Simple cange of power supply reduced temp by 2C

[JesseKnows]

I am running a VP6 2X700@933 (cB0) in a no-name case. I added an intake fan to the bottom front, and made intake holes for it. After a small improvement to the HSF (Intel stock for Celeron PPGA, lapped, instead of the PIII stock), the temp under full load was 37C.

I had the idea of improving the outflow. I found that another power supply I have, has air slots on the bottom as well as on the front. Since the bottom is where the CPUs are, I thought that would help.

I swapped power supplies, and the temp went down 2C! Now running 35.

Next step is more Hz and perhaps more V! The 1GHz mark looms close!

 

[Mikewarrior2]

Is the MB temp any cooler?

Is the case temp any cooler?

You could jsut be "manipulating" socket-thermistor readings. For example, if I take one of the Radio Shack Squirrel cage fans and point it at the base of the socket on a t-bird, i can drop temps 5-8C.


Mike

 

[smp]

Before I got my new system.. I downloaded AMD's whole PDF of "building an AMD based system".. they only recommend power supplies with the intake on the bottom.. mine's is on the back..
I got some blowholes though..
<edit>
but then you got Intel inside.. so I guess my comment is irrelevant
</edit>

 

[JesseKnows]

Mikewarrior2,

I really doubt that I am skewing the socket thermistor reading. The change is not near the socket base, but rather in the case environment. I wouldn't compare to your example.

The VP6 has no MoBo or case sensor, so I don't have an answer.:frown:

 

[Mikewarrior2]

you'd be surprised at how little, very subtle airflow changes chan effect socket-thermistor readings and not effect cpu core temp readings.


Mike

 

[JesseKnows]

Since I do not have the capability of measuring the real CPU temp (or do I? the VIA program which came with the VP6 uses the thermistors, but there is a diode in the PIII, what software would tell me the value there? I would prefer it anyway!), I must defer to your statement.

But pray tell, how would air flow reach the thermistor, which is between the MoBo and the CPU, and surrounded by the socket? There is surely no airflow there, unless through the possible cracks not closed by the socket itself - minuscule at best?

 

[Mikewarrior2]

Any airflow that hits the mb near the socket-air drops socket-thermistor readings. Another misnomer is that the socket is sealed. it isn't. Take a can of compressed air, turn it upside down, and spray at the bottom of the socket. YOu'll get condesnation inside the socket.

Remember that the thermistor touches cpu back-pcb with at most 1-2% of its surface. So 90+% is effected by socket air. ANy little bit of airflow that etiehr gets into the socket or cools the mb near the socket will drop the reading significantly.


Mike

 

[JesseKnows]

FWIW, getting some air circulated out of the warm pocket inside the socket would help cool the bottom of the CPU, thereby contributing to the general effort

More on the thread topic, we are talking about improved outflow through the power supply slits. These are sucking (warm) air out from the general area above the HSF. Hardly any flow directed at the socket area.

 

[Mikewarrior2]

Does the power supply pull air from above the heatsink/cpu area?

And no, cooling the backside of a cpu directly does not actually contribute to cpu cooling. THe PCB is heat resistant, and isn't a good heat conductor.


Mike

 

[JesseKnows]

Yes, the PS pulls air out of the case wherever there are slots, and the new PS has slots on its bottom, which is above the CPU area.

The new PS also has slots on its left side which is above the back of the motherboard. I am not sure what this would contribute. May be more relevant to cooling drives than MoBo components.

And, is it slots or slits or either?

 

[Mikewarrior2]

which is why I asked in the first place: Did your case temp drop. If so, that explains the CPU temp drop. If not, the socket-thermistors are being &quot;manipulated&quot;


Mike

 

[JesseKnows]

I see. Well, that question will have to wait until some manner of discovering the case temperature is devised - the VP6 has no case sensor.

And is there no way to interrogate the PIII's internal diode on the VP6 either?

 

[006agent]

just modded my psu...put a intake like the enermax and the ambient dropped 2c