Kisiizi IT Network Part 2 - Power over Ethernet
Written on Saturday 10 January, 2009
I had hoped to be reporting that the first part of the new wireless network, the central access point, had been completed and installed. But, as I’m increasingly finding in Uganda, circumstances conspire to confound even the most pessimistic schedules.
I’ve designed the system to have the access points themselves (Buffalo WHR-HP-G54 routers) as close to the antenna as possible, at the top of the masts. Why? Because the power of the transmitted wireless network signal is lost very quickly through co-ax (antenna) cable, so it’s a good idea to keep it as short as possible.
To keep the routers protected from the elements, I’ve improvised a cover for them, made from a large upturned plastic jerrycan. With the top (now bottom) removed and side vents added, it should provide sufficient protection from the rain while providing some ventilation. The equipment emits some heat (of which more in a minute), so I don’t think dew will be an issue. This design probably wouldn’t work where freezing is likely.
Getting power and network connections to the access points was going to be achieved using Power over Ethernet, which makes use of the two spare pairs in a CAT5 cable to pass DC to the router. To ‘break into’ the cable, I’ve made some injectors / splitters, with a voltage regulator on the router end to make sure that just the right voltage is reaching the access point. The excess voltage is dissipated as heat, hence the big heatsink. Serendipitously, this should reduce the chance of dew forming in the equipment.
Unfortunately, this is where my plan came unstuck.
The CAT5 cable spec says it is supposed to have a resistance of about 3Ω per 100m. My cable, bought in Kampala and notionally from a reputable Australian supplier, has an astonishing 26Ω resistance over 30m! That meant that my input voltage (12V) arrived at the access point at a mere 5V, not enough for the voltage regulator to work properly.
To combat the problem, I’ve had to arrange a second cable (notionally 1mm 2 flex, although I have my doubts about the actual size of this, too) to carry the DC. Absolutely not the most elegant solution, but the only way to get things going. This second cable is spliced into the CAT5 at each end, so we can treat the two cables as ‘one’, and my work on the DC injectors hasn’t been totally wasted. If we find a source of decent CAT5 cable I will swap this cable arrangement for one piece of cable. Can’t see it happening though.
It’s a source of continual frustration for me that manufactured goods in Uganda are generally rubbish compared to the UK. It’s almost as if the manufacturer of these cables offloaded ones which didn’t meet the western quality standards on Africa. It’s wrong, it’s a waste of money and a massive waste of time.
Hopefully we’ll be able to erect the mast on Monday. Only a week later than planned…
Read more about the first mast being installed in Part 3…