Paul wrote:

SC Tom wrote:
Recently (within the last couple of months), I have been having
sporadic, spontaneous shutdowns or reboots. The PC is only about 18
months old, and I have made no changes to it in the last year or so.
When it shuts down, I can restart it without waiting; it starts up fine
and immediately.
There are no entries in the event log other than for unexpected shutdown.
Core Temp shows no unusually high temps; it has never logged anything
closer than 35°C to TjMax.
The inside and vents are clean- no cat hair or dust. The fans are all
clean and running well.
CHKDSK shows nothing wrong.
MEMTEST86+ returns no problems.

My specs a
Lenovo H50-55
AMD A10-7800 Radeon R7 (12 Compute Cores 4C+8G) 3.50 GHz
12GB Ram
2Tb C: drive (came with the PC)
1Tb D: drive (used for data and movies, etc.)
SATA DVD drive
Nvidia GeForce GTX 750 video card
Corsair CX Series 600 Watt 80 Plus Bronze Certified Modular Power Supply
(replaced the original 180W PSU to power the video card- wasn't
impressed with the GPU graphics)
Logitech M705 mouse
MS SideWinder X4 keyboard

Windows 10 Home x64 V.1709 (Build 16299.125)

I'm getting ready to tear it down and check for puffed-up caps on the MB
and PSU, give it an extreme cleaning, and pull the heatsink and apply
new paste (replaced it about a year ago). Will reseat all cards, drive
connectors, and memory sticks. Might as well replace the CMOS battery
while I have it apart.

Anything else I can check while I have it apart?

Thanks for any suggestions :-)

A Google search suggests you should swap out the CX 600 and
test with a different PSU.

I've had a Corsair CX 600W in my salvaged PC for just over 4 years. No
one has absolutely no failures on their products, though. Corsair
doesn't build anything. They spec to the manufacture (Channel Well
Technologies aka CWT whose is the world's largest PSU maker and
contracts with numerous brands to build their PSUs). Corsair then slaps
on their "Corsair" sticker. Even FSP (aka Sparkle/Fortron) who builds
their own PSUs and also sells under the EVGA, Antec, OCZ,
Silverstone, Thermaltake, Nexus, and Zalman (maybe more) brands has had
some blooper models.

What I've seen of users of the Corsair CX600 where they complained one
unit failed, they replaced it with the same model, and that eventually
failed hints the problem was never with the PSU. "I used a spring-
loaded punch on a car's side window and it shattered. Replaced the
glass but the spring-loaded punch still breaks the window." The problem
wasn't with the window. Some users will report they see the fans
jiggle, the PSU comes up for a second or two, but then shuts off. That
could be a fan isn't spinning or the wrong type got used, like 3-pin on
a 4-pin mobo header, so the BIOS immediately shutdown the PSU because
it's prevent what it sees as a non-spinning CPU fan from causing the CPU
to burn up. I've seen fans that cease to send their RPM data so the
BIOS sees them as non-spinning. One wouldn't send RPM data except at
full RPM. On startup, the fans spin at max RPM and then get slowed to
reduce noise. Some users didn't connect the 8-pin cable to their video
card (because the foils in the mobo cannot handle the amperage). Some
mobos won't power up (shutdown the PSU) if there isn't a load beyond
some threshold, like 1 HDD + 2 fans or 2 HDDs or 1 HDD + 1 video card.
The load on the PSU is too light so the mobo (or is it the PSU?) thinks
there is a some problem and shuts down the PSU. Stable operation means
the PSU must experience a minimum load threshold. I've seen this min
threshold problem reported for Corsair PSUs; however, since I always
have 2, or more, HDDs (even when I added the SSD) along with using a
daughtercard video card instead of the dismal onboard graphics, I've not
encountered light-load shutdown problems with the Corsair PSU.

Can't really tell from the OP's specs if a 600W, any brand, would be
sufficient for the PC under full load of all components. While there's
the surge current on cold startup, the OP says his PC starts up okay.
Who knows what all he is running at the time of the shutdown. While
probably a bit overly safe, I get a PSU that handles twice the load with
the mobo fully populated (all connectors used to attach as many hard
drives, memory modules, dual video cards in SLI config, all daughtercard
slots filled, etc). PSUs wan in capacity over time so I want to start
big to make sure there is still plenty of over capacity 6-9 years later.
Maybe 600W really isn't big enough for his setup. A wattmeter (e.g.,
Killa-Watt) would show how much is going through the power cord, divide
by the PSU's efficiency rating (which is really on a benchmark spec on
the design and somewhat applicable to a new unit) to see how much power
his setup is consuming, then make sure there is plenty of reserve
capacity. Has he added any internally-powered components since
acquiring that pre-built?

I don't remember what Core Temp shows. The OP only mentions one
temperature which might be the overall or case temperature. Maybe he
needs to look at the individual core temps (there are 4 cores, not the
12 "compute cores" he mentions that AMDs spews because the other 8 come
from the GPU). I've found that my core temps for my Intel quad core are
not equal plus only 1 or 2 are actually inuse most of the time with only
short spikes in the other 2 cores. One core always ran hotter,
concerned me for awhile, but it's well within spec and seems to be its
natural behavior (that non-overclocked CPU is over 4 years old now).
TjMax means Tjunction temperature which is core temperature but the OP
lists only one. Maybe that's the max of all core temperatures. Looks
like his AMD CPU is rated for a TjMax of 71 C but he's only at 35 C
(which seems the unloaded running temperature so he doesn't use his
computer for anything but just leaves it powered on with only the OS
using the CPU?). Does Core Temp provide a chart showing history of
temperatures for all cores? If so, I'd watch that to see if there was a
spike in one, or more, core's temperature(s). I've never used the log
function in Speedfan to know it will save a history of readings into a
log file. Maybe Core Temp has a logging function. Once TjMax is hit,
the processor's THERMTRIP# signal will activate a shutdown. I don't use
AMD CPUs to know how accurate is their junction temperature sensors. I
don't trust any of them to be more accurate than ±5 C of the real
temperature.