is 100°C safe for mosfet?

[MESSERR]

i just overclocked my igpu (VEGA 11) from 1400mhz to 1750mhz on my budget build, and im just happy its running stable. But i noticed that the mosfet temp is reaching 100°C. So is it safe? or i should turn down the overclock so the mosfet doesnt running at high temperature.

[Haro]

What board is it exactly?  

Also what did you set the soc voltage to?

[MESSERR]

Just now, TofuHaroto said:

What board is it exactly?  

Also what did you set the soc voltage to?

AORUS B450 ELITE

1.4v on cpu and 1.25 on the igpu

[MESSERR]

[Haro]

Just now, MESSERR said:

1.4v on cpu

Under load? Or was that monitored at idle? 

 

Just now, MESSERR said:

1.25 on the igpu

 That's really high voltages for the Soc. 

Don't go over 1.2, try dropping it down to 1.2 and check, also try directing some airflow at the vrm and then monitor temps. 

100c on the mosfets isn't bad if it doesn't throttle the cpu. 

But for the most part it shouldn't damage the mosfets. 

[MESSERR]

Just now, TofuHaroto said:

Under load? Or was that monitored at idle? 

 

 That's really high voltages for the Soc. 

Don't go over 1.2, try dropping it down to 1.2 and check, also try directing some airflow at the vrm and then monitor temps. 

100c on the mosfets isn't bad if it doesn't throttle the cpu. 

But for the most part it shouldn't damage the mosfets. 

no i didnt touch anything on the soc, the soc is set on auto. but vcore is 1.4v and the voltage for the igpu is 1.25v

[Haro]

1 minute ago, MESSERR said:

for the igpu is 1.25v

The Soc has to do with the igpu. 

That's why the soc mosfets specifically are at 100c. 

But as I said. 

8 minutes ago, TofuHaroto said:

100c on the mosfets isn't bad if it doesn't throttle the cpu. 

But for the most part it shouldn't damage the mosfets. 

^. 

[mariushm]

The mosfets will be fine, they're rated for at least 125c, some are rated for 150c

Long term, it would be a good idea to get that down to around 95c ... over weeks and weeks of running at 100-110c the printed circuit board on which those mosfets are soldered will degrade from the heat ... but that's a long time, a few hours a day when gaming it's not a big deal.

 

You could just add a 92-120mm fan blow down on the heatsinks and that should get you a few degrees off.

[churblesfurbles]

The rating means nothing, rated for how many hours at 125c, and that's just a spec, was it made in china, a bad batch, its just odds when you run things at that ragged edge.  If you want things to last and be reliable, keep it cool.

 

Why is airflow so bad that a mosfet is hitting 100c?

[mariushm]

4 minutes ago, churblesfurbles said:

The rating means nothing, rated for how many hours at 125c, and that's just a spec, was it made in china, a bad batch, its just odds when you run things at that ragged edge.  If you want things to last and be reliable, keep it cool.

 

Why is airflow so bad that a mosfet is hitting 100c?

The temperature rating is for the actual mosfet chip soldered on the board.

 

They're made by American companies in various countries, including US, Taiwan, China and they're packaged (put in the case that is then soldered to the motherboard) in lots of other places like Hong Kong, Phillipines, whatever.

 

By the nature of how the silicon chip is designed, it can handle up to 125c continuously (there's no number of hours limitations as is the case with capacitors which have chemical substances inside which degrade with prolonged heat), provided the heatsink can take the produced heat and maintain the temperature below 125c.

 

Also most motherboards have temperature sensors and read the temperature and will throttle the vrm circuit if the chips get too hot.

In this particular case, due to his combination of voltage and gpu frequency (overclocking) the vrm outputs more heat than normal, it's overclocking after all. I would guess the chips go up to around 101c where the temperature protections trip and the vrm signals the gpu to reduce its frequencies momentarily. This reduces the gpus power consumption, which in turn means the vrm has to work less and this means there's less heat produced so more heat can be dissipated in a unit of time and therefore the vrm cools down to let's say 95c or less, at which point the vrm may signal the gpu that it's cool enough and it can ramp up ... and this gets repeated until you stop playing games or anything that puts high load on the gpu.

 

The motherboard is fine, the gpu vrm is otherwise cooled reasonably well, you can see there the average of 67 is a perfectly fine number.

If you want to be more comfortable and see smaller numbers, by all means, add a fan and get it to blow over the fins of those heatsinks around the cpu socket, and you'll improve the efficiency of the heatsinks, so more heat will be taken away, so the vrm will be cooler.

 

Or you could just go 1-2% less on voltage and frequencies on the GPU, it's not like you're gaining more than 0.5-1fps with this overclocking on a integrated graphics.

 

[churblesfurbles]

Yea I guess I was mistake on the life span rating, mosfets don't seem to have those.