cpu cooler off-label use

Mar 19, 2022 Last reply: 4 years ago 37 Replies

formatting link
This might not be as good as a gigantic machined heat sink, but it's easier.



The little guy will be an LM35 temperature sensor.


This is what the kids these days use to cool their gigantic GPUs, I think the mad-scientist green gives it a certain pizazz:

formatting link

lørdag den 19. marts 2022 kl. 03.23.47 UTC+1 skrev snipped-for-privacy@highlandsniptechnology.com:

if you look up the TPD for the CPU it is meant for you'd get an ideas of the likely performance

We can let this get pretty hot. 80 or even 100c maybe? We're going to test one in a mockup chassis and measure the real theta. The CPU power rating doesn't tell us a lot about theta.

lørdag den 19. marts 2022 kl. 04.01.17 UTC+1 skrev snipped-for-privacy@highlandsniptechnology.com:

sure, but CPU spec something like 50W TPD and Tj 100'C is a start

On a sunny day (Fri, 18 Mar 2022 19:23:37 -0700) it happened snipped-for-privacy@highlandsniptechnology.com wrote in snipped-for-privacy@4ax.com:

Not sure about the air flow Is there space between the fan and the surface?

PS was reading this yesterday:

formatting link

lørdag den 19. marts 2022 kl. 09.22.54 UTC+1 skrev Jan Panteltje:

airflow goes in through the sides of the heatsink between the fins, up through the fan and out the back

On a sunny day (Sat, 19 Mar 2022 04:27:44 -0700 (PDT)) it happened Lasse Langwadt Christensen snipped-for-privacy@fonz.dk wrote in snipped-for-privacy@googlegroups.com:

Thanks, got it

That particular cooler intakes on three sides and blows air out the fourth. That's ideal for my use; the hot air will blow over the Phoenix connector and out the back of the box.

formatting link
Some coolers are designed to inhale their own hot air!

I suppose these fans have low pressure. I know the axial fans are that way.

That cooler measured 0.24 K/W. The transistors are huge so the junctions won't get much above the heatsink temp.

I should be able to dump 200 watts.

lørdag den 19. marts 2022 kl. 17.41.48 UTC+1 skrev John Larkin:

the datasheet for the transistor says typ. 0.24C/W case to sink plus max 0.28C/W junction to case

Use the center legs to conduct, but rotate bodies to allow tabs to face the cooling surface.

RL

lørdag den 19. marts 2022 kl. 18.19.27 UTC+1 skrev legg:

the tabs are facing the cooling surface

I'll be running two in parallel some times, all four some times. And Tj max is 150C. Might work.

We'll run a realtime Tj simulation and shut down at computed 150C.

I'm sure it will be fine. It's not like he has to worry about the solder melting or anything. It's held together with screws.

What temperatures do CPU/GPU owners report when using similar cooling? I would not expect this one to be among the best. The fins are very short and the airflow is rather restricted. 200W is a lot even for a full blast CPU with overvoltage.

The transistors are on the bottom of the board, not sandwiched between the heat sink and the fan. The fins are very, very short on the top side of the heat sink, just under the fan which is moving air through the heat sink with a vacuum. Normally the fans blow into heat sinks, but in this case there's no room for that.

If this dissipates 200W per board and the chassis has say, 10 boards, that's going to be a metric s*it ton of heat to get out of the case, and even the room. A space heater is around 1.5 kW. Put your hand in front of that!

it's rated for 125W TPD and CPU are usually 100'C max Tj

0.29 C/W with the fan at max speed

formatting link

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required