XPS 15 9550 temperature observations (undervolt + repaste)

Are you poking fun at my technical drawing skills?!?

Regardless of temperature improvements, since the factory VRAM thermal padding is defective, I still think the repadding helps at the extreme limits. After repadding, the VRAM chip will have proper cooling to the heatsink. Thus, the VRAM chip should be able to continuously work harder at max temps. So you get significantly more performance at that temperature range.

The impact might be more significant as one throttling chip might knock all 4 VRAMs out, whilst fixing one VRAM's cooling could bring all 4 chips back to speed (but not quite sure how the GPU thermals address one excessively hot VRAM chip). . .

 

Finally had a night off from studying. Went to a buddy's house and played Rocket League for a couple of hours. The hottest the VRM sensor got up to was 82*C. No idea what it would have went up to stock, but my guess would be quite a bit warmer.

I will post a graph tomorrow with clocks and temps just for fun.

 

Sounds promising, I think I will give it a go.

 

What are the implications of changing the Windows power scheme on this TDP throttle?
Initially I switched to "maximum performance" which kept the CPU clock at base level at all times (thinking it would reduce these power related clock drops) but since it did nothing when the PL reduction kicks in, can we actually reduce our heat levels by allowing the CPU/GPU to downclock when the load is not as heavy (like main menu, idling etc in gaming sessions)? A bit less heat would help, but on the other hand this setting might also change the way fans operate?

 

So here is a little chart that I put together, it's a bit jumbled up but I think it gets the info out:



I made a %Max Clock - this just takes the cpu frequency and divides it by 3.5ghz to show what percentage of max turbo the CPU is running. I don't know what the drop in PL1 wattage was about, but CPU and GPU clocks were still near max.

 

Last week a Dell technician replaced the board and cooler of my XPS. I did a bit of testing with three scenarios, again playing Dota2. For the undervolting I used the settings CPU -170 mV, Intel GPU -100 mV and CPU cache -100 mV. The mod mentioned by @custom90gt is not yet implemented.

• red: undervolted
• green: repaste and new thermal pads, undervolted
• blue: repaste and new thermal pads

http://imgur.com/a/0wvoC

EDIT: repasting made a big difference, but I do not know if it was the thermal paste or the pads that did the trick, since I replaced them together. Undervolting does not seem to help.
EDIT2: after I got the new board and made the red recording I uninstalled Intel Thermal Control (or something like that) and reinstalled it again. No clue if that has something to do with the better performance.

 

Glad the replacement went well.

Funny, on benchmarks I was seeing most thermal improvement from undervolting (eg say 11*C from undervolting and 5* from repaste/repad). That is not atypical for 9550 users.

Also, for the 9550 I think you need the CPU and CPU cache undervolted at same amount. Check the ThrottleStop thread to confirm.

I remember you have complete deleted the defective Intel XTU.

Did you turn on ThrottleStop? It is a bit confusing. at the bottom right corner of the control panel, you will see a button that says "Turn off" when it is ON.

Also check to see that your undervolts are actually set. Click on the FIVR button and you should see the FIVR control panel. In the upper right corner you should see offset voltages with your three undervolt numbers (negative numbers).

Also review the simple ThrottleStop guide by Eason one more time to make sure you did not miss a step or misconfigure something. It is linked in his signature of anypost

 

@pressing yes, I will read up on ThrottleStop, since I am also a bit confused by what I found. But note that playing Dota2 for 30 - 40 minutes is not actually a Benchmark, since games play out differently every time. Additionally, room temperature was +- 2°C the same for all three test. Intel XTU is completely removed from my system, since I did not install it after I but a clean Win10 on my XPS.

Now to the funny news. Yesterday I played again some Dota2, with throttling to 1000 - 1200 MHz while CPU Temps were around 60 °C and the Ambient sensor reported only ~80 °C. So I do not know if the replacement of the board actually fixed the temp issues or if the throttling yesterday was regarding some software problems.

Also I complained to Dell about coil whine, since my board made buzzing noises. The new has a high pitched coil whine only when
connected to the PSU, this is driving me bananas...

 

What does your VRM sensor say when it throttles? I was able to play Rocket League for an hour at 3.3Ghz without any throttling.

I don't have any coil whine with my system so I can't comment there. I would say have them try again with replacement if that's the case.

 

With Ambient sensor I mean VRAM, or at least the small chip marked with a red arrow by GoNzO a few pages back. This was around 80 °C. Today I played again and the throttling went only to 2200 MHz, with CPU around 75 °C and Ambient/VRAM of ~95 °C. I am really confused by the variation in temperature and performance under comparable conditions.

Could you tell me how thick the thermal pad are, which you put on R22 chips? I would like to test this next before Dell replaces the board again ^^

 

Could undervolting also decrease the amount of coil whine noise?

 

try it and let us know

 

The thermal pads are 1mm thick.

 

custom90gt, GonZ0 and Brian Anderson- I really appreaciate your help and thank you for taking the time to document your VRM efforts here.

Tonight I took a stab at cooling the VRM.

1. MOSFETS
My thesis was that the MOSFETs are likely getting hotter than the R22 chokes; also heat is especially detrimental to performance of MOSFETs. Some roadblocks:

A. Opening the laptop reminded me once again of the staggeringly small scale of parts; the AO chips are about 3mmx5mm and the TI chips are just a little larger.

B. The distance from the MOSFETs to the bottom case looks to be approximately 5mm, stacking tiny pads which are taller than wide without any reasonable adhesive will be unstable. They will get quite inefficient at transferring heat when so thick and triple stacked

C. The MOSFETs have related ultra tiny resistors next to them that I worry will not properly covered by the thermal pads and might effectively be isolated from any air cooling at all. Someone who is really careful with very fluffy thermal pads might be able to make that work but not me

D. Given the tiny scale of the chips and the thick clay density of the Fujipoly 17 W/mK (which is all I have), making the pads too thick could, with a bit of pressure on the case bottom, knock a MOSFET off the board. I just don't have the tools to get the pad thickness just right

2. CHOKES
So now I understand why custom90gt decided to use thermal pads on the R2 chokes: they are relatively big, relatively robust, and close to the bottom case. Per custom90gt's recommendation, I tried 4 x 1mm pads and they all seemed to make good contact with the bottom case based on my distilled water test.

3. TESTING
I ran a few tests. On the "light" tests, the case got a bit warmer and performance improved a bit.

Unfortunately, on more demanding tests (e.g. RealBench StressTest) things fell apart. The case bottom got really hot really fast, especially under the VRM area. After ~5 minutes:

A. top of the keyboard (actually the plastic bar above the function keys and the carbon to the sides of the function keys) was hot but touchable. The heat was creeping down the keyboard after 10 minutes. You would not want that level of heat on the keyboard unless you lived in an igloo

B. back few centimetres of metal bottom case (imagine below say the f4-f11 keys) was hot, almost too hot to touch

C. most troubling was that the area on the bottom case between the fans and front of the rubber strip was incredibly hot, and I think that caused some real issues (GonZ0 described the death spiral when connecting the heat pipe to the case with thermal pads - heating up the case bottom such that the cooling fans were sucking in hot air, fighting a losing battle to cool the CPU & GPU. I could visualize that happening tonight)

Never had these heat problems prior to padding the chokes tonight

4. CONCLUSIONS
A. Using less efficient thermal pads for the R2 chokes might provide a better balance, some extra cooling and performance without overheating the case as custom90gt documented

B. High performance thermal pads for those running only light and medium intensity tasks could provide a small boost in performance when the VRM starts to heat up. However, with intense benchmarking and gaming, high-performance thermal pads on the four R2 chokes quickly overheated my case bottom, throttling the system towards 1Ghz. Also caused keyboard to get uncomfortably hot

5. NEXT STEPS
I removed the R2 choke thermal pads after a few hours of testing. Learned there is a lot of heat in the VRM department. Of course, will continue to think about this some more as there is some performance to unlock...

 

I put some 1.5mm pads (the SSD one) on the offending parts around the VRM sensor and will be removing later today. Not enough gain and yet again I came to the conclusion this isn't a gaming laptop and will throttle when pushed as expected.

 

Is it common for the fans to take a really long time to ramp down after heavy load such as gaming?

My temps are decent while gaming and they fall back to respectable idle temps but the fans are still blowing at full speed for upwards of 20 minutes after the clocks and temps come down.

Putting the computer to sleep obviously stops the fans and waking it up immediately after I have silent fans.

Sent from my LG-D852 using Tapatalk

 

Yes it has to cool the board down as well so it will stay high until all sensors are back in range.

 

Great report on your findings. I am surprised that for once my cheap thermal pads may be better than my Fujipoly ones (which were too thin for me to try on the chokes). I haven't had any problems with the laptop getting hotter than normal, and gaming for an hour didn't bother me either. I don't know how good of a test Rocket League was in comparison to say the Witcher 3 demand wise, but I experienced 0 throttling either CPU or GPU wise while playing it at 1080p. If I have some time i will try RealBench and see what happens.

So far using the laptop every day at school (often in my lap while sitting in the comfy chairs at the library) I haven't had any issues with the chokes sinked. I'll probably leave it like this.

 

Ok, thanks. I guess, I am coming from a huge gaming laptop that would have cooled a bit faster. So I was used to the fans stopping as soon as I closed the 3D application.

I don't mind the noise, so as long as it keeps things cool, no worries!

 

As it turned out, the system was undervolted for both test with the new thermal paste and pads. Apparently the changed voltage persists after turning TS off. I have now played again some Dota2 and recorded temps with HWInfo 64, this time I also report the VCORE, thus we can see if any undervolting is taking place.

• green: new thermal compounds and undervolted
• red: new thermal compounds

Undervolting helps with temps and core clock, according to my two games of Dota2. Interesting is, that the VRAM (Ambient) shows a higher temp while keeping a higher clock speed.

Undervolting did not change coil whine for me.

 

I thought ThrottleStop did the same to my laptop. It took me an hour of bumbling around to get that sorted but it has been rock solid since.

1. I think the fix was completely uninstalling ThrottleStop then reinstalling; only had to do that once.

2. If that does not work, I might have tried unistalling ThrottleStop, installing Intel XTU and running that for a few minutes, completely uninstalling XTU, then reinstalling ThrottleStop.

FYI - So that you can see the status of ThrottleStop, on the options pane under "notification area" select show CPU/GPU/GHz in the bottom taskbar. I also have an octagon shaped indicator in the taskbar that is red when ThrottlStop runs and green when not running; I think that octagon appears automatically when ticking said items in the "notification area" of ThrottleStop but don't remember. Regardless, based on the MHz indicator I know when ThrottleStop is running (e.g. in idle CPU is at ~3.1GHz when ThrottleStop is on)

 

You can install Dell command power manager and set thermals to performance. I thought it was software only but it changes the BIOS as I have formatted since I last installed it and the laptop never went over 4000rpm (it got very hot)
It turned out it was still set to cool mode so I set it back to optimised and it ramped to 4900RPM straight away!

It also turns out my last repaste was badly done as I was hitting 95c, the transfer was good a see through layer on the CPU but I tried MX2 and got the same then went back to the Gelid GP2 and shaved nearly 10 degrees off. Fekked if I am my hundreds of repastes know why.

 

[edited article link]

Optimized Balances performance, noise, and temperature.

Cool Cooling fan speed is raised to maintain a cooler system surface temperature. Your system may produce more noise and experience reduced performance.

Quiet Cooling fan and processor speeds are lowered to reduce fan noise. This configuration may also reduce system performance and raise system surface temperature.

Ultra performance Both processor and cooling fan speeds are increased to produce higher system performance. This configuration may also produce more noise and higher system surface temperature.


Can we make these changes without installing Dell Command Power Management? Or do we need to install it, change settings and uninstall?

I see a optimized [edited], cool, quiet and ultra performance on the dell article; are you now on the ultra performance?

I read a lot of people reporting very consistent performance with MX2, even a few years after pasting.

Some of the overclockers are reporting great results with "uber pastes" but heat issues after a few months so in a continual repaste calendar. I haven't noticed any significant degradation of the Gelid paste over the past few months, so we will see how that goes...

 

Just running optimised now & you can't make the changes in BIOS but the program doesn't run at startup, you load it, make the change and close it. uninstalling it doesn't change the setting you picked either.

As for the gelid, I think I just had a bad application?

 

Thanks GonZ0, I just bumped up to "ultra performance".

Now fans are a bit busier (a bit noisy). Intense CPU activity in a very large, real-time, highly latency dependent music synthesizer program has improved measurably. Concretely, dropouts and clicks from the CPU "falling behind" have fallen considerably. Not earth shattering but it allows me to reduce buffer size by 1/3, decreasing latency by a similar factor. YMMV on other software

[EDIT - to be clear, "ultra" did not improve performance by 1/3. System was almost good enough to reduce latency by 1/3, but "ultra performance" pushed it over the edge. Maybe a 5-10% bump up in this particular application at the extremes. Sorry for any confusion]

Interestingly, "ultra performance" removes audio clicking when using my new USB2 DAC (which has its own ASIO drivers by RME). Those problems existed this afternoon when not on "ultra performance":

1. When listening to youtube on M$ Edge at 1.5x speed
2. When changing brightness with the f11 & f12 keys

Unfortunately, listening via the internal speakers (with RealTek drivers) still have clicking in scenarios 1 and 2 above...

 

Restarting my XPS does reset the voltages, thus I do not bother reinstalling programs for now. The icons for the activity and temp are already in my bottom right corner

 

hmm ... I am definitely getting some CPU throttling while gaming, even on High Power settings. My i7 clocks drop from around 3400 (with turbo boost) down to around 1500 once the throttling begins, albeit keeping the CPU at a respectable 60ish degrees.

If I inhibit the throttling using Throttlestop, or by removing the Intel Thermal Framework, my clock speeds stay up at max but my CPU temps start reaching 90+ degrees. Not sure if that is a comfortable temp for the CPU.

 

Well it can do 90 but will BIOS throttle regardless of any software once it hits it's limit. Did you repaste?

 

This thread has really been confusing me after I left. I never had temp/throttling problems with throttlestop to undervolt 150mv. I peaked at 85 gpu and 78cpu with full turbo boost, no throttling with ts open. What's been the problem? Vrm heating up?

 

I have set the undervolt to -150mv and disabled turbo on throttlestop. Still, after about 15 minutes of gaming, my CPU clocks are dropping from ~2.2k to ~0.8k with TS open and "on", and getting major framerate drops. The CPU temp doesn't get much higher than 80 degrees before the throttling begins.

Am I missing something in TS that I need to adjust?? I'm on AC power, Windows power setting is set to High Performance, Dell Command Power Centre is sent to performance.

I have not repasted yet. I guess I'll have to try and see if that helps ...

 

Get the repaste done

 

Try setting clock modulation to 100% as well

 

As many people report here, including me, CPU temps are usually not the problem. Downlaod HWInfo64 and have a look at the Ambient, or VRAM temperatures. If this sensor reaches about 95 to 100°C the system will regulate the CPU down to produce less heat.

 

Is this the how to undervolt CPU ( core volt offset)?
I was able to keep Cpu temp 35c-40c on normal use, Under 60c when gaming
undervolt cpu 52c when gaming , 35c normal use
How ever 960m video chip is going all the way up to 90c is that normal ?

 

Did you adjust "offset voltage" for both the CPU core & CPU cache same amount? I think that is a requirement for the 6th generation i5 and i7 but not 100% sure. Review Eason's guide again just to make sure.

Also given the mysterious Intel ops, ThrottleStop has some well educated options that are designed to cover a lot of chips, but don't necessarily work as you might expect on chip X. Consequently, some of the options require a bit trial and error to see what is applicable for their system. "Disable Turbo" is one of those options; I didn't notice it was impacting my 6300HQ so don't check that box in ThrottleStop.

There are two well documented thermal defects on the 9550:
1. Lousy thermal paste on CPU & GPU
2. Defective VRAM thermal pad (one or two chips)

After these fix, users' systems seem to see improved benchmark thermals in the 5*C range. But more importantly, you clear thermal bottlenecks that throw you into throttling range way too early (e.g. one uncooled VRAM chip gets hot prematurely and throws the entire system into thermal panic)

There is a third issue but that is not really a defect as it is the nature of such a tiny system. Dell still could have done it better:

3. The voltage regulator module (VRM) gets pretty hot and throttles a lot of gamers. Cooling the VRM is a tough one. There is a recent post from a guy who really understood the issue; he just blew a small fan towards the f7 key. Still throttled from VRM but was much more useable. I think one of those thin USB cooling pads with a fan underneath would make a big difference, just make sure the fan blows in the VRM area.

Also you solve these problems and still only have 45W to play with so there is always a bottleneck!

Read the posts here at NBR for more details and keep us posted

 

EDIT: This post will be rewritten throughout this week (18/09/2016).
Upcoming changes: clearer text, Google Sheets document for the results
I welcome all your comments and results!


Went through the whole topic and combined some information:

1. Tools and Items

Spoiler: Tools/Items

(General) Required Tools:

• Torx T-5 screwdriver
• Philips screwdriver
• Flat-head screwdriver
• Plastic scribe

(Cooling) Required Items:

• Thermal paste: MX-2, MX-4, Gelid GC-Extreme...
• Thermal pads: ARCTIC, Fujipoly, Thermal Grizzly Minus 8...

2. Results
https://docs.google.com/spreadsheet...RbnUszcelJLdyMMHJxs/pubhtml?gid=0&single=true

3. Benchmarking and Undervolting

Spoiler: Benchmarking and Undervolting

Before starting your repaste/undervolt, you should benchmark your system and write down the results. This is useful to check whether or not your repaste/undervolt has had effect.

3.1 Benchmark software

Spoiler: Benchmark Software

• Furmark: do not run for this too long! Preferably 30 secs. [15] Can cause serious overheating and decrease the lifetime of your laptop!

• Prime95: most people in this topic used the small FFT's test

• LinX

• Realbench StressTest (by ASUS RoG)

3.2 Undervolt Software

Spoiler: Undervolt Software

There are two popular undervolt tools: Intel XTU and ThrottleStop. XTU is not recommended! This piece of software has significant bugs with the XPS 9550 and will throttle your system prematurely.

Thus, the recommended tool is ThrottleStop. A guide can be found at:

• http://www.ultrabookreview.com/10167-laptop-undervolting-overcloking/
• http://forum.notebookreview.com/thr...observations-undervolt-repaste.785963/page-46

Also note that you should not install both XTU and ThrottleStop simultaneously! These programs conflict with each other and will not function properly.

4. Repasting and Repadding

Spoiler: Repasting and Repadding

4.1 Repaste

Spoiler: Repaste

• Remove the bottom cover
• Take out the 6 screws from the heatsink
• Clean heatsink and CPU/GPUwith rubbing alcohol
• Apply a grain of rice sized dot of thermal paste in the center
• Put everything back together [7]

4.2 VRAM Memory Chips

Spoiler: VRAM Location

Spoiler: VRAM Information

The VRAM chips are not padded correctly in many (all?) XPS's. These should be replaced by thicker ones [11]. Thermal pads differ mainly in their w/mk value. Higher values will often result in a stiffer pad and could cause your heatsink to not fit properly (bent/flex).

VRAM Chips Size: about 14x15mm
Recommended sizes: 1.0 - 1.5mm (see picture which size to put on which chips) [12].

Note: the heatsink is very thin. Thus, it is very easily bent.

Results (will be rewritten):
Replaced the 4 vram thermal pads with Fujipoly 17 w/mk (3x0.5mm and 1x1.0mm for under the heat pipes) [16].
Re-padded a second time with thicker pads (4 vram thermal pads with Fujipoly 17 w/mk (2x1.0mm and 2x1.5mm for left side by the heat pipes) [17]. For [17], you might note that the second VRAM repadding with thicker pads had same results as first. Some people seem to have only one uncooled VRAM chip; others have two.

Spoiler: SSD Thermal Pad

Another thing I looked at was my 950pro SSD. My laptop came with the 32gb cache drive, so it did not have the thermal pad. Using AIDA64 I was idling at 40*C and hitting 63*C in ATTO. I put some 2mm thermal pad between the SSD and the bottom case and now it idles at 29*C and hit 39*C in ATTO. A fairly good reduction I'd say [1].

Spoiler: VRM Area

There has been some discussion on trying to cool down the VRM area (~page 50) which seems to be causing throttling for a few members. I have had good experience cooling the R22 chokes seen on page 27 with cheap 1mm thermal pads, however Pressing had a negative one using high thermal conductivity pads on his. YMMV [1]

Spoiler: PCH

TODO

5. FAQ

Spoiler: FAQ

• Can undervolting decrease coil whine? No [19, personal experience]
• My CPU is locked at <= 0,79Ghz! Remove your battery, hold down the power button for at least 5 secs. Put battery back, boot system. Your CPU should now be running at its normal speed again.

A. Useful links

Spoiler: Useful Links

Primer for removing thermal paste
http://forum.notebookreview.com/thr...ratch-your-cpu-die.752647/page-4#post-9662435

Primer for applying thermal paste
https://www.pugetsystems.com/labs/articles/Thermal-Paste-Application-Techniques-170/

Primer for thermal pads
http://www.overclock.net/t/1468593/r9-290-x-thermal-pad-upgrade-vrm-temperatures

Pictures of the XPS 15 9550 Internals
https://www.flickr.com/photos/gonz-/sets/72157661224025234/with/24025803239/

B. References

Spoiler: References

[1] http://forum.notebookreview.com/thr...rature-observations-undervolt-repaste.785963/

[2] http://forum.notebookreview.com/thr...undervolt-repaste.785963/page-5#post-10172570

[3] http://forum.notebookreview.com/thr...undervolt-repaste.785963/page-6#post-10175762

[4] http://forum.notebookreview.com/thr...undervolt-repaste.785963/page-7#post-10176778

[5] http://forum.notebookreview.com/thr...undervolt-repaste.785963/page-8#post-10179574

[6] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-11#post-10209497

[7] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-13#post-10214112

[8] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-16#post-10222404

[9] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-21#post-10231002

[10] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-21#post-10232284

[11] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-24#post-10236458

[12] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-25#post-10236759

[13] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-25#post-10238792

[14] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-28#post-10251902

[15] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-32#post-10263611

[16] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-34#post-10269283

[17] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-35#post-10272754

[18] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-45#post-10310951

[19] http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-52#post-10342760

C Acknowledgements

Spoiler: Acknowledgements

• All the NoteBookReview users who have posted their results
• pressing
• GoNz0
• custom90gt

 

Nice overview!
Just a note on the VRAM pads: since the cooling unit is so thin, it can bent or flex easily, thus pad thickness (or difference in needed thickness for the two pairs of VRAM) may vary from notebook to notebook. Also the squishiness seems to correlate with w/mk. A higher values means in stiffer pad, alt least for the ones I used. This could give you less headroom when you use too thick pads.

 

Nice post, I put a link up on the first page to your summary.

 

Ah Ha ! This seems to have done the trick. Adjusted the Cache as well and it seems that my CPU temps never went above 71 degrees and never throttled clocks during a 1hr Overwatch session.

Will repaste this week and hope for even more temperature improvements.

 

Has anyone put thermal pads on the chip right below the right fan ( https://www.flickr.com/photos/gonz-/24097906270/in/album-72157661224025234/ below the right fan). This reaches more than 55 degrees when I plug in my USB-c adapter and causes the fan to start running.

My thermal pads are arriving next month so I can't test it yet.

 

How are you measuring that chip's temperature? Is 55*C on that component really ramping up the fans? You could change the fan settings in the Dell Command Power Manager also.

As an FYI, several people have tried using thermal pads to the case bottom here with relatively disappointing results. A few have tried with the copper cooling pipes, just above the CPU/GPU, the chokes in the voltage regulator module. The case just got too hot in those examples (making it uncomfortable to use and crippling the CPU/GPU cooling sink). I believe everyone has removed the pads.

I think some of the factory m.2 drives have a thermal pad to the case which seems to work without generating much heat. Haven't seen any complaints there

 

Using HWinfo:


The first temperature is when the laptop start with the USB-c connected. (It's an USB-c to ethernet Anker). This causes the PCH to ramp up to 54-57 degrees and turns on the fan.
Immediately after removing the USB-c causes the temperature to decrease to 40-44 degrees within 2 minutes. Fans also turn off at that point.
Inserting the USB-c once again will immediately cause the PCH to return to it's higher temperatures.

 

I did some testing yesterday, windowed gaming with HWiFO64 in the background.

Ambient and PCH temps where averaging around 60c with the GPU under load, where the CPU was around 50c, and this was causing the fans to hit 4900rpm. Not sure how hot the 960M was cause that sensor wasn't being displayed, but I imagine it was around 65- 68c.

I will do another respaste this weekend, and replace all of the gpu ram thermal pads with thermal grizzly 8 (8W/m K) 1mm and 1.5mm. I will also remove the 2mm pads from directly above the CPU / GPU heatsinks and try placing them over the VRM thingys. I'm curious to see if it makes a difference, and will be easy to remove the pads if it doesn't help.

 

A lot of people tried putting thermal pads above the CPU/GPU and I thought everyone had removed them. I guess not! GoNz0 wrote a detailed summary why that was not optimal for him. (Something like the case bottom got hot, then the keyboard started to get uncomfortable. This also heated the air being sucked by the fans, indirectly crippling the CPU/GPU cooling system).

This month, I believe that custom90gt, GoNz0 and I all tried heatsinking those six R22 chokes in the VRM area. custom90gt has lower conductive pad that he still uses. I think GonZ0 removed his. I removed mine. For me the case bottom got really hot in a few minutes, etc...

Additionally, this does nothing to the MOSFETs which are likely hotter than the chokes.

FYI My thermal pads were 17W/mK but that really does not matter; just indicates there is a lot of heat in the VRM and that the case bottom is not an effective heat sink.

I am trying to figure an easy way to get some air circulation around the VRM but not having much luck. Thinking about:

1. an air intake with a hole in the case bottom

2. an air exhaust at the back by clearing the foam which is behind the VRM and infront of that grille.

Haven't figured how to get a small fan underneath yet...

 

I'm just lazy, thats all. I've been waiting for my new 1mm and 1.5mm thermal pads to arrive before opening the base and doing a full repaste. Also, because I have a laptop stand with fan the machine never touches my lap, so I don't care how hot the bottom is. But after reading this thread I had a closer look at HWiNFO and realised the 'Ambient' temperature is too high, and likely the cause of excessive fan noise and the occasional throttling when gaming.

A while ago I had an older generation xps 14, which had some serious over heating issues. I did a repaste and ended up drilling small holes all over the bottom cover, along the heatpipe, everywhere I could. I also removed all of the mesh covers from the grilled areas of the base cover. It made a huge difference to the cooling system, along with using a stand to keep it off the lap. But the 9550 has an aluminium base, so its not such an easy thing to do :/

 

I never removed mine. I haven't had any adverse issues and the VRM is a few degrees cooler than it was before. It's likely the mediocre thermal pads that I have because I do not have any issues with the laptop on my lap (it's been on my lap for the last hour while I study).

I also noticed a huge difference in putting a thermal pad over my 950 pro in my last 9550. I had the 1TB HDD version before and it didn't have the thermal pad. Putting one on the 950 pro when I got it kept it much cooler and improved performance in long benches.

 

Apologies!

 

Don't apologize, I haven't been super active in this forum lately. I do however wish I would have recorded VRM temps for a few days of normal usage to see if sinking the R22s had any real effects...

 

I have this problem as well. I have an USB-C to USB-A adapter and just plugging it in causes the temps to go up and the fans become very loud. The normal 2500rpm are not too loud, but attaching most USB-C accesories makes the machine very loud. This also happens with HDMI adapters.

Is this problem solvable or is this a bad design of the machine by Dell? Is it not possible to attach anything to the USB-C port without having extremely loud fans and heating up? I thought this was a problem months ago with the docks and it would have been solved by now. Is this also the case with the Dell XPS 13?

 

I just did a second repaste and replaced the thermal pads on the GPU ram. Temps have improved a little, no throttling during gaming. Ambient / 'PCH' temps around 50- 60 whilst gaming, GPU around 60 and CPU around 50, which is all very healthy temps IMO. This time I moved the thermal pads from directly over the GPU and CPU and placed these on the ends of the heatpipes next to the fans. I also placed thermal pads on the R 22s. CPU is undervolted by 160mV. I tried underclocking the GPU but this made a minimal difference to the temps.

The thing that still bugs me is the noisy fans. It cranks up to 4,900rpm very quickly, even with healthy temps as described above. HWiNFO64 has an option to control fan speed manually, but it doesn't seem to work. I'd be very happy if I could cap the fan speed at 4,000rpm or maybe 3,700rpm.