Isn’t the risk of damage with liquid metal too high? Why not just run the CPU without the lid?

Most likely you have some bad caps. However, my Operton 185 also stopped working pretty quickly due to the old thermal paste under the lid.

feipoa wrote on 2026-04-07, 21:34:

Isn’t the risk of damage with liquid metal too high? Why not just run the CPU without the lid?

Most likely you have some bad caps. However, my Operton 185 also stopped working pretty quickly due to the old thermal paste under the lid.

I didn't understand if the liquid metal can damage the lid or not, if not and it's properly applied I doubt it can damage anything.
Bad caps on the legendary dfi lanparty ut nf4 ultra-d? I hope not 😁

I needed to recap my socket 939 motherboard to improve stability.

As I understand it, liquid metal is electrically conductive. When the heat spreader is placed back on, the liquid metal
can get pressed out and into one of the numerous capacitor networks, thereby shorting out the CPU.

It is also very easy to have the heat spreader short on these capacitor leads. There’s only about 1 mm of clearance.

If you insist on placing the heat spreader back on (which I see no need for), please photograph and document your work here.

feipoa wrote on 2026-04-07, 22:10:

I needed to recap my socket 939 motherboard to improve stability. […]
Show full quote

I needed to recap my socket 939 motherboard to improve stability.

As I understand it, liquid metal is electrically conductive. When the heat spreader is placed back on, the liquid metal
can get pressed out and into one of the numerous capacitor networks, thereby shorting out the CPU.

It is also very easy to have the heat spreader short on these capacitor leads. There’s only about 1 mm of clearance.

If you insist on placing the heat spreader back on (which I see no need for), please photograph and document your work here.

Honestrly I am even unsure if I will be able to delidd it without killing the cpu.
I ordered the razor blades but I will try to delidd a cheap 939 cpu first.

I don't see how you can kill the CPU. All you can do is ruin a few capacitors.

Did you run the cheap 939 in your system to ensure that there's no other issues with your motherboard or heatsink?

A small update.
I did some research about the liquid metal, the lid of the opteron is Nickel-plated Copper so it shouldn't be an issue, but it may need a second application after 12 months from the first one.
The alternative I am considering it's the old good Artic Silver 5, I remember it lasted more than 10 years when I used between the the lid and the heatsink on my 775 build back in time and I bought several tubes before it went out from market.
Another alternative that I am considering and it shouldn't need any new application is the Honeywell ptm7950 that would be the definitive solution.

Soon I should receive the opteron where to try the first delid, according to the result I will proceed further witht the 180.

Chiming in with this cooler I use on my Opteron 170.

Originally cooled a 64 x2 5000+ (iirc Brisbane core?), didn't have any issues w/ the Opteron either.

Temps I recorded with it were around 45-50ish during idle, around 63-68 during load.

Socket 939 dual-core Athlon 64/Opteron doesn't require insane cooler to work normally, if properly delidded.

I delidded the opteron 180 successfully (and a x2 4200+), it wasn't so difficult. Remove the gunk from the lid is pretty easy with the blade, on the cpu though it's quite difficult and I tried to use some black silicone just on the angles of the cpu to seal it again after applying some thermal paste, it proved to be useless it came off when I took the heatsink away from the socket.
My next try will be with using SL620 Pattex black silicone, it is neutral instead of being acetic like the last one I used, but mostly to remove the old one from the organic package I will try Soudal SR100 which is a silicone remover stuff, I will obviously try it with a spare cpu and if it works proceed with the good ones.
As for testing the cpu with new thermal paste (I used the Artic silver 5 on the opteron and MX-4 on the x2), proved to be perfect, it was at 25°C on idle and goes up to 40 during load.

You don't even need to delid it unless the paste has gone bad. In that case you won't be able to stop it from overheating unless you delid it.

One of my favorite coolers for 939 was the AMD OEM Phenom II cooler. There are lots of options of course. I have a Thermalright XP-90C that I got with a 939 motherboard acquisition long ago. I also like Scythe Ninja. Any old gnarly beast is fun to use.

swaaye wrote on 2026-05-07, 03:30:

You don't even need to delid it unless the paste has gone bad. In that case you won't be able to stop it from overheating unless you delid it.

One of my favorite coolers for 939 was the AMD OEM Phenom II cooler.

The paste went "bad" it was just solid and the opteron was clearly overheating, I think it's normal after almost 20 years?

Nemo1985 wrote on 2026-05-07, 03:32:

The paste went "bad" it was just solid and the opteron was clearly overheating, I think it's normal after almost 20 years?

I don't know what the cause is but not all of them have this problem. Maybe it is inevitable though.

Why do you want to put the lid back onto the CPU, as opposed to running the CPU without the lid?

feipoa wrote on 2026-05-07, 06:03:

Why do you want to put the lid back onto the CPU, as opposed to running the CPU without the lid?

I suppose that to use the cpu without the lid it will be necessary to adapt the heatsink, other than risk to chip the die?
Not my cup of tea honestly.
In the meanwhile I received the products, after apply the gel and left on the silicone for about 15 minutes it came of easily with the nail.

It depends on the amount of spring force and the intended set distance of the heatsink clamp. The various heatsink/clamp/fans systems I tested had no issue without the Opteron's lid.

Have you managed to damage a CPU die from heatsink clamp usage? Much of the PIII series came without a heat spreader. I never damaged any.

If it is about aesthetics and collectable value, I get it. If not, then be [mentally] prepared for the CPU to overheat again in a few years, especially if you are in the habit of removing the heatsink, putting it back, removing the CPU, etc. Consumer Silicone glue has some give, and with repeated on/offs, it will loosen, which may not be visually apparent. The silicon then acts has a spacer, pushing against the heatsink's clamp, thus, preventing proper heat transfer. It might be OK if you leave the heatsink on and never remove it again. This was what happened when I did this operation. I ultimately removed the heat spreader after about a year or two and the problems went away.

Good luck.

For myself, I would prefer a Zalman CNPS9700, Zalman CNPS9900MAX, or Zalman CNPS 10X Extreme, running at 100 % speed in the most OC-time.
All of these coolers are still easy to mount onto a 754/939/AM-socket.
And, of course, with liquid metal between the lid and the CPU-core.
By the way, if you intend to use the pure CPU-core without the lid (in this case no liquid metal because of the danger of a shortcut), you can mount 4 black round "spacer" from an Athlon XP around the core.

I just thought of this, but if there was a 3D-printed jig like the one in the image that could properly align copper pieces, wouldn't it make reapplying grease easier while also relatively reducing the risk of core chipping or capacitor short circuits?

kagura1050 wrote on 2026-05-11, 08:58:

I just thought of this, but if there was a 3D-printed jig like the one in the image that could properly align copper pieces, wouldn't it make reapplying grease easier while also relatively reducing the risk of core chipping or capacitor short circuits?

The idea is good. The problem I see: if you have a printed jig, this will consist by a kind of plastic (as it is 3D-printed). In this case, near the CPU-core, maybe the melting-point of the plastic will be reached, as the core became very hot (even more if someone wants to overclock the CPU).

A similar jig already exists for the Athon-XP and the Athlon Thunderbird (so. 462) - but it's made by copper, not by plastics, for the reason of high temperatures, and for stability.
This copper-jig was so designed, that all the electric components next to the core are not covered by them. The thickness (heigh) was just a bit higher than the components, but a little bit lower than (or equal to) the core. I just found a picture somewhere in the web - see it attached.