Strange shutoff problem with Optiplex 745 Ultra Small Form Factor - RAM tests fine, 2GB shuts off under load, 1GB seems fine

Hi Kentchristopher,

Did you try a memory test outside of Windows, such as with the Dell diagnostics or a bootable memtest flash drive? I'd try that.

I did, both with the Dell bootable diagnostics and with memtest86 (on the Redo Backup bootable USB). It passed on both. And again, I also tried swapping out both DIMMs completely with another working pair.

Overheating?

• Faulty CPU fan?
• Failing heatsink ?
• CPU thermal paste dried out?

I ran a temp monitor and then put the CPU under load using Orthos stress test. Neither core ever went above 70C and the system didn't shutoff during the test. 

I'm actually having this problem with 2 different system: both 745 USSF systems. Both had their system battery go dead while in storage so I'm starting to think it's somehow connected to that (and based on the few other related reports I've found on the internet). Still strange.

Thanks to those of you who have given some input already. I just did some further tests to probe this mystery:

- I tried 3 different power supplies (these being Ultra Small Form Factor with the external power). No difference.

- I installed a fresh system battery (Sony CR2023 good 'til 2023). No difference.

- The side panel was open for all these tests (which should have helped if heat was a factor).

I've been able to consistently produce the shutoff (screen goes black, fan revs up high) by running Vista's Windows Experience Index assessment. The results:

- with 2GB RAM, shutoff occurs within 10 seconds (during the Aero assessment phase of the test) -- FAIL

- with 1GB (1st stick from the above setup) it makes it through the entire assessment -- PASS

- with 1GB (2nd stick from the above setup in first slot) it makes it through the entire assessment -- PASS

- with 1.5GB (adding 512MB DIMM) it makes it much longer into the assessment but still eventually shuts off -- FAIL

- with 1GB (2 x 512MB) it can sometimes make it through the entire assessment, but not all the time -- FAIL

The mystery continues... any suggestions appreciated.

If the motherboard batteries died and you've replaced them, BIOS was reset to the factory default settings. Do you know that the defaults are actually the correct settings for the hardware installed in these systems..??

Thanks for the input, Ron. I've tried manually resetting the BIOS to factory defaults and then running the WEI assessment. Still shuts off. I've also tried it with minor BIOS adjustments, namely turning on SpeedStep which is disabled by default. Still shuts off.

Today I applied new thermal paste to the CPU in hopes that it is in fact a heat problem. Booted up and ran the assessment with 2GB and thought I was getting somewhere -- it passed the test without shutting off! Then I ran it again: fail. Rebooted, failed again. This does lead me to think that it could be a heat issue (since the machine was cold on the first test from sitting overnight), which is supported by the fact that the fans rev up very high when the shutdown occurs.

The problems with this overheating theory are:

1. The CPU heatsink never feels hot to the touch when a shutoff occurs. Nor the hard drive. Both CPU fans are functioning fine. 

2. Why would the machine overheat with 2GB (2 DIMMs) of RAM but not with 1GB (1 DIMM)? 

3. Why doesn't the machine shutoff when I stress the system under Linux Mint 13 (Xfce) running a GL benchmark test and playing a 720p YouTube video in the background (with 2GB RAM installed)?

This suggests it might be a driver or OS issue. Out of curiosity I loaded Win 7 Pro 32-bit (instead of Vista), used the included drivers (all devices found) and ran the WEI assessment. Sure enough, it shutdown within a few seconds. 

Any other suggestions appreciated. This is truly one of the most baffling computer problems I've ever encountered. I don't need the machine -- but I am curious as to what's actually going wrong.

Twice as much RAM probably equals twice as much heat. USFF systems frequently have heat issues.

Look in Windows Event Viewer around the time of a shutdown and see if there are any error messages, like "Thermal Event" which would mean an overheating problem and for other error messages.

It might be the video card overheating when there's extra RAM.

Since it shuts down after fresh installs of several versions of Windows and your OP says it shut down with Mint, that still makes me think it's hardware.

Have you tested the hard drive? Boot from the Dell Resources (Utilities and Drivers) CD and run the diagnostics on the hardware. See if it either reports any HDD errors or shuts off during the testing.

I suppose it could be a motherboard problem with the memory controller and it can't manage 2GB of RAM so it just crashes.

After a crash, does it power off completely or just go to a blank screen? What color is power button after the crash (if it's on) and is it steady or blinking? Could this be a power supply problem and it can't take the drain with 2 GB of RAM but 'manages' with only 1 GB?

Photo with the heatsink covering what I think is the integrated GPU outlined in red, in case anyone can confirm this.

No "Thermal Event" or any other error in the Windows logs at the time of shutdown. Errors only come up on reboot because it wasn't shutdown cleanly. When it crashes, the screen goes blank with a "Cannot Display This Video Mode" message and the fans slowly rev up to maximum speed over the course of about 20 seconds. This may be the best evidence that it's a heating issue; a fail-safe to protect the CPU. The power light stays solid green and none of the numbered error LEDs (1, 2, 3, 4) light up on the back.

Just booted the Dell diagnostics again and ran both the memory test and hard drive tests. All passed. Ran all other tests while I was at it: all passed.

I don't think it's a power supply issue since I've tried 3 different power supplies and it crashes with all of them. 

So far I'm inclined to think this is a heating issue related to the integrated GPU. It can't be the CPU because I can run Orthos stress test and regardless of the test I run (CPU focused vs. some CPU/some RAM), the system doesn't crash and the CPU core temps never go above 73C. I've dealt with systems shutting down from CPUs overheating before and it only happened when the core temps approached 100C. 

There is a heatsink mounted on a chip right next to the RAM slots (which could explain the RAM connection) which I am guessing is the Intel integrated GPU. I removed this heatsink and the thermal adhesive tape, applied some Arctic Silver 2 paste and remounted it. For a moment I thought I had finally found the solution, as the Windows 7 WEI assessment completed without a crash. On the 2nd try though, sure enough, it crashed. 

If it is related to the GPU, it baffles me that I can play a 1080p YouTube video in linux while running an OpenGL benchmark in the background with no problems. Maybe this is a combination of heat and poorly implemented drivers for the Intel GPU in Windows Vista and 7?

Not sure what that is. And it's not identified on the motherboard drawing. Might be a heatsink on the chipset chips.

Do you have the video driver set to the correct ("native") resolution for the monitor you're using? Did the monitor come with its own driver CD and do you need to (re-)install the monitor driver? Are you connecting via DVi or DVI>VGA on the monitor? Can you test a different monitor on this PC?

What happens if you boot in Safe Mode (reboot and press F8).

On some systems, when the video driver fails to load, is corrupted or crashes, the fans rev up but it's not a heat issue.

Well, that's a little bit of progress...

Reboot and press F12 before Windows starts to load, assuming you can keep it running or with a fan blowing on that heatsink. Go to Diagnostics partition and run the fan tests. I'm wondering if the sensor in one of the fans failed so it's not speeding up when things start to get too warm so airflow isn't able to keep things cool.

If  a fan fails, you'll have to replace it with another Dell OEM fan because you need to get one with a sensor.

Is the heatsink just a solid block of metal? I wonder if it failed and/or if there's a later revision of that heatsink that's more effective.

Check the air vents on the tower to make sure none of them are clogged up.

Any room to add another case fan? Since you don't have any PCI slots you can't install a PCI fan card...

Oh, and search these forums to see if somebody else has a solution to heat issues in this OptiPlex USFF model :emotion-55:

Me again...

Read this thread. Don't know if there's room in the USFF for a Dell NY290 Hard Drive Blower Fan with 5-Pin Connector that's mentioned in that thread. Or if any of those BIOS settings changes or CPU replacement will help.

BTW: Is there an air sensor in the USFF? I wonder if that failed, got disconnected from the motherboard, or got dislodged from it's "normal" position so it's not sensing things correctly?

There's still hope...!

I also have the exact  same issue and look for the solution to the issue. I have the exact same model. I am quite confident it's display driver. Once I disable the display driver and use native window driver, mine passes all tests and everything works fine. My original suspect was memory and I have tried 2 and 3 GB which didn't make any differences.

Regarding the display driver, I have tried XP, Vista 32 bit driver which didn't make any difference. Particurarly once I run any apps using directx, the pc shuts off and fan starts running like crazy.  I am not sure if it will work fine using windows XP rather than windows 7

Is there any suggestion