Bill's Homemade 

Weather Instruments 

Observing the weather can be lots of fun and educational. There are all kinds of weather instruments available to collect weather data. This webpage provides information on how to make inexpensive weather instruments that I designed and built myself, spanning from 1980 to the 2020s.   I provide instructions on making the following sensors and instruments:

Electronic windvane with indoor, LED display

"Bi-plastic" thermometer from a discarded plastic bottle

Digital, electronic temperature sensor 

Electronic, carbon film, RH sensor (works well in detecting when fog is forming) 

Ceilometer for measuring the height of clouds

Electronic digital rain gauge with indoor display (And it's not a tipping bucket gauge)

Simple, low cost, near-infrared electronic snow depth sensor  (prototype)

Frost or Dew Sensor (will also detect rain and snow fall)

Homemade Electronic Wind Vane 

Instead of paying hundreds of dollars for a windvane that provides an indoor display of the wind direction, I decided to build one in 2004.  It would have a resolution of 22.5 degrees of azimuth and designed to last for years in the open environment.  

The heart of the weathervane design I provide here is the magnetic switch. These switches come in different sizes and are commonly found in burglar alarm systems and even some commercial electronic wind vanes.  When a small magnet passes over the switch, the switch closes.  This closed switch can be used to turn on a small light bulb or Light Emitting Diode (LED).  What I did was attach the magnet to a shaft that has a wind vane on top.  When wind rotates the shaft and the magnet swings over 1 of 8 switches placed on points of the compass, one or two Light Emitting Diodes (LED) mounted in a small box inside my house lights up and tells me which way the wind is blowing.

Putting it all together

I completed my weathervane in February 2004, and it's still working. Here are the parts I obtaned to build the electronic weathervane:

(1) Eight "press fit" Magnetic Reed Switches (shaped like small cans).  I have opened up one of them to see how they work and they are  mechanical  "reed-type" switches packed into the can and sealed with epoxy.  The switches I bought are rated to work down to at least -40F and provide millions of operations before wearing out. 

NOTE:  The switches I bought in 2004 worked for years, but two failed and I replaced them 2019.  Theses new switches only lasted a couple of years and failed.   They may not be rated to handle out-door weather conditions.  So shop around and find ones that can handle extreme conditions.   

These might work and are low cost:   Enclosed Alarm Reed Switches 

(2) Eight light emitting diodes or LED's, available at most electronic supply stores.

(3) Small electric on/off switch

(4) Length of 10 wire strand cable and another length of ribbon cable  to connect the windvane through a window and to the indoor display

(5) Battery holder for two AA batteries.  

(6) Plastic or wood box to house the LEDS and battery

(7) Plastic cylinder 2.5 inches wide (to house the switches and shaft), available at a plumbing supply store

(8) Plastic bottle cap about 1 1/2 inches in diameter

(9) Plugs made of plastic or wood for both ends of the cylinder.  I used a hole saw bit on my drill to make these.

(10) Five to 6 inch long, 1/4" bolt -Galvanized or brass

(11) Assortment of galvanized or brass bolts and washers for holding the vane to the shaft

(12) Small 3/4" length of ~1/4" ID pipe (acts as a bearing for the shaft as it exits the top of the cover)

(13) Six inch section of 1/2 inch aluminum pipe or sturdy plastic to make the windvane

(14) Small section of light weight plastic or metal sheeting to make the windvane

Using the diagrams and photos below, you can see how I put all these parts together.

Once you get the magnetic switches and LED's, don't run off just yet to get the other parts and start construction.  My design works, but after giving it some thought, you may come up with a design that is simpler, more weatherproof, and uses even fewer parts. Consider building the device using just magnetic reed switches and not ones placed in a small can. This will save you even more money and they may work better and longer.  And if you can find junk scrap parts, hardware, wire, etc, around your home (I like it when folks recycle!) you could build this weathervane for less than $30.  The key is designing the instrument so that the shaft rotates with the least friction and the magnet, switches, and circuit connections are protected from the environment (i.e., rain, snow, sunshine, etc). Also, be careful not to apply too much voltage as I have found that the switches will stick open or closed.  Once you have the unit built, install it so that it is perfectly level and use a compass to orient it to true North. Also try to locate it away from trees or other obstructions.

Operational notes

The windvane and switches have been working well since February, 2004.  The switches respond to changes in wind direction very quickly even in extreme weather conditions.  The instrument has functioned over a wide range of temperatures (between 105F and to 0F so far) and in heavy rain and snow.  Yet, if freezing rain is occuring the wind vane may get frozen in place until the ice melts. 

Early in its operation, two or three of the switches failed occasionally and then started working again.  I was not sure why.  I thought perhaps it was from too low a voltage or current in the circuit. The two AA batteries I used may be just enough for the circuit to work and when they drop just a little in performance a switch will not work.  Often when I passed an additional 1.5V through the switch circuit for a few minutes it works fine again. Another method I used to get the switch to work again is to simply replace the batteries with new ones.

On two occasions none of the switches worked.  At first I thought the switches got too hot in the Summer sun or the indoor display on the windowsill got too hot in the afternoon sun and either the LED's or batteries stopped working because of the heat.  In the end, I found that the wires wrapped around (not soldered) a metal washer in the base of the instrument got oxidized or dirty and that is what caused the circuit failure. So I had to clean the wires with sandpaper from time to time to get the circuit working again. So if you make this circuit yourself, I recommend soldering all outside connections..

If it is humid or foggy, condensation sometimes occurs on the shaft near the bearing where it later freezes, keeping the windvane from moving. To help keep this from occuring, I spray a little light oil (WD-40) on the bearing a couple of times a year.

Using a block of plywood to mount the switches was not a good idea.  I checked the instrument in March, 2007, and the plywood is slowly coming apart even though it does not get wet when it's raining or snowing. A block of plastic should have been used instead. I also checked the switches and the epoxy seems to have failed on one of them, allowing me to easily pull apart the switch and see how it works. Inside the can it's a reed switch.  I slid the switch back into the plastic can and it still works okay.

In April, 2009, I took the instrument down to replace the decaying wooden base, solder all outside wire connections, and give it a fresh coat of black paint.  See below for more information. Soldering all the connections solved the circuit problems noted above. 

In early 2014, the plastic bottle cap cover shattered while I was raising the shaft for lubrication with oil.  It was easily replaced with a new one.  

I noticed in 2016 and early 2017 that one switch would not light a LED if the magnet was just a tiny bit too far away. The result was a good measurement of due east winds, but no measurement (no LED turned on) for ENE or ESE winds . I removed the switch, rotated it a bit, and placed it back in.  It works fine now, but I wondered if this was a sign the switch was beginning to fail after years of use. 

In the Spring of 2019, after 15 years of use, a switch finally failed, causing the North light to stay on continuously.  I removed it and replaced it with a new one.  I also replaced the NW switch as it appeared to be not working correctly. I also replaced the cable from the wind vane to the indoor display.  

In December 2021, both switches replaced in 2019 failed.  These were from a different manufacturer and perhaps were not rated to handle outdoor extremes in temperature.  The other switches from 2004 were working fine.  I found two leftover switches that I had ordered along with the others in 2004.  These were installed and the windvane is working again.

And after more than 20 years old (2004 to 2025) the wind vane is still operating.  

Lastly, when installing the instrument on your roof  or elsewhere high above the ground, please be careful!  If you don't like heights, then get someone else to install your windvane.  Also please make sure it is properly grounded for protection against lightning (see my link below).

Schematic and Photos

Above is a schematic diagram of the windvane, without the protective cover over the magnet and switches.  Only 1 of the 8 magnetic switches are shown.  One lead from each of the switches are all soldered together to one length of wire.  This wire and the 8 remaining wires (9 total) from the switches are strung down the side of the house and into an indoor display of 8 LED's positioned in a circle inside a box.  Correctly setting the height of the magnet above the switches is needed so that two switches turn on when the magnet is between them.  If the magnet is placed too close, three switches will turn on.  If the magnet is too far away, the switches may not turn on at all.

If I ever build another electronic wind vane, I will not use reed switches epoxied into a can.   I will mount 8 reed switches (or maybe 16 switches for improved data resolution) like spokes on a wheel around the Phillips head screw in the center and have the magnet pass over them.   And they will cost less too.

Here is the weathervane base  (~2 1/2 inches wide) showing the placement of the 8 magnetic switches.  The switches are placed in a circle about 1 to 2 mm apart.  This way when the magnet passes over between two switches, two LED's on the indoor display will light up and  provide better wind direction resolution (i.e., NNE, SSE, NNW, etc., etc.,). Setting the correct height of the magnet above the switches has to be just right to make the switches act this way.  Also note the phillips head bolt in the center.  A drop of oil is placed in the "X" shaped head and the vane shaft (I sawed off  the top of the shaft bolt and filed it to a nice point) pivots on this.  Lastly, I used 3/4 inch thick oak plywood for the base, but a block of plastic would have been better and more weatherproof (see my notes below).

Here is the bottom of the windvane with the base removed. The white circle is the magnet inserted into a piece of balsa wood that was painted black.  Two bolts hold the wood piece to the shaft.  Note how the wood piece is aligned parallel to the wind vane. The base of the shaft was filed down to a nice point so that it would pivot easily in the phillips head bolt. The base of the magnet rides about 5 mm above the switches.  If you build this instrument, you may have to adjust this height based on the spacing you have between the switches (see the previous photo).

If you look deep inside, you can see where the shaft exits the top of the cover (PVC pipe fitting). The top of the cover was cut from a 1/2 inch thick piece of sheet plastic that fit snugly into the top of the PVC fitting. I then sealed the joint with silcone glue. Next, I hammered a small piece of brass pipe into the hole to act as a bearing for the shaft and the shaft fits nicely (not too tight) through it. I periodically lubricate this bearing with light oil

Here is the base of the instrument attached to the PVC pipe cover.  It's held in place with two small brass screws.

A total of 16 wires (2 per switch) come out from the base.  Eight are soldered to the cable (wrapped in tape), while the other eight are wrapped around a metal washer held in place with a central bolt and two brass washers and a nut. A black wire is also wrapped around the washer and run along with the cable to the inside of the house to complete the electrical circuit. In retrospect, wrapping the wires around a washer was not a good idea as the connection gets oxidized/dirty and disrupts the circuit. As I note later on below, I should have soldered the wires together and did just that five years after installation.

 This next photo shows the assembled windvane as it appears after being outside in the elements for well over a year.  Using a "U "bolt, the instrument is attached to a boom that is attached to a vertical pole on the roof of my house.

I used a white covering to help keep the switches cool in the Summer months.  To keep preciptation from getting inside the cover through the shaft hole,  I drilled a hole into a plastic salad dressing cap and placed it on the shaft. I later painted this cap black.  The wind vane  was then place over the cap and the two were bolted together.  I used plenty of silicone glue to seal the bolt on top of the shaft and all around it. The base of the cap rides just a few mm above the top of the cover The instrument as shown is waterproof, but I have noticed condensation build-up on the inside of the cover and along the shaft.  Yet, it has not effected the performance of the instrument.

The wind vane itself is made from a lightweight sheet of plastic (looks like cardboard) and aluminum pipe sealed shut at both ends.  I added a bolt at the forward end of the pipe to help balance the vane.  I painted it black to help melt any ice build-up on it when the sun is shining during the winter.

25 feet of cable connect the wind vane to the indoor display and two AA batteries (3V) is sufficient to light up the LED's.  If you use a longer length of cable, you made need to use additional batteries.

The next photo shows the top of the instrument with the vane lifted up so that you can see the shaft and bearing (This is what I do when I periodically apply oil to the shaft).  The bearing was hammered into the plastic top so that about 1/4 inch of it remained above the top.  This keeps rain water and snow melt from getting inside.  As noted earlier, the bottle cap rides above the top by just a few millimeters.  This also helps keep rain and snow from getting inside in the instrument. 

Here is the inside display showing two LED's illuminated, which indicates the wind blowing from the west-northwest.  Because the switches in the base are close together, the magnet will fire off  two lights as it passes over between two switches. This doubles the resolution of the wind measurement.  To make this work correctly you may need to remove and rotate the switches a bit after they have been inserted in the block base.  In addition, the magnet needs to be set at the correct height above the switches. 

A total of nine wires enter underneath a window (using using a small piece of  ribbon cable   would be better) and are conneted to a 9 point terminal in the back of the display to complete the circuit.  In the upper left of the diplay is a small switch to turn off/on the LED's.  Also behind the display I bolted a battery holder for 2 AA batteries.

By April, 2009, the wooden base was in poor shape and a number of the switches failed to work (owing to oxidation of the wire connections at the base of the instrument).   So I decided to remove the entire unit (windvane and display) and refurbish it.  Above is a photo showing the decayed wooden base and the back of the display showing how the wires are connected to the box.

Above is the new plastic base with the switches installed.  I used a hole saw to create it.  One wire from each switch (including the black wire)  is soldered to a wire loop in the center. In the past they were simply wrapped around a thin metal washer.  Over time oxidation occured and the circuits began to fail more often.

Above is the top side of the new base beside the old wooden one.  The new plastic base should hold up much better to the outside weather conditions. 

And here is a December, 2024, video of the wind direction display in operationn on a cold, windy day.  The green LEDs I used appear white, Outside temperature was 33F

Making a Homemade "Biplastic" Thermometer

There are many types of thermometers -- liquid in glass, bimetal, and an assortment of electronic-based sensors. In late 1970s, I was trying to make a homemade thermometer based on the bimetal concept of measuring temperature, but was having troubles joining two strips of different metals together.  In 1981, I wondered if two different kinds of plastic joined together would work the same way.  After giving it some thought and tinkering around,  I made a "biplastic" thermometer from a strip taken from a discarded plastic bleach bottle and an equal  length of "strapping" tape.  Just like two different kinds of metal,  the plastic strip and tape have different rates of expansion/contraction with changes in temperature.  And when they are fastened together the resulting strip deforms as the temperature changes.  

Putting it all together

The cost to make this simple homemade biplastic thermometer is just a few pennies.  Here is what I use to make this homemade thermometer:

(1) Strip of plastic from a bleach bottle,  2 to 4 inches long and as wide as the strip of strapping tape that will be used

(2) Length of strapping tape the same length and width as the plastic strip

(3) A cocktail straw

(4) Small block of wood 1/2" by 1/2" by 2"

(5) Small, stiff sheet of white cardboard

(6) A few thumbtacks

First, I cut out the plastic strip from a clean and well rinsed bleach bottle and affix the tape to the inside curve of the plastic strip.  Note: If you make the curve of the plastic into a tight semicircle before affixing the tape, the deformation (i.e., sensitivity)  of the strip will be much greater with temperature changes.  I have done this by wrapping the plastic strip around a piece of metal pipe, holding it in place with a hose clamp or something similar, and heating it in boiling water or in an oven at 250F for a short while.  This slightly melts the plastic and creates a tight plastic curve for me to attach the tape.

Next, I tape or glue a cocktail straw to one end of the plastic strip.  Using a thumbtack, I then affix the sensor to a small block of wood.   Finally, the block of wood is tacked to a small sheet of cardboard.  See the diagram below for more information.  Once completed, I test the instrument by simply breathing on it.  My warm breath makes the cocktail straw move significantly downward.

Calibration and Operation

To calibrate my homemade thermometer I use another commerical thermometer placed near my instrument.  Using a pen, I note the temperature and write its value on the cardboard sheet adjacent to the cocktail straw pointer.  I repeat this for every 5 to 10 degrees of temperature change.

The instrument works rather well for measuring temperatures from about 5F to at least  90F.  Compared to other temperature sensors, this thermometer has considerable mass, slowing it's response time to changing temperatures.  Accuracy can be as good as about +/- 1F around 65F and is lower (about +/- 3F ) near 10F.  The sensor can suffer suffer from hysteresis. and may lose its calibration over time.   These problems are reduced as the instrument is used over time and re-calibrated as necessary.  The instrument shown in the photos below has been tested in cold and warm conditions multiple times from 2014 to 2016 and the accuracy has been good.  

I have found that bleach bottle plastic and strapping tape works best, but try different types of plastics and tapes.  Also experiment with different shapes of the plastic strip.  Remember, the tighter you make the curve of the plastic, the more the cocktail straw will move for a given temperature change.  

Here is the homemade thermometer outside where the temperature was marked at 23F.  

Here is the homemade thermometer inside the home where the temperature was 67F.  Because the plastic was shaped into a tight circle, the pen arm moved downward over 2 1/2 inches over a 44 degree temperature change. Click the image for a larger view. I have taken this thermometer several times from a warm room to 10F outside and back inside again and it maintains it's calibration.  The sensor shown here is over two years old and over this time the sensor has possibly been "broken in", improving it's accuracy.

A Simple, Low-cost, Digital Electronic Temperature Sensor

Light Emiting Diodes (LED) come in many shapes and emit a variety of colors.  Another type of LED is the blinking or flashing LED.  These are LEDs with a tiny IC multivibrator chip embedded inside that cause the LED to start blinking when connected to a power source.  Blinking LEDs can be purchased for less than a dollar each and come in a variety of colors.

The flash rate of the blinking LED is not constant. It will vary with significant changes in the applied voltage (lower voltage = slower flash rate and vice versa).  However, in 2010 I found that the flash rate per minute varies almost linearly and accurately with changing temperature.  As the temperature decreases (increases) the blink rate of the LED increases (decreases).  Red LEDS blink the fastest, while yellow ones blink slower and green ones even slower for a given time range and temperature.

To measure temperature accurately with a blinking LED, a constant voltage source is required.  A 2 to 6V DC power supply from an AC wall outlet supply can provide stable voltage across a blinking LED placed in series with a 10 to 30 Ohm "limiting" resistor. If a battery used, the voltage can be stabilized by using a voltage regulator IC chip across the battery.

As the LED blinks the voltage drop across it varies.  To record the blink rate of the LED it can be built into a circuit that counts and even displays the number of blinks (and the temperature) that occurred over a time period such as one minute. I have inserted a blinking LED into a simple, audio-oscillator circuit. As the LED blinks on and off, the oscillator emits audible “beeps” to a speaker.  The software application or App, “LiveBPM”,  which displays the beats-per-minute of a song, picks up these beeps and counts and displays them as beats per minute (BPM). See Figure 1.  A calibration chart or table showing  beep rate versus temperature allows for the temperature to be determined from the display.

Figure 2 shows a plot of the blink rate per temperature change for two yellow blinking LEDs.  The LED was compared to an electronic thermometer placed near-by.  Note that the calibration is linear from at least +16 to near -20C.  Over this range, the rate of temperature change is about 0.95C/blink for a yellow LED.  The LEDs have been tested for months and the calibration remains stable. Using LiveBPM, one can detect temperature changes of around 0.1C.  The accuracy is around +/- 0.5C from at least +20 to -20C. The linearity of the calibration plot in Figure 2 probably falls off and becomes more curved as the temperature falls below -30C or warms above +30C.  The temperature response time of the sensor is not slow.  After removal from a freezer where it was  colder than -15C, the sensor recovered to room temperature  in just a few minutes.  Shaving away the LED plastic cover helps speed the response time. I am testing the LEDS over a wider temperature range and will post the results on this website.

What causes the LED blink rate to change with temperature is not clear. Temperature changes do affect the performance of diodes, resistors, and capacitors. These components are inside the LED and IC chip.   Another possibility is that the LED is physically changing (e.g., expanding and contracting) with temperature change and this changes the IC circuit, causing a change in blink rate.   The blinking LED sensor allows for simpler, lower cost electronic circuit designs to measure and display temperature.  Yet, I have found no formal papers or studies on this unique phenomena with blinking LEDs.  So, I have published a copyrighted paper.

Figure 1.  LiveBPM App display of "beats per minute"  However, here it is displaying temperature changes over a 30 minutes period from a blinking red LED inserted into an audio oscillator circuit.

Figure 2: Temperature calibration plot for two blinking yellow LEDs.  The x-axis is temperature (degrees C) and the Y-axis is the blink rate of the LED during 1 minute.  LiveBPM software was used to determine the blink rate of the LEDs.

I have posted other data plots here: https://www.instructables.com/id/Simple-LED-Temperature-Sensor/

References:

Light Emitting Diode: https://en.wikipedia.org/wiki/Light-emitting_diode#Flashing

LiveBPM:  https://itunes.apple.com/us/app/livebpm-beat-detector/id554766778?mt=8

Homemade Electronic Relative Humidity (RH) Sensor

Experimenting with "Wire Glue"  can be quite interesting.  The glue is water-based and contains very tiny carbon particle ("nano-carbon").  It's cheap and conducts electricity nicely when dry. I have been looking into ways in which this material can be used for making electronic weather sensors. Beginning in 2009,  I have made simple, inexpensive,  electronic relative humidity sensors with this glue.  At first, I applied the glue to materials that absorb (and  evaporate) water vapor in the air and expand and contract while doing do.   I thought that as the material absorbs moisture, it expands and the carbon particles move farther apart, raising the electricial resistance.  As it dries, the carbon particles move closer together lowering the resistance.  I have mixed the carbon glue with Elmers Glue as it absorbs and evaporates water vapor.  Later on in 2024, I  experimented coating just the glue on wood, paper and PVA (cold water soluble) plastic sheet.  Of all these materials, wood worked best.  

After the above experiments,  I  placed a strip of the carbon based glue on a glass jar bottom and gluing wires on both ends.  Unlike the materials I tried earlier,  glass does not absorb water vapor.    When I applied the Ohm meter, I got the same results as I observed above.  When I breathed on the glue strip the resistance rose significantly.  See the photos below.  Pure water has very high electrical resistance.  It appears that water vapor is absorbed by the carbon particles in some way, causing a rise in the resistance.  Water vapor evaporating from the carbon particles causes the resistance to decrease.  A thin glass rod or perhaps a glass microscope slide would make a better sensor but I did not have them for this test.  Total cost to make this sensor is just less than 50 cents.

Above is a overturned glass jar with a strip of the wire glue applied.  Wires were glued on both ends and I applied a strip of tape to hold the wires in place.  The photo on the left shows the electrical resistance at 363 Ohms inside my lab.  When I breathed on the sensor for about a minute the resistance rose to over 700 Ohms as shown.  After I stopped breathing on the sensor, the resistance continued to slowly rise for about a minute or so longer and then it began to decrease.    Response time may be improved by using a thinner coat of glue.

Below is an improved electronic RH sensor.  I painted, with a small, thin brush, a thin coat of the carbon glue on a glass eyedropper tube.   Wire leads were attached on each end and also coated with the glue. The sensor was mounted on a block of wood.   Two wires, from a  4-wire cable were soldered to the wire leads and the sensor was then placed outside inside a DIY ventilated covering.  The cable was run into my home and attached to an Ohm meter.

Calibration and Operation

Calibration of the carbon based RH sensors is not easy as I don't have an accuate, close-by RH or dew point sensor to compare it to.  I have a hand-held RH sensor with a display, but it is old and probably has a dry bias in the readings.  I have used NWS hourly ASOS weather observations of dew point temperature at two near-by airports ~5 to 15 miles away.  With this limited information I have noted the following for a glass based sensor.

When the atmosphere has an RH exceeding 90%, the resistance of the glass based sensor I made (your sensor may vary)  is more than 2,000 Ohms.  When  fog formed (100% RH), the resistance exceeded as much as 20,000 Ohms. With a 30 % or lower RH in the air, the resistance falls to around 1,000 Ohms or less.  Change in resistance follows well with changes in the atmospeheric RH.  Accuracy is probably around +/- 10 to 15% .  The sensor does a good job of detecting when the atmosphere is near or at saturation and this could be of use if one only needs that information (e.g, fog detection).  Response time to changes in RH with the glass based sensor is a few minutes or more and likely increases significantly in a very cold temperatures (<0F).  The RH vs resistance curve plot is logorithmic with low sensitively at low RH and high sensitivity at high RH.  And this curve is likely temperature dependent So far, I have been testing the sensor at temperatures in the +20 to +100F range.  Hysterresis probably occurs and it's unknown how long this sensor with work accurately placed in an outdoor environment   So far the sensor has worked okay for more a year.  

The sensor you see below was placed outside in March 2024 and has worked well,  But by December, 2024 it appears that in dry weather, the sensor now reports resistance values around 1,100 Ohms or less.  So there is some drift in the calibration.  

Electronic RH Sensor made from wire glue and a discarded glass eyedropper tube. Total cost for the sensor - a few pennies. I mounted this sensor outside in a white plastic container (opened at both ends) that was oriented horizontally on a pole.  After more than 12 months of use (March 2024 to July 2025), it still works quite well in detecting 100% RH and fog, as well as very dry air (<30% RH)

Above is a scatter plot of RH vs Ohms from the carbon film RH sensor.    RH was obtained  from hourly dew-point  observations taken at two airports ~5 miles east and ~15 miles west of my location and averaged.  Temperature needed for the RH measurement was obtained from a electronic thermometer placed near the RH sensor.  It is not sited well and likely has some solar heat contamination.  Above 2300 Ohms the RH is at or close to 100%.

Homemade DIY Ceilometer

In 1980, I built a homemade ceilometer that measures the altitude of cloud bases at night.  I built the instrument using a Edmund Scientific 18 inch parabolic reflector and a 300 watt clear (not frosted) incandesent bulb. The mirror comes with a central hole to mount the bulb and has a focal length of 4.5 inches.  The filament in the bulb was positioned to be at the reflector focal point.  This way, a beam of light is projected.  At nightfall, the mirror was then aimed at a 90 degree angle and a patch of light would be seen at the bases of clouds up to about 4,000 feet.  To measure the height,  I made a simple clinometer from a protractor, straw (site tube), and weighted string.  I would then walk a known distance (at least 300 feet, but the longer the distance, the better) from the vertical light beam and measure the angle from the ground to the patch of light on the cloud.  See the diagram below.

Using simple trigonometry, I could then measure the height of the cloud base above the ground. In the late 1980s, I tried a much brighter bulb and was able to measure altocumulus  cloud  base heights exceeding 10,000 feet up.  To get an accurate height measurement,  one night I measured the angle 3,300 feet away from the ceilometer. I called a friend at my house from a near-by payphone (no cell-phones in those days!) to turn on and off the light to help verify I saw the dim light spot on the cloud base.  That night the cloud had a base just  over 9,000 feet up.  I have to say it was a lot of fun.

If you build this device,  do not aim it at flying aircraft and shut the light off if aircraft approach.  Check local laws/restrictions to see if you can legally operate a ceilometer of my design in your area.   

I strongly recommend wearing safety goggles when using this instrument.  If rain or snow starts falling, raindrops or snow hitting the hot light bulb may cause it to shatter or explode.  A possible solution to this would be to place a thick plexiglass sheet over the ceilometer mirror and bulb and at a height where the hot bulb will not melt the plastic sheet.

Homemade digital rain-gauge with indoor electronic counter

For many years, I have placed a plastic, 4-inch-wide rain-gauge in my backyard to monitor rainfall for my vegetable garden.  But I always wanted a way to measure the rainfall amount and intensity while it is raining and have the data displayed inside my home.  

You will find electronic rain gauges being sold that can measure rainfall amounts to the nearest 0.001 inch.   Here is one selling for about $100:

ULTIMETER PRO Rain Gauge (peetbros.com) 

The gauge counts the number of drops passing through metal plates and this is converted into rainfall amounts.   They don't mention what electronic components are used, but it got me thinking about making such a sensor that counts water drops that can be converted into rainfall.

The Darlington Pair transistor has extremely high gain and can be used to detect very, very, low electrical current or conductivity as is found with pure rainwater.   Search the internet and you can read that a simple electric circuit can be constructed with a Darlington Pair to detect when rain starts. 

In late 2021, I constructed an electronic, digital rain gauge with an indoor counter display than records the number of drops touching wire terminals connected to a Darlington Pair transistor and counter circuit.  The number of drops can then be equated to inches of rainfall. Total cost: Less than $10.

Parts you will need,

Plastic Funnel  Two to four inches in diameter (I bought a set of four at Harbor Freight Tools for about $2.  One is a 2.5-inch funnel and that's what I used).  Note: the wider the diameter, the more accurate the rainfall measurement.

Funnel Support  This can be any weather proof container that the funnel can be placed in.  

Darlington Pair Transistor.    I suggest buying 10 of them, as your experimenting may cause the transitor to overheat and fail.  This has happened to me.

MPSA13 Darlington Transistor | Jameco Electronics 

Electric Counter.  I bought the one below.  There is no reset to zero button, but they work and are low cost.   Before a rain event,  write down the value on the counter and subtract it from the value recorded after the rain ends.

G21361 - (Pkg 2) Kessler Ellis E660AF 4.5 to 6VDC Impulse Counter (goldmine-elec-products.com) 

Some resistors ranging from about 10 Ohms to as high as 10 megaohms.  Or obtain some variable resistors.

Length of two-ply single strand wire.  I used some junk, two strand "speaker wire".

Small length of plastic or glass tube and a plastic cap or a pipette.  This is needed to help create a small, uniform sized drop. I used a clear plastic pipette and cut off the very top portion and bottom. The inside diameter of the tube is 4mm. I cut the bottom at a 45 degree angle to help the drop form and fall away from one side of the tube. It slid nicely over the funnel neck.  Here is a supplier, 

Amazon.com: uxcell 20 Pcs Plastic Disposable Pipettes 0.2ml, Clear Transfer Pipettes, 61mm Length, Liquid Dropper for Lab : Industrial & Scientific 

I have also experimented using a discarded glass eyedropper to create the water drops. The design goal is to have the drops fall away at the same size (i.e., volume) in heavy and light rain.  The drops created in an IV line drip chamber line may work well.

6V battery power supply.  I have used 4 AA batteries in a cheap, plastic holder.  In my first prototype design, I used a discarded AC to DC power supply, providing 6V and 500 mA.  This works okay, but I later tried a 6V, 300 mA DC power supply from Jameco and it works even better. 

AX06V300: Jameco Reliapro : Unregulated AC to DC Wall Transformer 300mA 6VDC : Power Supplies & Wall Adapters 

Rain Gauge.  You will need this to calibrate your digital gauge.  It should be at least 2 inches in diameter and can preferably measure rainfall to the nearest 0.01 inch.

Below is the circuit diagram and a diagram of how I attached the wires to the plastic funnel.

Construction and Operation.

You can build this circuit quickly on a small electronics bread board. This will allow you to try different resistor values and their placement in the circuit . To verify your circuit is working, place the two wires for the drop measurements in some water or touch them with wet fingers and the counter should increase the number display by one digit.   If it increases by more than one, try lowering the voltage a bit or add some resistors to the circuit as indicated in the  diagram below .  If it does not count a drop, increase the resistance.  With the counter I bought (see above) four, AA batteries work well.   A DC power supply providing about 6 V and 300 to 500 mA also works.  

Now as for connecting the wires to the funnel, your goal is to have the rain drop simultaneously touch both wires long enough so that the closed circuit can set off the counter.   If the drop hits just one wire or falls away too quickly, the counter will not activate. What I did was to expose one inch of bare wire on each contact.  They start about 2 mm apart but fan out a bit towards the end.   I also bend them downward at about a 30-to-45-degree angle.  When a drop leaves the base of the funnel it hits the two wires and rides down them briefly before falling away.   It also helps to keep drops from staying on the wires, causing a short circuit.  This configuration works for my design, but you may have to do a bit of "tuning" of the wires to get it right for your circuit.  

When you have your gauge done keep it all inside and test it.  Get a small plastic bottle and put a small pinhole at the bottom so that it creates drops of water every few seconds or more.  Fill it with rain or distilled water and place it in the funnel.  Avoid tap water as it has far higher electrical conductivity than rainwater.   Drops will fall to the base of the funnel and onto the wires and you can see if your circuit is accurately counting each drop that falls away.  If it does not work or a drop stays on the wires (causing a short-circuit), try adjusting them.  Also try changing the diameter of the plastic tube.   I'm still tinkering around with other ways to set the wires.

Once you have the wires glued and set correctly, glue a plastic beverage bottle to the base of the funnel so that it will keep the wires that run close to the funnel dry and to keep the wind out.   Next, affix the funnel to the top of a pole that can be stuck in the ground. Locate the gauge away from trees and other obstructions.  See the photos below.

To calibrate your digital gauge, you will need a rain gauge than can preferably measure to the nearest 0.01 inches.  If you don't want to buy one, you can make a rain gauge cheaply yourself. Search the internet for ideas.  Place both gauges away from trees and other obstructions.   With the 2.5 inchwide funnel and a 3 mm opening at the bottom, so far, my digital rain gauge measures about 11 to 12 drops for each  0.01 inch of rainfall.  This calibration was done in light rain to moderate rainfall  and my 4 inch rain gauge was a few feet away.  I was also about to discern changes in rainfall intensity by noting the time between drop measurements.   My circuit can count drops falling as fast as once or twice per second which corresponds to about 3 to 6  inches of rain an hour.   I also added wind shield around the base of the funnel so that high winds don't blow the drops away from the wires.  A section from a clear plastic soda bottle or food container works.  

During one rain event that produced 1.1 inches of rain  over a 24 hour period, the digital gauge measured 12 to 12.3 drops per 0.01 inches of rainfall as intensity ranged from very light to moderate.   I periodically compared the 4 inch gauge amount and the number on the counter.  Yet in one instance  a small droplet remained on the wires and the counter rapidly increased by 118 and then the drop fell away.  This introduced a ~10% error.   I am trying to replicate how this happened and correct it by resetting the wires below the funnel.  In another rain event, a water drop ran down the wires and stayed.  This caused  a short circuit, and the Darlington transistor apparently overheated and burnt-out.    

During some heavy rain events with larger rain drops,  there were occasions where the counter did not count all drops falling - some were falling to the wires much faster than once per second and two drops were on the wires at the same time. I think this was from some large rain drops falling directly down the funnel, (never touching it) and right into the pipette.   I may (not sure yet) have solved this problem by placing a small glass marble  (or you can use some other obstruction) at the base of the funnel.

I am currently experimenting with a new gauge design using a four-inch wide funnel.  The wider diameter increases the number of drops to about 27 for each 0.01 inches of rainfall (1 drop = about 0.0004" of rain).  See the photos below.   I might try  making a very wide gauge to measure the tiny amounts of water that come from falling drizzle or morning dew accumulation.

It is important that the funnel and pippete  stay clean.  Debris can get its way into the pipette and  create different shaped drops that cause the circuit to not work. A piece of window screen placed inside the funnel or placing a glass marble at the base of the funnel can help keep debris out.  It's also essential to  clean the exposed wires with steel wool every six months or so to keep the circuit working correctly.

A few final notes: If you are good with electronics, you could certainly design a better circuit to count water drops.  I like the Darlington Pair transistor as it is easy to build into a circuit, is cheap, and has long wires for easy soldering.  But they have pros and cons with their use.   You could try making an Op-Amp or other amplifier circuit that works better.  My design uses a low-cost, electro-mechanical counter, but if you get your digital gauge to work well, try to build a circuit to have the drops counted, converted to inches of rain,  and then displayed/plotted on your PC.  Also, the larger the diameter of the funnel,  the more drops (and counts) will fall for a given amount of precipitation, so that should be considered when building your rain gauge.  I used two wires to capture the drops and make a circuit.  Try other configuratons of the wires and try differnt kinds of conductors (small metal plates). And lastly, I have tried a plastic pipette and a glass eyedropper glued at the bottom of the funnel to create, as best as possible, drops of water that are uniform in size and don't flow out too quickly.  I am currently getting an accuracy of 5 to 10%, but it's less in very heavy rain.  I am continuing to experiment in this area and you should too. Have fun.

Schematic and photos 

The top photos show my first 2.5 inch digital rain gauge prototype under test.  If you look closely at the photo on the left, you can see the plastic pipette section over the base of the orange funnel and below it are the two wires.  I tacked a food container to the pole to keep wind away from the base of the funnel.  However, this does not keep the wires near the funnel dry.  If they get wet, they may cause a short circuit, and the counter will not count drops.

The photo on the right shows a better cover over the wires.  It's a plastic beverage bottle glued to the funnel.  This helps keep the wind off and keeps the wire running along the funnel dry. 

Near-by is the 4-inch rain gauge used to calibrate my digital gauge.  With this set-up, about 11 to 12 drops equal 0.01 inch of rain-fall in light to moderate rainfall

The bottom photo shows the electric counter with the circuit board taped to it.  You can see the Darlington transistor in the board.   I'll clean up and shorten the wiring and am looking for a small plastic case to insert the counter, board and wires,  To the right is a old plug-in DC power supply providing 6V and 500 mA.  I later tried a 6V power supply at 300 mA from Jameco.com and it works much better The circuit uses just a emitter to base bypass variable resitor set to a value of 1.75 megaOhms

Above is another prototype digital rain gauge under test in June, 2022.  I found an old four-inch diameter funnel and placed it in a discarded beverage cup (Dunkin Donuts!) to keep the wires dry and not exposed to wind.  I then caulked the funnel to the cup to keep moisture out.    I used a plastic pipette to generate drops and set the wires the same way as was done with the 2.5 inch gauge above.  I then placed it in another discarded container to keep it upright.  With this configuration, I get about 27 drops (+/- 3) for every 0.01 of rain.   And someday, I'll place the funnel in something more sturdy, weatherproof, and more "scientific looking" than discarded containers.

The bottom photo shows the circuit on another board.   Since the darlington transisitor can overheat and burn out, I insert one into a 4-pin wire socket and it can be easily removed and replaced..  The ground wire is inserted into this socket as well.   I found a junk 4-way release connector and the rain gauge sensor wires and the power supply wires are inserted into it. Two 1-megaohm resistors were placed in series to provide a 2 megaohm "bypass" resistor across the transistor emitter and base wires.  And I used a variable 5 megaohm variable resistor across the incoming wires from the gauge.  Later tests showed that a higher resistance reduces heating of the darlington pair.  And maybe later I will mount the board and counter into a box or something similar.

Above is the funnel upside down showing the pipette and the two wires.   I used a hot-glue gun to affix the wires to the funnel and pipette.  I made a ~45 degree angle cut at the base of the pipette tube to help drops form and fall away on one side of the tube.   Also note the small square of paper (from an egg carton)  placed on the tube.  On occasion a drop of water gets stuck at the base of the wires (or where they diverge), causing a short-circuit.  I think one possible cause is from a tiny amount of water running down the exterior of the pipette and getting onto the wires.  The paper is there to absorb the water.

Here is the rain gauge top flipped over (sitting on a fence) showing how the wires were affixed with glue and tape to the glass eyedropper tube.  The photo on the right shows the gauge top back in place.  A glass marble was placed at the base of the funnel to slow water draining down the tube in heavy rain events. It improves accuracy in heavy rain andit helps keep debris out as well. 

Simple Snow Depth Sensor (Prototype design)

After a snow storm, I like to to grab a ruler and measure the snow depth.  But I have always wanted  a way to do that with an electronic sensor of some kind that displays the results indoors and while it is still snowing.  There are DIY and commecial snow depth sensors that use ultrasonic devices or a laser to measure snow depth, but they are either complicated to make or too expensive.  

Liquid water and ice significantly absorb near-infrared  (NIR) light waves in a band occurring  around 1,000 nm.  In 2024 . I began experiments with NIR emitting LEDs  (peak output at 940 nm) coupled with an associated  NIR phototransistor  and Darlington Pair transistor amplifier circuit to detect falling rain and snow.  

I placed the NIR LEDS and the NIR phototransistor about 1 foot opposite and horizontal of each other and measured small fluctuations of the phototransistor current or voltage as snow or rain fell between them.  Later, I placed the phototransisitor in a pvc tube with a small lens at the other end to focus light on it.  Adding the lens greatly increases the sensitivity of the phototransistor.  I used a simple darlington pair transistor circuit  to amplify the signal. I'm still experimenting  to improve this circuit.

In early 2025, I saw that snowfall  was on the way and very quickly designed and tested a prototype snow accumulation sensor.    The design concept  was having a NIR LED light shine veritcally upwards through the snow layer and be detected by a phototransistor positioned above the fallen snow.   As the snowdepth increases the NIR light passing through would decrease and this could be measured and coverted to inches of snow

 I placed a tight circle of six NIR LEDS, aimed vertically,  under an elevated sheet of glass.  They were turned on with a 9V DC power supply.   At about 18 (and later at 12) inches above the glass sheet I aimed my phototransistor/lens tube downward.   The wires from it were brought inside to a darlington pair transistor/resistor amplifier circuit that was hooked up to a voltmeter.  See the diagram and photo below.  With just a clear glass sheet between the LEDS and the phototransitor, 12 inches apart, I measured over 6V across the resistor circuit inside.   As snow fell and covered the glass, the voltage slowly decreased as the snow got deeper.  At 4 inches deep, the voltage had dropped to about 0.8V.   I don't think the sensor in this configuration will work well at snow depths deeper that this.   

 You can buy NIR LEDS, phototransistors, and lenses at a number of locations for a few dollars or less. I bought a  pack of 10 at Amazon:

Amazon.com: HiLetgo 5mm 940nm LEDs Infrared Emitter and IR Receiver Diode 5MM Infrared Emission+Receiver Tube (Pack of 10Pairs) : Industrial & Scientific 

I bought the small lens (focal length 59 mm) at Surplus Shed. See the link below.  I used a darlington pair and a emitter resistor to amplify the phototransistor signal.  I measured the changing voltage across the resistor to observe increasing snow depth.  You could certainly build a better amplifier to make this sensor better detect snow.

Snow Depth Sensor Diagram Above.   I used a sheet of glass to accumulate the snow, but a sheet of wood with a glass covered hole to let the NIR light through will work too.  The phototransistor was fixed at the top of a wine cork that fit snugly inside a small length of PVC pipe with a 20 mm inside diameter.  A plano-convex lens with a focal lenght of 59 mm was affixed on the other end.  Moving the cork in and out, I was able to get a nice focus of the NIR light onto the phototransistor.    The NIR light source is a tight circle of six LEDS, connected in series, and powered by a 9V DC power supply.  I placed a resisistor in series to provide ~20 ma of current.

Above Photo:   Here is the set-up the next day after the snow had melted.  I built the prototype snow depth sensor rather quickly using whatever I could find around the house and get it all done before the snow began   A metal plant pot holder was used to hold everying thing together.  I used to a strip of plastic to hold the phototransistor/lens tube above the glass.  I placed of piece of aluminum and tape over the tube to keep it dry.   I stacked blocks of wood to get the NIR LEDS just underneath the glass sheet.   Yes, it's messy looking, but it worked.

Final Notes:

The snow depth sensor I designed here works nicely to detect the start of snow accumulation and increasing snow depth.   It should work to detect frost that completly covers the glass between the LEDs and phototransistors. 

Accuracy of the snow depth sesnor is not known yet The voltage reading vs snow depth is not linear and may not be repeatable for each snow event.  Snow flake size, for starters, probably will have an impact on the transparency of the NIR light.

The phototransistor detected the falling snowflakes in front of it causing the voltage readings to flucuate noticably..  

If the snow melts a bit, compacts, or refreezes, the NIR absorption is greater, impacting the measurements.  

Right now, it measures snow depth up to about 4 inches.  Brighter LEDS, bringing the phototransistor a little closer to the glass,  and improving the amplifying circuit indoors may improve the sensitivity at greater snow depths.  Trying light waves other than NIR may work better too.  I am also experimenting with sensitve photodiodes with a peak at 1,000 nm

It detects snow accumulation best at night with a dark sky and no other light sources around.

Outside cold temperature changes likely effect the performance of the phototransisitor and maybe the LED brightness

The phototransistor tube should be held in place with something narrow that will not obstruct snow accumulation below it.  

 I am experimenting to see if this configuration can be used to measure rainfall amounts.  One idea is a  NIR LED and phototransistor would be placed on either end the rain guage, The accumulating water fills the tube  and increasingly blocks the NIR light This change could be measured and converted to rainfall amount

The cost to build this snow depth sensor and experiment with it is low.  Have fun if you give it a try.

Frost or Dew Sensor (and it detects rain and snow too)

Search the web for "Rain sensor, Darlington Pair"  and you will see simple circuits and sensors to detect rainfall.   I am interested in dew and frost formation and when it occurs and dissipates.  In March, 2025, I improved the sensor to more quickly detect the tiny drops and ice crystals that form dew and frost.  It also will detect larger falling rain drops and snow flakes.  

To make the cicruit I used a Darlington Pair transistor, a blinking red LED,  a junked 3V DC power supply with 200 ma output, and two resistors.  The outside sensor is simply two strips of galvanized steel that are bolted to a block of wood and placed face up to the sky.   The strips are spaced very close --just the thickness of a sheet of paper-- and affixed above the block of wood, suspended in the air.  See the diagram and photo below. When a tiny amount of water or ice bridges across the metal strips, the LED begins to blink rapidly.  Total cost: a few dollars.

Above is a diagram of the Dew and Frost detection sensor.   This circuit works, but you could certainly make a better one with other configuations and resistor Ohm ratings.  I ran a strip of paper between the plates to get them very closely spaced together. Coating the strips in black paint may improve the black-body radation effect and enhance condensation on them.

Above is the dew and frost sensor outside.  It is next to my homemade sensor shelter housing my carbon glue, relative humidity sensor.  Two  galvanized metal strips, 1.75X 5 inches, were separately bolted to a block of wood and positioned about 1 inch above the block. They are suspended in the air and not sitting on the wood.  I also tilted each strip slightly to help condensate pool towards the gap in the plates.  I wrapped a wire around each bolt and fed them back into the house to the circuit.  Any type of non-corrosive metal should work.

The sensor shelter is from discarded container.  Both ends are open and each have a screen covering to keep insects out.  Out of view on the bottom of the container, I cut a round hole and placed  a small  12V "muffin" fan to blow air through the container.  It's not an optimal design, but it does the job.

 Links

Well, two of my favorite electronics suppliers, All Electronics and BGMicro, are now gone, but the companies below sell lots of electronics.  Also try Amazon.

Electronics Goldmine:  

Jameco Electronics:   

Electronix Express:  

Surplusshed  

When installing anything on your roof, make sure it is properly grounded to protect you and your home from lightning.

American Science & Surplus. Another source of interesting stuff.

My Other Websites 

A description of how I built a large homemade Dobsonian telescope 

My recipe for hot sauce Making your own hot sauce is easy and tastes great.

 Webpage Author: H. William James.  All images and text are copyrighted.  2004 - 2024.

Disclaimer Statement: The information on this website is provided for informational purposes only. The instruments and devices described by the author function as described, but your results may vary.  The author disclaims any liability for any damages, injuries, or any other losses of any kind you may incur.