[split discussion topic] Thermal interface material

split
cooling

#1

This is a new discussion topic split from an off-topic branch of an existing topic, discussing the merits and pitfalls of thermal paste versus thermal pads. Note that since the V is already in production, no changes will be made to the thermal interface material used at this time. That doesn’t mean we can’t share experiences, learn new things, and generally get our geek on!

Source topic:
Day 5 at the factory by @Konstantinos


Day 5 at the factory
#2

I’m not too convinced with the thermal pad.
I dont think theres ever been a laptop with thermal pad instead of thermal paste. OEMs generally just use a little too much thermal paste to account for inaccuracy in the installation, which, true that it has a small impact on performance, but generally still better than a thermal pad i believe

Reason being is metal-metal contact is still the best, and the thermal paste simply fills the gap where the metal doesnt come in contact with metal. Meanwhile, thermal pads simply block all the metal-metal contact.

Also, if its possible, Id be thrilled to see the Thermal Grizzly Conductonaut or similar liquid metal alternative, as its proven to give double-digit improvement in the delidding community


#3

This is something we could definitely try in the future but we would prefer to be on the safe side to make sure that all the people ordering V get consistently great performance! Also bear in mind that our cooling solution is anyways overkill and sustains way above 7W TDP :smiley:


#4

I had a discussion about thermal pads and their performance with Konstantinos quite some time ago. Konstantinos assured me that the V thermal pads are quite good quality with high watts per meter-kelvin (W/(m·K) values i.e. with good heat conductivity.

And from the teardown I learned that the thermal pads are pretty thin and get thinner due to the pressure between the CPU and heatsink for example.

It seems that the V has now three thermal pads - one between CPU and heatsink, one between Thunderbolt controller (Intel JHL6340) and heatsink; and one between the CPU on the motherboard reverse side and the heatpipe/casing.


#5

Hmm, thermal pads were used regularly just a couple of years back, especially in laptops where one heatsink usually covered multiple chips and pads being flexible the design was easy (you accounted for different chip height with different pad height) and the installation was fast and risk free (uneven pressure causing bad contact or chips get chipped if not having a metal heat spreader).
I hated it, as excessive heat stress did tend to cause the pads to dry out, but for an Y CPU it might be fine, besides do the Y CPUs have a metal heat spreader?


#6

Mobile CPUs dont have heatspreaders (IHS) for various reasons. Regarding the heat, while it is true that Y series produces less heat, but I dont think it really matters to the thermal pads, as the temperature still goes all the way up to the same, high point. Its not the heat, but the repeated changes of temperature that causes it to dry out. Correct me if Im wrong.


#7

Guy, let me kindly remind you that thermal paste dries out too.


#8

Thats true, but a dried out thermal paste would probably still work better than a brand new thermal pad. Correct me if Im wrong.


#9

I’m not so sure about that. Also, which would take longer to dry out? I think both questions depend on the quality of the paste/pad.


#10

Please create a “thermal paste vs thermal pad” thread. There’s no need to discuss this in threads about the V assembling because it can’t be changed already.


#11

You are probably (generally) correct. That said, if the hest spreader/heat sink is designed witha large gap (say 0.5 mm), then a thermal pad will probably work better.

Thermal grease works because it is thermally conductive when applied thinly. If it is applied too thick, it’s my understanding it does a really poor job and can risk shorting out the processor.

My guess -emdoor is more comfortable with pads, since it requires a less precise manufacturing process and gives reasonable results…especially with our huge coper heat sink.

I think Eve would probably benefit from a future optimization, but it’s clear that we have a really good (overall) design.


#12

You’re probably wrong, from my personal experience :smile: When my thermal paste dried out, my desktop CPU was constantly above 80C and IIRC it even started giving me BSODs. And i3-2100 isn’t exactly a powerhouse… The thing is that dried out thermal paste starts cracking and decaying. When I removed the heatsink it was like Great Canyon :smile:


#13

same experience here with dried thermal paste… three times already!


#14

Nowadays they make “noob-friendly” non-conductive thermal paste that doesn’t short anything.

But in the end, it all depends on what quality components you use. A very expensive thermal pad is in no doubt better than a cheap thermal paste and vice versa. So I guess the important thing here is price to quality ratio, because we don’t want to go overboard with buying super expensive paste. Since applying thermal paste is more difficult than a pad, I suspect the assembly would be more expensive. And Eve can use that money to get high quality thermal pads instead. Since we don’t know exact pricing, I don’t see much of a reason to guess.


#15

Gonna completely agree here. If we were struggling with the cooling system in any way, then I’d argue that we’d need to spend on getting quality control with high-quality thermal paste.

And by struggling, I don’t mean struggling with performance that OEMs generally struggle with.

Thermals affect a great deal. Battery performance, efficiency of the SoC ( based on my memory, the hotter a chip gets, the more efficiency you lose in power moving across the die), and finally performance in different workloads. It’s far more than simple performance that’s involved. It can affect consistency of the device in many respects under different workloads.

Our cooling system looks really overkill. Which is fantastic. I loved to see the thermal pads on the Thunderbolt controller, and CPU/heatsink + CPU/motherboard and heatpipe. We don’t just have good thermal pads, we have many, and massive heatsinks. This method will be more reliable with QC, give more reliable performance benchmarks, and probably cheaper long-run while still letting the V operate at cool temperatures across the board.


#16

I’ll correct you on that, as I’ve experienced it in person with a Dell laptop.

When the thermal paste dries out, it’s like not having any thermal paste at all. Temperatures skyrocket very quick and the CPU starts throttling even with low loads.

Still, I prefer thermal paste (on a device that can be opened for maintenance, of course).