1' specification, and you'd never have to worry about your temps again

*hint, it's EPUK*

Cheers

Gray

Thanks for the reply

 

[killingdjef]

Is the Prometeia properly installed and recognised in the control panel and control centre?

My first thought for the high temp would be contact, and second thought would be that the new motherboard doesn't read negative temps correctly - but you seem to have covered these

Control panel and control centre? I take you're talking about windows? I wouldn't know where to look for that else and no, it doesnt (and has never done) add my prometeia to it.

thanks for the reply

 

[killingdjef]

update:

I've reassembled everything again and this time the evap. touches the processor for 100%. Still my temperature readings in the bios say 25C.

 

[Gray Mole]

I sure sounds like the temperature response is correct for the unit, indicating that your contact is fine.

Your scale is all wrong on the mobo sensor but the actaul response is about right from idle to load.

So to me it very much looks like the only problem is the mobo's sensor is incorrect.

I would look into whether a new bios is available for your mobo, perhaps there is a known fault with it's readings when frozen? Worth a Google to check.

Otherwise I would overclock as if it WERE frozen and see if your stable overclock is 'up there' where it should be.

Oh, and the Mach1 has no control center, I think that Stocky was cofusing that with Mach2.

Hope it all goes well, but considering how much the display is warming under load I think the only problem is the usual dodgy mobo sensor.

Gray

 

[killingdjef]

I've tried it and I can overclock a little bit further.. but not greatly. I can up the FSB with about 10 FSB before I'm getting errors again.

Just for the record. With my air cooler I got 35 degrees celcius idle and 45 under load

Prometeia gives met about 23 degrees idle and 42 under load.

I fear, taken these results, the board sensor is acting correctly...

 

[Gray Mole]

Looking at it from an OC point of view then yes, it does seem like it's just not cooling.

However when you look at your display being 35c idle and 25c load with not that much overclock it really seems to me that it's transferring the heat well.

I'm not really sure what to recommend as you probably don't have the load testing equipment that I have here (I do a little phase stuff myself)

The only thing I could offer really is that if you were willing to cover the shipping cost, I could lap the evap, check the mounting hardware for security, and set it up for a load test to make sure it's all ok for you.

That's if you live in the UK that is 8)

Anyways you can PM me if you'd like that kind of help with it, but other than that I'm not really sure what to recommend.

Just a thought though before it comes to that, are you increasing the Vcore to get overclocked or are you trying to do that with stock voltage?

Do you happen to know the stepping code? If it's a mildly coldbugged cpu it may not clock well at all on phase, though normally it just doesn't work at all as opposed to clocking poorly when frozen.

The way the phase cooler is behaving is as if it's getting good contact, and with the evap at -25 load, the cpu should really be more like 10c at most, maybe 0-5c if lighter load (30-40c warmer cpu temp than evap temp under load is normal for heavily oc'd A64 cpu's, and yours isn't THAT heavily oc'd yet)

The only thing that I could think of is if it's only a partial contact, then the temp you're at might make sense, and the OC, while better, wouldn't be that great as the cpu temp isn't greatly colder. Partial contact could still pull the evap temp up and not give fully cold temps on the cpu I suppose.

Cheers and good luck. you find anything else post it and I'll keep checking and post if I think of anything else.

Offer stands anyways if it comes to it.

Gray

Gray

 

[killingdjef]

..

Argh.. I wrote a very long reply only for it to be in veign because the forum thought I wasn't logged in.. Im going to retype it, pardon me if Im a bit short or too direct.

Looking at it from an OC point of view then yes, it does seem like it's just not cooling.

However when you look at your display being 35c idle and 25c load with not that much overclock it really seems to me that it's transferring the heat well.

I'm not really sure what to recommend as you probably don't have the load testing equipment that I have here (I do a little phase stuff myself)

The only thing I could offer really is that if you were willing to cover the shipping cost, I could lap the evap, check the mounting hardware for security, and set it up for a load test to make sure it's all ok for you.

That's if you live in the UK that is 8)

Thats very kind of you to offer, however I live in the netherlands and shipping from here to the UK and back would cost a fortune. Nontheless, thanks for the offer.

Anyways you can PM me if you'd like that kind of help with it, but other than that I'm not really sure what to recommend.

Just a thought though before it comes to that, are you increasing the Vcore to get overclocked or are you trying to do that with stock voltage?

Do you happen to know the stepping code? If it's a mildly coldbugged cpu it may not clock well at all on phase, though normally it just doesn't work at all as opposed to clocking poorly when frozen.

I initially started on air with my processor and overclocked it to 285fsb and about 1,5 vcore. After that I attached the prometeia to it. Naturally, as you know, that would give alot more playroom but untill now it didnt. Now before this I used the opteron processor on my died DFI Lanparty SLI-DR eXpert board and got to 3,06 GHz on about 1,575 - 1,6Vcore.

It was stable enough to run 3dmark05. Also on the DFI board it showed a -20 degrees temperature when on stock vcore. When I had it overclocked to 3,06GHz the idle temperature was 0 to -5 and under load it would shoot up to 30 degrees. Alot.. but in comparison with the current events its nothing. Im practically ~0.1v lower though the temperature is higher.

Then my DFI board died because of a combination of condensation and too much sealing kit.

[quote[

The way the phase cooler is behaving is as if it's getting good contact, and with the evap at -25 load, the cpu should really be more like 10c at most, maybe 0-5c if lighter load (30-40c warmer cpu temp than evap temp under load is normal for heavily oc'd A64 cpu's, and yours isn't THAT heavily oc'd yet)

The only thing that I could think of is if it's only a partial contact, then the temp you're at might make sense, and the OC, while better, wouldn't be that great as the cpu temp isn't greatly colder. Partial contact could still pull the evap temp up and not give fully cold temps on the cpu I suppose.

Cheers and good luck. you find anything else post it and I'll keep checking and post if I think of anything else.

Offer stands anyways if it comes to it.

Gray

Gray[/quote]

Thanks again for your time.

A small update for tonight:

I've reassemled my rig again and made the thinnest line of sealing kit. I also placed a drop of thermal paste on the evap. I attached everything screwed everything pretty tight. Started my machine and immediatly entered the bios (my rig won't wait till -33 degrees anymore, strange enough...) As temperature was dropping to sub-zero the temperature of my processor gradually lowered to again 24 degrees.

Turned everything off, waited half an hour and checked for the thermal paste drop. It was spread out neatly in a circle like shape, so it does make contact.

Then I thought my processor's sensor might be broken so I set everything up with a 3800+ X2 processor and again temperature would hover slightly below 24 degrees (about 21). I blame the (lower) stock vcore for this.

So in short, the evap shows frost when powering it without the mounting kit.

Ive used 2 different processors with the same temperature result

I've made sure and had proof that the evap made contact with the processor which only leaves over the board?

Well, on my DFI board it would show -20 degrees untill after I used too much sealing kit. Since then the temperature has always been arround 25 degrees.

Im almost feeling removing the entire mounting kit and hold the evap on my processor manually.. although that might not be such a great idea... 8)

 

[Thickbrit]

File away half the feet of the front plastic mounting bracket and check for obstructions under the bracket before refitting with a single bead of sealstring,just to totally eliminate any chance of poor contact.Also,I think just a little more ceramique wouldn't hurt,by the sound of it.You can only get away with the little drop if you have mirrior finishes on both cpu and evaporator face.If you do decide to file the feet (i would),then you may want to file the philips head screws by the same amount,to prevent them screwing throught the bracket.

 

[killingdjef]

File away half the feet of the front plastic mounting bracket and check for obstructions under the bracket before refitting with a single bead of sealstring,just to totally eliminate any chance of poor contact.Also,I think just a little more ceramique wouldn't hurt,by the sound of it.You can only get away with the little drop if you have mirrior finishes on both cpu and evaporator face.If you do decide to file the feet (i would),then you may want to file the philips head screws by the same amount,to prevent them screwing throught the bracket.

What do you exactly mean with "fileing"? I fail to see what you exactly mean.

I've used thermal paste only on the evaporator. When I screwed everything together and disassembled it again both the evaporator and the processor's IHS were covered with thermal paste.

 

[Gray Mole]

If the thermal paste is spread out so it's thin enough to see though (super thin after mounting) then it's got to have a reasonable contact.

However, what Thickie is saying is a good idea, and just means that it will contact even tighter than normal.

You're just getting rid of a little of the plastic that would hold the head further away from the cpu.

Also, if you can pull out the screws, and add a couple of the same size washers to each (making sure they have the same number) it can give a lot more pressure and better heat transfer.

Now that you've described it's behaviour on the DFI it very much sounds like partial contact or inadequate pressure. Adding a couple of washers can really help with that.

Filing the plate and screws has a similar effect, but if you mounting kit is getting old or bent then grinding away a little of the plastic tabs might be the only way to get full contact back.

OR

Buying a new kit, which would also likely give you back the full pressure you used to have. I'd still like to see the washers 'maxed out' or enough on each side that it's just a little tight to get the screws started in the mount which gives you the max amount of pressure on the cpu.

The very first thing I would check now is that there's not any gouges or anything in the copper on the evap. if there is, lap it flat.

Then get the screws out, and add a couple washers, enough to make it feel like you've got to push, just a little, to get them to thread into the mount.

Try that, and see if it gives you full pressure.

That really sounds like it's what is going on. At -35 idle, and such a low vcore, you should really see 0c.

The display temp isn't totally accurate but I find about 5c to 10c off at most on the mach1. That still doesn't explain the massive change in idle to load though.

A new headkit would very likely take care of things, and it would also give you the peace of mind knowing that it's secure. However if you still have an issue after it's replaced the money spent won't make you happy so I'd suggest trying what you can to improve this one then look at a new one later on.

If I can get a kit out with a head (I know I've got a couple kicking around here) I'll get pics of what Thickie's talking about.

Gray

 

[killingdjef]

Time for an update again!

It's been over a month since I last touched my Prometeia, I guess I got a little burnt out over it and also got a bit busy. 1,5 week ago I decided to check the cooling temperatures on a different motherboard and amazingly they -were- subzero. So the conclusion I can get out of this is that the temperature sensor on my EPOX 9NPA+ SLI does not register subzero temperatures. (Might be handy for other users?)

However.. considering my luck and the way lady Fortune smiles upon me, I have once more another problem. It happend a day or two ago. Whenever I turn on my prommie, the temperature display goes "crazy". It goes from about 27 degrees to 25 to 22 to 24.. halts a few seconds and then drops rapidly again, halts, drops, rises etc. After a minute or two it stabalises at a certain temperature.

I discovered it a day or two ago when I had turned my PC on for a couple of hours. When I got back I noticed the loud noise my prommie made. This is the same noise it makes when it's in start-up phase to -33 but it wouldnt stop.

Now, I shutted down my system, waited half an hour, restarted and thats when the "crazy" numbering started. However I was a bit shocked to see that it would idle at arround 0 degrees celcius (it was -8 when I got home). So I turned off the machine again, screwed open the prommy, turned it back on (bout 15 minutes later) and checked on the fans. They were both running.. but -this- time the idle temperature was arround 10 degrees celcius. Again, I shutted down my prommy, waited a while and turned it back on. This time, strangely enough the idle temperature was 16 degrees on the display.

So.. that could indicate a leak in the hose I guess but when I checked the Bios the CPU was 15 degrees celcius, while the display read 16 degrees. Isn't the CPU usally warmer than the display? That has always been the case in past though.

So I decided to do some stresstesting in windows... the temperature of my CPU was more or less the same than before.. . So what can I conclude out this? A lose wire? A circuit board that is gone bad? A virus? hehe.. I don't know.

There are 2 causes I can think of. Everytime I switched on my PC for the past days the POST screen wouldn't wait for the prommy to cool down to -33. So I always pressed Del, let it cool down to -33 and then start windows. Everytime I did this though, the temperature would exceed 40 degrees and I'd get an alarm beep and the system would turn itself off. The second time I would try it, the prometeia would get to -33. Maybe this has "stressed" something or burned a fuse thru?

Second cause I can think off.. I moved the Prometeia from my room to the living room and back. I did this with the greatest care and waited atleast 12 hours before moving it forth and another 12 hours before moving it back (so nothing would still be frozen nor would any possible condensation spill on my VGA cards)

Ahh a long story again..

Hope you or someone else knows what would be the best way to solve it.

Thanks in advance!

p.s. Right now my prometeia is considerbly making alot more noise because of it being "stuck" in the start-up phase. Now I could live with the noise for the moment but wouldn't this somehow create alot of stress on that black gas container (heh, forgot the name, non-native english person here )

EDIT/Update: Whenever I turn on my Prometeia now, it shows 53 degrees (!) celcius on the display, followed by an "ER4" notification and shuts down.

Update 2: I removed the PWR-ON plug from my board and started my machine with the EZ-buttons. This way the prometeia would run without shutting down and amazingly after a while my CPU temperature was at its original temperature (15 in the bios) while the display went from 53 to 36. I think a board circuit is busted or something since my computer also booted without waiting for the prometeia to drop to -33. So this means im not leaking any cooling fluid.. but the controller -is- busted.

Which leads me to the question, how can I fix this? Probably order a new circuitboard or something? I live in the Netherlands so shipping my prometeia to Denmark/UK is pretty much impossible.

 

[fatty]

If you speak to phil he will be able to help with a new control board first I would make sure all your board connections are tight ie press them into tight into the plugs use a small terminal screwdriver to do this ....

Also try running your mach not connected to your motherboard see what happens

 

[killingdjef]

If you speak to phil he will be able to help with a new control board first I would make sure all your board connections are tight ie press them into tight into the plugs use a small terminal screwdriver to do this ....

Also try running your mach not connected to your motherboard see what happens

Ok! i'll check tonight or tomorrow if everything is connected correctly.

How do I run the mach not connected to the motherboard though? I have a 5V plug that's connected to my PSU and there's a small PWR-ON connected to my motherboard which I had to remove in order to keep the computer running (otherwise it would shut itself down due to the high but faulty temperatures)

 

[killingdjef]

I screwed open the plates and couldn't find any disconnected plugs whatsoever.

I did discover that the temperature sensor is located on the copper radiator. The sensor is of the type "Airpax 67L070". Is it possible that the sensor is busted?

To answer this I need to know if my problems correlate I would guess. So, is it save to assume that the skipping of the pre-booting phase (where the cooler cools to -33 before starting) and the faulty display of the temperature with the error code "ER4" are corelated?

I used the prometeia on a different board as well (DFI RDX200) and on that board it would wait for the prometeia to cool down before booting.

In addition, starting my computer setup now gives me a 75 degrees celcius on display. While after forcefully keeping it running it is down to 22 degrees celcius. (Real temperature is sub-zero)

 

[fatty]

try for a new controller board if that doesn't help then it could be the head temp sensor I dont think it will be the one on the condensor