revisiting the 434 MHz band

Werner Almesberger werner at openmoko.org
Sat Jul 24 02:13:34 EDT 2010


Wolfgang Spraul wrote:
> No no, not hookup to an FPGA.

I think a CPLD would be sufficient to beat the bitstream into
shape. This wouldn't have to be a large chip. What it would do is
just find out where the packet starts and ends, find byte
boundaries, and do a bit of buffering so that the CPU has
something it can work with.

You may be able to do those bit-level operations at low speeds
(802.11b) also entirely in software, but you'd really spend a lot
of resources for a task hardware is much much better at.

Maybe the DSP could do it, though. Also, maybe there are some
chips that integrate some of these functions.

> I definitely want to hook this up to a regular ASIC XBurst CPU.

I think this would be quite wasteful, even if it works. Kinda
like the ZX81, where the CPU spent most of its time refreshing
the screen, instead of running your program.

> I was encouraged to look into SiGe chips by Henry Hallam
> (http://www.pericynthion.org), who says he works on a solution with a
> Blackfin DSP and SiGe SE4120L to implement the GPS stack,

Okay, but the SE4102L outputs a digital Q/I stream. Even with
the MAX2830, you still need the ADC. Plus, unlike GPS, you also
have a transmit stream to take care of.

Regarding the ADC/DAC, in case you were thinking of using the
things you find inside the CPU: they only do a few kilosamples/s,
which is too slow as well. I don't know what sample rate 802.11b
or even 802.11g need, but I would be surprised if it was less
than a few hundred kSa/s, even if you do tricks like scaling
802.11b down to 1 Mbps. (Would be nice for Lies and Dece..., err,
marketing, though. Much like all those "USB 2.0" devices that
only do full-speed :-)

Last but not least, unlike WLAN, the GPS data stream moves at an
unhurried 50 bps.

> The one thing that worries me about IEEE 802.15.4 is the range. Only a
> few meters? That's what I still like about 433 Mhz - it seems to work over
> hundreds of meters, possibly even more...

Yes, range may be a problem. You can use more power, see e.g.,
[1], but then you're adding extra chips and you go deep into
RF design land. They also say they've achieved 378 m without
such aids [2].

[1] http://www.freescale.com/files/rf_if/doc/app_note/AN2975.pdf
[2] http://www.mwrf.com/Articles/Index.cfm?Ad=1&ArticleID=9379

433/434 MHz can go a long distance, but at least in the lands
under FCC's harsh rule only with crippling duty cycle
restrictions.

- Werner




More information about the discussion mailing list


interactive