Alienware m17x R4 - weird display issue

Hi guys.

I have quite weird and annoying issue on my Alienware: distortion of information on the display and by distortion I don't mean like swapped colours or something like that which would indicate broken/losen LCD cable or problem with GPU - the distortion is more like horizontal lines with partially duplicated content of the screen. At first I thought it maybe something related to Win 10 (like updates) and tried a few recommended things which didn't help. Then I wanted to confirm it's a hardware problem so I've run the built-in Alienware hardware test (on F12) - it was long-running process due to complete testing of the RAM, but it didn't show any errors either, so I tought "OK, then it must be OS issue of some sort", but then I've moved cursor to the right edge of the screen and it started flickering/distorting similar to what it did in Win 10! I've also booted from USB live linux and confirmed that it behaves exactly the same in terms of distortions on display as it does in Win 10 and in hardware test shell.

What I did so far:
- switched back from "only AMD 7970M" to "Intel GMA 4000 + AMD 7970M" - with integrated graphics the laptop became unusable: immediately after booting (during the boot everything seems to be fine) when Windows logon screen is shown, the display blinks and doesn't show anything, tough the backlit is working; I had to log in blind and switch to dedicated graphics only
- re-installing drivers for both cards didn't help, in Safe mode it's also flickering and as I said above it seems to be more like hardware issue since it's present not only in Windows
- on dedicated GPU I can still use the laptop, but I have to minimize the applications - once application is maximazed, I'm getting that distortion again; odd that if I move a window of some app to the right edge of the screen, it starts developing the above described distortion, but if the window is moved to the left/top/bottom edge, nothing happens and all works as it should; in KODI which is running full screen, I cannot reproduce this issue
- tried to check if it's an LCD problem via connecting to the TV as to the wireless display - didn't succeed, probably something with Win 10 settings; unfortunately I don't have a HDMI cable so wasn't able to connect to the TV directly, but I'm about to buy one to verify if it's LCD or not

Any thoughts anyone? IMHO it's highly unlikely that both GPUs have failed simultaneously, but if it's LCD, why I can only see the distortion if I either move cursor to some point on the right edge of the screen or if I move there an application?

UPDATE 1: tried to switch back to dual graphics mode and realized that the notebook immediately after powering it on starts to blink, I can only see the BIOS booting for a half of second and then it's gone and display just goes dark...if I'm entering the hardware testing shell or BIOS, screen flashes for a second or so and then can go dark for 10 mins, then flashes and so on. Clearly there's a difference between running with only AMD 7970M and dual mode - in dual mode flickering make NB unusable.

UPDATE 2: got a monitor with HDMI cable, situation is as following: with dedicated graphics only, upon connecting external monitor the message appears on it "Out of range" 0_o (using default 1920x1080), cannot get external monitor working in any setup - duplicating / extending / main + external / external only - none of them works; if switched to dual graphics mode, then the external monitor starts working, but the built-in gives up and goes dark as written above.

 

i have similar symptons in my M17xR3 with 6870 ( with "partially duplicated content of the screen" distortion). i thought it might be software or driver issue but after reinstalling driver it will still getting same distortion imsage after some time use(two or three days). so i replaced with a new 6870 and it works fine for several days. then it suddenly died with 8 beep in boot. after removing 6870 (i.e. without any discrete GPU card), the laptop can boot up normally(laptop screen can light up) but if replacing with old 6870 card the laptop screen won't light up (even pressing F2 in boot up, laptop screen is totally black) while external screens can light up without any more distortion. i can't figure out the reason.

 

Meanwhile I was trying to get more info on the subject by trying different things - I've done the display test (holding "D" key while powering on) and it seemed to be perfectly fine: different colours were displayed and NO distortion on the screen; also I've completely disassembled laptop, veryfied all connectors, re-inserted dedicated graphics card and...nothing. The issue is still there and seems to be that it's getting worse - on dedicated card only now there are even more distortions, the only way to use it is to use external monitor via HDMI. I've said yesterday that if I switch to dedicated graphics only, then upon connecting external monitor I see on it "Out of range", but yesterday evening I've tried to connected to the TV and it works absolutely fine, so there maybe an issue with monitor/resolution/refresh rate.
In any case what I've managed to confirm is that on both Intel+AMD and AMD only with external monitor I'm getting correct picture without any distortions. Furthermore, playing demanding games even for long time doesn't develop any issue - from this I conclude that AMD card must be OK. So if it's not a built-in graphics, nor dedicated and LCD test is fine + (!!! attention, this is important) there IS a difference in behaviour in dedicated graphics only and dual graphics - in first case I can see the picture, but very distorted, in second case it blinks when BIOS loads, then shows the picture of Windows 10 loading and when gets to the logon screen, display just goes dark and stays like that, periodically flashing the picture if I switch between external and built-in monitors in Windows via Fn+F6.

At this stage and with the facts above, I tend to make a conclusion that most probably it's a hardware failure of motherboard - could be that chip(s) around display port damaged and there must be a difference in how signal is going to the display port for singe and dual graphics mode, otherwise I'd observed the same distortions or no picture and it would be consistant for both modes.
I want to hear your opinions guys, because buying new motherboard for 300 EUR, puting it in and realizing that it didn't help is not an option for obvious reason

UPDATE: yesterday evening built-in display started to work again as if nothing has happened! I was playing some game on external monitor and built-in display came in, then flashed and went dark again, but after some time picture appeared again and it was working fine since then. My best guess is that using external monitor via HDMI caused some warm-up of the chips on the motherboard.

 

Now the problem is back How to make sure that it's really a faulty motherboard and not LCD? It just doesn't make sense - it's like twice I've managed to "cure" it with just attaching external monitor via HDMI and using it while ignoring flashing built-in LCD - how is that even possible? Any ideas, please?

 

The MXM card has separate outputs for internal screen/eDP and HDMI; it can be partially defective. Try setting it to Intel HD and physically remove the card.

Failing that; CPUs aren't immortal either, even though it's rare for them to fail. Get a cheap drop-in and try running that. A Celeron 1000M should do fine and at $1.48 isn't much of an investment. Always keep an assortment of low-end, same-era cpus for just that purpose ...

 

I'll try to physically remove AMD card tomorrow, but I didn't get it - do you mean that defect is in card or on the motherboard?

Never overclocked NB which of course cannot guarantee anything, but if it would be CPU, then why if I'm doing print screen during LCD flickering and inserting it into Paint it's completely without any artifacts? I would guess that in case of faulty CPU screenshot also had artifacts on it... Anyway, thanks a lot for sharing your thoughts!

 

OK, I've re-enabled integrated graphics to prove that it's still the same (unusable) - yes, after flashing BIOS loading screen, display went dark. Then I've tried to remove AMD card and here's what's happened: after booting Windows on log in screen LCD flashed once like when you have new video card and then kept working fine - no flickering/flashing. After that I've decided to re-insert the AMD card again to confirm it's partially damaged - but NO flashing/flickering, tried games also fine...the only one thing which was annoying is that something has happened to AMD driver apparently because I couldn't open AMD control panel and after booting I've got message "restart is required bla-bla PCI config changed" or smth similar to that - the thing is that even after reboot it was still appearing. I've downloaded AMD driver and updated, after reboot no warning about "restart required" (which was coming from AMD events service), BUT screen started to flashing again and within minute or so notebook became unusable. I've removed the AMD card again, but it's exactly the same - LCD remains dark, periodically flashing for a second once in 5-30 mins. So I've put the AMD card back, did login blind into Windows and with Fn+F7 disabled integrated graphics - this way I can use notebook somehow if I don't open applications full screen and don't move mouse cursor to the notification button in the tray...

Yesterday I've also tried to unplug built-in touchpad as I've seen some info on Dell forum that someone had some visual artifacts and unplugging touchpad helped to solve that - for me it didn't help. Still can't understand why there are no distortions till I put cursor in bottom right corner of the screen. Tried also Miracast from Android tablet to my Alienware's screen - all worked fine, no visual artifacts.

So where does it leave me? Unfortunately I don't have any CPU I can drop into my Alienware for test at the moment, but frankly if integrated graphics is disabled, HOW it can influence discrete AMD card?

 

Because the cpu handles the traffic, including that between card and display.

What removing the card does is ruling out a defect in the chip that shorts one thing or another, causing errant signals that, in turn, affect the rest of the system. Having it disabled software-wise does not prevent current flowing between the mxm and the motherboard, so the aberrations would also show up on pure iGPU while having the card still attached.

At this point I'd say gpu as the most likely cause. Might want to run without mxm at all for a week or so and see if the issue is 100% absent then.

Hmm ... something else; there might also be a problem with the power circuit on the motherboard. The gpu needs a lot of juice after all, so having it active might simply be the last drop that reveals an underlying fault. Could try dGPU, yet force lower clocks via software or merely by running on battery, assuming it won't run at its highest speeds then, of course.

 

OK, maybe I didn't make myself clear - I have removed physically MXM card and while it's "helped" for few minutes, I think it was more like coincidence, because now no matter if card is inserted into motherboard or not, integrated graphics virtually unusable. Btw tried running on battery already - makes no difference at all. Obviously I can't run for a week or so on integrated GPU since LCD is dark most of the time (i.e. unless I attach external monitor). Did some investigation and it seems that in case of HDMI/VGA output, chip Parade PS121 is used while for eDP it's more complicated: pair of Parade PS8321 (no idea why 2 of them - maybe because the same signal has to be on HDMI out?), multiplexer/demultiplexer SN74CB3Q3253 (marked as "BU253") and probably the main suspect - PI3LVD1012BE - which is the chip next to eDP and is responsible for switching between multiple LVDS sources, there was some info which links this chip to visual artifacts on LCD. Still thinking why using external monitor via HDMI after some time "fixes" the issue for few months...

 

The second DP mux chip was probably for eDP, even if this specific board doesn't use an eDP display (or even has that connector). You might have the empty solder pads for it though. Dell/Alienware didn't build or even design the motherboard; Compal or Quanta did and they supply ~75% of all laptops. For them it makes sense to design a generic design (large gaming, in this instance) and simple build variants of those, according to customer demands, whether it is Dell, Asus, Apple or whichever (sell the same thing twice).

These chips are fairly cheap, so can't hurt to order and swap them. Soldering these needs a hot air gun and a decent magnifier or a rework-station. If you haven't got these then a diy-electronics shop will do and shouldn't cost more than $25-50 or so.

That low-end cpu or a low-end gpu might be cheaper to rule out first. Always have a few of those nearby anyway; if they're not of use for this issue then for that inevitable next one that crops up.

 

There are 2 display ports on motherboard one next to each other - one is in use and one is empty. Since I have FHD LCD 60Hz without 3D, I assume LVDS port is in use and eDP is empty. Thanks for additional info about the stuff - much appreciated!

Which one do you mean? The PI3LVD1012BE?

I haven't worked with air gun yet, but radioelectronics used to be my hobby + learnt it in university (mainly processors and microcontrollers) so hopefully I'm not completely useless in that matter Thanks for tips!

While I agree that Celeron 1000M is fairly cheap to try, I fail to see how it can be the source of the problem here: screenshot is looking good - no artefacts, besides image via HDMI/VGA is absolutely fine - I would have guessed that in case if CPU was partially damaged (or rather the iGPU inside it), I'd see the same picture on HDMI. I didn't get the idea about low-end GPU though - as I said removing MXM card doesn't solve the problem at all.

 

Now this is getting interesting & annoying at the same time Guess what? I've "fixed" it, though I'm sure it will not last long (last "fix" helped from April till couple of days ago) since it's not really a fix.
Previously I said that last time after using external monitor via HDMI for some time (cannot say now for sure, maybe it was taking day or two) problem gone without any further actions, this time however, after studying schema (you were right - motherboard is developed by Compal), I've realized that there are actually 3 pieces of PI3LVD1012BE - on HDMI IN, on iGPU/dGPU selection and finally one to switch between HDMI in and iGPU/dGPU - so I thought "it'd be nice to sort out HDMI IN at this stage": when I connected another NB via HDMI to output it's screen onto my Alienware, I've got perfectly fine picture and moving mouse cursor didn't cause any troubles at all (before test the problem was still there of course). I thought that's OK, I didn't expect really HDMI IN to have any influence on this and decided to hit Fn+F7 to re-enable iGPU and re-test just in case - now the interesting part: when I've unplugged HDMI cable before even enabling iGPU, I've realized that problem was gone!!! After enabling iGPU and rebooting all still is fine - no distortions, like this nightmare has never happened. What is annoying is that surely it is NOT a fix and it's just a question of time when it's back. This proves that LCD has no problem at all and because the "fix" as previously has to do with HDMI out/in (though HDMI IN proved to be VERY efficient because it only took few mins or so to "fix" the problem ) I would say that somehow one of the PI3LVD1012BE or other chips around HDMI (excluding PS8321 as it turned to be it's purely related to eDP port which is not used with current LCD) are producing this glitch.

Could it be that problem lies somewhere in the nearest passive components next to those chips (mainly I'd suspect capacitors)? Can someone explain me what the hell is going on here - how on Earth simply connecting something to HDMI IN could fix it?!!

 

Straws here, but perhaps setting the HDMI-in switch allows a short somewhere in the chip to drain harmlessly. Or indeed a faulty capacitor nearby. They can be checked with a multimeter, but that is a tedious process, the smds being hard to make a good connection too. Not to mention there's dozens to check.

One of those PI3LVD1012BEs might be swapped, but better know which one then. Ran into this neat graphic of a mux setup:



With this particular setup you'd suspect the top-right PI3LVD1012 switch to be faulty. It is the only chip active in iGPU, dGPU and iGPU+dGPU modes, yet not in the HDMI circuit. Perhaps after using HDMI-in just once that resulted in that bottom-right HDMI switch (PI3WVR12412) getting stuck in the route CPU/GPU -> HDMI -> LVDS:



That would also explain why hooking up an external HDMI monitor helped alleviate the issue. But only if the LVDS signal really passed over HDMI then. This setup might not use that option. There might also be a difference between simple monitor duplication and extending in the way the circuits are activated.

That is one crazy route though, passing 6 ICs. Looking at that diagram it's no wonder there's so very few mux systems ...

 

Looking at capacitors at those switches...well, my understanding is that they are supposed to filter out some "noise" on the +3V (in parallel 4.7u + 0.1u + 0.1u = 4.9u to the ground), so probably I'd ignore them for now. With those switches (PI3LVD1012BE) it's somewhat complicated, but looking at the scheme you've posted as well as the real NB scheme from Compal (which differs in that it doesn't have PI3WVR12412 and HDMI IN and OUT are looking truly separated), I'd blame that top-right or the next one after it (graphics-HDMI IN to LVDS). Apparently something "stuck" and logic failed, though I thought it's more like "it's either dead or alive - but not the God damn Schrodinger's cat" Knowing that those ICs are cheap (~2.5$/piece on eBay) it seems to be no-brainer to buy 3 and see how it looks after replacement. Will order them and patiently wait till the glitch appears again - sounds like a plan t456, thanks for all your help and provided info!