If you read my SD2008 review, you will notice that opening the SR2016 is very similar. The front plate pops off easilly, then the top panel can slide towards the back then can be removed.
The difficulty is in removing the front panel completely. It is easy to open
it halfways but the front is latched in place and then some tough plastic tape
is glued on top. Using a blade or a Philips screwdriver, the plastic sheet can
be lifted enough to allow the blade in the latch to raise the plastic bit, then
the top plate comes off. It's quite a bit of work and you may break off the
cover if you're not careful.
The motherboard features 5 main ASICs and two fans. All 5 ASICs have a passive dissipator glued on top so I was not able to see any manufacturer's reference.
I suspect that the central asic, slightly larger that the other 4, is the actual switch fabric, complete with the 16 MAC subsystems. The remaining 4 ASICs of comparable size are layed out near the connectors towards the back, and I assume they must be have 4 PHY subsystems each. That makes 16 ports, which is what this switch provides.
The motherboard sports a number of references which look like part numbers, such as:
A/N: PCGB1706-000-JH300 P/N: GB1706 Rev. JH START: 2003/07/03 [X] GB1706 [ ] GB2706
For all of you Google freaks out there ;)
The SR2016 features an internal power supply. The PSU is auto-ranging 100-127/200-240V, 1.0A/0.5A, 50-60Hz and has a standard computer equipment connector. Plugging this device anywhere in the world should be possible.
The magnetics (the rows of RJ45 connector if you will) are from Tyco, ref. Mag45 or 1615074-1 or RJS08-8G07, whichever one applies. The clocking is thanks to a Lattice ispMACH LC4128V asic.
This is a loud switch. The noise is somewhat different from the stock SD2008 but is quite loud nevertheless. A friend of mine who hosted a LAN party a few days ago and to whom I lent the switch, commented that noise was an issue. Hence my motivation to try to run the swith quieter.
All ASICs being already passively cooled in addition to the forced airflow from the fans, I thought to myself that the airflow was probably important for the built-in power supply. Indeed the PSU doesn't have a fan or even a proper casing, it's just built-in in a corner of the case with very little protection around it.
Removing the fans was as easy as unplugging the connectors. Then I let the switch run open for 3 days, and it still works. I did NOT put the case back together: in my tests, the switch was running with the case open. I chickened out closing the case becaude of the fire hazard there might be running this switch with no forced airflow. What if the internal PSU overheated and caught fire...
After a few days on my desk and several crude pics snapped with a cellphone, I am putting the switch back together for testing its performance in the lab.
Onto Part 2: Performance Testing >>
gaetan@soltesz.net