Appreciate the feedback. I just wanted to mention, I have some things on order (taking forever) and will update once I can do some more detailed testing.
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Fri Jul 08, 2022 11:01 Post subject:
Im looking at your previous image and spotted on of the the wireless radios winth the antennal cable plugged into it and it has no thermal solution at all, given the r9000 propensity for randomly dead wifi radios I would consider slapping a passive heatsink ontop of them. Cant see the ote radio its likely under the GF grill
Some small updates. Previous testing was @ 1Gbps, I've now configured one of the 10Gbps ports on my main router for Wifi (been wanting to do this for a couple months lol) so now I have a full 10Gbps link. Spoiler alert, the SFP got just as hot running 1Gbps as it did 10Gbps.
SFP+ Ethernet Transceiver
Link = 10Gbps, CAT8 shielded (overkill but all I buy now)
Testing = Basic, copying files over the link via windows, 5Ghz WiFi (rather than testing idle).
Running for at least 30m before measuring. Temperature readings taken with a physical probe. Laser readings were way off and unusable. A/C is on 49% Humidity / 72F Temperature.
#1 WiiTek SFP-10G-T
Max temp recorded 135F/57.2C
Notes: Looks and feels "cheaper" than the 10GTek. Slightly smaller. Slightly cooler RJ45 temps.
#2 10GTek ASF-10G-T
Max temp recorded 140F/60C
Notes: Feels like very nice quality but runs a bit hotter. I didn't conduct a series of tests but felt that 5C was considerable enough to be outside of expected variance. RJ45 temps measured quickly as I removed the CAT8 cable were 114F/45.5C.
Time to cool it down (maybe). I am working on stacking the A/C infinity fans in a push/pull config. Now that I have the fans in my hand, my original thought isn't going to work but there is enough space behind the router due to the angle so I can orient them another way in push/pull.
As I mentioned, this is sort of for fun but overkill aside, if I get get this link to work with a reasonable comfort level that I am not melting the R9000, I would indeed love to use the 10Gbps link permanently. I don't know if this is going to make much difference, perhaps I am just cooling the shield but I'd like to think that it will be helping at least a little. I also enabled the built in fan so that Mr. Foreman might (strong might) get slightly cooler.. doubt it does anything for the SFP+ but who knows..
Im looking at your previous image and spotted on of the the wireless radios winth the antennal cable plugged into it and it has no thermal solution at all, given the r9000 propensity for randomly dead wifi radios I would consider slapping a passive heatsink ontop of them. Cant see the ote radio its likely under the GF grill
I would actually pull the complete George Foreman heatsink assembly & friends and apply better thermal paste/pads, if it currently has thermal pads, the paste will likely need copper shims to bridge the gap, or really high quality low thermal resistance pads.
this is only the Wilocity WiGig module with the connection cable for the active phased array antenna
pretty useless stuff, since you can't buy 802.11ad WLAN cards
You can only harvest the cards from old devices.
And there are different revisions of the cards and each revision has different specific antennas.
and not all cards support WiFi - some of them only support WiGig Bus Extension (WBE)
The other two radios are on the right side under the big heat sink.
And they don't have a normal antenna design either, they also use active antenna modules.
And as you can see the board has 16! antenna connectors
And at this point you could suspect the hardware design problem.
All other QCA9984 radios that I know of do not use active antenna modules and they do not have thermal problems, they work fine at very high temperatures (80-130°C).
Another theory would be that the SoC has a design flaw.
Joined: 08 May 2018 Posts: 14125 Location: Texas, USA
Posted: Thu Jul 14, 2022 16:43 Post subject:
Some of those "antenna cables" are not RF signal paths, they are for powering the leds and other bits on the pcbs in the antennas. If I am not mistaken, only two leads per antenna are for RF signal paths, the others serve those other purposes. _________________ "Life is but a fleeting moment, a vapor that vanishes quickly; All is vanity"
Contribute To DD-WRT Pogo - A minimal level of ability is expected and needed... DD-WRT Releases 2023 (PolitePol)
DD-WRT Releases 2023 (RSS Everything)
----------------------
Linux User #377467 counter.li.org / linuxcounter.net
I've installed the A/C infinity fans in push/pull around the transceiver shielding and the good news is that the temps have dropped down to 84F/29C. The bad news is that these fans, even when are low, are super loud. I didn't test further because they are simply abnoxiously loud, even with just one of them running. I have some other A/C infinify fans (140mm I think) and they are whisper quiet so it wasn't expected to be this bad. I'm going to give these I try. I didn't know Noctua made 5V USB compatible fans. Noctua NF-A4x10 5V PWM, Premium Quiet Fan with USB Power Adaptor Cable
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Sat Jul 16, 2022 8:11 Post subject:
Fans are going to be loud(ish) depending on their size and RPM, but there is always some sound proofing. Also ensuring they are not loose can help reduce vibrations.
When you're more or less done would be nice to see some pictures of the finished result.
And 29C which part (this is quite a drop), is this general temperature or SFP port only? When enclosure/case is open vs closed matters to final temperatures. Closed will reduce airflow dispersion rates.
SFP port side... If you have push/pull its going to force hot air wherever the pull direction. Since Hot Air wants to travel upwards, you can also try push/push assuming there's a fan either side left/right, this will force the hot air to be ejected upwards faster in theory.
Nothing like a couple of tests to prove/disprove.
I wonder if the fan header is fully operational on R9000 or if it just supplies steady voltage or is just a 3pin. Back to looking at pictures.
Joined: 16 Nov 2015 Posts: 6410 Location: UK, London, just across the river..
Posted: Sat Jul 16, 2022 11:58 Post subject:
huh, you can even use an adapted water cooling from VGA and CPU/motherboard or just put your unit in DA'freezer...
just for the record...my unit light's off never used 5Ghz been in a busy 20 PC+ office + some wifi clients under ambient temps..and small build in fan running non-stop...(settings 25-30C) unit temp never shows more than 50-60 C max.. fingers-X its still up and running..radio is turned off 0h-07h....
when i used SFP port it was around 65-70C and port was hot, so i mounted a small heat sink on the top of the port inside..and used unit open with no case cover...ambient temp in the office is never more than 27C summer time..
and yep i know R9000 probb XR700 is a bad design anyway...there was a discussion that there ware 2 different batches of those R9000, but couldn't find that post any more...nor i remember what was the small difference mentioned...there.. if im not wrong was the flash chip or some other hardware but im not sure so...
I guess if you use the unit with case open and big fans on the top with their own PSU...it will keep cooler enough but..no idea why those SFP ports are so hot...i've never tried with optical SFP to see the Temp difference..if any... _________________ Atheros
TP-Link WR740Nv1 ---DD-WRT 55179 WAP
TP-Link WR1043NDv2 -DD-WRT 55303 Gateway/DoT,Forced DNS,Ad-Block,Firewall,x4VLAN,VPN
TP-Link WR1043NDv2 -Gargoyle OS 1.15.x AP,DNS,QoS,Quotas
Qualcomm-Atheros
Netgear XR500 --DD-WRT 55460 Gateway/DoH,Forced DNS,AP Isolation,4VLAN,Ad-Block,Firewall,Vanilla
Netgear R7800 --DD-WRT 55460 Gateway/DoT,AD-Block,Forced DNS,AP&Net Isolation,x3VLAN,Firewall,Vanilla
Netgear R9000 --DD-WRT 55363 Gateway/DoT,AD-Block,AP Isolation,Firewall,Forced DNS,x2VLAN,Vanilla
Broadcom
Netgear R7000 --DD-WRT 55460 Gateway/SmartDNS/DoH,AD-Block,Firewall,Forced DNS,x3VLAN,VPN
NOT USING 5Ghz ANYWHERE
------------------------------------------------------
Stubby DNS over TLS I DNSCrypt v2 by mac913
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Sat Jul 16, 2022 12:15 Post subject:
@Alozaros In the freezer never (sure way to kill it fast the humidity is not good for tech), but you sure can dunk the lot in mineral oil assuming the cable jackets can take it and there's no rubber in there, thought that's a little extreme for a router.
Joined: 08 May 2018 Posts: 14125 Location: Texas, USA
Posted: Sat Jul 16, 2022 12:18 Post subject:
the-joker wrote:
I wonder if the fan header is fully operational on R9000 or if it just supplies steady voltage or is just a 3pin. Back to looking at pictures.
All of my notes are on the laptop that is at the venue I'm working today, but if my short-term memory hasn't gone to sh*t, it is a fully-functional 3-pin header (see picture again) with on/off and speed control capability. My R9000 is sitting in a box awaiting new wifi hardware and other bits and pieces. I'm guessing output power is the reasoning for some of the RF shielding, but most PCs don't even bother. Perhaps the proximity of components on the SoC board dictates, but even an rPi doesn't shield wifi cards. Maybe at some point I'll post revival / rebuild pics for everyone to gawk at. Don't hold your breath _________________ "Life is but a fleeting moment, a vapor that vanishes quickly; All is vanity"
Contribute To DD-WRT Pogo - A minimal level of ability is expected and needed... DD-WRT Releases 2023 (PolitePol)
DD-WRT Releases 2023 (RSS Everything)
----------------------
Linux User #377467 counter.li.org / linuxcounter.net
The proper way to remove the front panel on the R9000 is to turn the unit over and peel back the rubber guards. There are (4) screws and you'll need a star bit. After that you can gently pry it off (do not pull on the front "wings", pull UNDER the mesh screen).
The overkill plan:
(1) Upgrade stock fan (went with a Noctua NF-A4X20 PWM 5V)
Purpose: To force some additional air toward the SPF+
(2) Delete front panel screen mesh
Purpose: Fit larger fan, improve airflow/avoid dust
(3) Install 30mm/5V fan on top of SFP+ (pull)
Purpose: To suck air out of the SFP+ port
(4) Install (2) Noctua NF-A4x10 40mm 5V USB fans push/pull
Purpose: Heat removal for SFP+, transceiver & RJ45
More updates coming soon, just waiting for glue to dry.
Last edited by Laithan on Sat Jul 23, 2022 2:28; edited 3 times in total
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Fri Jul 22, 2022 5:15 Post subject:
Well... Not bad. that new fan is a little bit on the tall side, personally I would have gone slimmer and bigger, there is always mounting a bigger fan by making an adapter to attach to heatsink in some criss/cross fashion to attach to the fan, but lets see what the temperatures say.
Overall still a better job than default. I hope the results will speak for themselves.
Looking forward to see the finished product.
@brainslayer has one of those clocked to 2 GHz but his case is open so ventialtion sint so much of a problem and his is running at 40C