Expert opinions about a hardware failure?

[GreyGeek]

At the price of usable laptops ( or good used ones) it is not worth buying a mobo and doing all that risky manual labor to install it IMO.

[Qqmike]

I like selecting the key components: motherboard, CPU, memory, power supply, HDD.

(Additional testing: I swapped the 2 memory cards all around in very possible combo, using channels 1 &2, and the two slots.)

[woodsmoke]

Qqmike...

There is really nothing quite like firing up metal that one has actually laid hands on and smelling...a little bit of ozone...not in the morning...lol...maybe 1:00 a.m...lol

woodlikeshardwaresmoke

[kc1di]

unfortunately Most Power companies are not liable for your equipment the best insurance for that situation to to have good UPS battery backup that will insulate your equipment from surges and outages. They need to be powerful enough to allow you time to shut the computer down the regular way.
In any event good luck in your search for answers.

[Qqmike]

Possibly Solved -- we'll see ...

As I said above, I did the following:

Reset CMOS.
No go.
Reset CMOS again, this time removing the battery while doing so.
No go.

I consulted a friend and PC expert I know, and among other ideas he had, he suggested I completely re-flash the ASUS BIOS file. I did that, and so far--after 3 days--it has fixed the problem. I'm cautiously hoping this fix holds (as there could be other factors involved, as well). But everything is working now as it should.

He had a lot of technical reasons why this might work, involving how video is initialized and how things are initialized from stored CMOS files.

Btw, if you are an ASUS fan, you already know this ... that flashing your motherboard UEFI-BIOS is a piece of cake these days with ASUS:
Download the ASUS xxxx.cap BIOS file you wish to use.
Put it on a flash drive.
Plug the flash drive in to the PC.
Startup your PC and enter the firmware setup menu.
Locate Advanced -- Tools -- EZ Flash menu.
Select that .cap file from the flash drive (which will be listed in the EZ Flash menu).
Wait until it is read.
When prompted, you are done and can exit with Saving.
Reboot.
Again, enter BIOS setup and load optimized defaults and Save.
Re-boot and customize your BIOS settings.
Re-boot, and it should work as you want it to work.

[jlittle]

That's really great, but I'm puzzled.

Your post #15 said it didn't go at all and your initial post said "it starts up but no display on the monitor.". How do you "enter the firmware setup menu" if there's no display on the monitor? It now sounds like the OS gave a black screen, or the boot loader (most likely grub) wouldn't start. If you could get to the firmware, the next step would be to try boot from a Live USB.

[Qqmike]

jlittle, the original problem was this:

If the PC is OFF, it wouldn't start up correctly in this sense: When I press the Power On button on the PC, it seems to start OK--fans, CPU, lights, are OK--but there is no video appearing on the monitor. If I shut it back down, then try again to start it, it may or may not start up correctly. I had to keep trying to start it until finally it would start correctly WITH video to the monitor. At first, it only took one to three tries to get it to start right. Gradually, it took more and more tries. To get it to start correctly with video finally took, maybe, 10 tries! So, I did that at the end: I tried about 10 times to start it, and it finally started correctly with video, I entered the ASUS UEFI BIOS and re-flashed the entire CMOS load. That seemed to fix the problem. You are right: if you can't get video, then you can't see what you are doing to enter the BIOS setup to do anything.

If you could get to the firmware, the next step would be to try boot from a Live USB.

Why a live CD? To diagnose it further? I already knew that the PC was not initializing correctly at the level of either the motherboard hardware circuits or the firmware on the motherboard. Btw, when it was broken, and when I would--by luck--get it to start correctly with video, the PC would run nicely, indefinitely, as long as I didn't shut it down. If I shut it down, then it might take forever to get it to start up again correctly with video.

Am I explaining this so it makes some sense?

So far, it is starting correctly after several days of watchful waiting.

[jlittle]

Thank you. I thought #15 meant the 10 tries failed.

If the screen remained completely blank the boot drive might have been faulty or the boot loader corrupted. Some screens take a while to start showing anything, and some firmwares can be set to not display anything, or POST very quickly.

Again, I'm very pleased at your success.

Regards, John Little

[Qqmike]

So after a couple weeks of running this PC, it appears that the fix posted in #20 above worked: I completely re-flashed the ASUS BIOS file.

Makes you wonder, what other more subtle PC problems might be remedied with a complete BIOS re-flash. For many users, that's kind of a scary thing to do, so you hesitate to recommend that someone do it to fix some subtle, otherwise unsolvable PC misbehavior. Personally, I keep that BIOS file handy on a flash drive right here, and will not hesitate to use it again as a last-ditch effort.

I did, however, upgrade my power supply. I only have two PCs here, my main 2015 PC and my 2009 back-up PC. I only own two power supplies: a 2015 SilverStone and a 2005 Antec TruePower 2.0. The SilverStone now goes into the 2009 back-up PC. Then, I did upgrade the PSU for my 2015 PC (the one I just fixed). (Btw, as I posted above, the PSU was not causing the problem behavior I posted in the OP--I tested that.)

I bought this PSU for the 2015 PC:

EVGA SuperNOVA 750 G3, 80 Plus Gold 750W, Fully Modular, Eco Mode with New HDB Fan, 10 Year Warranty, Includes Power ON Self Tester, Compact 150mm Size, Power Supply 220-G3-0750-X1

https://www.amazon.com/dp/B005BE058W..._t1_B010HWDP48

That price at Amazon bounces around. up and down; I got it at $99 last week.

I mentioned I was prepared to build a new PC if necessary, so I'll drop my component list in case anyone is interested. A Mainstream, non-gamer, ATX build, probably $650-$750 (I would try to re-use all my other stuff but might add a HDD to this list):

Motherboard:
ASUS Prime H270-Plus LGA 1151

CPU:
Intel® Core™ i7-7700K Processor, 8M Cache, up to 4.50 GHz, Socket: LGA 1151

Memory:
Crucial 16GB Kit (2 x 8GB) DDR4-2400 UDIMM

Power Supply:
EVGA SuperNOVA 750 G3, 80 Plus Gold 750W, Fully Modular, Eco Mode with New HDB Fan, 10 Year Warranty, Includes Power ON Self Tester, Compact 150mm Size, Power Supply 220-G3-0750-X1

Building is easy, as you DIY guys [includes gals] know. The key is the motherboard, and the key to choosing that is to first choose your chipset. (Choose your chipset by googling and reading recommendations/reviews. Use wikipedia for reference, also.)
The motherboard tells you what CPU and memory to consider. Just check your three 2-way compatibilities:

Motherboard <--> Memory (type of memory)
Motherboard <--> CPU (minimum BIOS, s-spec)
CPU <--> Memory (memory voltage called for by CPU)

Website configurators (at the motherboard, the CPU, and the Memory websites) help with memory and CPU.

For now, no need to build yet! Hoping to run this 2015 PC at least another two years, maybe another two-three years or longer.

Thanks, again, for all your input.

[Snowhog]

So after a couple weeks of running this PC, it appears that the fix posted in #20 above worked: I completely re-flashed the ASUS BIOS file.

So, issue SOLVED?

[Qqmike]

So, issue SOLVED?

Yes! I remembered and then forget (twice) to mark it Solved. Done now. Thanks.

[GreyGeek]

FYI - an issue that is characterized by requiring more and more reboots to reach an active display, as time progresses, can also be an indicator of an HD or file system going bad.

[Qqmike]

FYI - an issue that is characterized by requiring more and more reboots to reach an active display, as time progresses, can also be an indicator of an HD or file system going bad.

In this case, I don't think so ... based on how abruptly this started (i.e., immediately after the electrical power failure); and based on my hard stress testing. Cause and effect. Snippets, off-the-cuff, from a PC guy I know:

... we tend to think of data in terms of alphanumeric symbols (ones and zeros). But the computer's CMOS memory consists of arrays of little capacitors (cells) and data is represented as the specific arrangement of cells that hold an electrical charge and cells that do not hold an electrical charge within the array (charge/no charge).

Charge/no charge is not absolute...there is a threshold value (which I used to know) above which the cell will be read as "charge" and below will be read as "no charge" even though a residual voltage may (and most likely) be present. Of course, the further away from threshold in either direction that a cell's voltage lies gives more consistent results and normally there will be a big enough delta to keep the error rate acceptably small.

So, say your CMOS gets zapped by a current spike (or even the notorious cosmic ray) and as a result certain cells that are supposed to be read as no charge pick up a small additional residual voltage and are now very close to the detection threshold. The machine's ability to sense at threshold isn't absolute either, so at that level it might detect a "charge" or it might not. Also, the ability to detect a charge near the threshold is likely affected by environmental conditions and other things the machine is doing at the same time. As a result, unpredictability and inconsistency reign. Unfortunately, the bootstrapper is really stupid even by PC standards and there is little or no error checking or error reporting at that stage. It either works or it doesn't.

Now, if the spike or cosmic ray or whatever raises the voltage in a given "no charge" cell by a considerable amount over threshold the machine behavior (and failure) will be consistent. With PCs, as in real life, it's the corner cases that get you.

[Qqmike]

and, GG, there's a bit more thinking to it, that somehow I truncated off the above post:

The idea to try a complete reflash of the UEFI load came from thinking about how a power surge below the threshold of physical damage might affect the machine. The video subsystem displaying the principal symptoms gave some weight to the possibility of a surge coming from in through the video port.

I know that a PCI slot video adapter can cause all kinds of shenanigans if the manufacturer get the BIOS wrong (c.f., some ATI cards and open-source Linux drivers) or if the onboard BIOS gets corrupted, but I wasn't sure about integrated video since I don't use it and haven't researched it much.

Turns out the first document I looked at outlined the bootstrap routines in the UEFI code that initialize the video subsystem. Some of the data comes from the EDID chip in the monitor and some of it comes from the data stored in the UEFI CMOS memory. If the data in either source gets scrambled by a power spike it could explain the unpredictable behavior your system was showing. Unfortunately, the computer isn't smart enough to perform error checking on the data stored in the CMOS chip at that stage of the boot process.

At that point I recalled that ASUS motherboards are mildly notorious for resetting only the user configurable options when you reset the BIOS//UEFI to default settings from the setup menu. That led me to the possibility that perhaps one or more of the non-user video related settings stored in the CMOS might have been scrambled by a spike. The way you reload those is to reflash the entire BIOS/UEFI package, hence my suggestion.

Plus, it's a relative low-effort and no cost thing to try. Hope it holds.

I see you noticed the cheap-ass USB 3.0 motherboard headers that ASUS used in that vintage of their motherboards. I completely agree with your assessment of their quality, ease of use and potential to cause reliability and assembly problems. More recent motherboards have gone to a completely enclosed socket and keyed connector that does a better job of supporting and retaining the plug and making it impossible to orient the connector the wrong way 'round.

The UEFI CMOS does store some settings that control aspects of the motherboard's USB3 support. So yes, it is very possible that if a spike munged video data it could also have affected USB setup data and that reflashing UEFI package would correct the situation. [Note: When trying to re-flash CMOS with a USB3 flash drive, it wouldn't read two of my USB3 ports; had to try a USB3 port on the back of the case. Thus, the comments about how USB3 may have been affected. Following the CMOS re-flash, the two bad USB3 ports came alive.]