2009

Sat, Oct 10

<01:07 EEST> I seem to have a peculiar mix of good and bad luck, when it comes to hardware failures. About a week ago, one of my power supplies got busted, a third in a row of less than a year. With the same diagnosis as the past two: bulged electrolytic capacitors at the output filtering stage. This time, it was the power supply of an external Firewire hard drive, and it was a fairly quick and easy fix.

As before, the most tedious part of the fix was opening the glued/molded case without too much damage. I was amused by the ROHS logo on the underside; so this is how you reduce hazardous waste, by making electronics single-use and harder to repair. In the good old days (TM), a radio came with a complete schematic diagram, in addition to the vacuum tube types and locations marked on the case. The latter was so that everyone could change broken or worn-out tubes, and you did not have to be an electronics expert. Today you can be an expert, and not be allowed to fix what you own, due to laws like the DMCA. At the same time, you have increased environmental awareness, and increasingly throwaway product designs.

In an unrelated technical topic, my GSM/3G provider announced a free trial upgrade to my 3G account. My 384 kbps wireless is now temporarily upgraded to 1 Mbit/s, and a little testing was in order. A bulk download from a local Linux mirror averaged to 121 KB/s, while my likewise 1-megabit ADSL only turned out 103 KB/s.

After considering the multitude of factors involved, such as protocol overload, I recalled that ADSL often uses an ATM layer. This alone wastes about 10 % of the capacity, since each cell of 53 bytes carries 48 bytes of payload. I would also think that 3G uses more advanced protocols overall, being a newer technology. Moreover, the PPP connection used in 3G can use old tricks like deflate compression.

The problem I see here is that consumers are being sold one megabit of Internet in either case. These are the line speeds, and the available speed also depends on the technology. Of course, line speeds are quantized so that it is practically impossible to provide the equivalent available rates. There are also huge practical differences between ADSL and 3G, for example, and the exact data rate is not the deciding factor. Nevertheless, I feel a little deceived when my ADSL, which is better than 3G in so many ways (except mobility :) is slower at the same nominal speed.