Weather sensor pod on roof. The rain collector
moved to here.
One of the nodes on the home control network is my weather
system tracks 8 parameters. They are : wind speed, wind direction, rain
fall, outside temp, outside humidity, inside temp, inside humidity, and
barometric pressure. I then derive three additional parameters, which
: wind gust (max winds over a period of time), the outside dew point,
the rain fall rate.
The weather node
The weather node is shown above. It has an LCD display with the name of the parameter
shown on the LED display. Currently, the outside temperature is shown.
At the top left is the switch to change the parameter shown, the middle
top has the inside temp and humidity sensor. On the bottom left (red
is the power cord, the middle bottom is the Home Automation Network
connection, and the bottom right is the fiber optic connection to the
sensor pod on the roof.
The external parameters are collected by an 18 pin PIC inside
instrument on the roof. The data is relayed via wire and fiber optic
(the latter for lightning protection) to my inside node which displays
one of the above parameters on a four digit 7-segment display and a 16
character LCD panel. The inside PIC also handles communication to my
conformant network to allow the home control computer access to this
data.If you wish to build part of the system, here is how I would
you go about it:
Anemometer and Wind Vane
The design of both these items is from the Electronics Now, Oct. 93
. They are made from commonly available items such as
40 PVC tubing. I purchased some hard to find parts from FE, but 90% of
the items were bought at my local home center (Lowe's
I made two small changes to the anemometer, I used a Hall effect sensor
instead of the magnetic switch, and I used a 1/4" bolt instead of the
bolt as the main axle. I also used 1.25" pipe instead of the 1.5" pipe
throughout, and added a lightning arrestor to the top of the whole
I estimate the output of the anemometer to be approximately 0.476
per mile an hour of wind. The friction is low enough that it start
at 1-2 mph.
Rain fall gauge
The design is similar to the EN article, except I adapted the whole
to fit inside a 1.25" 'T'. This makes it much more compact, and allows
it to be fastened by suitable plumbing to the above two instruments on
the roof stalk. It has a resolution of 1/32" of rainfall, which is a
less sensitive than original sensor since I used a more compact
was afraid of saturations in heavy rain, and for cosmetic reasons.
The microprocessor inside the weather pod. The silver RJ11 cable
the left leads
to the optical link and power supply,
the ribbon cable on the right
leads to the transducers.
I decide using the plain old LM34 as the temp sensor. This outputs 10mV
/ F, and by setting the reference input on the A/D on my PIC to 2.55
it allows me to directly read degrees F with no analog processing
I purchased the sensor from Digi-Key. By biasing the LM34
I can read temps down to -25F.
This is the moisture sensitive capacitor that detects humidity. I
this from Newark
about $9. The circuit I use is the
as in the EN article. My PIC counts the frequency directly, and results
in good accuracy since my only temperature sensitive components is the
humidity sensor and two 100ppm/C resistors (the chip oscillator has a
low temp sensitivity). Vishay Corp. part #2322-691-90001
in Jan 2016 shows that this unit is discontinued. An
alternate source is Digi-key
part no BC1333-ND (courtesy
The micro installed into the tube of the weather sensor pod.
The result is a neat installation that is easily removable.
This sensor is one of the Motorola absolute pressure sensors, also
from Newark. It comes with a small port so you can attach some tubing
it. With a gain of about 5 or 6 on the output, you can get 10mV/0.01"
This should be plenty for weather prediction/ logging applications. The
sensor is Motorola #MPX5100ASX and costs about $27.00. The sensor
is made of a semiconductor, and has temp compensation and and
amp built right onto the same die.
The kit is sold by Digi-Key
kit. It includes fiber, one transmitter, and one receiver in the SMA
of fiber optic holder. Operation is rather simple. You forward bias the
LED, and the optical transistor on the other end conducts. By suitable
biasing one can use this to replace an EIA/RS232 line. Since air breaks
down at about 3 megavolt/meter, this will afford me some protection
If you would like to build your own, I would suggest that you lookup
above mentioned article. It also contains the address of Fascinating
from whom you can buy the hard-to-find parts such as the bearings and
dual wiper pot for the wind vane.
In addition I would like to add that I appreciate all the
email that I received from my request for discussion. I think the most
important 'lesson' from all of it is to take lightning seriously.
Now the fun part of collecting and analyzing data begins. I
noticed a few well known correlations among the various data points: -
at night the temp falls, the reverse is true during the day
- with no rainfall or strong wind, the falling temp at night is
by rising humidity. - rapid changes in barometric pressure is
by strong winds.
The local NBC
used to offer 24 hour history plots
of the parameters that I am measuring. When comparing the data, one is
struck by the similarity in their appearance.
Humidity chart at WRC4 for 3/26
Humidity chart from my weather station for 3/26
Updates in text
Long term update
- March 11 1996. The weather station starts continuous
- May 1997. After more than one year of
operation all is
well. One minor glitch is that the rain fall sensor's sensitivity has
a little. However since the sensor is so high up, I won't be able to
to it easily, and I will have to live with it.
- December 1998. The weather station was
removed from the roof
for service during the fall of 1998. In general, the condition of the
pod was good, the following items needed to be fixed:
- The thin PVC parts were
pitted. These included the
cups and the rain funnel. These were replaced.
- The pot that registers the
direction was damaged, probably
movement of the wiper. I would say the pot has lasted 2 years.
- I found out why the rain
working. After a while
and dried onto the rain sensing spoon. This added weight stopped it
tipping. The only good long term solution would be not to use a rain
at the top of the house, but near ground level where it can be easily
- The initial installation
and removal of
the sensor pod required
someone to go up on the roof. This was undesirable to me. To address
I made a small vent window in the side of the house which I could reach
by climbing on the trusses of the roof. In this manner, I could access
the pod from relative safety from the attic of my home.
- March 1999.
stopped working. I was able to easily
take the sensor pod down from the roof to find out why (thanks to the
window mentioned above). It turns out that the ball bearing that I used
for the anemometer was damaged (after 3 years of operation) and needed
to be replaced. This episode made me realize that I needed a
ball bearing, and I was pleased to find out that Fascinating
Electronics (FE) was still operating, and they are now web
Since 1996, they have upgraded various parts of the weather sensor. In
particular, the anemometer cups are deep drawn aluminum, and the wind
is also aluminum. This new ensemble looks
nice, and I will probably upgrade parts of my system in the
I ordered two wind vane wiper pots and two ball bearings.
- April 1999.
The new ball
bearings arrived, and I
the faulty unit.
- September 2001.
ball bearing gives out
again, this time
after two years. It is replaced with a spare unit.
maintenance, I also replace the wind vane wiper pot. Four
bearings and wiper pots are ordered. Over the past year, the
temperature sensor inside the pod has become flaky, so I solder all the
connections on the wire-wrap board to fix this.
- March 10 2002. The PVC
anemometer cups degraded to the
point that they flew off! I called Fascinating Electronics,
have improved the overall unit again. This time, they added
drawn aluminum cups that should last forever. In removing the
from the roof I realized that its great weight was a big problem, and I
modified the unit to make it lighter and easier to handle. I
the lightning arrester and moved it down, and I removed the rain
That latter part never worked right, as dirt would foul up the unit
a few months. I also shortened the stalk slightly.
changes, made near the six year anniversary of the weatherstation, made
the unit much easier and safer to handle.
New modified weather station pod. Lightning arrestor and rain
- July 23
After missing the rain gauge for more than a
a new one has been built at ground level here.
to the roof:
- Replaced ball bearing in
vane. This unit has lasted 8
years. I have a few of these bearings left.
- Replaced potentiometer in
with a sealed unit from
Fascinating Electronics (first time sealed type used).
- Soldered PIC into a socket
all previously unsoldered
wire-wrap connections due to temp sensor erratic going from max to min
readings. It is clear that in outdoor environments, wire wrap
construction will only last a few years.
- The humidity sensor has
working, but since I was unable
to find the spare one in my parts bin, so it was not changed it out.
- October 30 2006.
Roof pod stopped working.
- Replaced crystal because
broke due to corrosion.
- Resoldered in 10k pot (it
- Replaced humidity sensor.
Image of the sensor electronics taken in 2020. After almost 25 years on the roof
everything is quite corroded.
- June 2020.
Total redesign of the roof sensor unit. After almost a quarter century
in the elements, the electronics inside the roof sensor unit (so-called
Weather Slave) have corroded and stopped working.
built a new electronics unit and moved its location into the attic so
that it is less prone to corrosion. It can be operated either in
the attic or sitting at the Weather Node for easy code development
(using no additional wiring). Instead of using UV erasable device
(PIC 16C71), it uses a Flash based 16F877. Much easier code
development and In-Circuit Debugging is very handy. See image
- Removed the wind
vane and I decided to focus on digital only interfaces. Thus I
eliminated the outdoor humidity (sensor would fail after a few years
due to temperature swings), and also the wind vane (the dual
potentiometers would fail due to the constant wiping back-and-forth).
Also eliminated the roof temperature as it was of limited value.
only sensor remaining on the roof sensor is the anemometer, and it
connects to the Weather Slave via a three-wire cable to drive the hall
effect sensor in the cap of the spinning sensor. It is now also
much lighter and safer to handle for me than the original design.
weather slave now has its own 1-Wire network and collects temperature
data for the attic zone, and the heat pump air handler unit.
Initially this function was handled by the Thermostat Node,
but I found that running such a long 1-Wire network proved unreliable
(too much capacitance on the line). So I greatly reduced the
length of the network for the Thermostat Node and the temperatures for
the Heat Pump are now monitored by the Weather Slave. These
temperatures are the Intake, Exhaust and Attic Temperatures.
New version of the weather slave node electronics.
Home Automation Main Page