ben-wpan: status

Werner Almesberger werner at
Fri Sep 17 21:40:59 EDT 2010

Since it's been a bit quiet about ben-wpan lately, here's a status

I'm now examining the board's behaviour with an USRP2. The USRP2 is a
device for software-defined radio, but it also has some of the
functionality of a spectrum analyzer or an RF signal generator. There's
quite a lot of learning involved there, not only on the RF side, but
also on gnuradio itself.

Needing a few more atusd boards for experiments, I made a new pair of
them, this time half a millimeter narrower, so they now protrude only
about 1 cm from the case. I also added a little "tail" to the antenna,
which Joerg has suggested for tuning. (atusd uses a thinner PCB than
the board the antenna was designed for. The antenna length therefore
needs to be adjusted. Unfortunately, the antenna documentation doesn't
say how, so some experimenting will be necessary. This will be a good
warm-up for the antenna tuning that will also be necessary when adding
some sort of plastic case.)

Alas, when making the boards, the tip of my last 0.0125" drill broke,
and I found out the hard way that the rest of the drill, while still
making nice holes, will do its best to damage whatever it pinches
through at the end, yielding defective vias. This in turn caused power
distribution issues, resulting in transmit power drooping during a
frame. It was instructive to analyze this problem with the USRP2,

Finally, after fixing the boards, I started to measure received
amplitude as function of the sender's frequency. The senders being my
atusd boards, and receiver being the USRP2 with a common WiFi antenna
attached. The results so far are quite unstable, suffering from sender
and receiver being too close to each other and also from the proximity
of a large bag of water - yours truly, sitting at the desk - causing
major distortions.

So I'll move the test setup to another room next week, where I have
more space than in my office and where nobody will come close.

Some of my scripts and results are here:
The step-* curves are a bit misleading. While they are neatly
reproducible in one set, the same configuration will yield quite a
different curve the next day, even with me not being in the room. I
hope these issues will go away after moving everything to the other

On the software side, I'm waiting for the 2.6.36 kernel to be released
and linux-zigbee to update their branch. That way, I'll have 802.15.4
from linux-zigbee and Lars' Ben code in the same tree, without having
to mix things myself. Then I'll gradually shift things over from my set
of user-space tools to the kernel drivers. This will open the path
towards running IPv6 over IEEE 802.15.4, and the interrupt support in
the kernel driver should help to make things more efficient.

- Werner

More information about the discussion mailing list