?

Log in

No account? Create an account

Josh-D. S. Davis

Xaminmo / Omnimax / Max Omni / Mad Scientist / Midnight Shadow / Radiation Master

Previous Entry Share Flag Next Entry
GS108
Josh 201604 KWP
joshdavis
Ok, so my failed GS108 that I've not RMAed, I took apart.

I found a couple problems.

A) There is a heat-sink, but the fins are perpendicular to the case vents (and airflow).
B) The heat sink is held on by a thermal pad by very low force springs
C) There's no fan for the heatsink or for the case, so the airflow stagnates.

It turns out that the Linksys, Trendnet and Netgear switches are all the same HW. BCM839x switch on a chip, BCM87xx PHY, and some circuit traces.

The reason the 100mbit switches were so much more reliable is because the 100mbit roboswitch chips from broadcom are rated for "industrial temperatures" which they list as -40C to +85C.

The BCM839x gigabit RoboSwitch chips are NOT rated for "industrial temperature". They're rated for 0C to 40C operating. Their storage temps are -40C to +70C.

Just so you know, 40C is 104F. Hot, but not SUPER hot. Easily reachable under load.

Of course, replacing the heat sing with a copper, active cooling, low profile heat-sink and fan would void the warranty, but it would increase the life span a whole bunch.


  • 1
104 F is NOT acceptable. Thats less than outside. I have one of these in my garage (but its the "green one" that powers down ports link lights on).

Whats the issue with using low force springs to hold the heat sink on?

Yah, my switches are all v2. I'm hoping the gs116 replacement is a v3 (Green).


As for the springs, I'd really like them to have some amount of force. The springs in use just keep it stuck to the rubbery, thermal pad but don't provide force to keep it compressed. So, shipping, long-term wall-mount, etc could leave air voids under the pad (appeared to be the case in my GS108).

  • 1