The Story of Herb and Jr.
(Marin County Hang Gliding Association)
Herb, our original windtalker died November 1995. Herb, Jr. was built over
the course of mostly two months but not installed until April 1996. Junior
resides upon Middle Peak (near East peak) about 2 miles east of the hang
glider launches on Mt. Tamalpais.
The original (Herb) windtalker was a direct offspring of the Ed Levin and
Mt. Diablo windtalkers of recent years. (Mt. Diablo has recently been changed
to a Litek system, I believe). The main guts of the original system was
a Z80 microprocessor, on-board EPROM's, a dedicated voice synthesizer chip,
and part of an answering machine. The board was wired-wrapped (not reliable)
which was the source of many problems over the course of time. Another
problem was the design of the interface with the tower; basically there
was none. The extent of the computer really did go up the tower to the
instruments. The RF interference was extreme because the instruments were
mounted 3ft from two antennas, although in reality good RF blocking should
have been fairly easy.
At the time, the club decided that buying a commercial windtalker (such
as the Litek) did not seem to be a very good option because of the limited
user-serviceability, unsure turn-around time for repairs beyond our immediate
control, and most of all, incompatibility with the existing tower location.
Update 12/15/97: Reinstall Jr. He had his annual physical.
Cleaned up a lot of the quick fixes and put it through hardware & software
tests. Update 8/27/97: Jr. is fully functional again. Finally
found some defective components but still decided to modify the circuit
for more reliability. Update 8/14/97: Went up to take a look
at what might be wrong with the telephone circuit. Rain, power outages,
& generator noise kinda made things tough. Still had an intermittent
problem. Update 8/11/97: Kevin put up the anemometer and
the wind speed is working again. Update 6/18/97: Looks like
JR's anemometer decided to leave the tower. New cups need to be installed.
It's going to be down for a while until we can get back on the tower. Update
6/16/97: Going up to East peak with Lauretta on 6/17/97 7pm. BTW,
JR has not experienced any glitches since the soundblaster card was replaced.
Update 2/8/97: Last week was a bit of a kluge so I put in
some better components for the telephone circuit, including replacing a
sticky relay (must have somehow disturbed it last week). The software was
updated and a new soundblaster card put in. Update 2/4/97:
Junior got hit with some nasty stuff over the phone line. The telephone
circuit was fixed. Now it's back to the old problem of finding the source
of the intermittent computer crashes. Still, if it fails it'll reboot for
the next day. Best we could do for now. I'll be up there again this week
to swap out the sound card which I believe is the problem. Update
1/1/97: Happy New Year! Well, Junior rang in the New Year with
some amazing winds to 95 MPH just around midnight.
A description of Junior is in order. Junior is a 80486DX2-66 computer
with 4M DRAM for its memory and virtual disk. The system is booted off
a conventional 3.5" floppy disk. No hard drive is used to avoid mechanical
failures. The sound is generated by a Sound Blaster 16 card. Once booted
up, there are no moving parts other than the power supply fan and the small
relay for the telephone connection. The C programming was custom to Junior
and with some very minor supporting Microsoft (DOS) software. The advantage
in this system is that only very common components (other than the interface
board) is used. Even the interface board comprises of components that are
very common and can mostly be picked up at a Radio Shack, Jameco, etc.
The interface card was constructed specifically for Junior. It comprises
of the telephone interface, temperature measurement and A/D conversion,
and interface to the wind instruments on the tower. The tower instruments
include an anemometer and a weather vane for wind direction. The detection
scheme is solid-state using an optical encoding method. The signal travels
down the tower and underground to the building where the computer is housed.
This is about 200 ft away. The signals off the tower are spike protected
(lightning, etc.) and signal processed. The front end is optically isolated
(6kV) from the computer.
The tower is about 60 ft. tall with the wind instruments at the top; thanks
to Kevin Young. The anemometer is calibrated to +/-3% accuracy with a precision
much less than that. The wind direction is set to geographic north and
is accurate to +/-5 degrees. The raw resolution is only 22.5 degrees (360
degrees/16 segments) but averaging (dithering) makes up for this. The temperature
measurement (+/-1 degree F) is not so ideal. The temperature sensor is
poorly located above the door on the main building. It probably reads inaccurately
(too high) in the late day when the west-facing dark bay is illuminated
by the setting sun.
The current algorithm for calculating the numbers generated by Junior
is the following: The current condition is from the last actual single
measurement. Every 5 seconds a reading is performed with the one-minute
averages stored in a data array. For the "15 minute average",
the last fifteen one-minute averages are utilized. The wind speed is just
an arithmetic average. The wind direction is a normalized vector average
(ie. average direction, evenly time weighted, not weighted by wind speed).
The spread in wind direction is computed from the one-minute average entries
to avoid transient swings in the wind direction. The "one hour ago"
condition is calculated using twenty one-minute averages from 50 minutes
to 70 minutes ago. This gives a very reliable average for trend information.
The very last temperature after the "good-bye" (John's voice)
is the lowest temperature for the last 12 hours. This is a good indicator
of the presence of an inversion layer which tends to keep things warm on
the mountain (~2000') overnight. The very last number is the number of
calls (modulus 200).
Remember that the winds can be higher and different from the HG launches
due to the higher altitude and different location. It also tends to experience
higher indicated NW winds due to a venturi effect of the terrain. In the
summertime an extreme inversion layer can set up by the marine effect.
Many times it can be northy on launch but southy at the beach. In addition,
don't ever trust the temperature readings when the sun is west of high
noon (ie. in the afternoon until sunset). The temperature sensor is in
the upper part of a *dark* bay and gets very hot.
Obviously, since we are dealing a standard PC, the system can be configured very
easily for a number of different functions. A future possibility is a "high-tide"
warning indicating time and height, but is mostly a concern for the winter months
when the beach is washed away. A paging service is very easy and if enough interest
can support the extra telephone cost then it can be done. That's all for now.