[wxsig] AAG TAI603 first impressions

[William Beals]

Folks:

I've been meaning to do a write-up on the new AAG TAI603 weather sensor, but
wanted to get some of the basic communications with the device going before
I did.  I got some of those basics working yesterday and today.  This
write-up does assume you are reasonably familiar with the previous versions
of the temperature/wind-speed/wind-direction sensor.  In the interests of
full disclosure, the unit was provided to me by AAG for the purposes of
updating the T238(+) firmware to support it.  I have a long and very
positive history with AAG and a particular person there.  Without AAG, the
T238(+) projects would have died a long time ago for lack of sensors.

WHAT YOU GET
============
Main body.  Pre-assembled except for wind cups and wind vane, and 25 (est)
foot pigtail Cat5E cable.
Wind Cups (and hardware)
Wind Vane (and hardware)
12" long 1"x1" aluminum rod for mounting the weather head (nice!)
RJ45 extension adapter
RJ45 Y adapter
Extra 5 meter Cat5E cable
USB to 1-wire Adapter (big, approx 4.75"x3.75"x1.25")
Wall wart for above adapter (12V at 500ma)
USB cable for above adapter 
AAG Stickers for weather vane (nice to leave 'em separate and let customer
decide!)

All in all, a very complete kit.  All you need to get things running on a
bench is to download the software.  To do that you do need to register with
an email address.  For the final installation, all you need is the extra
cabling and pole to mount the extension arm on.

MECHANICS
=========
The main sensor body is clearly different, changing from the opaque beige
plastic to a transparent plastic for the LEDs.  It is roughly the same shape
and dimensions still.  The plastic parts for the wind cups as well as wind
vane are still the same beige plastic as before.  The older plastics have
clearly stood the test of time for most if not all of us, so hopefully the
clear plastic will do the same.  The case itself is three-piece, the top and
bottom parts both mount to a center ring piece, all clear.  The pieces twist
together and are held in place by extremely small screws.  Very sturdy.

The hardware supporting the wind vane and anemometer appear to be the same
high quality stainless steel as before too, also a good sign.

The mounting pole is the same 1"x1" square hardware as before.  The only
difference is the retaining bolt is vertical instead of horizontal.  

WIRING
======
A big change is the use of a pigtail instead of a connector for the weather
head.  It may be a challenge for some to get the other end of the cable into
a protected environment before using the extender adapter to continue the
cable run.  The cable uses a twisted pair each for Ground, +5V, and +12V.
The fourth twisted pair is for data and return of the 1-wire bus.  As best I
can tell, the weather head uses only the +12V power bus.  We've had issues
with Ethernet cables on long runs before, so hopefully the Cat5e will be
better than the Cat5 people have had trouble with before.  Hooking up to the
T238(+) was not difficult, I soldered the GND, +5V, and +12V pairs to the
terminals of the regulator on the T238 and then ran the 1-wire bus to an
RJ11 connector and plugged it into the T238s 1-wire port. It is clearly
messier than before which required just plugging in the RJ11 jack, but it
still works well.

Hooking up the PC interface is straight forward too, plugging in the RJ45
connector into the adapter, plugging in it's wall-wart and then hooking up
USB.

TEMPERATURE
===========
As with the previous versions of this sensor, temperature readings are not
accurate at all due to solar heating.  Adding to the problem are the LEDs.
Just sitting on my workbench at an ambient 74F, I decided to just leave all
LEDs at full power and the temperature reading stabilized at 120F!  Stick
with an external temperature sensor as we've discussed many times on this
reflector.

WIND SPEED AND DIRECTION
========================
A nice change from earlier versions is what appears to be hall effect
sensors instead of reed switches for both wind speed and wind direction.
The mechanical setup is the same as before, with a small board with magnets
on it inside the sensor body.  

AMBIENT LIGHT
=============
The weather head does have an ambient light sensor.  It's primary purpose is
to sense if there is enough daylight around to turn off the LEDs.  I has no
known calibrated outputs.

LEDS
====
Very hard to ignore, but fortunately the default settings are off.  There
are a lot of them and they are very bright.  They are also very focused, so
even inside the case with a slightly textured surface, the beams are more
than bright enough to make a pretty good flashlight and easily create a
visible spotlight pool of light up over 20 feet and in general light up a
room bright enough to read by.  Comment from my wife was that they indeed
could be useful.  We are under final approach for a local general aviation
airport, so we can have red lights on our house just like all the commercial
buildings closer to the airport.

COMMUNICATIONS
==============
This is by far the most radical departure from the past.  Instead of just
"dumb" sensors at the head, they have a micro to do a lot of the signal
conditioning and the information returned over the 1-wire bus is already in
real-world units.  All communications is through a DS2750 1-wire chip.  It
provides the temperature reading directly.  Wind data, LED commands and
ambient light data are relayed between the on-board micro and the 1-wire bus
using a 16-byte RAM buffer in the DS2750.  Here is the tough part.  The
DS2750 has only one 1-wire interface, so the on-board micro controls an
analog switch that selects which interface (the micro's or the external one)
control the interface to the DS2750.  There is no negotiating over the
switch, according to AAG the on-board micro takes over control of the bus
about once every two seconds to read any commands, and place any data being
requested by those commands.  Much of the work for me is handling this
interface where the comms can (and will) get interrupted on a regular and
relatively unpredictable basis.

Making reliable communications even tougher, the DS2750s 1-wire interface
protocol no longer supports some of the strict data CRC rules that earlier
devices had, so it is actually quite tough to know if the data has been
corrupted or not.  For the temperature data coming directly from the DS2750,
this is particularly difficult.  For the data from the on-board micro, AAG
have added a CRC to every data packet, so hopefully this data will be easier
to detect problems with.

PC SOFTWARE
===========
The PC software consists of three parts.  There is a Silicon Image driver
for a USB-to-UART chip (guess what's inside the adapter!), a National
Instruments "Data Socket" driver, and finally the AAG application itself.
The software is still pretty rough, I'd call it late alpha or early beta
software--not unlike what I unleased on many of you early adopters of the
T238!  It does the basics of reporting wind speed, direction, temperature,
and LED controls.  It has logging and graphing capabilities, but they are
still kind of buggy.  I'm being told they are still working on the software.

Hopefully that is a helpful write-up.  Back to coding for it!

will


==============================================
This email sent using 100% recycled electrons.
(Minimum 30% post-consumer content.
==============================================
Will Beals