Power supply issue

Hello,
yesterday my m17x R3 shut down while using it, I thought it was too hot, so I opened it and cleaned the fans (I do this every few weeks), but as I wanted to restart it, it didnt react. Then I saw that the LED at the power cable wasn't on. I pulled out the cable, put it back on power and the LED went back on. But when I connect it to the Laptop it goes off. I suspect it is a short circuit in the Laptop because the cable seems to work and to "save" itself as it shuts off. Can I do something to save it or is it destroyed? The Laptop is 2 years old and the warranty was only one year Long.

 

Hi, it does sound like a short somewhere. Does it not work on battery?

You could try to remove everything you can (CPU,GPU, memory, drives, etc) and see if at least the power cord LED remains on and it gives you a beep code. If not then it has to be either the PSU or the motherboard, but if it does not start on battery then the PSU is unlikely. Without someone to loan you a PSU I have no idea how to test it

 

The problem is, I don't have a battery because the PSU once fell down, got destroyed and grilled the battery with it. I bought a new PSU, but no battery because I don't really need it and it was expensive as hell. This happened a year ago so these should be unrelated issues. Everything has worked fine until now.
Also I don't have anyone to lend me a PSU so yea. I will try removing the parts tomorrow, thanks for the quick response.

EDIT: I removed the parts as you said and it made me happy to see that the Motherboard isn't toast it is the gpu that makes the PSU "save" itself by shutting off.
Question 1: Can I remove the gpu and the Laptop will work without it? (because it has internal one in the processor)
Question 2: I had a Nvidia 560m and was very Content with it, should I buy the same gpu or can i upgrade it?
Question 3: If I can buy a better one which one would you recommend?
Question 4: what is a good shop for replacement parts?
I got the Laptop on 17th october 2011, it has an i7 2630qm (i think it's qm), also i exchanged the hard drive with a crucial M4 256gb, rest is stock. Bios is still A08 because back when i bough the SSD I read that the newer BIOS doesn't work well with SSD.

Also just out of curiosity: what could cause the gpu to suddenly be destroyed? Could a current fluctuation be the cause? Because 2 chargers for mobile phones got destroyed in my house too.
Again thanks a lot

 

That is at least a relief indicating the NV card BUT it still could be the PSU being unable to supply the extra current the dGPU draws causing it to shut down.

1. as long as it's not a 120hz/3D display then it should work fine without the dGPU.

2. Your upgrade option for the R3 is limited to a 580m for full functionality (unless you want to change to ATI in which case I don't know the models). If you get a GPU that was not sold with the R3 like a 680m for example you need an unlocked BIOS to disable the on-board GPU for it to work. Obviously loosing optimus (auto GPU switching).

3. See 2

4. I have only seen eBay being used so have no commendation as to a supplier. You could get it from Dell but probably V expensive. Main thing is to get the x-bracket, the bit that attaches the card to the heatsink, specifically for the new card. The heatsink can be reused but I know the 580m heatsink has some thermal putty on one edge for the VRM chips and if your 560m does not have this then it may be hard to find. Easiest way is to get a replacement heatsink and bracket from Dell. It comes with all the pads/paste/putty pre-applied and is a straightforward job of just screwing it together. The last quote I saw from Dell was $24.00 for the heatsink so not too painful.

A good supplier will have pics of the pad positions - or you could plough through this thread for pics - http://forum.notebookreview.com/ali...x-580m-78c-throttling-see-post-191-later.html I haven't seen anyone complaining of overheating for a long time so either they are all gone now, or more likely poor factory pasting, as was the case on my R3, was the main cause.

Since you have had two other device chargers fry then you have to suspect the same thing has happened to this one - but you do know the MB is probably OK.

The BIOS problem you mention is to do with the interface speed. The R3 was supposed to have SATAIII (6Gb/Sec)on the drive bays. It did not work well on the second bay and dell eventually gave up trying to fix it and downgraded the BIOS to SATAII (3Gb/Sec). It's unlikely your HDD is a SATAIII device so no benefit there. You can use the IRST software (should have an icon in the taskbar) to see the speed the drives are running at. When I had an R3 I did a lot of testing trying to get SATAIII to work and my experience was that an SSD is so fast anyway I could not see any difference in use once it was working at SATA III.

I'd start with another PSU before spending a bunch of cash on a GPU only to find it does not help. Time to get a surge protector as well methinks!

 

Posting this from my Alienware now, I removed the graphics card+heatsink and everything works fine. I think it is unlikely that the PSU has the issue because as I said, when the gpu is in the laptop the psu shuts itself off when just connecting it (I can't turn the laptop on obviously). That the PSU could be broken seems strange to me could you please elaborate? If it is really the issue, I could order one from ebay for 30-40€ so that wouldn't be a big problem.

The options with the 580 and 680 both sound interesting but the 680 is too expensive for me. Also I have problems finding sellers of either 580 and 560 in Europe so I will see what I'll do in case the PSU is okay and I find sellers.

Thanks a lot for your help

 

What I'm thinking is a surge has damaged the PSU. The GPU is a 100W maximum card!!! Probably drawing 40-50w just at startup (guess). I think the chances are that the extra current draw is causing the PSU to trip out. No way to know for sure without another machine or PSU to test but with those others you mention it leads to the cheaper option to try first.

I did find this (ended) GENUINE ALIENWARE M17X-R4 M18X-R2 MXM GTX 680M RARE UPDATED VERSION 20HTK | eBay . The listing had good info about heatsinks and vBIOS versions, and offered to supply the HS a separate part or assembled. Looked like they did other cards as well. Never followed it up but might be worth a 'contact' as they are in Europe.

 

580m should be fine, at least you are able to boot it up.
Make sure you put some good quality thermal compound on that new card.

 

Guess I'll have to buy a PSU then. I'll keep you updated when it arrives.

 

It may sound crazy, but maybe even a bios update may create wonders sometimes...

BTW: When I had my R3, the latest bios update (I think was A10) could allow you to switch off from bios either of graphic cards (internal or dedicated). Someone with this model, please correct me if I am wrong.

 

Yea but I can't even plug in the cable if the graphics card is in the laptop, how should i deactivate it then^^

 

If the PSU does not help then at least you should be able to sell it quite quickly - there's always people looking for a spare PSU

I do recall from back in my R3 days reading that dell added a bios graphics card switch on one of the later releases, but I never looked at it since disabling the on-board on my model (120hz) bricks it - been there done that

 

MickyD you're right. BIOS version A12 has the Graphics Card switch - I think the options are labelled as PEG or IGP in BIOS.

 

Thanks for the sanity check

If it works in the way I've been reading you don't get the option to actually turn off the dGPU. It's either the iGPU with optimus switching, or just the dGPU?

 

Ah yes, I just checked my BIOS options, it's PEG or SG, so I think that means the NVidia card and Optimus Switching respectively. I have the TechInferno svl7 modified "A12 BIOS with SATA III tweak", so that has extra options in the BIOS, and one of them is to enable integrated graphics as the primary graphics adapter, so that might work for him if he flashes that one?

 

Might be worth a try but I think the default is the iGPU as primary, and the dGPU still gets activated only it switches to it when required?

My big concern is the if the OP makes any mistake and sets the dGPU as primary it will be bricked (if the dGPU IS faulty) as well as still cause a power loss? Would require the card to be removed and a BIOS drain or blind flash .

Could well be worth a try though with care, but I'd still wait for the replacement PSU just to eliminate that - remember he has had two PSU's on other devices go out so power supply is questionable, and all it will achieve is the machine may start but still not have a dGPU - same as removing it.

 

Yep, I think your recommendations make more sense, I would go with your recommendations!

 

Regarding the BIOS gpu switch: Is it even possible to turn it off without the dGPU built in? Should the problem not be in the PSU

 

No idea I'm afraid but I would recommend that at this stage all you do in the BIOS (if anything) is 'default' it. Hoping the PSU is the issue although a faulty dGPU is still a possibility ;(

 

bad news: I got a new PSU and does the same thing as the old one. Guess I'll have to buy a new card
Is the heatsink for the 560m and 580m identical?

 

That's a shame, but worth the try. The heatsinks wont be identical but close enough to not be a problem. Main thing is to try to get the x-bracket with the new card. The height of the posts is specific to the core thickness and is what sets the pressure of the heatsink onto the core. It should come with a new card. The 560 might be the same core (research needed) in which case all you need to do is identify the pad layout and thickness's (which again may be the same).

See this thread, http://forum.notebookreview.com/ali...x-580m-78c-throttling-see-post-191-later.html a long trawl but I'm sure there are pics and diagrams in there. Personally, I'd just go for the dell new part (number in post #1). I was quoted £24.00 for it and it comes with all the pads, paste, and putty pre-applied so just a screw together job.

Good luck

 

I decided to buy the 560m again because I couldn't find a cheap 580m (only 300€ and up + fees which is too expensive for me).
It should be fine if i just replace the pads+paste on my heatsink right? How do I find out which thermal pads I need to buy? I removed the heatsink from the card and there seem to be 3 different pads.
Or should a just contact dell for those pads?

You're a saint for helping, thanks a lot!

 

The pads are reusable so you'll only need to take them off the old one and put them on the new. Watch out for the putty as this can get in the way if not placed correctly (assuming it has the same as the 580m). Just get some thermal paste and clean the core contact areas with some good spirit (IPA). Easiest thing is those medical cleaning swabs for injection sites.

You'll probably see a lot of different recommendations on the make of paste but IMO the cheaper stuff is OK for most people. I found a difference of around 5C between the best at the time and a cheap one. Easily a variation that could have been down to me and not enough to make a difference unless I wanted to over volt it and push it to the limit.

I don't know if dell sell the pads as a kit but worth asking anyway, they cant be too expensive? I got some sheets from a site that specialised in PC water-cooling blocks.

At a guess you have 1mm, 1.5mm and 2mm pads although I have read you can stack then without issue. A really OCD way to see the exact thickness you need is to assemble with a small ball of putty at the centre point of each component without the pads (not the core). Disassemble and measure the thickness, add a small amount a for the pad thickness as they compress - a small amount of talc stops the putty sticking

 

Excellent idea on the balls of putty method, I didn't think of that when I did mine!

 

Last question: would this work as replacement of the old ones?

 

That's the stuff, only you will probably need 1.5mm as well for the thicker ones and maybe even 2mm. You can safely double up the 1mm ones but I check the clearances with putty (and judicious use of a vernier calliper) and round up to the nearest .5 mm.

I am VERY OCD though....

haha.

 

Yea i did that and even though they look different they're all 1mm thick. Could it be that they become 0.5mm thinner with Age and pressure?

 

I'm sure there were two different thickness's on mine but as it was pre-padded all I had to do was put it together. And not the same card anyway

If you're saying the clearance with putty is 1mm then you need 1.5 to allow for compression. They do loose some thickness in use, that's the dents you see but they are around .25mm or less thinner. The reason for pads is that the components are not manufactured to the same tolerances as a core and can come from various suppliers. They will be within a tolerance so a compressible pad will suffice. Also those components get hot over a period of time so speed of dissipation is not as critical. A core can jump 15c~20c in seconds so it needs a speedier transfer.

Main thing is that the pads are getting some compression but not too much that it might hinder the core pressure. Screw it down, leave overnight and open it up looking for dents maybe, (and check out how a small amount of paste on the core gets spread).

If the original pads are in one piece then I'd just just reuse them.

 

Good you told me, 1mm might have been too thin.

imgur: the simple image sharer this is a pic of my heatsink, the right yellow one Looks like it Needs replacement also they're a bit dirty.

 

Hey, just had a look at the pic of your heatsink. The large VRAM green pad to the right of the core is not placed correctly. I can tell by the indents in the pad that the entire surface area of the VRAM chips are not being covered - specifically the pad is too far located to the right. This is also evident in the green pad being ripped & broken down in two areas where the pad has impinged on 2 other chips that it shouldn't be touching - pad is too far to the right.

 

Further to Robo's comments, I am also wondering about the core contact pattern of the paste? Looks very much like there is different pressure across it. The lower half looks perfect to me with barely any left, and evidence of squeeze-out so not a case of too little paste .

Those squeese-out areas Robbo talks about (top and bottom of right hand bank of memory) are what is probably interfering with the core contact, as well as stopping the pads reaching the components.

Fix that and you should be golden

 

Well it came like this. But I had never had Problem with overheating etc. so it was okay I guess.
Thanks a lot for your advice, as soon as i get the components my Laptop should be as good as new

 

Wow, it's a bit disappointing that a heatsink can be shipped like from Dell with the pads misaligned like that. Well, it should be better when you get your new GPU & move those pads around into the correct position like you say.

 

Not that unusual though. The 580m overheat problem was mostly down to poor paste/pad jobs from dell - or in my case a a pad was missing altogether!
I had a feeling that the Chinese factory had put the children on this task (jk)...