Home | Research | Publications | Teaching | Personal
Linksys SD2008 Gigabit Ethernet Switch
Go back to...
Intro
Around the beginning of 2008, I bought a Linksys SD2008 gigabit ethernet switch. It ran very nicely and allowed me to transfer files at around 300MBit effectively. I did notice that it ran quite hot, though. Since it had a metal casing, I supposed that was all calculated…
Problems
…wrong! In July 2008, the thing popped. Only my laptop would connect at 1000MBit, but my other two computers would either connect at 10MBit or not connect at all. I contacted Linksys and went through the RMA process. I sent my switch to Linksys (at own cost). So far, so good. The RMA status webpage announced that the switch, which was a SD2008 v.2.1, would be replaced by a v.3.0. Great! Problem solved!
Nope, not quite. Linksys sent me back my own v.2.1 switch, polished up and with its rubber feet re-attached (they had moved due to the heat it generated). I actually got a bit worried that I hadn't done my testing properly: were my two computers broken instead of the switch? I settled that question by connecting my computers directly to my laptop, which ran again at 1000MBit.
Hmmm, I did not feel like wasting more money on sending the device back to Linksys again, so I did some more googling for people who shared my fate. I found some nice links in the end:
The first link made me realise that the original model had a fan(!). It also showed nicely how to pry open the box without breaking it. As I had already given up on the warranty process, I happily tore off the “warranty void” seal and set to work. This is what the PCB looks like:
This is where the second link comes in: it mentions bulging capacitors, which indicate that a capacitor is broken. If you look carefully, I had one of those bulging capacitors too: C251, right next to the heatsink. It was a 470uF, 25V, 105°C rated cap:
Solution
Ha!! I quickly checked the opening hours of my local electronic component store (“Joenit”), saw that they were still open, and rushed out the door. Less than half an hour later, I was back with three replacement caps, which were only rated at 85°C, but the cost of those was low enough to take the risk.
I set to work: soldered off the old cap and soldered a new one. I tested the board with all computers: everything ran at 1000MBit again! Some speed tests later, I was convinced that everything was back to normal. Just out of curiosity, I felt with my fingers how hot the heatsink actually gets: ow! Hot! After transferring only 600MB and without any casing around it!
That settled it: I was going to mod the case with an external fan. I fished an old harddisk cooling bay with two front-mounted fans out of my hardware graveyard. Then, I took the top shell of the switch to the basement and drilled and sawed away at it. I fixed the fan to the top shell with screws and soldered the fan leads to the PCB where the power comes in:
This is the result when closed back up:
And this is where it goes in my setup:
I should mention that the modded switch did generate quite a bit of fan noise (comparable to normal computer?). In any case, it no longer gets hot (the case feels cool, even after copying a number of large files).
The fan I've used is a 12V, 0.05A fan, which means it has some 240 Ohm resistance at 12V. The power supply to the switch is also 12V and can deliver 1A. After some testing, I've decided to add 110 Ohm resistance in series to the fan. I used two 220 Ohm resistors in parallel, which should also help dissipating the 0.27W that they have to deal with (I believe the small resistors can only take 0.25W). The result is less noise, while still remaining cool.




