Return to the archive index

Re: wearable data bus with power?

From: Asher <>
Date: Sat, 15 Nov 2003 23:37:15 +0000

Jesse Hammons wrote:
>My intuition is to use USB cabling, since it already supports
>transmitting power and data on the same cabling.  But I don't want to
>have to support the actual USB protocol (to expensive).  From what I
>can tell, I might be able to use RS-422 to make a differential serial
>signal and run that over the USB cabling.

You need to watch the impedence of the twisted-pair cabling you're using
with differential signalling (and ideally with RS-232 as well, but that
seems to work over any old cable in my experience). If USB cabling
doesn't have the right impedence then just get a roll of double
twisted-pair cable and run data over one pair & gnd+power over the
other. Remember to correctly terminate the cable with the appropriate
resistors. If you want a bus type network (multiple devices connected to
the same cable) then look at RS-485 - this is much the same as RS-422
but the drivers can be put into a high-impedence state for when
something else is driving the line.

If you're producing wearable stuff then check the power consumption of
the drivers, some Maxim RS-485 drivers I used in a body area network
(BAN) prototype (which was intended as a protocol demonstrator only and
so didn't actually need to be wearable (which was good, since each
interface card was 4"x6")) used a whopping 50mA each, which is rather a
lot on a microcontroller wearable setup.

If you need to be able to do collision detection on the network
(obviously only necessary on RS-485 type nets) then an easy way to do it
is to take the input going to the line driver and XOR it with the output
from the line receiver. With suitable clocking to allow for the
propagation delay through the line driver you get a signal which can
produce an interrupt on a collision.

Since you'll inevitably get power drops and noise on the long power line
in your network it's a good idea to have the network voltage level
somewhat above the voltage level required by each device and then use a
DC-DC convertor in each device, plus a good smoothing capacitor, to get
a nice stable power supply.

>I would be interested in hearing about how people have solved the
>problem of running data and power along the body.  Anybody been able to
>have a device on your foot send power and data to a device on your
>hand?  What datarates did you get and what cabling did you use?

The first BAN demonstrator I built, back in 1998 for a company internal
research show, used unshielded wiring, non-differential 0V/5V
signalling, a single +5V supply with no regulators on each device, and
had a 9600bps data rate. This worked well over a two-week exhibition in
a large room full of unshielded prototype electronics. The demonstrator
was built into a 7' tall wooden cutout of a person and so had some
rather long wire lengths.

Tests with RS-485 drivers showed that they worked up to megabit speeds
over a 20-30' length (I think, this was several years ago) of
deliberately tangled twisted pair. (We then moved our protocol
development work onto networks of PCs linked over ethernet and I stopped
needing to do hardward dev. work.)

>I suppose wireless would definitely work but that is much more
>expensive and much lower data rate.  And of course the power line is
>still required.

Wired connections between wearables are a much overlooked idea. If the
wiring can be built into items of clothing then the wired network
becomes invisible and you have the huge advantage of not needing
batteries in each wearable device, which means that they can be very
small. You do then have the problem of linking the devices to the wires
(choose your plugs and sockets carefully) and the wires between the
various items of clothing together. You'll soon get tired of having to
unplug things like jackets before you can remove them so perhaps the
best solution would be a wired network with very short range wireless
links to bridge the gaps between various items of clothing.

Asher.

--
Subscription/unsubscription/info requests: send e-mail with subject of
"subscribe", "unsubscribe", or "info" to 
Wear-Hard Mailing List Archive (searchable): http://wearables.blu.org
Please, *PLEASE* don't subscribe through a forward/expander/false domain

+Previous Message in Thread | Next Message in Thread

From Wear-Hard Mailing list Archive (WH)
Maintained by R. Paul McCarty

Archive created with babymail