As the digital silicon densities grow larger and the power density goes way down, it is possible to make low enough power DSP chips with enough processing power to do things like CDMAone, GSM, etc in DSP code. thier idea is not a new one, been kicked around for about 10 years now. It is just becoming time that this may be realizable and still be able to compete in talk time, standby time, chip size, chip cost, etc. Lots of factors.
now I can see how this can be done with the digital portion of the phone...this is not hard...but the Rf radios are so totally different in requirements, it would mean getting alot closer to Steinbreckers design where you just modulate and demoulate directly from digital all the way to 2 times the RF frequncy. this would be the only way to make them really work. but there would be 1 million times more information to process at that speed.
Example..consider that my phone is working at 1900 MHZ, then I have to sample at 3800 Mhz to trap the waveform. this means that I have 3.8 billion samples a second to work on. then I have to figure out how to desample or filter this down to baseband(digital) at the 9.8 million samples per second that CDMA takes. this takes ALOT of processing power.
for the present radio, you take the 1900 Mhz carrier and down convert it with a sideband carrier to an intermediate frequency of 200 Mhz, then sample it at 9.8 Mhz. by knowing exactly what you were doing and taking into account alot of math and RF complexities, you were able to reduce the number of samples by more than a million times. this saves alot of processing power. This was done by a simple mixer and filter which takes only a couple of milliamps rather than some giant high power processor taking 100's of milliamps (or even a few amps...)
This is not to say that technology can;t get there in the next few years. At 0.1 micron technoloy (year 2001), you begin to approach the ability to do this.
(BTW - Saw the REAL Mike Doyle at a longboard contest last week...or was it JMD?) |
