Flakey Acer desktop

glennd

Well-Known Member
Reaction score
2,527
Location
South West Victoria Australia
I got a curly one and need some ideas.

Acer Desktop
CPU: Intel Core2 Quad
MB: Acer G4T/G43T-AM3V:1.1
RAM: 2 x DDR3-1333 2GB

A week ago I rolled back Win10 to Win7 and it appears successful. This week customer reports BSOD and brings it in to the shop. First time I powered it on, the BIOS gives a continuous beep and doesn't post. Eventually it started and booted Windows. It ran ok for 10 minutes or so and then BSOD code 1e ntoskrnl.exe KMODE_EXCEPTION_NOT_HANDLED. Event viewer shows the same bsod 3 days earlier.

CHKDSK reports no problems. SFC found and fixed corrupt files somewhere in Winsxs. Did a malware scan and cleaned up a few things. I left it running overnight. This morning Windows was locked up. Reset and boot again. Again Windows locked up during login. Boot up UBCD and ran Memtest86+ which immediately crashed and went crazy. Swap out the ram modules, no change. Two new ram modules, no change. Remove the video card and run onboard video, no change.

Plugged in a new PSU and now Memtest is working. Call the customer and say it needs a new PSU. Customer says go ahead. I install the new PSU and plug everything in and Memtest crashes again, the same as before. At this point I realised I tested the new PSU without plugging in the peripherals, only the motherboard was powered on. I unplugged the hdd and now Memtest is working, plug in the hdd, memtest crashes, unplug the hdd, Memtest works. :confused:

Booted into PMagic, smart shows no errors, smart short test passes, running the long test now but so far no errors.

As best I understand it, the hdd is causing the computer to lock up randomly or throw bsods and stops memtest from working at all.

EDIT

Specs on the two PSUs:

Old:
+5V / 20A
+5V (SB) / 2A
-12V / 0.5A
+3.3V / 20A
+12V1 / 10A
+12V2 / 13A
Total Output Power: 300W

New:
+5V / 18A
+5V (SB) / 2.5A
-12V / 0.3A
+3.3V / 22A
+12V1 / 21A
+12V2 / 18A
Total Output Power: 500W
 
Last edited:
==================================================
Dump File : 060916-15007-01.dmp
Crash Time : 9/06/2016 10:55:25 AM
Bug Check String : KMODE_EXCEPTION_NOT_HANDLED
Bug Check Code : 0x0000001e
Parameter 1 : 00000000`00000000
Parameter 2 : 00000000`00000000
Parameter 3 : 00000000`00000000
Parameter 4 : 00000000`00000000
Caused By Driver : ntoskrnl.exe
Caused By Address : ntoskrnl.exe+6f3d0
File Description : NT Kernel & System
Product Name : Microsoft® Windows® Operating System
Company : Microsoft Corporation
File Version : 6.1.7601.23418 (win7sp1_ldr.160408-2045)
Processor : x64
Crash Address : ntoskrnl.exe+6f3d0
Stack Address 1 :
Stack Address 2 :
Stack Address 3 :
Computer Name :
Full Path : C:\Windows\Minidump\060916-15007-01.dmp
Processors Count : 4
Major Version : 15
Minor Version : 7601
Dump File Size : 275,472
Dump File Time : 9/06/2016 10:56:46 AM
==================================================

==================================================
Filename : ntoskrnl.exe
Address In Stack : ntoskrnl.exe+66dfe
From Address : fffff800`0324c000
To Address : fffff800`03832000
Size : 0x005e6000
Time Stamp : 0x5708972e
Time String : 9/04/2016 3:46:22 PM
Product Name : Microsoft® Windows® Operating System
File Description : NT Kernel & System
File Version : 6.1.7601.23418 (win7sp1_ldr.160408-2045)
Company : Microsoft Corporation
Full Path : C:\Windows\system32\ntoskrnl.exe
==================================================

I should point out that once I started booting into UBCD and memtest, Windows is out of the equation.
 
Last edited:
Boot a linux live cd/usb stick and see if your still getting problems. I normally boot up to a desktop then open a term window and take a look at the boot log by doing: dmesg |more .

See if you still have problems when running linux.
 
I'm with Coffee, You need to determine if it is:
Hardware issue
OS/software issue.
Both
If hardware is ok, I would do a clean install.
Rollbacks may not always work "correctly"
Start a process of elimination with hardware 1st.
Don't chase your tail, if the hardware is bad.

Also, how was it acting before the Win 10 upgrade/rollback?
 
Did you clear the bios when replacing the cpu? we have had a few weird issues in the past and now we always clear the bios and start with a clean slate when installing or replacing a chip. We also stopped rolling back OS and just do a fabs backup and a new install because we had so many rollback issues.
 
You Said memtest locks when HD is plugged in did you try another HD and see if it locks with a different drive plugged in? Could be a Board issue on the HD that is causing feedback on the Power lines or something.
 
This might sound dumb, but check the power supply again. What type of psu is it? Quality does make a difference regardless of specs. Also, I find that when I install a PSU, I've had better luck getting units with a single rail design instead of dual rails. Too many problems like you are experiencing, the machine locks up or freezes, or powers off because it can't pull enough from a PSU, even a brand new unit that was a split rail design. Whenever I started using single rail designs from Corsair, EVGA, etc, I find I have a lot less issues like those.
 
Thanks to everyone for your suggestions. I tried every single suggestion that I could but nothing changed the fact that memtest will crash when a hdd is connected. Obviously booting to UBCD discounts Windows. I swapped different power supplies, hard disks, cables, sata ports to no avail. I tried @ComputerRepairTech's suggestion of just the sata power and not the data cable. Memtest will crash if both are connected but not if just the power cable is connected which kind of tells me it's not power related but signal related. Possibly the onboard sata controller.

The comment from @myleso about clearing the bios reminded me that when i was looking to fix the high cpu fan problem i reset the bios to default settings which fixed the fan problem. So I went looking in the bios for anything hdd related and came across a setting called "Onboard SATA Mode" which was set to "RAID". I don't know what it was set to prior to my resetting the bios to defaults. However, I set it to "AHCI" and Memtest is not crashing anymore! Yay!

I'm of the opinion that there were several independent issues at work here. Following the Win10 upgrade/rollback there were several Win7 system files that were trashed which probably caused the BSODs and was fixed by SFC. The lockups and Memtest crashes were caused by the incorrect sata mode setting.

So I've let Memtest run a full cycle with no errors. I booted Win7 fine, it installed its AHCI drivers, which tells me it hasn't been in that mode before (?), let it do its restart and apparently it's all good. I'll let it run today to see if it locks up.

This is what makes Technibble special! When you work in isolation as I do, the ability to brainstorm with such a large group of bright people throwing around seemingly random ideas, something is bound to lead you to the right answer.
 
Back
Top