Concern about x58 VREG temps w/ WC

[Buckeyes1995]

Hey all.. just got my initial system up (still lots of work to do).. running a 930 on an EVGA 3X SLI at stock speeds.. CPU temps are about 17C deg .non-load.. assume is pretty good.. but VREG, non load is 50C already.

I'm using the Bitspower X58 MOSFET block... question.. are these normal temps? Should I try to remount the block? I'm going to run some LINX runs today at stock speeds to see how it does before messing around with overclocking.

Thanks!

PS. Wasn't sure if it was better to post this in the LC section or the Intel section above.

[clone38]

Sounds like the blocks not mounted correct as when i had my 3 way with bitspower blocks i was running about 30c odd take it off and reseat it or you may not have screwed it down tight.

[Buckeyes1995]

Hrmm.. ok...

Do you recall.. did you use the 2 1.1mm spacers between the bottom of the BP block and the motherboard? I did, and I notice that there's a little 'wiggle' on the block.. could very well be a contact issue.

Also.. I mounted mine like the stock tray was mounted.. I did NOT put thermal material on the row of square capacitors (i assume they're capacitors) that the block cutout hovers over.. did you? Am I supposed to?

[Buckeyes1995]

I just don't get it.. I reseated the block (twice).. final attempt i removed all the standoffs and found small screws.. the block is literally against the mosfet chips with a thermal pad in between.. still hovering at 50C idle.

CPU & NB both around 20C idle.. these are all at stock settings, no overclocking yet.. I can't find anywhere on the net where anyone has got near 30C VREG... I'm really confused.

PS. Is the thermal pad supposed to "melt' onto the chipset? I noticed that when I went to reseat the board it was just as good as new.

[quattro_]

If you can't see any kind of imprint on the mb side of the thermal pad then no you don't have any pressure on it .

[gmat]

50C idle on water is normal if your settings are like 1210Hz for CPU VTT regulators, and everything is overclocked.
And 17C-20C is very cold. It's actually below ambient... You're using chilled water ? What do you use to monitor these temperatures ?

[Buckeyes1995]

50C idle on water is normal if your settings are like 1210Hz for CPU VTT regulators, and everything is overclocked.
And 17C-20C is very cold. It's actually below ambient... You're using chilled water ? What do you use to monitor these temperatures ?

Yeah that's what the EVGA ELEET util was showing for the 'cpu temp'..

Here's the screen shots of the temps & info... Im a tad new to OCing so Im sure i have a lot more to improve upon... Wasn't sure what you meant by CPU VTT.. I have a voltage setting for that.. but not sure how to set the speed?

Anyways..

Windows clean boot... IDLE:

http://www.herro2u.com/haveblue/base.jpg

http://www.herro2u.com/haveblue/basemon.jpg

http://www.herro2u.com/haveblue/basevolts.jpg


And now after 20 Linx iterations...

http://www.herro2u.com/haveblue/vreglinx.jpg


Does this seem reasonable? Should I take the MOSFET block off again and try to reseat?

[mwlhrh]

hey buckeyes;

we have slightly different equipment, but I think you may need to remount that block. My VR and NB even at load usually are below 40C at a 4.5 OC with the VTT and pwm turned up. I think you are right about poor contact...the thermal pads won't melt per se, but will have the impression of the thing they have pressed on. There shouldn't be any wiggle at all...it should be a "snug" tight.

by the way, nice OC on that 930...should be able to push it quite a bit farther given those temps and your Vcore.

rob

[noob eater4726]

I was getting 50 C idle as well with the same setup - EVGA X58 SLi + Bitspower block - even what the VRegs was in its own loop with a 240mm rad. :shrug:

[gmat]

Must be the bitspower block making poor contact.
I have almost the same motherboard (X58 Classified), but my VREG idle is in the 40's, that's the full load which is at 55C, and i upped the "VREG frequency" (without that setting full load is about 47C), since i'm running high VTT and high RAM clock. I use EK full cover block though.

If the block wiggles it definitely doesnt make good contact. Are you using a back plate or spacers to prevent motherboard from flexing ? Could be a small amount of flex that prevents the middle regulators from making proper contact. And yes the thermal pad isnt supposed to melt away, but you should definitely see the imprint of regulators in it.

[clone38]

Eleet is very buggy tbh mate what does it say in the bios?

[H20Cooled]

Eleet is very buggy tbh mate what does it say in the bios?

+1

Check the bios as many have reported issues with the temps outside the bios.

[Buckeyes1995]

Yeah I'm kind of at my wits end here.

I re-seated the board, AGAIN.. I did notice that the pad had no impressions in it.. so bad contact.. this time, I removed all spacers & laid the pad directly on the MOSFET chips (before I stuck it on the block then seated the block).. Right now, IDLE BIOS temps have it hovering around 50C.. I ran a 10 Linx iteration and it hit low 60s.. so, some improvement but certainly not the sub-40C temps you guys are experiencing.

I thought my problem before was due to the spacers.. so I removed those.. There is no backplate that comes with the BP MOSFET x58 SLI board.. just the block, two screws, 2 1.1mm spacers, and 2 2mm spacers... and I seem to have lost one of the tiny 1.1mm spacers..

This is starting to drive me nuts

[Buckeyes1995]

I suppose the good news is that after 80 runs of Linx.. the ELEET reported temp is down to 66.. from 77 last night.. so re-seating helped.. just certainly not to the level you guys have reported.

[gmat]

Yeah i'm using spacers on the EK block tough, so i guess that was your problem, using the right spacers / tightening might improve things. You should really try & check you have a proper imprint on the thermal pad (around center especially, if the imprints are shallow there it means the PCB is being warped, and that you need less pressure on the screws).

[mwlhrh]

buckeyes;

other considerations would be slightly shorter screws or washers on the other side of the MB, or buy some thicker thermal pad. If the BP block came with 1 mm thick pad, buy some 1.5mm EK thermal pad...worth a try.

rob

[Vinas]

Stack the thermal pad. It works.

[Buckeyes1995]

Yeah i'm using spacers on the EK block tough, so i guess that was your problem, using the right spacers / tightening might improve things. You should really try & check you have a proper imprint on the thermal pad (around center especially, if the imprints are shallow there it means the PCB is being warped, and that you need less pressure on the screws).

Hrmm.. maybe it is a warping issue.

Are you saying I could be warping the area around the MOSFET by the screws holding the MOSFET in? Or perhaps I'm causing warping by the other blocks (CPU & NB/SB) being too tight?

[gmat]

It's a warping issue if you notice the center of the thermal pad has shallower or no marks left by the VRM's, compared to the edge. First suspect is the waterblock above them being tightened too much. EK gives spacers so we cant warp the board but if you dont use any spacer, you have to hand tighten the screws properly up so that the board doesnt flex. It wont require much force to over tighten them...

[Waterlogged]

From the pic in your worklog, it almost looks like the block is mostly resting on the chokes. I think stacking the thermal pad should fix that.

[Buckeyes1995]

It's a warping issue if you notice the center of the thermal pad has shallower or no marks left by the VRM's, compared to the edge. First suspect is the waterblock above them being tightened too much. EK gives spacers so we cant warp the board but if you dont use any spacer, you have to hand tighten the screws properly up so that the board doesnt flex. It wont require much force to over tighten them...

On my first attempt I used both sets of spacers included with the Bitspower (the 2mm spacers I used UNDER the board between the screw had and the board.. and then the 1.1mm spacer (which I assume is the height as the MOSFET chips) on top.. that resulted in the 'wiggly' block.

I then removed all spacers and got smaller screws.. the block is now secure (ie.. it doesn't wiggle) but I suspect when I remove it tonight I'll find there's still not much of an indent on the pad.


From the pic in your worklog, it almost looks like the block is mostly resting on the chokes. I think stacking the thermal pad should fix that.

yeah I think that's my next thing to do.. I'll stack the old pad that came with the 3X SLI and the Bitspower pad and see how that does.

Thanks all for the advice.

[Buckeyes1995]

Ugh..

I have no clue what is wrong. Well I know, but I don't know how to fix it.

Re-seated the heatsink..again..

This time.. I used 2 thermal pads.. and the correct screw/spacers that came with the BP..

10mm Screw --> 2mm standoff -> motherboard -> 1.1mm standoff -> 2 thermal pads (the one that came with the stock heatsink and the one that came with the BP block) -> BP Block

I have literally made my temps worse.. now LINX'ing it I hit 72.

GRRRR

[Vinas]

My next question would be, what kind of temps did you get with stock?

[Buckeyes1995]

hey buckeyes;

we have slightly different equipment, but I think you may need to remount that block. My VR and NB even at load usually are below 40C at a 4.5 OC with the VTT and pwm turned up. I think you are right about poor contact...the thermal pads won't melt per se, but will have the impression of the thing they have pressed on. There shouldn't be any wiggle at all...it should be a "snug" tight.

by the way, nice OC on that 930...should be able to push it quite a bit farther given those temps and your Vcore.

rob


+1

Check the bios as many have reported issues with the temps outside the bios.


Must be the bitspower block making poor contact.
I have almost the same motherboard (X58 Classified), but my VREG idle is in the 40's, that's the full load which is at 55C, and i upped the "VREG frequency" (without that setting full load is about 47C), since i'm running high VTT and high RAM clock. I use EK full cover block though.

If the block wiggles it definitely doesnt make good contact. Are you using a back plate or spacers to prevent motherboard from flexing ? Could be a small amount of flex that prevents the middle regulators from making proper contact. And yes the thermal pad isnt supposed to melt away, but you should definitely see the imprint of regulators in it.


My next question would be, what kind of temps did you get with stock?

I never ran the board with the stock cooler.. but I'll try and run it tonight at stock settings.

[mwlhrh]

buckeyes;

Any chance there is a protective transparent plastic sheet on the bottom of the mosfet block (very unlikely, I know) or something in the block obstructing or diverting flow from the 'cold plate' of the block? I don't know your loop order, but if the CPU, which has good temps, is in the MB loop, it implies either poor contact/mount or poor heat transfer of the block or something borked with the voltage regulators.

one other thing...and I say this in complete deference to Waterlogged...I read a test between stacked thermal pads vs thicker thermal pad and the thicker pad performed quite abit better. I can't remember where I read it, and, in all truth, I may have even misinterpreted the results...Waterlogged is my superior in all things computer. Still, before something drastic like an RMA of the board or the block, might be worth a try of a thicker thermal pad.

rob

[Buckeyes1995]

buckeyes;

Any chance there is a protective transparent plastic sheet on the bottom of the mosfet block (very unlikely, I know) or something in the block obstructing or diverting flow from the 'cold plate' of the block? I don't know your loop order, but if the CPU, which has good temps, is in the MB loop, it implies either poor contact/mount or poor heat transfer of the block or something borked with the voltage regulators.

one other thing...and I say this in complete deference to Waterlogged...I read a test between stacked thermal pads vs thicker thermal pad and the thicker pad performed quite abit better. I can't remember where I read it, and, in all truth, I may have even misinterpreted the results...Waterlogged is my superior in all things computer. Still, before something drastic like an RMA of the board or the block, might be worth a try of a thicker thermal pad.

rob

Def. good advice.. i think it's a contact issue as I'm not getting impressions in the thermal pad (I haven't taken it off since my last post, but I suspect I know what I'll find)...

Also.. most of you replying here and over at EvGA are running Classifieds.. could it be a 3X SLI non-classified issue?

And yeah.. no plastic on the MOSFETS :)

[mwlhrh]

hey buckeyes

i just was looking at your build log...it looks like you've applied thermal pad to the VR pieces closest to the cpu socket, but did you apply thermal padding to the blocks farther away? I know when I placed my EK full WB, there were two long thermal pads that ran over the tops of the two rows of VRs:

http://i627.photobucket.com/albums/tt359/mwlhrh/IMG_1429.jpg?t=1272333647

probably a silly suggestion, but just a thought.

rob

[Buckeyes1995]

hey buckeyes

i just was looking at your build log...it looks like you've applied thermal pad to the VR pieces closest to the cpu socket, but did you apply thermal padding to the blocks farther away? I know when I placed my EK full WB, there were two long thermal pads that ran over the tops of the two rows of VRs:


probably a silly suggestion, but just a thought.

rob

On the 3X SLI there's only one row of MOSFETS.. so just have the one BP block

[mwlhrh]

oh, ok...sorry about the silly suggestion...I've had no experience with the x58 SLI board.

As an experiment, I just ran some full memory LinX with my 4.5 overclock; I let the room warm up to 25C and turned the fans to 900 rpm...the highest VReg temp on eleet was 52C load, 36-40 idle. I don't know where your OC is now and how much you've adjusted the bios VTTs/pwm, but looking at your build log, we have about the same cooling capacity (2x 3.120), same pumps, similar fans, so I would expect your VR temps to be similar, or maybe a little better if this review is to be believed:
http://www.overclock.net/water-cooling/564301-evga-x58-classified-waterblocks-roundup.html...I know we have different EVGA boards, but they shouldn't differ in temp that much.

i'll keep thinking about it...

rob

[Buckeyes1995]

oh, ok...sorry about the silly suggestion...I've had no experience with the x58 SLI board.

As an experiment, I just ran some full memory LinX with my 4.5 overclock; I let the room warm up to 25C and turned the fans to 900 rpm...the highest VReg temp on eleet was 52C load, 36-40 idle. I don't know where your OC is now and how much you've adjusted the bios VTTs/pwm, but looking at your build log, we have about the same cooling capacity (2x 3.120), same pumps, similar fans, so I would expect your VR temps to be similar, or maybe a little better if this review is to be believed:
http://www.overclock.net/water-cooling/564301-evga-x58-classified-waterblocks-roundup.html...I know we have different EVGA boards, but they shouldn't differ in temp that much.

i'll keep thinking about it...

rob

Nah no silly suggestions.. appreciate your support :)

I have a 4.2 OC now.. stable as a rock.. 800 mhz on the PWM

[mwlhrh]

buckeyes

any chance the company who sent you the blocks accidently sent you the mosfet block for the classified BP-WBPEI58DP2 instead of the block for the 3X SLI BP-WBPEI58DP? by looking at pictures, they look about identical.

rob

[Waterlogged]

buckeyes;

Any chance there is a protective transparent plastic sheet on the bottom of the mosfet block (very unlikely, I know) or something in the block obstructing or diverting flow from the 'cold plate' of the block? I don't know your loop order, but if the CPU, which has good temps, is in the MB loop, it implies either poor contact/mount or poor heat transfer of the block or something borked with the voltage regulators.

one other thing...and I say this in complete deference to Waterlogged...I read a test between stacked thermal pads vs thicker thermal pad and the thicker pad performed quite abit better. I can't remember where I read it, and, in all truth, I may have even misinterpreted the results...Waterlogged is my superior in all things computer. Still, before something drastic like an RMA of the board or the block, might be worth a try of a thicker thermal pad.

rob

Honestly, I'm not really too keen on the stacked thermal pads myself but it is sometimes required in a rare case or two to get good contact with a poorly design/manufactured component.


@Buckeyes1995

Next time you mount the block, can you get some clear close up (as close as you can) pics?

[Church]

Waterlogged: just an idea .. but shouldn't some copper plate of needed thickness with TIM paste aplied from both sides work even better then thick thermal pads?

[Buckeyes1995]

Honestly, I'm not really too keen on the stacked thermal pads myself but it is sometimes required in a rare case or two to get good contact with a poorly design/manufactured component.


@Buckeyes1995

Next time you mount the block, can you get some clear close up (as close as you can) pics?

Will do.. as I lay in bed last night I was going through my head what could be going on (yes.. clearly I need a new hobby).. anyways... it has to be a tension thing.. I suspect when I take the block off tonight I'll see there are no impressions in the pad.. if there are.. then well I'm at a loss.

If it is a warping issue, could the CPU or NB/SB blocks being tightened too much cause the area where the MOSFET is to be pre-warped? A visual inspection of the board shows some warping but not by the MOSFET.. none that i can see anyways.. but I probably have tightened the screws too much.. I'm going to back them off a little and see if there's any difference.

[Waterlogged]

Waterlogged: just an idea .. but shouldn't some copper plate of needed thickness with TIM paste aplied from both sides work even better then thick thermal pads?

:yepp: Absolutely but, not everyone has the tools, skills and willingness to do that much modding to get something to work.


Will do.. as I lay in bed last night I was going through my head what could be going on (yes.. clearly I need a new hobby).. anyways... it has to be a tension thing.. I suspect when I take the block off tonight I'll see there are no impressions in the pad.. if there are.. then well I'm at a loss.

If it is a warping issue, could the CPU or NB/SB blocks being tightened too much cause the area where the MOSFET is to be pre-warped? A visual inspection of the board shows some warping but not by the MOSFET.. none that i can see anyways.. but I probably have tightened the screws too much.. I'm going to back them off a little and see if there's any difference.

If you have a short straight edge (like a 6" metal ruler), lay it on the MOSFET's and you should be able to see if the board in indeed bowing or not in that area. I would also check MOSFET to each choke height on the board and the compare that with what is on the block. Another thing to check with the block off and the straight edge is, see if there is anything else in the area that protrudes from the board higher than the FET's themselves. I still think that your either hitting the chokes or something else on the board closer to the FET's.

[Buckeyes1995]

:yepp: Absolutely but, not everyone has the tools, skills and willingness to do that much modding to get something to work.



If you have a short straight edge (like a 6" metal ruler), lay it on the MOSFET's and you should be able to see if the board in indeed bowing or not in that area. I would also check MOSFET to each choke height on the board and the compare that with what is on the block. Another thing to check with the block off and the straight edge is, see if there is anything else in the area that protrudes from the board higher than the FET's themselves. I still think that your either hitting the chokes or something else on the board closer to the FET's.

When you say Chokes, do you mean the row of square capacitor-looking things RIGHT beside the MOSFET, between the MOSFET and the CPU? And by FETs? you mean the mosfet chips?

[Waterlogged]

When you say Chokes, do you mean the row of square capacitor-looking things RIGHT beside the MOSFET, between the MOSFET and the CPU? And by FETs? you mean the mosfet chips?

:yepp: & :yepp:

[Buckeyes1995]

ok.. the struggle continues..

I redid the block once again.. this time... I replaced the pad with a new one.. used short screws, and no spacers.. verified (see pictures) that block is making good contact (at least from what I can see)..

I'm really starting to wonder if the difference is the 3X SLI board vs. the 3X SLI CLASSIFIED board.. appears the CLASSIFIED has 4 rows of MOSFETs and everyone I've seen respond is using a full block (I'm not sure why anyone would use 4x MOSFET blocks).

With only have two screws to tighten down the block, I'm not sure how you could make it have any better pad contact..

Anyways.. here's some pics.. I'm all out of ideas, guesses, etc... I would REALLY love to hear from someone with a NON CLASSIFIED 3X SLI that is experiencing better temps than me.


OLD mounting - stock screws & offsets.. DUAL Pads.

Top view.

http://www.herro2u.com/haveblue/premount.jpg

South view.. looks like clean contact

http://www.herro2u.com/haveblue/premount1.jpg
http://www.herro2u.com/haveblue/premount2.jpg

NOT resting ok the Chokes..

http://www.herro2u.com/haveblue/premount_noc.jpg

Depressions in PAD, looks like good contact..

http://www.herro2u.com/haveblue/padstacked.jpg

Naked MOSFET..

http://www.herro2u.com/haveblue/naked.jpg

New Pad..

http://www.herro2u.com/haveblue/newpad.jpg

and finally new mounting on the north side (i forgot to take the south side picture which you can see better, but trust me.. it looks like there's plenty of contact)

http://www.herro2u.com/haveblue/north.jpg

[justin.kerr]

on the pic that says "Depressions in PAD, looks like good contact.." the far right looks like partial contact? it kinda looks like even a different shape, but hard to tell in the pic/
I might have missed it, but have you posted the load temps?

[Buckeyes1995]

http://www.herro2u.com/haveblue/Capture.JPG

[justin.kerr]

that is warm, but not that warm. on air I have seen them around 90C. seems like 60's would be more acceptable though. but the temp sensor could be inaccurate.

[Buckeyes1995]

Yeah I'm just wondering if I should get past it and continue with the build.. Not sure how much higher they will go as push past 4.2Ghz though or 933 PWM.

[Vinas]

I think you're safe either way. I'm wondering too how accurate the temp probe is. Can you get a probe in there and test it?

[gmat]

If you see good contact with the block then you're getting proper cooling (if your water loop doesnt have any issue, of course).
Having less regulators means they are heating up a lot more for doing the same job, that makes sense. I guess that's partly why we have to pay extra for the classified model...
Go on with your build and see what temps you get past 4GHz, if you start hitting 90's that means you've hit the limits of this board (doubt it, i didnt see much temperature increase on VRM's up to 4.3GHz).

[Buckeyes1995]

I think you're safe either way. I'm wondering too how accurate the temp probe is. Can you get a probe in there and test it?

Ya know I probably can.. I have 4 little flat probes that came with my FC-5.. maybe i can put one of them on one of the MOSFETS.. I've got a new thermal pad coming from Frozen this week.. going to try to stack those and use with the short screws that I'm using now.. see if that helps.. when I remount this weekend I'll give that a shot.

[Waterlogged]

Well, with those pics, it looks like everything is sitting good and good contact is being made on most of the FET's, which proves me wrong. With the exception of testing with the stock cooler, I think you've done everything else possible to eliminate mounting of the block as the source of your temps. The only other thing I think it could be is a wonky probe, certainly wouldn't be the first one. . .nor the last for that matter.

.
.
.
Hm, just had a thought occur to me. Any chance the block will mount without the thermal pad? If you've got some Arctic Silver Ceramique, put that on the FET's and see what kind of contact you get. Check everywhere to make sure there is no contact with anything else though.

[shazza]

I've been following this thread for a while - I'll just say that, with the EVGA X58 board (first one out), an i7 920 C0, and the Bitspower mobo blocks, my Vreg temps were regularly in the 55-60 degree range at idle (running at 3.6-3.8 GHz, Vcore around 1.30). At 100% load, they hovered around 65-70 degrees C. This compared to 80+ without water cooling the blocks. The mobo blocks were on their own loop with the XSPC RX240 rad. My ambients tended to be on the warm side - 25 degrees C+.

Every setup is different, but I will say you can bring the temps down some with some airflow. Note that I ran with 70 degree Vreg temps for 6 months almost 24/7 at 100% load while crunching - never had an issue.

I'd say stop worrying about it and move on, but I'm not expert at this stuff - only speaking from my experience.

[Buckeyes1995]

.
Hm, just had a thought occur to me. Any chance the block will mount without the thermal pad? If you've got some Arctic Silver Ceramique, put that on the FET's and see what kind of contact you get. Check everywhere to make sure there is no contact with anything else though.

Yeah that's a good idea.. I have some OCZ Freeze that I could try. I've read a lot about doing thermal grease on the MOSFETs vs. a pad and people have generally been against it due to the MOSFETs not being all exactly the same height.. but certainly worth a try.


I've been following this thread for a while - I'll just say that, with the EVGA X58 board (first one out), an i7 920 C0, and the Bitspower mobo blocks, my Vreg temps were regularly in the 55-60 degree range at idle (running at 3.6-3.8 GHz, Vcore around 1.30). At 100% load, they hovered around 65-70 degrees C. This compared to 80+ without water cooling the blocks. The mobo blocks were on their own loop with the XSPC RX240 rad. My ambients tended to be on the warm side - 25 degrees C+.

Every setup is different, but I will say you can bring the temps down some with some airflow. Note that I ran with 70 degree Vreg temps for 6 months almost 24/7 at 100% load while crunching - never had an issue.

I'd say stop worrying about it and move on, but I'm not expert at this stuff - only speaking from my experience.

Good to hear about your temps Shazza.. I'm seeing the same thing at a 4.2 OC with an 800 PWM.. you're probably right.. I'm certain I'm wasting more time on this than I should. .thanks for the reply :)