Bought a used HP Elitedesk 800 G1 SFF (basically a small form factor corporate desktop) over the summer with the intent of using it as the DVR for a set of IP cams.

It's got an i5-5490 3.3GHz CPU, 4 GB RAM and an Intel HD Graphics 4600 display adapter.

I powered it on initially when examining it, had no issues that I could tell, powered on fine, no BIOS errors, booted to Win 7 desktop.

It sat in the box for several months while I prepared to move.  After moving, I unpacked it, took the original HDD out, put a 4TB HDD in and installed Windows 10 and Blue Iris camera manager software on it. No issues with the OS/software install.

Set it up with one camera and left it on overnight.  Came back the next morning to find it powered down and with the power switch LED flashing a sequence of 4 flashes.  There was still some system power as the power LED was flashing, the ethernet port light was on and the keyboard num, caps and scroll lock lights were all on.  Wouldn't power on again until I unplugged the power cord from the back of the power supply and plugged it back in again.  After doing that, it booted back up without issue, other than recognizing it had not been shutdown correctly previously

Did some googling, found a page that suggested that sequence indicated a power supply problem.  Bought a replacement power supply, hooked it up and left the PC running overnight again.  Next morning, same thing, powered off, 4 flashes.

I installed the test software HeavyLoad which supposedly maxes out the CPU and GPU to stress the system.  Running a HeavyLoad test will crash the system in anywhere between 2 and 10 minutes - the display goes blank, it sounds like the CPU fan briefly goes to max RPM, and the power LED flashes 4 times (along with 4 corresponding beeps that repeat just twice).  There are no error messages or blue screen, it's just suddenly lights out.  The CPU cooler is barely warm to the touch at this point, so I don't think it's an overheating issue.

What else should I be looking at?  I can pick up a replacement motherboard for around $35 - that will be my next step in the absence of anything else to try.

Sounds like an overheating issue.  That's the only thing that would cause it to power down.

CPU Cooler may not be getting hot, but the CPU may be - you may need to clean/reapply thermal paste.

There are a bunch of things that can cause a system to shut down.  I'm not familiar with 10 but I'm sure it has Event Viewer.  That's where I'd start.  From what you have told me though, it does sound hardware related.  I'd look at the heatsink and make sure it's seated properly with a compound.  Most computers will shut down before the CPU fries itself.  Bad power supplies can also go lights out without warning. 

EDIT:  Confirmed.  HP error codes, 4 blinks is a thermal issue.

45 minutes ago, Krdshrk said:

Sounds like an overheating issue.  That's the only thing that would cause it to power down.

CPU Cooler may not be getting hot, but the CPU may be - you may need to clean/reapply thermal paste.

This first - I had a similar issue on an older machine as a file server here - the paste if you can believe it dried out over the years and was non conductive.

It's probably overheating like mentioned. Check that the cooling fins are clear of dust and the fans are operating.

Could be anything from a thermal event to bad capacitors. also check the mainboard and any pci cards you have for leaking or swollen capacitors. they don't have to be leaking to be bad, swollen is just as bad as leakers...

Initial thoughts were overheating, psu, or bad capacitor.

Look for cpuid (i believe) HWmonitor and install, see if anything is spiking to unsafe levels or if any fans are at 0rpm that should be spinning. If you look up the cpu, you can find thermal specs from intel.

Not sure of what options are in 10, but control panel > power options - take a look it isnt set to any stupid settings.

Check that there isnt a cable in any fans, or a covering of dust blocking any heatsinks.

The above pics are helpful. Look for that, but also look if any caps are at an angle other than perpendicular to mobo. Sometimes they go from the bottom instead. Anything like that is gonna cause problems.

Update:

Reading elsewhere about some things to try, I removed and re-seated the RAM, then started it back up and ran Heavyload again.

Heavyload has now been running for over 2 hours, still running ok.  This is the first time it's run for more than 10 minutes.  Going to keep it running for a couple of days to see if it powers off again.

Got my infrared thermometer out of the garage - it's showing the CPU cooler heat pipes at 104 deg F, motherboard near the CPU socket is 110, Software monitor is showing CPU at 140.  Don't think overheating is/was the issue.

These HP Elitedesk 800's were introduced ~2014 - it should be clear of the "capacitor plague window".  However I did check all the visible "round" caps, all look sealed/flat on the ends.

CPU fan is spinning, and generally the interior of the case is pristine, no dust/lint buildup I can see anywhere.

Windows event viewer has a bunch of critical events about recognizing that it rebooted after failing to shutdown, but there is nothing I can find where it logs anything before the "abnormal" shutdown.

Well, that is promising, and I hope it is as simple as reseating RAM.

As for caps, almost any of that style can go bad under the right (wrong) conditions, though specific years and batches have been far worse than most.

Spontaneously powered off again at about 5 hours into the test.  Restarted everything, and then it did it again about 5 minutes into the next test.

Going to try a replacement motherboard now.

+ 10 on overheated CPU. Verify that thermal paste is still exist on the heatsink and it’s fully attached to the CPU. Also, once overheated it may start snowball effect to other components on the MB


Sent from my iPad using Tapatalk

CPU never went above 145 deg F during 5 hour 100% CPU utilization test.  I am very skeptical it is overheating.

Before you go through all the trouble of swapping out mobos, go for the low hanging fruit.  The power led code (4 blinkie blinks)  is telling you it's a thermal issue.  btw, onboard thermometers are not exactly precision instruments.  Just pull the heat sink, scrape it and the cpu,  and put a dab o' fresh thermal past on there.  That's a whole lot less work, is a logical step in diagnostics and is exactly what I'd do based on the 15 years experience I had as as a computer tech.

Will try new thermal paste as soon as time allows.

May I ask where you were able to find info that 4 beeps/flashes = heat issue for HP’s?

12 hours ago, Scorpio64 said:

Before you go through all the trouble of swapping out mobos, go for the low hanging fruit.  The power led code (4 blinkie blinks)  is telling you it's a thermal issue.  btw, onboard thermometers are not exactly precision instruments.  Just pull the heat sink, scrape it and the cpu,  and put a dab o' fresh thermal past on there.  That's a whole lot less work, is a logical step in diagnostics and is exactly what I'd do based on the 15 years experience I had as as a computer tech.

 

Exactly what I would do and I've been putting PCs together since the late 80s.  Measuring the cooler temp doesn't give you a great indication of what the CPU is doing, specially if it's not transferring heat due to lousy thermal conduction.

14 minutes ago, DirtyDigz said:

Will try new thermal paste as soon as time allows.

May I ask where you were able to find info that 4 beeps/flashes = heat issue for HP’s?

google --->  HP support site.

https://support.hp.com/us-en/document/bph07107#AbT1

That's alot of electricity for that function.  Get a new Pi they run windows or just download DVR software for Linux.  The cost of Pi is cheap and way cheaper yet to run.  

1 hour ago, Scorpio64 said:

google --->  HP support site.

https://support.hp.com/us-en/document/bph07107#AbT1

Yeah, that’s where I ended up too.  

I believe this model was manufactured in 2014, so you have to go to the second section for pre-2015, where 4 beeps/flashes has a different meaning.

38 minutes ago, Rob0115 said:

That's alot of electricity for that function.  Get a new Pi they run windows or just download DVR software for Linux.  The cost of Pi is cheap and way cheaper yet to run.  

I have a Pi 3 that I play around with, but I'm skeptical that it's up to recording 6 1080p cameras, doing all the motion/event analysis, and re-broadcasting the simultaneous feeds to other devices.

2 hours ago, DirtyDigz said:

Yeah, that’s where I ended up too.  

I believe this model was manufactured in 2014, so you have to go to the second section for pre-2015, where 4 beeps/flashes has a different meaning.

Before I looked up the HP info I had a feeling it was either the CPU overheating or the graphics card, which is probably on board.    When I went to the HP site, I just guessed the age of the machine based on specs.  Still, both error code designations fit my initial gut feeling.  If you happen to have an old graphics card laying around in a drawer, or one you could borrow out of anther PC just to diagnose your machine, that would be super.  So, if you could just go ahead and uhh, disable the on-board graphics and throw a graphics card into an expansion slot, that'd be greaaaaat,  Soooo, yeah, let's do that and you submit your TPS report so we can see the result.  M'kay.

1 hour ago, DirtyDigz said:

I have a Pi 3 that I play around with, but I'm skeptical that it's up to recording 6 1080p cameras, doing all the motion/event analysis, and re-broadcasting the simultaneous feeds to other devices.

For what it's worth, I played around with a Blue Iris recorder for a while at work.  It worked great with 2 cams and fell flat on its face with 8.  I ended up buying an NVR for that set of cams.  It's been running for several years now without issue. 

FYI when dealing with cameras - know one thing - storage is the small piece of the puzzle

 

On 12/21/2017 at 4:54 PM, USRifle30Cal said:

FYI when dealing with cameras - know one thing - storage is the small piece of the puzzle

 

What's a bigger piece?

Per Malusa: if thermal conduction between the CPU and the heat sink is under performing, or non existent, your cooling fins will be... cool!

Try the paste route first.

Update:

- Removed CPU cooler from CPU, heat sink compound looked healthy and still pliable.  Wiped it all off anyway, cleaned with rubbing alcohol, added new heat sink compound, reassembled.  No effective change in temps or behaviors - still spontaneously powered off anywhere between ~5 - 10 minutes.

- Don't have a spare video card to run instead of onboard graphics. 

- Found a "bare bones" version (motherboard/case/power supply) of same PC on Ebay for $50 , bought it.  It arrived yesterday.  Swapped CPU, Cooler, RAM and HDD into the bare bones system.  Reset CMOS and powered on.  Other than alerting to a change in installed memory there were no issues, booted right up into Windows 10.

Been running the Heavyload test for 21 hours now, no hiccups that I can see.  Will keep running it for 2 more days. Max CPU temp so far is 147 deg F.

Every time I had that same issue in past, I just replaced ps with a new higher wattage unit.
It was never anything else.


Sent from my iPhone using Tapatalk

2 hours ago, DirtyDigz said:

What's a bigger piece?

Processing, video is a hardware hog.

While archival storage is all most worry about and rightly so, horsepower for video processing is the most important .

 

If the chipset motherboard ram bus etc are not optimized for video processing.  You will have bottlenecks and issues.

 

There is and stark difference using a machine has can be used as a file server and one that is setup for video processing.

 

Also the minimum requiremet most VMS software solutions show, are the bare ones minimum to barely run. Always run recommended specs. ....

1 hour ago, USRifle30Cal said:

Processing, video is a hardware hog.

While archival storage is all most worry about and rightly so, horsepower for video processing is the most important .

 

If the chipset motherboard ram bus etc are not optimized for video processing.  You will have bottlenecks and issues.

 

There is and stark difference using a machine has can be used as a file server and one that is setup for video processing.

 

Also the minimum requiremet most VMS software solutions show, are the bare ones minimum to barely run. Always run recommended specs. ....

Roger that - I didn't pick this HP PC at random, researched it at ipcamtalk forum - Consensus was that it was a good for up to ~16 cams with blue iris.

Hope the new mobo solved the issue for ya.

What do you mean by alerted to change in installed memory?  Does the mobo not handle as much memory?  Just curious.

 

And it's great how far video processing has come.  I remember the fist video I ever tried to edit in Premiere.  Was a short ~4 minute 640x480 video.  Render time on a Pentium II 266 was somewhere around 24 hours.  And then the video didn't work anyway.  Wonder how many seconds it would take these days on even my wimpiest of systems...  First video stream recording (live TV) I did was on a 500MHz system, and any kind of fast motion caused the video capture to stumble and hang up for a second or two because it was so close to the edge of unusable - long before hardware acceleration of encoding on video cards.

I’m assuming the motherboard had a different amount of ram in the bios confit than it does now. Thus it told him during POST that there is a different amount of memory.  It probably update and moved on or he had to manually set it in bios setup. Not an issue.