It's not that often that a guy who once advertised himself (tongue firmly in cheek) as "the much renowned Electronic Genius" comes asking hardware help. I need your help.

The following is happening in Zoom. I can spend hours on a browser, & that's ok, but Zoom triggers misbehaviour after ~1 hour. My Samsung NP350C laptop dies instantly. That's the fault. Let me eliminate what I can.

• I don't think it's heat related, because behaviour there is an orderly shutdown, and this is disorderly - it just goes off. The one heat sensor is in the cpu.
• I don't think it's mains related. The battery is nearly useless, but it does run the box for a short while. So if it was mains related, I'd see an orange light while the battery was recharging after a poweroff. Instead, I see no light, followed by a green (= fully charged) light and I can power up again.
• For the reason just stated, I don't think it's an external power supply issue either. That power supply just provides a 19V feed. It's stepped down inside the box.
• It's unlikely to be plugs or dodgy wires. The laptop is a 17.3" thing, and I can't carry it with one hand. I use an external keyboard & mouse, and it sits motionless for the last few years.

That left me thinking maybe some internal heat-related hardware fault. I say heat-related, because if it happens, for example, after an hour on Zoom and I reconnect, I won't get another hour, but maybe just 20 minutes. Today when it went off the second time, the battery 'charged' light returned straight away, indicating a power supply interruption, I think. I don't know enough about laptop power supplies to be categorical. My guess is that it's in the stepping-down bit of the power supply (19V==> 5V, 3.3V, etc).

EDIT: This is no tragedy. It gives me an excuse to buy something decent!

Old hardware guy here. Just some thoughts.

I would remove the battery and see if the system will run any more reliabily without the battery. They draw current to re-charge, and that adds a little heat.

Has this system got a heat sink on the cpu? Many do not, and I havn't seen a lot online about this machine. If yes, whaen was the last time the paste was re-done?

You would have to open it up, however, if you have a good meter, or scope, check what the adapter is putting out.

The problem sounds load related, I dought zoom is to blame. If possible, can you monitor the CPU temp?
This may be caused by ageing parts, like caps.

Last thought, sometimes it isn't worth the investment in time to fix an older laptop. I sounds like a "luggable"; and you wouldn't be too sad to see it replaced.

How did Zoom Video Communications respond when you reported the bug in their software?

Or do you think it's acceptable for software to cause overheating, when using a browser (which presumably includes watching videos) doesn't?

Thanks for quick reply, camorri.

Yes, charging the battery is a draw. There's a BIOS setting to charge only to 80% and I have this set. On this one led, charging is orange, and I basically never use it on battery, so I never see it charging. It might gain me 5 minutes, no more. Personally, I think the battery points up the fault nicely, as it happens. Power has gone, but the battery has not been drained at all. That detail points at the stepdown circuitry for 5V & 3.3V in the laptop?

CPU problems can be excluded, since that issues "shutdown -h now", and my box dies instantly. I will try to monitor cpu temperatures next meeting.

Zoom is probably not to blame but it's cpu wasteful & gpu heavy on 2 windows with different moving displays. I see it drawing supply current more than overheating my cpu. There is a miserable cpu heatsink, but it's only a 35W APU cooled by a fairly gutless fan - typical cheap laptop. It does make noise once the cpu goes over 70º. 90º is critical, and I presume that's orderly shutdown time.

Cooling is near zero, but it usually handles cpu stresses. If the CPU & GPU get busy together, that's
a big issue for me.

I fully agree that it's not worth fixing old laptops, especially with only one hand. Hence my edit. I have a meter, but I can't safely handle two probes, or do basically anything.

I don't think it's a Zoom bug. This laptop worked fine on present versions. Now it shows a fault on Zoom. I'm sure I could dream up other workloads, like streaming 2 videos while compiling or making a big squashfs. It's just been happening on Zoom - that's why I mentioned it. Zoom is fairly cpu & gpu intensive, and therefore power intensive.

FWIW: I've observed this same 'instant off switch' behavior with Zoom running even in-browser, on a desktop machine with more than sufficient cooling and power. It (Zoom) does create a higher workload, but you won't get complaints from me about writing it off as junk-ware either - I'd say try something akin to your proposed scenario and see if it goes down, but unless you can specifically observe it overheating or similar I'm inclined to blame the junk-ware. If you're allowed choice, perhaps consider an alternative (like Jitsi or [I forget the name of Google's service]).

Are you saying previous versions worked fine on the same laptop, but now you've got a new version and the problem has started?

Ok, so do that and report back on the outcome...


Google Meet (formerly Hangouts). There's also BigBlueButton, GNU Jami, Microsoft Skype, Cisco Webex, etc.

Hangouts - yeah that was it.

And on FaceTime, I believe they announced its coming to Linux/others later this year via browser-based client.

In addition to what others said:
no doubt Zoom uses a lot of resources - CPU, RAM... - I'd try to reproduce the problem with any program that uses that much. I recommend stress for finetuning resource hogging.

No, I haven't changed the software at all since last April. The problem I have is for about 3 weeks.

(On possible alternative workloads to stress the thing.)
Right, I'll try and cobble up an equivalent workload to get the current problem another way.

Very Interesting, because it lets me know I'm not the only one out there. A two hour meeting knocks 30-50% battery from a phone or tablet, which is a considerable chunk in 2 hours. The instant-off has to be the internal psu, as I can't think of anything else to do that, On a box, it could be mains, but with a laptop + battery it has to be the internal psu. I wonder is there an (undocumented) thermal cutout in that?

The meetings go out on Zoom. There's not a choice.

Got some stats on the offending box:

• I have a twin core box, where most applications max out @200%, but the odd one (e.g. mksquashfs) does up to 400%. Zoom sits at 110-140% mainly, but drops off to 60% when I drop to an agetty tty.
• Zoom runs 2 windows which appear on my 2 screens and are not minimized. One is the speaker, one is gallery view. I have no practical way of measuring current. My guess is that the load is borderline for the box, which has been going since early 2013.
• CPU temperature goes quickly to 80-83º (critical=90º) but the cooling appears to hold it there. It could go critical after an hour, I suppose. There was only a 15 minute meeting today. My fan goes to full speed over 70º. Anyhow, hitting critical issues a 'shutdown -h,' whereas my power drops without discharging the 19V battery. That makes me suspect the 5V/3.3V psu.

So, significant cpu & gpu loads are applied. I'm sure the bigger the window, the heavier the load. There is no power supply cooling at all - the one fan cools the APU. The load would probably generate significant current in the APU.

One other thing: there's been no maintenance in the way of dusting, replacing thermal paste, vacuuming, etc, because since 2015 it simply hasn't been practical for me to open the thing.

I can't minimize one window, e.g. the speaker window; If I do that, everything Zoom related minimises. I can mess with the 'participants' window.

Right - the heat issue:

CPU rarely goes below 60º when warm. I put this line in one terminal and ran it from a kernel top source dir I ran top in another terminal, and sensors in yet another.

Then I ran the kernel compile, ran one fullscreen video in waterfox, and another in vlc. I gave it an hour. I got up to 80º quickly, but didn't get past 88º. top reported a ≤200% load most of the time, heavier than zoom for certain. But it never hit critical (90º). I had a distro kernel config making all the modules, not my usual lightweight kernels. Leaving the compiles going with no videos running got me up to 87º, so I can conclude my sucky Intel HD4000 graphics doesn't suck power That pours cold water on my heavy current theory

Making a large squashfs would thermally trip it all right, but I managed that already by blocking a cooling vent and it shuts down. Mksquashfs gets me ≤400% cpu load, and will definitely cause it to boil over. But there's no good reason to do that.

So, it it a Zoom crash, or is it a hardware thing? And why has Zoom's various iterations worked without a fault since March 2020, and now is throwing me out every meeting? That includes,btw the present version working since April.

EDIT: Ambient rarely varies much - 5º to 15º ten months of the year; it's showing 16º now with highs of 20º some time today. So it's not a factor.

Could also be the GPU; not sure how to test GPU overload though.

I went sitting in X with 2 different fullscreen videos running on my 2 screens. That was the only way I could think of, short of setting up some gpu sweater. My box is vulnerable enough on heat that I will overheat it today, to ensure it does issue the shutdown -r and not just die. But I'd have to hand it in somewhere, a big deal for me. There's 3D tests you can run online, which would probably be good.