Question PC doesn't boot upon pressing power switch but flashes during PSU power on - Is it RAM, Mobo or PSU ?

[AdLibTact]

System
Mobo: B650M Aorus Elite AX v1.3
CPU: Ryzen 7 7700X
GPU: 1660 Ti
RAM: G.Skill Aegis 2 x 16GB (max 6000Mhz)
AIO: Thermalright Frozen Prism 360
Case: Montech Air 903

Installed Windows 10 Pro fine, activated EXPO profile 1 fine, didn't OC anything or touch any other bios settings.

3-day old rig was working fine 2 nights ago. I wanted to add a SATA SSD & a HDD so I connected them to the SATA ports. Mobo didn't light up (+ no post). I disconnected them, then mobo lit up with red DRAM light & "no signal" on monitor.

I reseated RAM & GPU, red DRAM & No Signal continues.

This morning, I took out ram, used eraser to rub the RAM & GPU inserts & used thin brush to ensure contact surfaces are clean.
Now the mobo doesn't even light up...

I ended up uninstalling everything from the case, & reinstalling everything to ensure all connections are correct & properly inserted.

I followed the stickied thread's instructions on everything, e.g. careful seating, connectors, GPU to monitor, flashed bios (no components, just mobo & cpu power connectors), extracted cmos battery + jumped cmos pins (always psu unplugged, both as an empty board, and with only cpu & AIO connected) etc.

Also tried solutions from YT like unplugging PSU & clicking pw sw for 30s, trying both ram sticks on each slot 1 at a time.

Did barebones set-up - CPU, CPU cooler, 1 ram stick & GPU (no iGPU), clicked switch (also tried jumping w screwdriver).
0 response from mobo despite flashing during PSU power-on

Have spent 20+ hours over past 2 days, would appreciate any help from anyone who has fixed this problem!

 

[rgd1101]

brand and model of the psu?

 

[AdLibTact]

brand and model of the psu?

Antec EAG 650W PRO Modular (Mobo, cpu & gpu power cords are attached to psu) I installed the 2 extra SATA drives (1 HDD 1 SSD) with 1 accessory drive power connector (so 1 connector for both drives) that can be added onto the back of the PSU). I searched 2 wattage calculators sites & both told me i'd need 450-500W, so i'd be extremely confused if it shorted my psu .-.

 

[JayGau]

Used an eraser to rub the ram and gpu connectors? Why did you do that? I understand you are desperate but this could have done way worse than good. The eraser can damage the gold plating on the connectors.

Your post is a little confusing and it's hard to know what the current state is. What does "0 response from mobo despite flashing" mean? Are the fans start spinning when you turn on the pc? Is the Prism 360 RGB light up? Maybe a video of what happens could help people here to figure out what is going on.

 

[Misgar]

This morning, I took out ram, used eraser to rub the RAM & GPU

I only do this as a last resort on DIMMs with very badly tarnished contacts or when the underlying metal pads have started to oxidise/corrode (happens in humid sub-tropical atmospheres). Rubbing the gold contacts could generate hundreds or thousands of Volts, leading to permanent ESD damage. Not a good idea for static sensitive RAM or CPUs,

I disconnected them, then mobo lit up with red DRAM light & "no signal" on monitor.

In isolation, I'd say the mobo was objecting to the 6000MT/s overclock if that's the speed you were using. AMD recommend maximum DDR5-5200 on the 7700X. Anything higher is not guaranteed by AMD, but down to the Silicon Lottery, the quality of the CPU/RAM/BIOS and any manual tweaks you make to RAM timings. Did you run a full pass of MemTest86 (several hours) to confirm stability with the XMP 6000MT/s overclock? No telling if it really was stable otherwise after only 2 days.

I installed the 2 extra SATA drives (1 HDD 1 SSD) with 1 accessory drive power connector (so 1 connector for both drives) that can be added onto the back of the PSU).

Could you be a little more specific about the "accessory drive power connector"?

Was it something like this Corsair modular SATA power cable?

Or was it a cheap-and-cheerful SATA splitter/adapter cable, which sometimes catch fire due to short circuits inside the moulded SATA housing?
https://nerdtechy.com/best-lp4-molex-sata-power-adapter

flashed bios (no components, just mobo & cpu power connectors),

I've been fortunate and never never had to do this. I'm fearful that something might go wrong and I wouldn't be able to see any error messages on the monitor. If the BIOS worked for 2 days before fitting the SATA drives, I don't think it was the problem.

Have spent 20+ hours over past 2 days, would appreciate any help from anyone who has fixed this problem!

After everything you tried, it's difficult to guess what has gone wrong. It all started when you connected the SATA SSD and HDD. That shouldn't be a problem. I have some PCs with 10 SATA HDDs and SSDs, but your system hasn't recovered from connecting two drives.

I normally grab a spare system with similar components and transfer parts over from the bad system, one at a time.

I'd probably move the RAM over first, because it's the easiest option. I'd test it at stock JEDEC 4800MT/s with MemTest86. No need to try XMP 6000MT/s. Just make sure the RAM passes MemTest86.

I'd move the GPU over to another PC and see if it works. A slight risk, but it's probably OK.

Next I'd use a multimeter as a quick test on the PSU, but only to confirm the +3.3V, +5V and +12V rails were present. Then I'd fit the PSU in my oldest, least important, most expendable PC, cross my fingers and switch on. No point killing an expensive PC with a bad PSU.

If I felt the risk worth taking, I'd transfer the CPU to a compatible motherboard and switch on.

For the next test, I'd fit a known good CPU in the suspect motherboard. Again, it's a slight risk.

On to the SATA SSD and HDD. I'd attach these, one at a time to a very old sacrificial PC. If the drives are deadm they could take out the PSU or mobo.

By now the bad component(s) should have been identified.

Assuming you don't have a spare system, is there a local computer repair shop nearby? Another option is to start buying replacement parts which could get very expensive. You stand the risk of further damage if you buy new components, or move parts which may already be damaged to a new system. Finally there's the new PC solution, but hardly desirable.

 

[AdLibTact]

Used an eraser to rub the ram and gpu connectors? Why did you do that? I understand you are desperate but this could have done way worse than good. The eraser can damage the gold plating on the connectors.

Your post is a little confusing and it's hard to know what the current state is. What does "0 response from mobo despite flashing" mean? Are the fans start spinning when you turn on the pc? Is the Prism 360 RGB light up? Maybe a video of what happens could help people here to figure out what is going on.

It appears from dozens of YT vids and hundreds of comments/upvotes that rubbing the RAM plates helped users boot immediately after, so I tried it. Thanks for the insight though, I won't rub them again of course.

0 response from mobo means nothing happens audibly & visually upon powering on the case. "..flashing" refers to the split second flash that I mentioned above, which occurs upon switching on PSU. You're right, I'll post a video when I return home later today. Thanks for your suggestion!

 

[AdLibTact]

I only do this as a last resort on DIMMs with very badly tarnished contacts or when the underlying metal pads have started to oxidise/corrode (happens in humid sub-tropical atmospheres). Rubbing the gold contacts could generate hundreds or thousands of Volts, leading to permanent ESD damage. Not a good idea for static sensitive RAM or CPUs,

I see... so that's why it's not recommended. Thanks for explaining, I'll keep that in mind

In isolation, I'd say the mobo was objecting to the 6000MT/s overclock if that's the speed you were using. AMD recommend maximum DDR5-5200 on the 7700X. Anything higher is not guaranteed by AMD, but down to the Silicon Lottery, the quality of the CPU/RAM/BIOS and any manual tweaks you make to RAM timings. Did you run a full pass of MemTest86 (several hours) to confirm stability with the XMP 6000MT/s overclock? No telling if it really was stable otherwise after only 2 days.

I didn't run MemTest86 after turning on EXPO profile 1. I thought the RAM would be stable bc I have a memory of reading somewhere that switching on default expo profiles would increase the RAM speeds to the max supported clocks of the mobo. I think that's definitely a plausible theory though...

Was it something like this Corsair modular SATA power cable?

Yes! I think mine had 1 less of those rectangular SATA plugs

I've been fortunate and never never had to do this. I'm fearful that something might go wrong and I wouldn't be able to see any error messages on the monitor. If the BIOS worked for 2 days before fitting the SATA drives, I don't think it was the problem.

Is that so... now I'm confused how so many users (tens of thousands) across the world seemingly fixed their DRAM lights with a simple bios flash. I read on some other forums (not Tom's) that some newer bios updates caused some instability to ram instead, so they tried flashing back to older versions & their systems booted normally. Regardless, I take it that you wouldn't recommend that, so I won't do that

Assuming you don't have a spare system, is there a local computer repair shop nearby? Another option is to start buying replacement parts which could get very expensive. You stand the risk of further damage if you buy new components, or move parts which may already be damaged to a new system. Finally there's the new PC solution, but hardly desirable.

Yea unfortunately I cannot afford a spare minimal system after building this new one. I'm thinking perhaps I'll need to refund the RAM... Will update here if there're improvements to my situation. Much appreciated, mate thank you

 

[Misgar]

I see... so that's why it's not recommended. Thanks for explaining, I'll keep that in mind

Cleaning RAM contacts with an eraser is common practice. Instead, Corsair recommend using a Q-Tip and Iso Propyl Alcohol, but I suspect this is aimed at home PC users.
https://help.corsair.com/hc/en-us/a...ice-for-cleaning-the-pins-on-the-DRAM-modules

Neither of these practices is 100% safe. You may never have a problem, but if you do "zap" your DIMM (or CPU) with static, they might fail at some time in the future (or not). ESD damage is insidious. It can take months or years before an affected part stops working.
https://mbsitem.co.uk/identifying-esd-damage-using-an-electron-microscope/

Having worked in the Aerospace industry for multi-billion dollar companies, we had special laboratories and work areas, with anti-static floor tiles, benches, work surfaces, and special tools, plus we wore ESD shoes, ESD overcoats, ESD wrist straps, ESD gloves, etc., when working on components and complete assemblies. You don't want components dying due to static damage when they're thousands of feet or hundreds of miles above the Earth's surface.

On home computers it doesn't matter a jot if a RAM chip or CPU dies due to mishandling. If you're flying in a plane at 30,000ft and something goes wrong in all the redundant systems at once ( highly unlikely) from poor handling practices during assembly, it's more serious! If a multi-billion dollar satellite fails due to unnecessary static damage, it's a waste of time and money.

that switching on default expo profiles would increase the RAM speeds to the max supported clocks of the mobo

Switching on the default XMP profile sets the clock speed to the maximum supported by the RAM, not the maximum speed supported by the mobo. If you look closely at your mobo spec, you may find it supports 7000MT/s or even higher speeds.

You bought 6000MT/s RAM, so presumably, when the BIOS checked the SPD (Serial Presence Detect) chip in each DIMM for the highest XMP speed, it pulled out the Primary, Secondary and Tertiary timings for 6000MT/s and programmed the motherboard accordingly.

Unfortunately, you cannot expect 6000MT/s to work on all 7700X CPUs, in all the various motherboards available and with every DIMM manufactured with a 6000MT/s XMP setting. The critical thing to note is that AMD do not guarantee operation of the 7700X above DDR5-5200.

As you may have spotted, 6000MT/s is some way above 5200MT/s. Just because your mobo and RAM can be clocked at 6000MT/s, doesn't mean they'll work at 6000MT/s with an AMD 7700X. It's common experience that AMD CPUs do not clock as well as similar Intel CPUs, when setting XMP/EXPO memory overclocks. The Intel 14900K is guaranteed by Intel up to DDR5-5600, 400MT/s faster than the 7700X.

In practice you can often run much faster XMP/EXPO speeds than the figure guaranteed by AMD/Intel, but there's no automatic guarantee you'll be successful, especially if you've bought a memory kit that's too fast for your CPU. Some motherboards overclock better than others. Some "Golden Sample" CPUs overclock better than others. It's called the Silicon Lottery.

It's common to find intermediate XMP settings in SPD, so your DIMMs may have options for 5000, 5200, 5400, 5600 and 5800MT/s, as well as JEDEC default (typically) 4800MT/s and the maximum 6000MT/s. To invoke these slower (and potentially more stable) XMP speeds, may require some manual intervention on your part, to select them in the BIOS.

Is that so... now I'm confused how so many users (tens of thousands) across the world seemingly fixed their DRAM lights with a simple bios flash.

It's not re-flashing the BIOS that concerns me. I do this on a fairly regular basis. What does worry me is flashing the BIOS when the CPU and more especially the monitor aren't there.

With a "normal" flash operation, you can see you're loading the correct BIOS file, by checking the drive and file name on the screen. You can monitor the progress bar during update and read the text messages that appear on the screen. You can breathe a sigh of relief after several reboots when the message "BIOS update successful" appears on the screen.

With a "blind" update, designed for use when your BIOS does not support the latest CPU and you don't have an older (compatible) CPU fitted, you're working in the dark. You cannot be certain the computer is loading the correct BIOS file. There are no reassuring messages on the screen to show if the update is proceeding normally, or has come to a grinding halt due to corruption of the BIOS file on a dodgy USB memory stick. All you have to go on is a flashing LED.

I believe it's easier to "brick" a motherboard with a blind update that with a normal update.

I'm thinking perhaps I'll need to refund the RAM...

There's a very good chance there's nothing wrong with the RAM. The only problem is you've been trying to run the memory too fast. A bit like flooring the accelerator in a car on a long downhill stretch, with the engine rev limiter disabled. You may simply be pushing your system too fast and into the unstable red zone.

If you can work out how to change XMP speeds in your BIOS, try 5200MT/s. It'll probably be stable. Next try 5400MT/s. Again it should boot into Windows. Next, try 5600MT/s. It might be OK. If so, try 5800MT/s. If 5800 crashes, reduce the speed to 5600 and then boot from a MemTest86 USB memory stick and run a full scan (several hours). If there are zero errors, you're probably good. If you get even one error, reduce XMP one step to 5400MT/s.

The idea is to increase the XMP overclock slowly until instability appears, then back off the setting to obtain reliable operation.

This all assumes you can get the PC to boot up at all.