Cooling Solution for PWMIC Area in DFI Boards
- Thread starter Saiyan
- Start date
- Status
, i hate noise:no: . How much mm fan is that??? Is there anyway i can find some silent fan??
...Chaos... I think you are not stressing your board enough or Oskar-Wu gave you a special one 
... as mine and almost everyones PWMIC gets upto 50-55c if not actively cooled. I have seen people gettting 60c on PWMIC and then rig freezes. This doesnt fit your statement, friend. I would like to know who told you that Chipset requires more cooling then PWMIC?
... as mine and almost everyones PWMIC gets upto 50-55c if not actively cooled. I have seen people gettting 60c on PWMIC and then rig freezes. This doesnt fit your statement, friend. I would like to know who told you that Chipset requires more cooling then PWMIC?Harshal said:Chaos... I think you are not stressing your board enough or Oskar-Wu gave you a special one
Eh well maybe dual core chips cause the PWMIC chips to heat up more cos of all the extra current flowing. I'm running my single core chip at 2.6 and well it doesn't even touch 45 :S. The chipset on the other hand reaches 50+ when running anything 3d or priming. The chips in the PWMICs are rated for operation upto 80 degrees. I don't see any reason why they should freeze at 60.
Darklord
Galvanizer
Well Chaos,i can very much vouch for the fact that the PWM area heats up like anything.
Also i never used Dual Core on it.
Also the Smartguardian thig can act weird at times.
Hand technique is the best way to find out...lol
Also enabling SLI jumpers and using the second slot makes both slots work @ 8X.
I tried that when i had this board.
Also i never used Dual Core on it.
Also the Smartguardian thig can act weird at times.
Hand technique is the best way to find out...lol
Also enabling SLI jumpers and using the second slot makes both slots work @ 8X.
I tried that when i had this board.
blueyesdude
Contributor
Well for my dfi ultra-d,
the chipset temp is 52c max at 100% full load on both cores,while the pwmic goes to 62c.
Havent experienced any freezing as of yet.
the chipset temp is 52c max at 100% full load on both cores,while the pwmic goes to 62c.
Havent experienced any freezing as of yet.
blueyesdude
Contributor
U talking abt fans.
I have 2 120mm fans@140cfm
3 80mm sunon fans and 1 92mm panaflo fan in my cabinet.
My gfx card dropped a superb 10c at full load due to that.
Also chipset went from 58c to 51c at load.
No effect on pwmic since it is located at a wierd location.
Maybe i will introduce 1 more 92 mm panaflo on top of the pwmic area.
Man that makes 7 fans running in my cabinet.Lol.
I have 2 120mm fans@140cfm
3 80mm sunon fans and 1 92mm panaflo fan in my cabinet.
My gfx card dropped a superb 10c at full load due to that.
Also chipset went from 58c to 51c at load.
No effect on pwmic since it is located at a wierd location.
Maybe i will introduce 1 more 92 mm panaflo on top of the pwmic area.
Man that makes 7 fans running in my cabinet.Lol.
The pwm Voltage regulator transistors are rated for more than 110C approx. (in theory atleast)
But the ambient conditions play a very important role in the efficiency of semicoductors.
The temps greater than 25C and humidity greater than say 50% will start decreasing the efficiency of these parts.
hence it would be prudent not to exceed say 65C on max load .....
NB:
1) The onboard sensors are rarely very accurate hence it is always better to take relative measurements. for e.g. measurements at idle and at load. the difference between them will be quite accurate.
2) The high temps will not damage the pwm transistors per se but the radiated heat will affect the other nearby components, thus reducing their efficiency / tolerance too...
@ Saiyan: try to get a couple of 50mm fans. they should cost about Rs.30 /- max each, and strategically place them over the pwm heatsink and one over the heatsink towards the bottom right hand corner of the m/b (to the right of the ram slots) this part is the Vdimm controller circuit.
But the ambient conditions play a very important role in the efficiency of semicoductors.
The temps greater than 25C and humidity greater than say 50% will start decreasing the efficiency of these parts.
hence it would be prudent not to exceed say 65C on max load .....
NB:
1) The onboard sensors are rarely very accurate hence it is always better to take relative measurements. for e.g. measurements at idle and at load. the difference between them will be quite accurate.
2) The high temps will not damage the pwm transistors per se but the radiated heat will affect the other nearby components, thus reducing their efficiency / tolerance too...
@ Saiyan: try to get a couple of 50mm fans. they should cost about Rs.30 /- max each, and strategically place them over the pwm heatsink and one over the heatsink towards the bottom right hand corner of the m/b (to the right of the ram slots) this part is the Vdimm controller circuit.
Whoops missed this thread.
The "PWM IC" term used in SmartGuardian is a bit confusing, since the actual PWM IC controller is not the one under the heatsinks. Its the rectangular IC higher up:
But the thermistor sensing the temp for SmartGuardian's PWMIC temp is located near the top of the heatsinked area. If i'm not mistaken, the heatsinked area is part of the PWM circuitry, just that they're the voltage regulators controlled by the PWM IC. Again its not exactly measuring the temps of the devices under the heatsinks, but just the general temps around that region:
- As per PWM IC Temps, i'm running a pretty normal 43C here too. But yes I fully agree that temps are very dependant on the proc you are running - I have seen temps shoot up to the 50's with a Dual-core processor and even with the single-core procs based on the older 130nm process on the same board/same setup.
- People running watercooling setups are expected to see higher temps as well owing to the fact that there's not much air flowing over the motherboard area from a traditional air-based cooling setup (The PWM IC sensor is very close to the CPU socket area).
- However I have noticed that the heatsinked area is pretty cool to lukewarm at most, so not sure what exactly is causing the higher temps being reported considering the sensor is not even in direct contact with the heatsinked area and depends on radiated heat rather than conductive heat.
- And I agree with Deejay about paying attention to cooling the vDimm Mosfet at the absolute top-right corner of the board as that is what tends to show more heat than the others. With normal RAM its ok, but should you decide to use high-voltage Winbond based memory then you'd need to pay careful attention to that area. A lot of reasons why people had their Winbond based memory dying on them was cause the VR would go wacky with the higher temps.
The "PWM IC" term used in SmartGuardian is a bit confusing, since the actual PWM IC controller is not the one under the heatsinks. Its the rectangular IC higher up:
But the thermistor sensing the temp for SmartGuardian's PWMIC temp is located near the top of the heatsinked area. If i'm not mistaken, the heatsinked area is part of the PWM circuitry, just that they're the voltage regulators controlled by the PWM IC. Again its not exactly measuring the temps of the devices under the heatsinks, but just the general temps around that region:
- As per PWM IC Temps, i'm running a pretty normal 43C here too. But yes I fully agree that temps are very dependant on the proc you are running - I have seen temps shoot up to the 50's with a Dual-core processor and even with the single-core procs based on the older 130nm process on the same board/same setup.
- People running watercooling setups are expected to see higher temps as well owing to the fact that there's not much air flowing over the motherboard area from a traditional air-based cooling setup (The PWM IC sensor is very close to the CPU socket area).
- However I have noticed that the heatsinked area is pretty cool to lukewarm at most, so not sure what exactly is causing the higher temps being reported considering the sensor is not even in direct contact with the heatsinked area and depends on radiated heat rather than conductive heat.
- And I agree with Deejay about paying attention to cooling the vDimm Mosfet at the absolute top-right corner of the board as that is what tends to show more heat than the others. With normal RAM its ok, but should you decide to use high-voltage Winbond based memory then you'd need to pay careful attention to that area. A lot of reasons why people had their Winbond based memory dying on them was cause the VR would go wacky with the higher temps.
- Status