Re: [uracoli-devel] Question regarding LQI and ED values [Update]

[Joerg Wunsch]

As Eric Jennings wrote:

> Really interesting antenna design.  Did you try to see what your
> maximum working distance was between two boards (despite the fact
> that the design goal only needs them to be a few meters away from
> each other?)

No, I didn't.  I only made radiation measurements in comparison with
other boards and antennas, using an older spectrum analyzer as test
receiver.  As I've got a few spare PCBs, I could run a test series
with different lengths of the antenna traces.

> I've been curious how trace antennas do vs the
> chip/whip/etc antennas w/ these chips.

Chip antennas are in general shortened antennas, i. e. they are
smaller than lambda/4.  By the laws of physics, a shortened antenna
cannot achieve the same efficiency as a full-length antenna.  The main
purpose of the chip antenna is to reduce the environmental influences
(like your hand in the vicinity of the antenna) on the antenna tuning,
and to include the tuning reactances into the antenna itself, so the
transceiver sees standard 50 Ω again.

The most efficient chip antenna design I've seen so far is that of the
Zigbee modules.

Full-length whip antennas are naturally better.  You have to divide
them into two basic types:

. lambda/4 antennas, perhaps with some minimal length reduction by
  forming the antenna wire into a spiral; as they are monopole
  antennas, they require a ground plane of lambda/2 diameter (or more)
  as their counterpart; efficiency is as good as a dipole (2 dBi) if
  this condition is met, but can drastically (-10 dB or more) drop if
  it isn't

. lambda/2 antennas, this is the classic dipole (2.15 dBi gain); does
  not have a groundplane requirement, but is twice as large as the
  monopole antenna; also, it needs a symmetrical feed and some
  impedance transformation to be matched for 50 Ω circuits; this
  transformation can be done with a piece of coax wire, google for
  "sleeve antenna"; all this can be integrated into a single whip
  device

As a special feature with the Atmel transceivers, they've got a
symmetric RF port with nominal 100 Ω impedance requirement.  A plain
open dipole has about 120 Ω, and thus fits almost optimal to the
transceiver.  That's what I've been using on the Tic-Tac-Toe board.
You don't need the balun, and get a good antenna for cheap.  (However,
it's really advisable to still spend the Pi-type output filter in both
RF branches, RFP and RFN, to improve the spurious emission
supression.)

So far the most efficient (and still cheap!) antenna I've seen is the
one described in Atmel's appnote AVR2006.  Of course, it's a
directional antenna, antenna gain can only be achieved by bundling the
radiation in a certain direction.  (In contrast, you can get antenna
losses even without the antenna being required to be omnidirectional,
or to radiate any of the energy at all. ;-)  According to my
measurements, the main directional radiation of that antenna is as
good (or even more) as the one of much longer lambda/2-style whip
antennas.  If I compare it to the plain dipole of the Tic-Tac-Toe
(with an assumed gain of 2 dBi), the 6.5 dBi claimed in AVR2006
appear to be realistic.

Just for fun, I once soldered a "quick hack" Yagi antenna for 2.4 GHz,
by taking antenna design data from amateur radio literature, and
resizing them for 2.4 GHz.  You can find the picture (and some
explanation in German) here:

http://www.mikrocontroller.net/topic/247618#2543650

It's got a proud length of 300 mm (mounted on a cardboard back
obtained from an ISO A4 paper bloc).  The original design was said
to have 14 dBi gain, my measurement got it to approximately 12 dBi.
Seems realistic enough, I'd attribute the missing 2 dB to the poor
construction (and missing balun), but also shows that you have to
spend quite some material to get *real* antenna gains.
-- 
cheers, J"org               .-.-.   --... ...--   -.. .  DL8DTL

http://www.sax.de/~joerg/                        NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)