Interesting thing to consider when overclocking a quad

[cmdrdredd]

Maybe it's not our memory, our CPU, our CPU temps, or our memory timings.

It seems that PWM temperatures are contributing to BSOD and crashes under load when overclocking a quad CPU on the Abit IP35 boards.

http://www.xtremesystems.org/f...d.php?t=147163&page=58

 

[SerpentRoyal]

I've warned people about not spending to much $ on fancy heat pipe. Best cooling is with a small custom heat sink at the MOSFET region and a low speed 80 mm Panaflo. My E4300 @ 3.46GHz maxed out at 52C PWM under S&M's heat mode (85F ambient). Quads may bump temp to about 65C.

 

[myocardia]

Not only that, guys, but with most motherboard failures that are overclock related, PWM temps weren't kept in check. Draw whatever conclusion from that that you want, but I've noticed it happen over and over.

 

[Hauk]

Serpent, you may recall the poor contact patch because heatpipe was slightly bent on my IP35 Pro. I wonder how widespread that issue may have been. Proper reapplication may have helped, PWM temps soar still.

You're right cmdrdredd, PWM temps needs to be added to the list!

 

[Shimmishim]

steelsix - you're not the only one with a slightly bent heatpipe assembly. i noticed that it was quite difficult to remove the pushpins when i was originally trying to reapply some new TIM.

i've gone ahead and bought some nuts, bolts, and washers to redo the assembly without the pins. i'm hoping it'll provide better temps

 

[Diogenes2]

How do you monitor PWM temps ?

 

[bryanW1995]

uGuru does it for ip35 pro. I've never checked on others but I imagine that most good mobos have similar software.

 

[SerpentRoyal]

Originally posted by: SteelSix
Serpent, you may recall the poor contact patch because heatpipe was slightly bent on my IP35 Pro. I wonder how widespread that issue may have been. Proper reapplication may have helped, PWM temps soar still.

You're right cmdrdredd, PWM temps needs to be added to the list!

Be sure to lap the heat sink to achieve a flat surface. Apply a thin layer of paste on top of each output device. Put the heat sink back on and verify that you see the footprint of each output device on the back of the heat sink.

Sometimes, it's better to remove the heat sink and blow cold air over each output device.

 

[Hauk]

Originally posted by: SerpentRoyal

Originally posted by: SteelSix
Serpent, you may recall the poor contact patch because heatpipe was slightly bent on my IP35 Pro. I wonder how widespread that issue may have been. Proper reapplication may have helped, PWM temps soar still.

You're right cmdrdredd, PWM temps needs to be added to the list!

Be sure to lap the heat sink to achieve a flat surface. Apply a thin layer of paste on top of each output device. Put the heat sink back on and verify that you see the footprint of each output device on the back of the heat sink.

Sometimes, it's better to remove the heat sink and blow cold air over each output device.

What did you end up doing for your IP35-E? Did you just keep heatsink off and use active cooling? How do your PWM temps look?

 

[lopri]

While I use a different board (EVGA 680i), the issue discussed in this thread is one of the main reasons I went with blow-down HSF with my Q6600. (Scythe Andy) AnandTech HSF reviews still 'downplay' the blow-down type of heat sinks because they don't net you as high overclocks (80% CPU load at that :roll: ) as the L-shaped ones do. But the true value of the heat sinks like Andy Samurai is that it moves air around CPU socket area. (NB/memory/MOSFET) And often times these components get very hot without users' knowledge.

This doesn't mean that L-shaped HSF are inferior. They are indeed superior than blow-down type heat sinks in CPU cooling itself. But when using these popular L-shaped HSFs, often times the NB, memory, or VRMs don't get any air-flow at all. From my experiences, the 'death ratio' of mobo/memory was much higher with L-shaped HSFs. This is before I learned how hot things get around CPU socket and now I always make sure to have air-flow there, or go with a blow-down HSF.

 

[SerpentRoyal]

Originally posted by: SteelSix

Originally posted by: SerpentRoyal

Originally posted by: SteelSix
Serpent, you may recall the poor contact patch because heatpipe was slightly bent on my IP35 Pro. I wonder how widespread that issue may have been. Proper reapplication may have helped, PWM temps soar still.

You're right cmdrdredd, PWM temps needs to be added to the list!

Be sure to lap the heat sink to achieve a flat surface. Apply a thin layer of paste on top of each output device. Put the heat sink back on and verify that you see the footprint of each output device on the back of the heat sink.

Sometimes, it's better to remove the heat sink and blow cold air over each output device.

What did you end up doing for your IP35-E? Did you just keep heatsink off and use active cooling? How do your PWM temps look?

Lapped the Mosfet heat sink because all the MOSFETs were co-planar. My Big Typhoon with 120 x 38 mm Panaflo provides sufficient air flow to keep the PWM temp under 52C with 85F ambient. CPU is E4300 @ 3.46GHz. I could also run without heat sink with an 80mm low speed Panaflo strapped to the Big Typhoon. PWM is around 48C.

 

[amking]

Originally posted by: Shimmishim
steelsix - you're not the only one with a slightly bent heatpipe assembly. i noticed that it was quite difficult to remove the pushpins when i was originally trying to reapply some new TIM.

yep my buddy and i both used ip35 pro's on our new builds last weekend and his heat pipe was bent a decent amount, which caused the northbridge heatsink to not sit completely flat. since he wasnt really overclocking a lot (400fsb) or anything and the thermal pad deal was still bridging the gap, he decided to not worry about it and start building rather than rma.

otherwise the board has been great.

 

[cmdrdredd]

Originally posted by: lopri
While I use a different board (EVGA 680i), the issue discussed in this thread is one of the main reasons I went with blow-down HSF with my Q6600. (Scythe Andy) AnandTech HSF reviews still 'downplay' the blow-down type of heat sinks because they don't net you as high overclocks (80% CPU load at that :roll: ) as the L-shaped ones do. But the true value of the heat sinks like Andy Samurai is that it moves air around CPU socket area. (NB/memory/MOSFET) And often times these components get very hot without users' knowledge.

This doesn't mean that L-shaped HSF are inferior. They are indeed superior than blow-down type heat sinks in CPU cooling itself. But when using these popular L-shaped HSFs, often times the NB, memory, or VRMs don't get any air-flow at all. From my experiences, the 'death ratio' of mobo/memory was much higher with L-shaped HSFs. This is before I learned how hot things get around CPU socket and now I always make sure to have air-flow there, or go with a blow-down HSF.

This is why I run my Panaflo H1BX at 100%. Loud sure for some people, but it pushes tons of air around the area and yes it does hit the PWM area.

 

[Hauk]

I've decided to do this bolt-mod
using Arctic Ceramique and a 40mm fan. Will run fan through mobo @ 60-70% to keep it quiet.

The quest continues...

 

[bryanW1995]

Originally posted by: SerpentRoyal

Originally posted by: SteelSix

Originally posted by: SerpentRoyal

Originally posted by: SteelSix
Serpent, you may recall the poor contact patch because heatpipe was slightly bent on my IP35 Pro. I wonder how widespread that issue may have been. Proper reapplication may have helped, PWM temps soar still.

You're right cmdrdredd, PWM temps needs to be added to the list!

Be sure to lap the heat sink to achieve a flat surface. Apply a thin layer of paste on top of each output device. Put the heat sink back on and verify that you see the footprint of each output device on the back of the heat sink.

Sometimes, it's better to remove the heat sink and blow cold air over each output device.

What did you end up doing for your IP35-E? Did you just keep heatsink off and use active cooling? How do your PWM temps look?

Lapped the Mosfet heat sink because all the MOSFETs were co-planar. My Big Typhoon with 120 x 38 mm Panaflo provides sufficient air flow to keep the PWM temp under 52C with 85F ambient. CPU is E4300 @ 3.46GHz. I could also run without heat sink with an 80mm low speed Panaflo strapped to the Big Typhoon. PWM is around 48C.

I've never checked pwm temps until now. I'm running small ffts and it is sitting on 50c. I'm a little worried about airflow to the nb but based upon these posts 50c doesn't look too bad.