Hi- I had my c4 hub mounted up with some velcro, and in pulling it off the wall, I accidentally pulled the bottom of the device off too. There's a metal plate in there that appears to have been attached to the plastic bottom of the hub, and block of purple foam like stuff.
What function does this plate serve? Is it a heat sink? The purple stuff doesn't look like any heat transfer compound I've ever seen before. I'm asking because I kind of broke it off when the bottom came off. The device seems to run without the plate and foam, and hasn't overheated or anything yet, but the days are getting warmer... thx.
I have a C-4 that I use as a Development hub and I just now popped the back off. The blue cube stayed behind, in my case. The temp before this, was 65C. In order to get some load on the CPU, I'm doing an upgrade to the latest. I'll let you know if I decide that it's time to put the lid back on.
However, it was pointed out to me that the metal plate is magnetic and thus is not a good heat sink. it's shape is pretty poor in that regard as well. It certainly would not hurt heat dissipation, but it's not helping much either.
Update: I took it a step further.. I popped that metal plate off and put the lid back without the blue cube or the steel plate. The temp was 75C after the reboot, but has dropped 10 min later to 62C.
The pad is 0.2" tall and 0.48" square. It fills the gap between the plate and the CPU, where the cpu puts a small indent in the surface. Indicating it's filling, but not providing a significant pressure.
I believe it is a simple weight so the wires don't have so much control of the box. Being steel of some sort and the foam pad being so thick I would guess it hinders cooling (circulation) more than helps.
I put my hand on the C-4 and at 71C reported, I'm barely able to feel it being warm... so it's sitting a degree above hand skin temp.
It's fractionally heavier than the C-7 that it's sitting next to, and I'm guessing that's because of all the USB A, HDMI and mem slot hardware. And the C-7 is really very near to the feel of the C-4... skin temp.
Most likely to be the case. The foam block looks like it's there to prevent the pcb from moving as there seem to be no screws or catches holding it down.
There's at least two screws holding the PCB to the larger portion of the plastic case. The 'base' perhaps?? If the smaller piece that pops off is the 'lid' then what remains is the base
Ah - couldn't see the screws in the original photo (which I thought would have been odd). In that case the block is probably supposed to be a thermal pad (which would be of little value given its thickness). I see thermal (gap filler) pads can be quite thick compared to what we are used to seeing - the important point being some amount of compression is required which is what we see in the photos.
In 2018 when I got my first Hubitat (C-3) there was the usual, expected, conversations about the internals and someone went to the trouble of detailing most of the specs... CPU, Mem, there's a HDMI port so that was discussed etc. The blue foam block and steel plate was covered too.. but I remember none of the details.. but there was supposition it was a specific brand of thermal pad. From then on, I've had it in my head "heat dissipation"... up til last week when someone said the metal was steel. Thump on the forehead because, yea, it's way heavier than aluminium would be.
Just checked my C3 and its running 59C at 22.8C ambient. As I recall, the datasheet for the Amlogic S905 specifies max of 125C Tj (temp sensor is integrated on the die), but who knows what their assumptions are for longevity... lower is always better.
Seeing that photo I'm tempted to dissassemble the cover/heat sink and drill a few reasonably sized holes in the cover. There is very little provision for airflow in that tiny box.
For what it is worth, variation in temperature causes failure, not the absolute temperature.
Temperature is a combination of ambient temp and computing load, so adding better cooling will extend the life. However I think the unit will be obsolete long before it fails.
Well, it has been more than twenty years since I last ran a failure rate projection for the mainframe designs I worked on. I just plugged numbers into a tool (the product of the RAS and thermal analysis gurus of my former employer). Ambient temperature was a key input, as was thermal cycling; but the systems I worked on typically didn't get turned off very often.
The physics of failure modeling back in the '90s was rooted in MIL-HDBK-217 and the classic bathtub curve with its ELF, IFR and wearout phases. It was a bit more advanced than the simplistic '10°C Increase in temperature reduces the life of electronics by half' trope of the Arrhenius equation. Still, in manufacturing it was routine to induce early life failures in key components (like SRAM caches) with elevated temperature stress at the factory-- we literally had fixtures called burn-in ovens for this.
Nowadays there's a discipline of reliability physics considering a dozen or more degradation mechanisms to address the root causes of failure in semiconductors and how it manifests in actual use. But old habits die hard; I still like to keep my electronics cool... drilling a few holes might make me feel better, though I agree with your take on obsolescence.
And the C-7 is really very near to the feel of the C-4... skin temp.
