Re: Computer turns on and then just dies

On Sat, 23 Nov 2024 14:28:54 -0600, VanguardLH wrote:

> > I'm going to replace the thermal paste of the CPU first and see what
> > happens.
 
> Make sure to use non-metallic paste.  Use ceramic.  Since you're not
> overclocking, you won't achieve a one to three degree decrease in
> temperature, anyway.  Not overclocking already reduced heat generation.
> Also, make sure it is thermal PASTE, not thermal adhesive which is an
> epoxy.

I already ordered Arctic MX-4. What can I say? I'm an impulsive buyer.
(-:

> You'll need a piece of mylar or other thin plastic to apply the paste.
> A credit card is too rough on its edges.  You want a shear coat, not an
> opaque coat.

I'm hardly an expert, I just saw a YT trying one dot, 5 dot, cross and
then some tool to spread the paste, like what you're saying. Some guy
commented that one dot is the best and that using a tool causes air
bubbles. (?)

> I've seen many techniques suggested for applying the correct amount of
> thermal paste.  One is to use a piece of thin mylar to spread around the
> paste to coat a thin layer you can see through.  Another is to put one
> drop in the middle of the CPU plate, or 5 tiny drops in the center and
> at midway to the corners, to push down the heatsink to rotate and spread
> the paste.  If any oozes out paste the CPU plate, you put on too much
> which also means there is likely too much between CPU and heatsink.

I think I'm going to go with the one dot method while keeping in mind:
less is more. I'll practice on a napkin first.
 
> You'll need isopropyl alcohol and a plastic scraper to remove the old
> paste.  Make sure the CPU plate and heatsink are devoid of any old paste
> or thermal pad, and are clean.  After cleaning, keep your fingers off.

Done.
 
> Considering your lack of electronic test equipment, I suspect paste
> might not be the way for you to go.  A thermal pad gets squeezed between
> CPU plate and heatsink, won't ooze out, and liquifies (melts) under
> pressure and heat.  Considering how many users apply way too much paste,
> and if not overclocking, a pad is just as good.  I've never gotten into
> pads, but https://www.youtube.com/watch?v=2BhKx0iQ4K8 and
> https://www.youtube.com/watch?v=niAQs8dZohE look interesting.  Despite
> the claim reusability of the graphic pad, I'd never reuse one.  I'd
> replace it just I would replace paste.  I'm not building and rebuilding
> hundreds of boxes.

One of the few computers I assembled myself had AMD CPU's with a
heatsink that had such a thermal pad. I'm just going to with the paste.
It's not like it's a brand new computer.
 
> Personally I'd go with the PSU substitution suggestion first.

So you're conspiring with Paul eh? ;-) I already removed the paste, so
I'll have to replace that first. If that doesn't work I'll come back to
ask if the even older PSU that I've got is compatible.

>  The load
> a PSU can handle with proper regulation varies with age, and crappy PSUs
> age faster.  Replacing thermal paste isn't a job for a newbie.  I always
> lap the CPU plate and heatsink to get them as flat as possible for best
> mating since metal-to-metal contact is by far the best method for
> thermal transfer.  However, rare few users lap their CPU and heatsink.
> I don't overclock, but do want thermal transfer to work best, and want
> the least amount of paste since it also ages.  

I understand your concern, but it's not a new computer. For me it's more
for the fun of it. If I fail, then I fail, but I will still have learned
something and that's worth something (for me anyway).
 
> Although the focus has been on the paste, another facet of concern is if
> the clamping of the heatsink is even.  You don't want to cinch down one
> side more than the other.  That will squeeze out more paste on one side,
> but leave too much on the side with less pressure.  You want paste to
> substitute for air, not thermal paste to be the primary transfer media.

Thanks for the advice. I've been practicing on the clamping of the
heatsink.

-- 
s|b