Take a look at our
ThinkPads.com HOME PAGE
For those who might want to contribute to the blog, start here: Editors Alley Topic
Then contact Bill with a Private Message

W530 overheating - remedy?

W530/W540/W550 series specific matters only
Post Reply
Message
Author
WizardOfBoz
Freshman Member
Posts: 104
Joined: Mon Mar 16, 2015 9:22 am
Location: Philadelphia, PA

W530 overheating - remedy?

#1 Post by WizardOfBoz » Wed Apr 22, 2020 8:51 am

A recent post by Natakranta probably answers my question, but here goes. I'm running a W530 with a 3820qm, k2000m, and 32Gb XMP. Win 10 Pro. In a dock. My CPU temp (via Core Temp) is peaking in one CPU at 95C. The dust from the heat sink has been blown out pretty well (I think). So

1) Tj is 105C, so I'm thinking that 95C is way too hot. Should max out at ~80C or so. Yes?
2) Is repasting the indicated remedy?

Thanks.

axur-delmeria
Senior ThinkPadder
Senior ThinkPadder
Posts: 2707
Joined: Mon May 28, 2012 5:49 am
Location: Metro Manila, Philippines

Re: W530 overheating - remedy?

#2 Post by axur-delmeria » Wed Apr 22, 2020 9:58 am

Properly clean the heatsink. Remove it, then separate the fan from the rest of the heatsink. This allows unimpeded access to the cooling fins, where dust builds up over time. Of course, you need to repaste as well.
Daily driver: X220 4291-C91 i7-2620M

Backup: X601 Core 2 Duo T8100
Toy: X60F Core Solo U1300
On loan: X220 4291-P79 i5-2520M
In pieces: two retired but working X61Ts
RIP: 760XD 9546-U9E; X61 7676-A24; and a BOE-Hydis HV121P01-100 in failed SXGA+ mod
:cry:

WizardOfBoz
Freshman Member
Posts: 104
Joined: Mon Mar 16, 2015 9:22 am
Location: Philadelphia, PA

Re: W530 overheating - remedy?

#3 Post by WizardOfBoz » Wed Apr 22, 2020 12:23 pm

Well, I repasted. Min temps still a bit too high for me: about 46C. Of interest, though is that core 1 (which had been running the hottest) is now about the same as the other cores. I'll do some work this afternoon, and we'll see if I get anywhere near 91C again. Thanks for advice.

I purchases a W541 to upgrade and update a little bit. May have been propitious.

WizardOfBoz
Freshman Member
Posts: 104
Joined: Mon Mar 16, 2015 9:22 am
Location: Philadelphia, PA

Re: W530 overheating - remedy?

#4 Post by WizardOfBoz » Wed Apr 22, 2020 12:25 pm

Anyone else running a 3820qm in a W530? What does core temp say for your temps?

WizardOfBoz
Freshman Member
Posts: 104
Joined: Mon Mar 16, 2015 9:22 am
Location: Philadelphia, PA

Re: W530 overheating - remedy?

#5 Post by WizardOfBoz » Wed Apr 22, 2020 1:12 pm

Ok, the min temps have gone down to 43,47,44, and 46 according to Core Temp. I think that's about 5C better than before.

Still, I'd like to see 39C or so at idle. Guess its not to be. Will report future anamolies.

axur-delmeria
Senior ThinkPadder
Senior ThinkPadder
Posts: 2707
Joined: Mon May 28, 2012 5:49 am
Location: Metro Manila, Philippines

Re: W530 overheating - remedy?

#6 Post by axur-delmeria » Wed Apr 22, 2020 1:37 pm

WizardOfBoz wrote:
Wed Apr 22, 2020 1:12 pm
Ok, the min temps have gone down to 43,47,44, and 46 according to Core Temp. I think that's about 5C better than before.

Still, I'd like to see 39C or so at idle. Guess its not to be. Will report future anamolies.
Liquid metal is the next step if you want lower temps.
Daily driver: X220 4291-C91 i7-2620M

Backup: X601 Core 2 Duo T8100
Toy: X60F Core Solo U1300
On loan: X220 4291-P79 i5-2520M
In pieces: two retired but working X61Ts
RIP: 760XD 9546-U9E; X61 7676-A24; and a BOE-Hydis HV121P01-100 in failed SXGA+ mod
:cry:

WizardOfBoz
Freshman Member
Posts: 104
Joined: Mon Mar 16, 2015 9:22 am
Location: Philadelphia, PA

Re: W530 overheating - remedy?

#7 Post by WizardOfBoz » Wed Apr 22, 2020 9:22 pm

I think that the temps will be ok for a while. Nothing above 79C today, and before the re-pasting one core got up to 91C.

I found an interesting reference that analyzed some of thermal compounds. It's here: https://www.nrel.gov/docs/fy08osti/42972.pdf

RealBlackStuff
Admin Emeritus
Admin Emeritus
Posts: 21597
Joined: Mon Sep 18, 2006 5:17 am
Location: Dublin, Éire
Contact:

Re: W530 overheating - remedy?

#8 Post by RealBlackStuff » Thu Apr 23, 2020 12:47 am

That doc is "slightly" outdated...

WizardOfBoz
Freshman Member
Posts: 104
Joined: Mon Mar 16, 2015 9:22 am
Location: Philadelphia, PA

Re: W530 overheating - remedy?

#9 Post by WizardOfBoz » Thu Apr 23, 2020 9:05 am

RealBlackStuff wrote:
Thu Apr 23, 2020 12:47 am
That doc is "slightly" outdated...
It was also more addressing automotive electronics which run at higher temps than computer CPUs. Still, I got a lot out of it. First, the fundamental physical properties (like thermal conductivity) don't change. Second, the methodology was (I think) sound and fair and gave a useful way of thinking about thermal management in CPUs. Third, it included phase change materials (like liquid metal). And it also pointed out some materials I hadn't considered before, like Shinetsu G-751 and Dow Corning 5022.

Since the stuff I actually used, Arctic 5, was analyzed I was very interested to learn that the manufacturers claim of 8.5W/(m K) was overstated by a factor of nearly 10. Even so, I was satisfied with the results: I got a drop in maximum temperature of about 15C.

Because the thermal conductivity of Arctic 5 was pretty modest, but the performance proved adequate, I became interested in how the paste works, how better pastes affect thermal management, and what the limits of the effect of pastes are. It turns out going from a bad paste joint (poor quality paste of low thermal conductivity, perhaps with loss of mechanical contact) to a good one (better paste, better contact) gives you a big drop in temperature. Going from a good paste to a great one may not have much effect at all.

I could do a long derivation, but an approximation suited my goals. The temperature difference between the transistors and coolant in the heat pipe (Delta T) can be approximated by an equation accounting for q, the heat generated and transferred out, and a term accounting for heat transfer (between transistors and CPU platen, and between platen and paste, for example, and generally labeled using the letter "h") and thermal transfer through bulk material such as the platen, or the paste. The term accounting for transfer through bulk is t/k, where t is the thickness, and k is the thermal conductivity. One could approximate the transfer between material 0 (transistors) through the platen (material 1) and paste (material 2) and heat pipe (material 3) to the coolant (material 4) as

Delta T = q*(1/h01 + t1/k1 + 1/h12 + t2/k2 + 1/h23 + k3/t3 + 1/h34)

What the heck does this mean? Let's look at the equation components. We assume that the heat generated by the chip, q, is constant. You want a small delta T (transistor temperature kept close to the coolant temperature). That's the Delta T. You can see that Delta T proportional to the terms in the parentheses. Those are individual thermal resistances. The setup is like a layer cake. Transistors generating heat at the bottom, a layer of aluminum (the platen) on top of that, the paste on top of that (frosting! Mmmmm!), and the heat pipe on top of that. So if any of the individual layers in the parentheses is really big (that is if we have a layer that's highly resistive, or insulating) relative to other terms, the other terms don't matter much. Likewise, making a small term even smaller may not matter. This tells us that if (for example) we lose mechanical contact between platen and paste (due to aging or poor quality paste) then the local heat transfer coefficient h12 gets very small and the term 1/h12 becomes very large. SO NOTHING ELSE MATTERS. If we apply new paste and it has pretty good thermal properties (the thermal conductivity, k, is high) and good mechanical properties (the viscoelastic properties are such that the paste spreads out an thickness, t, is low) then the term t2/k2 is small and the delta T depends upon all the other terms in the parentheses. That is, you could make the paste infinitely good (k2 infinite, and t infinitely small), and there would still be a significant delta T determined by the other resistances. Another way of looking at this is that if we have pretty good paste, with low t2 and high k2, then t2/k2 may be pretty insignificant - and spending money for fancier/better paste may not have any noticeable effect!

One last thing: it seems to me that the loss of mechanical contact is the big issue to avoid. Assuming that you have cleaned out the heat pipe heat transfer fins, if you have high temperatures, it's probably due to loss of contact. So almost any paste that replaces that poor contact with a good one is going to give you better temperatures. The quality of the paste, beyond its ability to make contact, may not matter at all!
Last edited by WizardOfBoz on Tue Apr 28, 2020 10:59 am, edited 1 time in total.

jursky
Freshman Member
Posts: 62
Joined: Mon May 07, 2018 7:31 am
Location: Kosice, Slovakia
Contact:

Re: W530 overheating - remedy?

#10 Post by jursky » Thu Apr 23, 2020 12:20 pm

Am I only one who considers 60-69°C a normal temperature at idle (without active fan cooling, eg. at 0 RPM)? :D

I am using tpfancontrol on all my ThinkPads (X230 with i7-3520M and W530 with i7-3840QM) with a setting that does not spin the fan up, not until it reaches at least 66°C, making both of them super quiet machines (and I had them like that for a long time (W530 for 4 years, X230 for 2 years) so I'd say it definitely does not make them worn out quicker; of course I change the thermal paste after a year (roughly)).

Anyway, because of that I had to limit my W530 (i7-3840QM) to 90% of maximum CPU power (effectively disabling Turbo Boost) because otherwise it would take the fan (which would be at 0 RPM) too long to spin up in order to cool the CPU (running at 100% with all cores, which would be able to hit 90-103°C temperatures in less than 2 seconds). Ever since I created a custom power plan (again, Windows) limiting CPU power (Power plan settings > Change advanced power settings > Processor power management > Maximum processor state > "90%" or another value under 100% (so to disable Turbo Boost)) I hadn't a single problem with temperatures (87°C at max? But that's only because I am too lazy to re-paste the machine. literally, to take apart W530 in order to change the paste is a nightmare... repasted it would be like 67-70 at max). Of course, performance might be a little bit limited (lacking few Mhz, or even Ghz), but W530 with a Quad-core processor is even nowadays (2020) pretty powerful machine, I'd say - I personally can't complain, I've been using it like that for years now and all good.
X230 i7-3520M (Nitrocaster Rev.6 + 13.3" FHD 99% sRGB IPS panel SHARP LQ133M1JW02).

WizardOfBoz
Freshman Member
Posts: 104
Joined: Mon Mar 16, 2015 9:22 am
Location: Philadelphia, PA

Re: W530 overheating - remedy?

#11 Post by WizardOfBoz » Thu Apr 23, 2020 1:13 pm

Jursky, I don't know if 66 C is normal, but my W530 (w 3820qm) was up in that range, with one core going up as high as 91 C. 91 C is unacceptably close to the Tj (105 C) of the chip. After repasting, it's down in the low 50s, with no temp above 76, and all cores about the same. So even if 66 is normal, I think 51 C is better!

It took me about an hour to do the re-paste. Remove the battery, then the memory cover, then the keyboard (2 screws and one flat cable), them remove the hard drive cover and hard drive, then remove the palmrest/bezel (many screws, including one in the hard drive bay), then remove the left speaker (two screws). At this point I differ with a lot of on-line DIY youtube videos. There's a couple of cables that route over the heat pipe/fan assembly. One has a plug just about in the center of the unit. This must be unplugged. Another (actually, its two wires, not just a cable) runs from the main body/screen joint into the body of the laptop. These are usually shown being completely disconnected. Don't need to do this! Just unplug the first cable and remove it from the cable hooks on the heat pipe. The UNHOOK the two wires but don't disconnect them. Once unhooked, you can just move them out of the way. Then you can remove the heatpipe/fan assembly. Clean the old gunk off with isopropanol. There are two masks made of Kapton (thin stiff plastic sheet). These are a PITA. I pulled them off and cleaned them. But putting them back in: arggh.

After reassembly, I'm using Core Temp and HWMonitor to get the temps. They give nearly the same numbers. HWMonitor also gives you the GPU (in my case, for the nvidia K2000M chip) temperature. I'm pretty happy I did this. That said, I'm switching over to a W541, probably this weekend. A bit faster chip, a few years less old, and a cool numerical keypad. Yay. So the W530 is to be a spare, and can be used to allow clients to run my models without having to give them source code. A second w530, which was a spare, will be put into service to replace the "family" (wife's) laptop. And the T510 that is now the family laptop will be sold.

Since I'm an extrovert working out of my home, one more thing I'll share. My son moved out West a couple years ago and he wants his old gaming desktop (a 3770-based unit). So that gets shipped to him. I'm going to try to score a dual Xeon Dell Poweredge T7910 cheap. Looking for a unit with 64Gb of DDR4 memory or more, and 2690 or better Xeons. That gives me 40 cores to model Covid and immune system response.

Post Reply
  • Similar Topics
    Replies
    Views
    Last post

Return to “ThinkPad W530/W540/W550 Series”

Who is online

Users browsing this forum: No registered users and 11 guests