BCM4709 Temperature

Post new topic   Reply to topic    DD-WRT Forum Forum Index -> Broadcom SoC based Hardware
Goto page Previous  1, 2, 3  Next
Author Message
Bride
DD-WRT Novice


Joined: 01 Dec 2021
Posts: 16

PostPosted: Mon Dec 20, 2021 5:58    Post subject: Reply with quote
to be precise, cut the aluminum lid on the red line and removed the part in yellow.

Uptime 50 min, 67C , room 19C

Useful photos web link

https://fccid.io/PY315200309/Internal-Photos/Internal-Photos-pdf-2709621



2709621-7 - Copy.png
 Description:
 Filesize:  1.3 MB
 Viewed:  1552 Time(s)

2709621-7 - Copy.png


Sponsor
Bride
DD-WRT Novice


Joined: 01 Dec 2021
Posts: 16

PostPosted: Mon Dec 20, 2021 13:33    Post subject: Reply with quote
Uptime 5 hours

Room temp 17.5

Temperatures
CPU 68.0 °C / WL0 55.0 °C / WL1 51.5 °C / WL2 49.0 °C

Until spring/summer time, I can define myself satisfied Razz
the-joker
DD-WRT Developer/Maintainer


Joined: 31 Jul 2021
Posts: 812
Location: All over YOUR webs

PostPosted: Mon Dec 20, 2021 17:49    Post subject: Reply with quote
Bride wrote:
to be precise, cut the aluminum lid on the red line and removed the part in yellow.

Uptime 50 min, 67C , room 19C

Useful photos web link

https://fccid.io/PY315200309/Internal-Photos/Internal-Photos-pdf-2709621

I wouldn't have cut anything, clearly that aluminum acts as both a shield and shim between the components in the middle section etc and the black heatsink altogether work as a thermal mass to draw the heat away, cutting it will reduce the efficiency of the mass.

The thermal paste in between the components and this aluminum should have been reapplied and contact checked.

IMO this is a terrible design to begin with, the heatsink should be in direct contact with the chips for maximum efficiency, and this is why the side areas have and indentation where the heatsink has matching areas that sink to that depth.

Unless there's an active piezoelectric thermal solution, the present design is deeply flawed.

With this said, it would be a perfect piece of kit to demonstrate a better engineered thermal solution.

Anyway, 17.5C is within the temperature range suited to enhance natural Circadian Rhythms, maybe you will have a inspired dream that will guide you to a better solution Wink

Glad you sorted it in the end.

Take care.

_________________
Saving your retinas from the burn!🔥
DD-WRT Inspired themes for routers
DD-WRT Inspired themes for the phpBB Forum
DD-WRT Inspired themes for the SVN Trac & FTP site
Join in for a chat @ #style_it_themes_public:matrix.org or #style_it_themes:discord

DD-WRT UI Themes Bug Reporting and Discussion thread

Router: ANus RT-AC68U E1 (recognized as C1)
kernel-panic69
DD-WRT Guru


Joined: 08 May 2018
Posts: 12482
Location: Texas, USA

PostPosted: Mon Dec 20, 2021 18:56    Post subject: Reply with quote
Bride wrote:
I opened this baby again and removed the fan I added previously. Removed the big black heat sink. Then I removed also the aluminum lid on the CPU part (yellow area, tks to TweakTown for the pic) and replaced the old thermal pad with a 5mm grey one, directly in contact with the big black heat sink, without the aluminum lid. For now, after 34 minutes of uptime, room temp 19C degrees and CPU temp 64C degrees. Let's see

IONK wrote:
Bride wrote:
without the aluminum lid
the aluminum lid is for RF shielding, so probably it's better (for nearby electronic equipment, living beings, and for the router itself) to fill the gap between the chip and the lid with thermal pad, and keep the lid closed, instead of removing it.

the-joker wrote:
Bride wrote:
to be precise, cut the aluminum lid on the red line and removed the part in yellow.

Uptime 50 min, 67C , room 19C

Useful photos web link

https://fccid.io/PY315200309/Internal-Photos/Internal-Photos-pdf-2709621

I wouldn't have cut anything, clearly that aluminum acts as both a shield and shim between the components in the middle section etc and the black heatsink altogether work as a thermal mass to draw the heat away, cutting it will reduce the efficiency of the mass.

The thermal paste in between the components and this aluminum should have been reapplied and contact checked.

Arctic Blue thermal pad. I don't think paste would be good here, since I don't think it is used in the original configuration, but I have to agree that this is a horrible design to begin with, but due to RF shielding requirements, it is what it is.

_________________
Official Forum Rules, Guidelines & Helpful InformationFirmware FAQInstallation WikiWhere Do I Download Firmware‽
DON'T use Chromium-based browsersRTFM/STFW TL;DR is NOT an excuse. • Why Should I Care What Color the Bikeshed Is‽
Please DO NOT PM me with questions; Ask in the forum. ---------------------- Linux User #377467 counter.li.org / linuxcounter.net
Bride
DD-WRT Novice


Joined: 01 Dec 2021
Posts: 16

PostPosted: Tue Dec 21, 2021 5:29    Post subject: Reply with quote
tks guys, I thought that the aluminum lid was acting also as shield from other radio interference, but in terms of thermal conductivity, is not useful.

If you try (and I tried) to put a metal foil between a normal CPU and the heat sink, instead to improve the heat dissipation, you reduce it. CPU-Thermal paste-metal foil-thermal paste-heat sink. Too many layers in terms of heat dissipation. So it's exactly the same here. Keeping in contact the BCM CPU with the black heat-sink, you can only improve the heat dissipation. For example, there is a procedure called direct-die contact in normal desktop CPUs, for maximize the overclock potential.

Anyway, as someone said, this router is good but with e very bad cooling system. I am just sharing these "experiments" as online knowledge, maybe useful, maybe not.

I cut that lid in a way that you can apply it back as before, quickly and safely. I am totally not concerned about what I did.

Previously I tried to keep the aluminum lid, replacing all the thermal pads with GELID high-conductivity ones, but right now, with the black heat sink in direct contact with the BCM CPU, I reached the lowest temp values.

We will see in the next step

Uptime
21:09

Temperatures
CPU 69.1 °C / WL0 55.9 °C / WL1 52.0 °C / WL2 49.8 °C

Room
18.7C
IONK
DD-WRT User


Joined: 19 Aug 2011
Posts: 484

PostPosted: Tue Dec 21, 2021 5:56    Post subject: Reply with quote
the thing you did to your rf shield is similar to "CPU delidding" https://www.pcgamer.com/delidding-your-cpu-is-scary-but-worth-itand-surprisingly-easy/. This mod is very common to a few generations of Intel CPU when Intel tried to cut cost and used a cheaper process instead of soldering the silicon to the metal cover. Some extreme overclockers even throw away the metal lid and put the heatsink directly on top of the silicon die. There's significant reduction in temperature, but I don't think your router can pass FCC test anymore without the lid.

p/s: weird... no comment about the oversized photo attachment... let me try...



04.png
 Description:
source: https://www.ekwb.com/blog/what-is-delidding/
 Filesize:  46.13 KB
 Viewed:  1406 Time(s)

04.png



_________________
▫ RSS feed for DD-WRT releases (2022): https://politepol.com/fd/BU5UpNmf6NRW
easyddup - A simple upgrade/downgrade utility by yoyoma2
kernel-panic69
DD-WRT Guru


Joined: 08 May 2018
Posts: 12482
Location: Texas, USA

PostPosted: Tue Dec 21, 2021 6:08    Post subject: Reply with quote
There's no image resolution limitations anymore, but if people upload 5MB files, it's going to load slow as sh*t. That was part of the reasoning for limiting the resolution in the first place, aside from skewing the page formatting.
_________________
Official Forum Rules, Guidelines & Helpful InformationFirmware FAQInstallation WikiWhere Do I Download Firmware‽
DON'T use Chromium-based browsersRTFM/STFW TL;DR is NOT an excuse. • Why Should I Care What Color the Bikeshed Is‽
Please DO NOT PM me with questions; Ask in the forum. ---------------------- Linux User #377467 counter.li.org / linuxcounter.net
Bride
DD-WRT Novice


Joined: 01 Dec 2021
Posts: 16

PostPosted: Wed Dec 22, 2021 13:55    Post subject: Reply with quote
IONK wrote:
the thing you did to your rf shield is similar to "CPU delidding" https://www.pcgamer.com/delidding-your-cpu-is-scary-but-worth-itand-surprisingly-easy/. This mod is very common to a few generations of Intel CPU when Intel tried to cut cost and used a cheaper process instead of soldering the silicon to the metal cover. Some extreme overclockers even throw away the metal lid and put the heatsink directly on top of the silicon die. There's significant reduction in temperature, but I don't think your router can pass FCC test anymore without the lid.

p/s: weird... no comment about the oversized photo attachment... let me try...


sry for the oversized photos, probably I made a mistake uploading it.

Yes, I know, that's exactly the delidding process, I used it with few desktop CPUs.

About the lid, I agree that I am affecting the FFC standards. In your opinion, the daily home use can be affected? the black aluminum heat sink in fact is not acting as shield?

About temps, I can only see benefits.


Last edited by Bride on Wed Dec 22, 2021 14:08; edited 4 times in total
Bride
DD-WRT Novice


Joined: 01 Dec 2021
Posts: 16

PostPosted: Wed Dec 22, 2021 13:56    Post subject: Reply with quote
kernel-panic69 wrote:
There's no image resolution limitations anymore, but if people upload 5MB files, it's going to load slow as sh*t. That was part of the reasoning for limiting the resolution in the first place, aside from skewing the page formatting.


I will pay attention about it
IONK
DD-WRT User


Joined: 19 Aug 2011
Posts: 484

PostPosted: Wed Dec 22, 2021 14:24    Post subject: Reply with quote
Bride wrote:
In your opinion, the daily home use can be affected?
some children say their mom's microwave oven makes the WiFi slow. some people claim extended exposure to high power RF can cause cancer. Some say they have ED because they put their mobile phones in their pants pocket. But that's for self-reading, I don't think we should discuss about that topic here.
Bride wrote:
the black aluminum heat sink in fact is not acting as shield?
Definitely no if it's not connected to the ground plane of the PCB. I don't have the router so I can't tell if the black thing is connected to ground or not. Even if it's connected to ground, there are still small gaps at the sides.

p/s: it seems like someone added the auto shrink function to the image attachment, that's why the large photos can't kill the layout anymore, so the mods relaxed the rules.

_________________
▫ RSS feed for DD-WRT releases (2022): https://politepol.com/fd/BU5UpNmf6NRW
easyddup - A simple upgrade/downgrade utility by yoyoma2
Bride
DD-WRT Novice


Joined: 01 Dec 2021
Posts: 16

PostPosted: Thu Dec 23, 2021 14:49    Post subject: Reply with quote
IONK wrote:
Bride wrote:
In your opinion, the daily home use can be affected?
some children say their mom's microwave oven makes the WiFi slow. some people claim extended exposure to high power RF can cause cancer. Some say they have ED because they put their mobile phones in their pants pocket. But that's for self-reading, I don't think we should discuss about that topic here.
Bride wrote:
the black aluminum heat sink in fact is not acting as shield?
Definitely no if it's not connected to the ground plane of the PCB. I don't have the router so I can't tell if the black thing is connected to ground or not. Even if it's connected to ground, there are still small gaps at the sides.

p/s: it seems like someone added the auto shrink function to the image attachment, that's why the large photos can't kill the layout anymore, so the mods relaxed the rules.


no, the heat sink is not connected to the ground plane of the PCB, tks for the details
mono
DD-WRT Novice


Joined: 09 Dec 2006
Posts: 31

PostPosted: Mon Dec 27, 2021 2:00    Post subject: Reply with quote
Umm, a piece of metal does not have to be connected to a ground plane to still block a substantial amount of RF, as long as the gaps are far smaller than the wavelength. It works better if it does but this is more about shielding cabling not RF in all situations. I suppose the point is, the heatsink isn't meant to be a shield but would be to a significant extent if it came to that.

Think microwave oven door shield, that piece of metal screen has no ground on most microwaves.

I just discovered something very interesting.

It's a BCM4708 instead of 4709, but here goes...

My Netgear R6700 was running in the '70C range, about 75C average, without P2P and no QoS running yet.

I know that's not alarmingly high but still, wanted lower. So, I have the means, took a hole saw and put 2" hole in the cover, screwed a 50mm fan on.

The thing is, I have to find some connectors to do this, want it easy to swap replacement fans in (I had several new spares of same fan), so project completion put on hold. I put the cover back on the router and powered router up, fan not running.

With the fan body mostly (more than 2/3rds) blocking the 2" diameter hole, and fan not even running, CPU is now about 8C cooler running. Ambient/room temp didn't change.

Edit: Not sure why it took so long but many hours later it has settled to 71C with P2P load, so closer to 5C, not 8C lower.

My theory is there wasn't enough additional venting to make this much difference, that I got most of that improvement by merely tightening down the case screws more than they were, so it compressed the thermal pad more. * I could be wrong*, but I think it is worth trying, taking the screws out, wiggling the PCB a tiny bit and putting it back together with screws as tight as reasonably possible.

I'm kind of wishing I didn't even bother with the fan/hole, just drilled a grid of little holes in the lid like I usually do with routers and switches, but I've got the fan on there now, debating whether to use existing 12V input from the wall wart and current limiting or instead keep it entirely external, use a 5V USB phone charger wall wart to power it, then with entire thing external, never need to open the case again to swap or clean the fan.


Last edited by mono on Mon Dec 27, 2021 12:40; edited 2 times in total
IONK
DD-WRT User


Joined: 19 Aug 2011
Posts: 484

PostPosted: Mon Dec 27, 2021 5:39    Post subject: Reply with quote
mono wrote:
Umm, a piece of metal does not have to be connected to a ground plane to still block a substantial amount of RF, as long as the gaps are far smaller than the wavelength.
I totally agree about this point. EMI shielding can be in the form of sheets or foams besides the boxes. However, when designing PCB, everyone tell me to connect the shielding box to the ground.

mono wrote:
Think microwave oven door shield, that piece of metal screen has no ground on most microwaves.
Ground can also mean the 0V potential of the equipment, not the earth ground. In the photo I found, I can see that the mesh is actually a whole piece of metal up to the hinge, but I don't know if there's any electrical connection from the hinge to the ground plane of the microwave oven.
_________________
▫ RSS feed for DD-WRT releases (2022): https://politepol.com/fd/BU5UpNmf6NRW
easyddup - A simple upgrade/downgrade utility by yoyoma2
mono
DD-WRT Novice


Joined: 09 Dec 2006
Posts: 31

PostPosted: Mon Dec 27, 2021 12:48    Post subject: Reply with quote
^ I've taken oven doors apart to get the screen out, there wasn't any connection to the hinge, screen entirely held within the plastic frame.
IONK
DD-WRT User


Joined: 19 Aug 2011
Posts: 484

PostPosted: Mon Dec 27, 2021 14:03    Post subject: Reply with quote
mono wrote:
^ I've taken oven doors apart to get the screen out, there wasn't any connection to the hinge, screen entirely held within the plastic frame.
It seems that the bent shapes at the edges of the metal mesh act as the "wave choke".
There's also a discussion here saying that
the_emi_guy wrote:
First, there is a common misconception that a microwave oven is a Faraday cage. It is not. Not even a close approximation. Go in your kitchen and open the door of your microwave. Look at the mating surfaces between the door and oven. There is no bare conductive metal at all, usually power coated sheet metal or plastic on both surfaces. It is essentially *wide open* from an RF perspective except right at 2.4GHz.

Attenuation at 2.4GHz is achieved through the use of metal features inside the door that make the slot between the door and the oven look like a quarter wave reactive choke. 2.4GHz is essentially reflected back into the oven from the door/oven seam. This is why the door is so thick.

The amount of attenuation at 2.4GHz is chosen to address human safety limits for RF exposure, with no consideration for RF communications. It is generally 30-40dB,

At 30dB, 600W on the inside, enough to cook your chicken, becomes 600mW on the outside, about the same amount your cell phone radiates.

This is a *lot* of attenuation when viewed from a human safety perspective, but only a modest amount of loss from an RF communications perspective. RF link budgets for WiFi can be > 100dB.
It seems the microwave oven door doesn't work the same way as the EMI shielding on PCB.
_________________
▫ RSS feed for DD-WRT releases (2022): https://politepol.com/fd/BU5UpNmf6NRW
easyddup - A simple upgrade/downgrade utility by yoyoma2
Goto page Previous  1, 2, 3  Next Display posts from previous:    Page 2 of 3
Post new topic   Reply to topic    DD-WRT Forum Forum Index -> Broadcom SoC based Hardware All times are GMT

Navigation

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum