XPS 15 9550 temperature observations (undervolt + repaste)

Joe - here is a post to get you started on cleaning the old paste off...

 

A bunch of thermal photos of the 9550's twin from Precision thread follow. BOKEH really heats up his laptop with Furmark, running the laptop closed, etc.

FYI - The 130 watts etal. is power at the wall outlet. He is generally running laptop on a desk, but quickly turns it upside down for a few photos.

 

Here is my throttling-thwarting idea (which I hope to implement shortly): connect the chokes to the heatpipes using something (a thermal pad?) that doesn't touch the case (or is thermally insulated from the case).

Here is why I think it might just work: I need to cool the VRM area because I am certain that I get severe PL1 throttling the instant that the VRM sensor hits 100 C (which happens to me during games that stress both the CPU and the GPU). At the same time, when I place a thermal pad between the chokes and the case, the case heats up immensely. Not only is this uncomfortable, but also this causes various other sensors to heat up, which in turn leads to PL1 throttling due to one of the other sensors reporting high temps even while the padded VRM remains relatively cool. Meanwhile, my CPU and GPU temps remains comfortable (low to mid 70s), so the heatpipes/heatsink assembly have some cooling capacity to spare.

Does anyone have any suggestions on the best way to thermally couple the heatpipes/heatsink assembly with the VRM chokes, without thermally coupling the chokes to the case? What about adhering a thin (thin enough not to touch the case) thermal pad over both the chokes and the nearest section of heatpipe? If I wanted to put an insulator between this thermal pad and the case, what insulating material could I use?

 

After installing Dell Command Power Manager I discovered it was causing intermittent throttling, for no apparent reason. I wanted to cap the fan speed to reduce system noise (after repasting and undervolting my temps are within healthy ranges, so no need for 5,000rpm).

During light gaming (Hearthstone) the CPU would sometimes throttle to 800Mhz, and the GPU VRAM would also often be throttled to 800Mhz, despite the system being well within healthy temperature ranges, and the CPU was underclocked to avoid excessive power draw. (Note that Hearthstone is pretty light duty for the GPU, average GPU load is 50%). So the the system should not have been throttling, and I had not experienced throttling prior to installing Dell Command Power Manager. Changing the fan profile from Silent (which caps the fan speed) to Optimized (default) or High Performance would often stop the throttling, but not always. Also, throttling would almost always occur on battery, and on a number of occasions I experienced throttling (for no reason) when using the Optimized or High Performance profiles.

After uninstalling Dell Command Power Manager the throttling went away instantly. This time I did some more intensive testing using Unigine Heaven for 30 mins+ whilst bench marking. I can report ZERO THROTTLING, so I would strongly recommend anybody experiencing throttling to check if Dell Command Power Manager is installed, and if so, uninstall it to see if this helps.

It's also interesting to note, and GoNz0 mentioned this earlier, after uninstalling Dell Command Power Manager my fans remained capped to a maximum of 4,100rpm, so I now have the best of both worlds. Make sure to select your desired fan profile before uninstalling.

I originally discovered this because I was testing the difference between having thermal pads on the R22s vs not. I found from my testing with Unigine Heaven that placing thermal pads on the R22s gives around -8c to the 'Ambient' sensor, and +1 or 2c on the GPU and 1+ or 2c on the PCH. So I think this is worthwhile, no to thermal pads directly over the CPU / GPU, but yes to pads on the R22s. This time around I recycled the cheap, stock 1mm pads from the VRAM, which I chopped into quarters and double stacked onto each of individual the R22s. It was a bit fiddly, but worth the effort, and from feeling the bottom of the machine it was not as hot as last time I tried this using 2mm Thermal Grizzly 8 pads.

One other thing, if you can live with a capped CPU speed of 2.0Ghz (or 2.4Ghz), during gaming, this can reduce the PCH and overall system temperatures due to the reduced power draw. For less CPU intensive games the FPS difference is negligible.

EDIT: ok I feel so dumb right now. After posting the above I had a game of Hearthstone and the CPU was throttling to 800Mhz very quickly. This is after extensive testing yesterday without any throttling. The only thing I can think of is that I've rebooted the machine since yesterday, when I uninstalled Dell Command Power Manager. I am sure this is some kind of driver bug which is related to Dell Command Power Manager, but uninstalling it didn't fix the issue. I might have to reinstall it, change the profile back to Optimized and then uninstall again. But I much prefer the lower fans speeds so I am kind of torn now, and I am not so confident this would fix the problem. Perhaps the throttling is bios related?

EDIT 2: ok so the throttling is definitely software or driver related. After posting the edit above I managed to trick the system to stop the throttling. Don't ask me why this worked, but I loaded up Hearthstone, windowed, so I could keep an eye on the clock speeds etc. The first thing I did was underclock the GPU by 100Mhz, and the throttling was gone instantly. I played for around 10 minutes with zero throttling, and then set the GPU clock back to the default and played for about 2 hours, with no throttling.

 

I would put dell command back on and set it to the normal profile as it caps CPU to keep the fan capped at 4k
Took me a fair while to figure this out.

 

uninstalling the dell command gave me 1h+ of battery life extra.. It somehow messes with the power consumption. I would never see anything below 4W when idle, now i have a stable 3.2

 

OK, dont know why this works, but underclocking the GPU seems to stop the throttling, even if you change the clock speeds back to default (but not when on battery). So I did not have to reinstall Dell Power Command, and my fan speed is capped at 4,100rpm and no throttling (with AC power). I found this out by accident when bench marking.

 

I recently reapplied thermal paste and cleaned out the dust recently and I have been getting around 70-75C while gaming. However, I have the following questions:

1) My thermal pads are becoming less sticky and have collected dust on it - does this affect temperatures? Should I replace them? They also do not cover the GPU cores completely.

2) Where are the R22's and should I be using thermal pads on them? I see mixed advice - pressing says that most take the thermal pads off after people realize it makes the bottom of the case very warm.

3) I was not careful with my laptop and stripped a screw of the heatsink so that it pops up slightly there and is not completely screwed down - would this be a contributing factor to my higher temperatures? If so, how can I resolve this?

4) Are there only positive benefits to undervolting given I don't take it too far? I've seen the spreadsheet and it seems people do around 150-180m.

 

1) Are you asking about the GPU VRAM themal pads? They are not very sticky out of the factory. Since they are covered on both sides they should not have dust. Ideally, they should cover the VRAM chips completely. There are some users who have recommended replacing one or two of those thermal pads with thicker ones; that is process and the benefits are discussed earlier in detail in this thread.

2) The chokes are silver boxes labeled R22 are clustered roughly below the f7 key "above" the CPU. If you look at the very first post of this thread you will see clear pictures of them. I think only one user kept themal pads on (everyone else who tried them removed them AFAIK).

3) That stripped screw is a problem. The screw should be softer metal than the "female" metal base on the motherboard so that might mean you just need a new screw. Otherwise, rethreading or replacing the "female" threads will be a bit of work m8!

The following preliminary thoughts are at your own risk of destroying your computer. They are based on my experience with heavy equipment, not delicate electronics. So get a local computer tech/shop to fix this. He might first find a way to remove any metal scraps around and inside the offending hole. Maybe a magnet and / or a specialized vaccumm?" (is the "female" hole closed at the bottom or could metal flakes have fallen through to the other side of the motherboard?) Then look at the "female" threads to see if they look stripped. If not, maybe test a "good" screw in the offending "female" thread. If that works, try buying a new identical screw. Repaste once again as the heatsink was removed.

The screws seem to be identified and sold by these guys
http://blog.parts-people.com/2016/04/18/dell-xps-15-9550-p56f001-heatsink-removal-and-installation/

4) Undercoating should help prevent rust on the bottom of your car hehehe.

For undervolting, start with the primer-article by user Eason and then search from there. The i5 and i7 will have different undervolting potential. And everyone's system is different. Based on my own experience and what is documented on this thread, I think typical undervolt ranges for the 9550 are a bit lower than you are suggesting (maybe closer to -130 to -170mv). A few guys had their i5 up to -180mv but subsequently reduced that I think. A couple of the pros here do not undervolt.

 

Need pics of the damage.

Sent from my SM-G920F

 

I actually had a screw pop through the motherboard- i.e. the little part where the screw top catches on the mobo actually pulled itself through the mobo! Nice and clean- no way to fix it. Dell sent a guy over to swap my mobo.

 

Hey Eason - out of curiosity, how was that female receptor designed? Is it like a rivet with flattened metal on the opposite side of the motherboard? Is the other side a flat, wide diameter nut? Is it just glued in?

What did Dell say about that defect? Did they pushback much?

 

Does anyone amongst the forum experts who has successfully undervolted and repasted the cpu of their 9550 fancy helping out a fellow member and doing another, namely mine? Had planned to attempt it myself but haven't managed to and could really do with having it set up and ready to use (clean install and all known fixes installed etc) as currently studying for my first I.T. certification.

Based in Oxfordshire, I'd drive to you and would obviously be willing to pay something for your time.

You'd really be helping me out.

Lmk.

Thanks.

 

You can see what it looked like on the right side of this pic, if I understand you correctly: https://img2.parts-people.com/ppc-dell-parts-image.png

I think on the other side it was a flat glued in nut. The pressure from the heatsink (maybe vram pads were too tall?) pushing up against the screws pulled it through the mobo I think.

Dell actually came to replace my mobo for the headphone jack bug, and when they opened my laptop they saw that the heat sink had popped off. My Thai isn't that great, but he just seemed amused about it.

 

Figured I may as well look into which thermal paste and pads to order. Currently looking at Thermal Grizzly and Arctic. Will there be much difference between them?

Paste:

Arctic MX-2 - £3.95 (8g)
Arctic MX-4 - £3.95 (4g)
Thermal Grizzly Kryonaut - £5.80 (1g) or £14.11 (5.5g)

Pads:

Arctic - £5.18 (0.5mm)
Arctic - £5.10 (1mm)
Thermal Grizzly Minus Pad 8 - £4.74 (0.5mm)
Thermal Grizzly Minus Pad 8 - £5.82 (1mm)

Prefer to only purchase once, so whichever one has provided the best results.

I already have ArtiClean Thermal Material Remover and Thermal Surface Purifier. Is there anything else that will be required to carry out this modification?

Thanks.

 

I suggest you read this thread from start to finish and by the end you should be able to decide for yourself

 

Thanks for the quick reply, and sorry for the late post!

1) I am referring to the thermal pad on my SSD as well as the ones on my GPU VRAM (I've only included a picture of the one on my SSD). From my screenshots, you can tell it doesn't cover my SSD entirely (should I just stretch it out a bit?) and are covered in dust. The same is with the thermal pads on my GPU VRAM.

2) Got it - I'll just stay away from this one.

3) I've attached pictures of what is actually two stripped screws on one side.. I'm pretty terrible with my hands and definitely won't try to do any further damage.
Eason mentioned that dell came to help replace a motherboard. Because my laptop is still under warranty - would I be able to have dell help me with these stripped screws?

4) I undervolted my computer to some recommended settings -140/-140/-100 and its lowered my temps by an average of ~8C with max of 80C when playing games, and will be looking to overclock it just a bit as well.

Thanks for all the help - really appreciate it.

 

You have not stripped the screw the mounts pulled out. this is a full rebuild as it requires a new palm rest, Dell will replace this under warranty so long as you don't say you did it, tell them it is overheating and let the tech come, he will "discover" the pulled mount and order a palm rest. You could wash the SSD thermal pad but don't stretch it or it wont contact. It doesn't have to fully cover to work, it does a good enough job as it is.

Do not use it for anything that will cause it to heat up as the heatsink isn't fully in contact.

 

Is that faster mount pressed into the board at the factory like a rivet might be? Do they engineer it to release before breaking the board (like they engineer the screws of a softer metal than the mount?)

That is a dusty environment! To help prevent that, I clean around the desk and vacuum the floor frequently, I purchased an air purifier, and I take the laptop apart on occasion to carefully remove dust.

 

I don't have premium support unfortunately so my only option is to mail it in. Therefore I won't have the luxury of explaining that I didn't do it to an on-site technician. On the other hand, I do have the Accidental Damage Service which does cover "drops and falls" which I explained to the phone rep. However, I have previously gone in already and scraped off the stock thermal paste and reapplied new paste - would they realize this and perhaps null my warranty/service?

Should I not sent it in and try to get it fixed instead at a reputable laptop repair shop?

Edit: Thought I would try to get Premium Support except it is $200 for 1 year

 

Dunno, doubt it as it isn't repairable once pulled out.
That pad looks like it's been used to wipe out a vacuum cleaner

No, don't worry about it. I doubt they would give a **** at the local repair centre. Let me put it this way, when we did contracts for manufacturers my boss made it very clear that we should get things fixed regardless unless it was water damage we couldn't remove evidence of as it took longer for no extra money to pratt around filling in forms, contacting the support of said company to see what they want to do, taking pictures, emailing pictures, chasing up on the phone and so on when the time could be better used to fix another laptop!

This ain't going anywhere near a Dell service centre, it's going to a company paid to do the work on behalf of dell. The part shouldn't fail, the part shouldn't pull out as it is only holding a heatsink spring, the screw should strip long before that happens.

As for finding a reputable laptop repair shop the best case would be they want to replace the motherboard £$£$£$£$
Worst case is they try and glue it in and it lasts days if not weeks.

Considering Dell allow you to repaste and take them apart and you have accidental damage you have nothing to lose.

Get it sent in!

 

I bumped into a good interview with thermal user IUNLOCK on youtube. For those who don't know him, he has some great thermal paste posts here at NBR. Including the crazy / dangerous new liquid metals. Check them out...

http://forum.notebookreview.com/thr...5r3-disassembly-repaste-guide-results.797373/

http://forum.notebookreview.com/thr...t-vs-cool-laboratory-liquid-ultra-pro.791489/

 

Wow he's had great results. The conductonaut looks like good stuff, I may order some...

 

I was thinking the same but want to better understand the performance of liquid metal pastes after several months. Misapplication is also dangerous!

 

I have no worries about the application process. I typically repaste my cpus/gpus every few months so the long term isn't typically an issue for me. I would be curious how it would hold up after a year or two though.

 

Let me know how you get on. Risky with a delidded cpu as any spillages could fry the board. Only to be done with a board under warranty

 

Last spring when I was shopping for the latest and greatest thermal paste, I read a few posts about liquid metal performance declines after 3-6 months. But there just wasn't enough reliable data to draw any conclusions. If you search the threads now, I would guess there is a lot more data available...

The liquid metal thermal performance looks staggeringly good so that might become more popular as the "hot" new nvidia GPUs get traction...

 

Just a quick question. When you are talking about "VRAM" and "VRM", are they the same thing?

 

Different. VRAM = video RAM

 

You could have asked google you know

 

XPS 9550 i7 960M 512GB FHD

I did repaste 8 months ago but these days the thermal performance became worse (I was using some really crappy thermal paste). So I just repasted with MX-4 today. I also ordered some thermal pads and are waiting for them to arrive.

The results (tested with prime95 small FFTs) are nice:

Stock - about 92*C
Undervolt by 170mv - 80*C
Stock voltage with repaste - 82*C
Repaste and undervolt by 170mv - 73*C

The only thing annoying me now is that the fan noise is quite noticeable even when the temperature is relatively low (40*C). Changing "Thermal management" in Dell power manager doesn't really help.

I am not sure if that is because of VRM or PCH temperature.

I also noticed that on the motherboard there is a bare chip with a glossy surface. It is not attached to any heatsink. What is that?

EDIT

More about the fan noise.

Changing "Thermal setting" to "Quite" only can temporarily lower the fan speed (like for 10 seconds or so). After that the fans speed up as usual.

 

Based on the reported results of other NBR members, 19*C decrease is a good result of repaste + undervolt.

Are you doing your undervolt by ThrottleStop? If so:

1. as long as you don't enable the "clock modulation" or "set multiplier" it seems you can try:
control panel>all control panel items>power options> (choose either Balanced or Power Saver)

2. in ThrottleStop in the "TPL" window you can enable SpeedShift. There are two check boxes. Also the min should be 1 and the max should be max turbo rate of your processor (I think your 6700HQ is 35). Then you can adjust the EPP which determines processor performance. 0 is max performance and probably max fan. The Highest setting is max energy savings and probably min fan (it could be 255, 100 or 64 but you need to check on your system). Out of the factory, Intel seems to guide manufactures to set at an average setting (say 127 or 50 or 32).

3. Check out the ThrottleStop Guide and the 9550 Speed Shift thread for better info. Getting your system tweaked takes a bit of trial and error

http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/
http://forum.notebookreview.com/threads/dell-xps-15-9550-skylake-speed-shift.796891/

If you are using Intel XTU for undervolting, note there are some conflicts with the 9550. It put my laptop (and a few others) into premature power throttling. Remove that software from your system.

Make sure your paste job was "even". Look at the temps of the cores and make sure they are pretty close (say 0-4*C difference but I'm not sure what the tolerance should be). Look at the temps of other sensors to see if something else is getting really hot when your CPU is around 40*C; those could be triggering the fans. Did you properly replace the VRAM thermal pads and make sure the heatsink is making good contact with the pads?

Also note that the temp sensors are not so accurate at 40*C since they are "tuned" for precision closer to the 100*C range where things start to have problems. Regardless, I don't think your laptop should be running more fans after a good paste job.

 

Thank you for these helpful tips. I guess you meant "Speed Shift" in the TPL window.

I enabled it days ago using EFI shell which overwrites a byte in the firmware to enable Speed Shift natively in BIOS (tho the setting is still hidden). I believe you know that post in the Speed Shift thread.

The SpeedShift tweak significantly improves performance (in terms of responsiveness) in power saver plan. But not so noticeable in balanced or high perf plans.

But I restored the initial value (turned off speed shift) because it drains more battery than without it. I think it drains 15-20% faster in power saver plan. I go to library a lot so battery life is a big concern to me.


The fan noise it a bit better than before repasting, but still not so quiet as my wife's Macbook pro 15 (2015) under similar workload. Macbook seems to prefer quietness than lower temperature? Personally I don't care too much about the heat (unless it is unacceptable high) so I like the way Macbook handles fan speed more.

 

Check your paste job, no good having a wild guess with a new paste m8, it is a pain but bolt it down then take it off again and check the spread, clean up and try again until you know how to do it for that particular paste. took me about 4 attempts with MX2 until I was happy and it has been great since. It isn't like I don't know what i'm doing as I have done hundreds of repaste jobs in my line of work!

 

Thank you for the advice. I tried 3 times but the fan noise is still there.

Here are the temp and fan speed in idle. The temperature in my house is 25*C.

http://m.imgur.com/pXlaRnF

http://m.imgur.com/YnslSmh

EDIT

You may have noticed the core temps are uneven. It is not likely due to uneven pasting but seems to be uneven 6700HQ heatsink.

Here is a thread talking about that.

http://forum.notebookreview.com/thr...ven-core-temps-due-to-uneven-heatsink.797477/

 

Yeah the whole core temps uneven thing should be ignored unless you have a >10 degree celsius variation and even then it will normally be 2 cores higher to show it's lifted on one side. Considering we have a delidded CPU as well it is rare for it to happen and such a small die that it should spread out anyway. If you find a 10 degree difference under load it's normally time to panic.
Without even clicking that thread I am giving a professional opinion on laptop thermals from my Toshiba laptop training and on top of that a decade over overclocking and water cooling pushing chips well beyond the designated specs.

Also a CPU's thermal sensors are not calibrated, they have no screw to turn to adjust the resistance


All the above aside your temps are absolutely fine, the fans are at idle so nothing untoward with that. I assume it is light load causing them to kick in or poor ventilation?

A quiet room doesn't help but if in doubt go find a 5 year old laptop and it will drown out your 9550 fans

 

Quick question. My ArtiClean just arrived in the mail, was left outside in the cold for awhile and froze. Is it still safe to use?

 

It's from the Artic!!!!

 

Not sure why you are asking that question in this thread, try emailing the manufacturer's.

 

Let it thaw and it'll be fine. Hope it didn't explode.

 

Stock compound, undervolted CPU: 92 degrees C
Arctic Aluminum compound, undervolted CPU: 76 degrees C

Much better.

 

Is that Arctic Alumina, MX-2 or MX-4?

Thanks.

 

I'm not sure which one it is, it is from this link:

https://www.amazon.com/Arctic-Silve...qid=1482260497&sr=8-3&keywords=arctic+alumina

I found the paste in a cardboard box in my basement from my old PC building days haha

 

Has anyone here tried reversing the fan intake to exhaust fans? Just wondering if someone has tried to.Thanks for the help.

 

Sounds impossible.

Sent from my SM-G920F using Tapatalk

 

Those fan blades are optimized for spinning in one direction so volume in reverse would be low. And the engineered intake/exhaust will be comprimised in reverse.

To GoNz0's point, there is feedback wire to the fan monitoring and control so that would cause some headaches to reverse.

 

I bumped into an interesting video this week.

Apparently Dell engineers recently have organized long calls with "expert users" to improve some new Alienware laptops. Issues included thermal paste, heatsink alignment, screen lines...

Based on the calls, it looks like Dell already changed production with improved thermal paste (from some stamp) and reduced the thinner thermal pads to improve heatsink alignment.

More details in the video...

 

Have those who tried liquid metal thermal paste noticed performance has deteriorated over time? Or have thermals remained stable over the months?

 

I'm still not clear on what to do with the VRM area. It seems obvious on thermal images that it gets hot.
(Did anybody do a thermal image during a load test with the bottom open?)

Regarding getting a little air there, Dell seems to have closed the central part of the grille of the bottom chassis deliberately, leaving it open only above the fans? Anybody tried opening it?

Regarding cooling the chokes without termally coupling to the case, is this what Dell did with the pad in the red right square in the image below
https://imgur.com/a/Gr8IE
What's below it, some chokes and a thermal pad? (I recall someone removed it and put thermal pads)
This probably reduces the temperature of the hot elements below. But I'm not sure if sth like this could reduce long-term heat buildup in the top-central VRM area, which is not close to any vent. I assume taking some of the heat to the case should work better.

 

Earlier in the post GoNz0 noted that the insulation at the center of the grille is to prevent hot exhaust from being sucked back into the machine. Also opening up one "exhaust" port would probably require a corresponding "intake" port and perhaps some means of moving the air

Earlier in the post there are quite a few people who tried taking heat to the case via thermal pads and all but one removed them. I tried thermal pads to the case both for the VRM chokes & the CPU/GPU heatsink; the case and keyboard got superheated and you can see the consequences above. One guy above did heatsink the chokes with low-grade thermal pads and got pretty good results...