So I take it nobody else has ever heard of this failure either? I've tried to google it and search the forums. There's nothing. I think algorithms aren't helping either because all I find are pages about POST codes indicated by the indicator, not that the POST indicator LEDs are failing. Also, it is day 2 and they haven't responded to my ticket yet. That said, i i's still before noon their time so I'll keep my East coast pants on lol.

Uh oh!
I'll be starting mine up tomorrow or Friday.

I purchased mine on 7/28/21. Agree with you it just may be a bad batch/something like that. Also not looking forward to dismantling or any downtime.

Yup, Oof-meter at max. I called in and they said I was right that the only explanation is a hardware failure and that it would be extremely rare. It doesn't effect performance, but if I want it fixed, the only way to do that is RMA the board.
 

frankd3
Those dates mean that it's possible the whole production run may have defective led readouts.
Oh great.

RMA's would come out of same production run.
Time for the DARK! maybe

I was thinking the same thing:
 
Either they have a bad batch of indicators...or the design is bad and they're over-driving the LEDs, which eventually causes failure. Either way, all the boards would have this issue. So the only way to fix it (without changing the hardware of course) would be to reduce the brightness or how often it is on. I think there's a setting in BIOS to determine what is displayed, but I don't know if you can make it do POST codes then nothing. I'll have to investigate that for my replacement, because if my board did this, and a new one bought on 7/28 did it...it must be pretty wide-spread on these. I can get by without having the CPU temp constantly shown, which is something I've done for years and years now, but I'd like to be able to see the POST codes when needed.
 
Edit: Just for clarification, the purpose of my conjecture above is twofold: 1: I'm an engineer. I cannot avoid trying to solve a problem like this, even if it means discussing temporary work-arounds or stop-gap measures. Yes, if there's a hardware problem, even something small like this, the proper fix is new hardware and/or a design change that prevents the hardware damage from happening. 2: For anybody else with these boards, it may be important to consider trying to adjust settings if possible as a preventative measure so they don't lose the functionality, or even for the folks at EVGA to update a BIOS if they have any intensity/brightness control under the hood (I'd be surprised if they did though, but who knows) to turn them down maybe 15-25%. It really doesn't need to be that bright, at least in my opinion.

Not exactly. I don't think it was "too bright", I was just saying it could be a little dimmer and still totally fine. Also, by over-driving, I'd suggest current. It takes both, but you control an led with current and let it take as much voltage as it needs for said current. This type of circuit is generally kept very simple: LED, resistor, switch: for each segment. The order doesn't really matter and the switch is digitally controlled (and some kind of small transistor or mux gate either passing or not passing the current, not a mechanical switch actually making and breaking contact). So you send a signal, "switch" closes, current passes, light comes out (because the pixies are angry as one of my favorite youtubers Ave likes to say). This happens for each segment, but they're generally tied to a single voltage supply. These aren't super specific LEDs, they'd have a max current rating. So you'd know your source voltage, subtract the estimated forward voltage of the LED, divide by the current you're trying to put through the LED and that would give the resistor value for each one. I don't know what the specific current rating is for each segment, but say for example it's 15mA. If the LED turns on around 1mA (just another example), you can pick a current anywhere between 1 and 15, the more current the brighter. So let's just say they picked 10mA, that should be fine forever and maybe meets whatever brightness specification they wanted.

So when I say they may have driven them too hard I imagine a few scenarios.
1: some engineer says "this is the simplest part of the whole circuit, but it lights up, pretty", glances at the spec sheet because they don't want to spend a ton of time on the simple stuff, thinks it's a 15mA part, picks a resistor for 10mA, but buys the 10mA model indicator. They're running right at max current, and sometimes, if the board is warm and runs for a while, they start failing. Woops. Didn't mean to, but accidents happen.
2: indicator the engineer wants is 15mA, resistor designed in sets ~10mA, when boards go to get made, the indicators are out of stock. Months have passed since the design. Somewhere in the alternate part selection process either the wrong indicator is ordered (but looks correct) or the wrong part number is on the bill of materials for a typo, or the distributor puts the wrong parts in the bag and nobody notices. (I have seen this sort of thing happen, and pretty much every electronic part I try to order these days is in stock when I put it on the schematic and out of stock when we go to order it so last minute substitutions are happening all over the place).
3: counterfeit parts do exist and sometimes make their way into distribution. One place I worked got a whole reel of about 1000 power mosfets that were counterfeit. They worked... For a few minutes...then they didn't. Idk how it's profitable to spend the time tooling up fabrication of components that look and sometimes perform almost as good as the real ones, but it happens. I think it's pretty rare that the parts get into the field, but it can happen.
4: More simple: there's a manufacturing defect in the indicators, but no human mistakes were made. Everybody thought it was fine and it usually is, but if you have enough tolerances and circumstances pile in one direction, you get some latent failures.
5: ESD will do that to an LED sometimes. Seems fine, lights up, but what you can't see is the internal stress that was created that will cause a part to die early. So if there was some unseen ESD event on a tray of indicators...trouble can occur. Most manufacturers are super strict about ESD protocols for this reason.
6: sure, voltage regulation problem. Say there's a specific regulator driving these and somehow the 3.7VDC regulator gets bought instead of the 3.3 (for whatever similar reason as above). Now there's more voltage across the resistor, more current in the LED, eventually fails. Voltage regulation could be a DC-to-DC chip or simple as a zener with a capacitor holding a specific voltage; maybe one of those is just a little bad for some reason...like I said, "simple circuit" lol. Nothing is really simple when you build a lot of them.

So, "too much" just indicating that the segments/LEDs couldn't do what was asked of them anymore, but could be for a great number of reasons, plenty of which could basically fall to "luck" one way or another and not negligence. Also could be negligence lol, no idea. When I said possibly the design is bad, I mean that either someone made a mistake, or something got built wrong making the design not work. I would be absolutely shocked if someone who knows how to make the rest of the actually complex circuits work didn't know how to do the simple math above. I would be less shocked if a wrong or bad part was accidentally bought, unknowingly received, or damaged somewhere along the way.

At the end of the day, sometimes hardware fails. I'm not happy about it, but I'm not going to call anybody out or anything. I like a problem to solve, but I wasn't looking for a blame game. I don't know where blame lies and I have no reasonable way or reason to find out. I'm sure they'll take care of it, I just like trying to figure things out sometimes.

Edit: I wrote this on my phone. Please excuse grammar, abbreviations, and typos lol

lol I couldn't sleep (unrelated). Also, it has been a few days of bad sleep and allergy meds (related to the not sleeping), so proof reading is...not so great. I just read through that and am surprised at how not-terrible it is.
 
Then last night I had a dream that I literally woke myself up laughing to and I can't quite figure out why it was so funny, but it still makes me laugh. I have to take my CPU out to take a picture of the socket for the RMA process. So in my dream, this was done by random people in lab coats on some sort of game show. the one person is being super careful, "easy does it", removes CPU from the board (which somehow the CPU has pins in this dream), looks at the CPU very carefully, it is pristine, then person B smashes a watermelon into it ala Gallagher and destroys the CPU and motherboard. Insert crowd laughter and my own laughter literally waking me from said dream. Why is that so funny? I have no idea, but I'm still laughing so I just typed this up hoping someone else laughs too.

This exact same thing happened to me too. I got my RMA back the other day and am hoping this one doesn't exhibit the same issue. Mine was fine for the 1st few days. 

added this video to the first post that shows the behavior:

frankd3 
BTW, is there any way I can find out where each mb temp sensor is located on the board?
 
And also, were the leds on the I/O cover supposed to light up on first boot? Mine didn't.
I'll see if loading that software activates it.

I don't know the answer to the first question but the second one I'm pretty sure that LED doesn't turn on by default and you have to do it in E-LEET.

kalsim
Mine started to fail. More disappointments.

Yeah, that's 4 that I'm aware of. Definitely a problem there. I suspect there are at least a couple folks at EVGA starting to ask how this could happen lol.
 
They haven't shipped my cross-ship replacement yet. (so 4 days of back and forth for the support tickets, and on the 3rd business day since the RMA was requested (7 business days total) and they still haven't shipped the replacement). I know someone was asking how long the RMA takes. I'll post the total once it is done. They could still ship it today, it's only 10:25am their time right now (early afternoon for me on the east coast).

hope it all goes well

No dice. I'll drop a picture in tomorrow, but there was a bent pin on the cpu socket of the replacement board. Glad I checked before I got too far into disassembly of my computer. I had only unplugged a couple things before I thought I had better check it first.

In any case, I'm sending the "refurbished" board back. I've fixed bent pins on a socket at work before (A mechanical engineer wanted to take some pictures and measurements of the motherboard, gave it back to me with bent socket pins lol), but I am not fixing it on a board from the factory that is supposed to be replacing my defective board.

kalsim
Mine arrived yesterday the same exact way. Gonna inspect it in the morning. Also, B0baganooshm, how can you tell it was a refurbished board from the part number?

Sorry, that was an error on my part (I just edited above). I meant to say serial number. There's a thread here (as well as a few others a google search turned up) that seem to come to the conclusion that when the 11th digit is a 9, it is a recertified part. Also, my board is more than 30 days old, so I knew they'd be sending a refurbished part. My grumps were that it was not actually "refurbished" and in fact damaged.
 
Here's a picture:

If you open that in a new tab it's a big picture and you can see the bent pin (indicated by the red arrow). It looks much worse from the side, as the pin is kinked and twisted downwards, but I couldn't get the camera to show good detail (focus!) from a close angle. You can clearly see it is off-kilter from the top, so the picture suffices, but from a diagonal it was very clear. It even had potential to short once the other pin was pressed down towards it.
 
The other issue I've seen with this kind of spring-copper (yes, they are coated with gold too, but it's a thin layer) is that if you bend them a few times the stress makes them considerably weaker. Sometimes you can just bend them back and they'll be perfectly fine, but if you're a little off or the springiness is effected from the multiple creases you get bad contact with the CPU pad and that can lead to intermittent failures and problems. By "bend" I mean a dramatic bend/crease/kink that changes the shape, not normal flexing.
 
I will add that as soon as I found this, even though it was 7:30pm my time, it was 4:30pm Pacific time and Daniel at customer service picked up the phone and helped me out when I called. He was very kind and with one email containing pictures he set up another RMA. The only real issue is that now I have to wait for this board to get back to them, they give me my collateral back, then take new collateral, and cross-ship me a board again.

frankd3
"grumps"?
 
Your a better man than I am, B0ba.
I'd be thoroughly pissed.
All my hopes are with you that EVGA will eventually make it all right.
 
When it eventually happens to me I'm not sure what I will do.

Yeah, I think at a different time I would have been the same. I've been working on acceptance being a more regular part of my response to things I can't fix. It is always hard for me to not try and do better, or fix something, but when it is totally out of my control, what does getting super angry help? (nothing at all). My main point was that the terms are pretty clear and known that if you need a replacement within 30 days you get a new product replacement, but after 30 days, you usually get a recertified product. So it was known before I requested an RMA that I'd get a recertified product. I just really hoped it would be in good shape. I read my previous post and it looked like I was whining about getting a refurbished board so I wanted to clarify. I mean yes, I had hoped that I'd get lucky and they'd just send me a new board, but I knew that it was like 95% chance of a refurb. I've had two other replacement parts from EVGA over the years and both were flawless so I had hoped for more of the same experience.
 
Also, it helps that the problem I have with my board is not causing it to be unusable. If it was a brick I would probably be quite a bit more frustrated lol.