C7 vs C8 temperatures and load - What are you seeing?

Anyone logging your hub temperatures, did you see a pretty dramatic increase in the new hub vs the old?

I'm curious if I have something going on, or are others seeing the same thing and it's probably expected.

My new c8 runs a solid 20 degrees hotter or at least reports that than the C7 did. All apps and devices are the same hub in the same exact location.

You can see the point where I migrated in the temp graph.

image

I do see the cpu load is slightly higher with the new hub too on average starting at the same point in time.

image

1 Like

Are you running Wifi or Ethernet?

1 Like

Ethernet

My gut feel would be to start by looking at the device / app stats to try and understand which might be contributing to the elevated load. But on the surface it would seem odd to have such as dramatic change in temps.

I'm going to guess you hadn't setup prioritised QAS for your old C-7, without having set it up myself either... But worth mentioning in case you had.

At a glance, nothing in device/app stats stands out. Granted I can't compare to what I saw on the C7, but everything looks about the same.

I'm not even sure what QAS is. :slight_smile:

That answers that question :-). My understanding is that you can prioritise network traffic either by device or by service, e.g. streaming video, etc. Never set it up myself...

I'd try a reboot if you haven't done one recently, just to reset everything... Beyond that it's a fact finding mission to work out what is causing the additional load.

Ohh QOS. No, I haven't done anything like that.

I've been rebooting a lot lately, I'm going down that rabbit hole of zigbee issues on C8 now. I was more curious if the load / temp thing was being seen by anyone else.

It's been a constant since the day I migrated, so my assumption was the new hub had a new background process running, or maybe even just a different chipset that reports temps.

My bad, QAS = QOS.... Not sure what I was thinking there :slight_smile:

2 Likes

With my C7 and now C8 in exactly the same location, same environment, same apps, etc., the C7 consistently ran 84°F - 86°F. The C8 always runs 104°F - 106°F.

1 Like

Maybe it is worth bringing @thebearmay back into the conversation (and maybe gopher), about whether the temp reading could be taken from a different reading in the hub? My understanding (from other linux-related setups) is that the temp readings can be taken from different parts of the board like CPU and memory, as an example. Could the HE endpoint be taking the reading from a different location now in the C-8? Or event between platform versions?

1 Like

Okay, so you are seeing the same thing I am..

That's actually a little bit of a relief. :slight_smile:

1 Like

Mine was running 12-14 degrees hotter as well

I put thia fan (very quiet) under the c8. But did remove the top metal grate and replaced it with this plastic one. The fan is rated at 18db at low speed. Dropped.my temps by 14 degrees into the range of my c7. Cpu seems to be the same as my c7.

https://www.amazon.com/dp/B00G05A2MU?psc=1&ref=ppx_yo2ov_dt_b_product_details

https://www.amazon.com/dp/B09NV4JH1S?psc=1&ref=ppx_yo2ov_dt_b_product_details

2 Likes

I am seeing something similar as well. My hub has gone from staying in the mid to lower 80's to the mid to upper 90's.

That said even 110 degrees F isn't really bad for electronics. Is lower always better for the most part yes, but these devices are designed to run allot hotter then 110 degrees as well. The old C4 use to run between 160 and 180 degrees F.

1 Like

That's definitely true, my C-4 definitely ran / runs much hotter than my C-7 / C-8 even with minimal integrations running. 20 C difference, if not 30+C.

1 Like

Im more of a noob for Hubitat stuff beyond my own setup. How do you log and monitor the temperature?

Where is the sensor providing the reading?

170F isnt even 80°C. No clue what process the various ICs are on, but if thats on-die that doesnt seem generally bad.

I opened my C7 and noticed none of the ICs had heatsinks. If we could see the waveform patterns, maybe from FCC filings, or of the Hubitat Board itself, we couks deduce which ICs could get a thermal pad or paste and a small heatsink. Looks like maybe flash on the bottom, wireless within the large surface tracing on the board, and processor(s) far from those.

Longer screws with some sleeve/standoff to stilt the unit could be a good first step.

Monitor:

The endpoints can be found in the code for that driver. Many who graph or log use WebCORE to do it.

This takes some work to set up, but gets you pretty graphs that are easy to customize...use the "cloud" version of the setup if you go this way.