Tracing the Sun’s Journey using Shadows

March 29, 2026March 29, 2026Ashish Derhgawen Leave a comment

A few months ago, I visited Jantar Mantar in Jaipur, where I was fascinated by all the giant sundials. After I got home, I went down a rabbit hole of sundials and astronomy. For a sundial to work properly, it has to be aligned with true north and adjusted for your latitude. This allows the sundial to compensate for the monthly variations in the Sun’s apparent position in the sky (its declination).

Typically, a sundial indicates the local solar time. However, a sundial could potentially also tell you the approximate date and month. This felt like a fun little problem to work on. I soon realized this wasn’t a unique idea and had already been done. The Cadran solaire Papillon in Paris is a sundial that is based on this idea:

To understand how this type of sundial would work, a basic understanding of astronomy is necessary. The Sun’s declination changes throughout the year as it moves between the Tropic of Cancer and the Tropic of Capricorn. As the Sun’s position changes, so do the direction and length of shadows that it casts. In a sundial, the object that casts a shadow is called a gnomon. Because both the direction and length of a shadow depend on the Sun’s declination, the shadow already carries information about the time of year. The paths traced by shadows on the ground are a direct record of the Sun’s journey across the sky! It’s remarkable how much we can learn about planetary motion simply by observing something as ordinary as a shadow.

I came across this website that has printable sundials that show the shadow path for different months. This seems very similar to the one in Paris. However, I couldn’t figure out how the shadow paths were determined. For me, that was the more interesting problem. Of course, it could be done by mapping the paths for an entire year, but that would be a lot of work. Moreover, it wouldn’t work if I were to move to another location.

I soon realized that solving this problem would require an understanding of both astronomy and mathematics. On the astronomy side, it meant understanding how the Sun’s position changes throughout the year, as well as concepts like the celestial equator and latitude. The mathematics, however, is more demanding. I hadn’t really used higher mathematics since college, apart from occasional applications in electronics projects. I had to step back and revisit vectors, trigonometry, and linear algebra. I actually enjoyed getting an opportunity to study mathematics again after many years.

The first step to mapping the Sun’s journey through the sky using shadows is identifying the parameters that influence shadows:

Time of day: Shadows are longest in the early morning and late evening, and shortest around solar noon.
Latitude: The farther we are from the equator, the longer shadows tend to be for the same solar position.
Solar declination: The Sun’s apparent position shifts throughout the year, causing seasonal changes in shadow length and direction.
Gnomon length: The height of the object casting the shadow directly affects the shadow’s length.

Our task, then, is to develop equations that relate these parameters and allow us to map the Sun’s apparent journey across the sky using the shadows it produces.

Two Ways to See The Sky

To describe the position of the Sun as a vector, we have to first define how we “look” at space. By looking at space, I mean the coordinate basis vectors that define the space in which the Earth and Sun exist. One way is to describe this space from Earth’s perspective; the other is to describe it from our (local) perspective. We use the “topocentric equatorial coordinate system” for Earth’s perspective as it is anchored to the geometry of the Earth itself. To describe space from our perspective, we use the ENU (East-North-Up) coordinate system. The topocentric equatorial frame essentially tells us how the time of day and solar declination influence the solar vector. The ENU frame tells us how latitude influences this vector.

So, why use two coordinate systems to describe the same vector? In the topocentric equatorial frame, it is easy to describe the motion of the Sun from Earth’s perspective. The Sun’s path is simply governed by Earth’s rotation. However, this frame does not tell us how latitude influences the vector. We experience the sky through the ENU frame, which is tied directly to the ground beneath our feet. Where we stand on Earth determines how the sky appears to move above us.

In the topocentric frame, the axes are:

Z-axis → up along Earth’s axis toward the celestial north pole (parallel to Earth’s axis).
X-axis → lies in the celestial equatorial plane, pointing toward the local meridian.
Y-axis → points East.

In the ENU frame:

E-axis → East (same as the Y-axis in the topocentric frame).
N-axis → points towards the north pole (but it is not parallel to Earth’s axis).
U-axis → points straight up (zenith).

Understanding the Language of Shadows with Mathematics

I used to dislike mathematics when I was in school, and I wasn’t very good at it either. I started enjoying it much later in life, when I could study at my own pace without the pressure of exams. Mathematics can be quite enjoyable if you don’t let school ruin it for you. To understand the Sun’s motion, let’s start with the topocentric equatorial frame and define the Sun’s rotation as seen from Earth’s perspective. This will enable us to describe how the solar vector changes in relation to the Sun’s declination and hour angle. The hour angle is the angle of the sun from our local meridian. Since a day is 24 hours long, the Sun takes 24 hours to cover 360 degrees. So, every hour, it covers 15 degrees. The formula for the hour angle (H) is:

Throughout the day, the Sun appears to rotate about the Z-axis in the topocentric equatorial frame. If you think about it, at noon, the Sun lies on our local meridian, and therefore its Y component (which points towards east) is zero. Let’s model the solar vector at this special time:

The next step is to generalize this for any hour angle (H). In the equatorial frame, changing the hour angle means rotating the vector about the Z-axis. The rotation matrix is:

I had to watch a few linear algebra videos on YouTube to understand rotation matrices and how they transform vectors. That’s beyond the scope of this blog post, though. We can rotate our solar noon with this rotation matrix:

So, now we have an equation for the solar vector from the Earth’s frame of reference, which includes the declination and hour angle. Wonderful! Now, the last piece of the puzzle is to figure out how this vector looks from our local perspective using the ENU frame. This is where latitude finally comes into play.

To translate our solar vector into the ENU frame, we’d have to first define the ENU basis vectors in the equatorial frame. The East vector is the same as the Y-axis in the equatorial frame, so it would remain unchanged. The East and North vectors are essentially the X and Y basis vectors (in the equatorial frame) rotated by our latitude. To visualize this, imagine looking down from the Y-basis vector and rotating the X and Z vectors by the latitude (𝜙):

We can now define the ENU vectors in topocentric equatorial space:

The ENU basis matrix is:

This matrix encodes the observer’s latitude and defines how the global (equatorial) frame is oriented relative to the local one. The next step is to project the solar vector, which we defined in the equatorial frame, into this coordinate system. In other words, we are asking: what does the Sun’s direction look like when expressed in our local frame?

So, now we have a representation of the solar vector that contains everything we need: hour angle, solar declination, and latitude. Excellent! The top row represents the “E” component of the solar vector, the next one represents the “N” component, and the last one represents the “U” component.

Altitude and Azimuth

The solar altitude is the angle between the local horizon and the sun.

The vertical component here also equals the solar “U” component in the ENU frame. So, the altitude is:

The azimuth is the angle between the “E” and “N” components:

This needs to be corrected based on the signs of E and N so that it lies in the correct direction:

These cases are the equivalent of using the atan2 function in languages such as Python.

Once we know the altitude and azimuth, we can determine the length and direction of the shadow using trigonometry. A shadow always points directly opposite the Sun, so its azimuth is simply reversed. The shadow’s direction and length can therefore be expressed as:

To test the equations, I calculated where the shadow of a 4cm gnomon should fall on a piece of paper at a specific time. Seeing the shadow tip land exactly where the math predicted was really exciting! I later wrote a Python script to generate lines for different months. My Python scripts are in this GitHub repository.

One interesting result is that the line corresponding to the March and September equinoxes becomes a perfectly straight line, whereas the lines for the other months produce curves. Why does this happen? There are a few ways to explain this. An intuitive explanation is that the shadow is a two-dimensional projection of a three-dimensional motion (the Sun’s path across the sky). Only under special conditions does that projection reduce to a straight line. Another explanation comes from the mathematics itself, where the equations simplify in a way that removes the curvature entirely. I’ll leave that as an exercise for you to explore!

Dharamshala

November 2, 2025November 4, 2025Ashish Derhgawen Leave a comment

After my trip to Buran Ghati, I went to Dharamshala with my family during Diwali. My wife and I were joined by our parents. We stayed in Dharamkot, a quiet, peaceful village perched above McLeod Ganj. McLeod Ganj is often called “Little Lhasa” because it is the heart of Tibetan culture in India, and home to the Dalai Lama himself. We were incredibly fortunate to have a once-in-a-lifetime opportunity to meet him in person!

I really love the calm, meditative atmosphere of Dharamshala. I would spend my days working from a Tibetan cafe near the monastery. The cafe drew an interesting mix of travellers and monks. I enjoyed listening to the conversations. One afternoon, I noticed a few people playing a board game with black and white stones on a wooden board. There was something about the simplicity of the game that grabbed my attention. The simplicity of stones meeting the infinite complexity of the board was fascinating. I later learned that the game was called Baduk (or Go). The game is deeply rooted in Chinese and Japanese culture, though I had never heard of it before. I’ve been fascinated by the game and have started learning it myself.

One morning, we set out on a short hike to Gallu Falls. Dogs in these mountains are incredibly friendly and helpful. They often accompany travelers and guide them to their destination. They seem to somehow always know where you want to go. I asked a dog to take us to Gallu Falls, and he immediately took the lead. After a few hours of hiking through a beautiful forest, we reached the warerfall. The water cascaded into a small pool that was crystal clear. We couldn’t resist taking a quick dip, even though the water was uncomfortably cold. Our furry friend waited for us, and once we were done, he escorted us back down the trail.

Later that week, we went on the famous Triund trek. It’s about a four-hour climb from the Gallu Devi Temple in Dharamkot. The trail wound through forests with beautiful views. When we reached the top, we were greeted by sweeping views of the majestic Dhauladhar Range. The snowcapped mountains were gleaming in the evening light. We camped overnight beneath a star-studded sky before descending the next morning.

Buran Ghati

November 2, 2025November 3, 2025Ashish Derhgawen Leave a comment

I went to the Buran Ghati trekking route in September. My friend Anish and I traveled from Bangalore to Shimla and met the rest of our gang there – Omkar, Shashank, Nilesh, Yash, and Udit. We spent a day in Shimla, exploring the town and relaxing. The next day, we left for Buran Ghati. The Buran Ghati trek begins in the quaint little village of Janglikh, near Rohru, and passes through lush forests, meadows, and glacial lakes before crossing the dramatic Buran Pass at an altitude of about 15,000 ft. The trek ends in the town of Barua and covers a total distance of approximately 50 km.

The drive from Shimla to Janglikh is an adventure in itself. Hours of twisty roads that turn endlessly for hours and really test your stomach. Our drive came to a halt a few kilometers before Janglikh, where a bridge across the Pabbar River had collapsed. With no way for vehicles to cross, we shouldered our packs, crossed the river on an alternate footbridge, and hiked the final stretch to reach the village. Janglikh lies at an altitude of 9,200 ft

After spending a night at the TTH base camp in Janglikh, we made our way to the next campsite, Dayara Thach. The trail from Janglikh was beautiful and easy to follow. We passed through beautiful settlements that had houses made of wood and rocks. In the morning, women from Janglikh were taking their animals out to graze. We had cellphone service for a few hours into the trek before we lost it for good. The trail gradually climbs into a dense forest of oak and pine. A few hours into the walk, we went above the treeline and into a vast expanse of beautiful meadows. The Dayara campsite was about 8.5 km from Janglikh at an altitude of 11,100 ft. We spent the rest of the day socializing with the other trekkers, exploring areas near the campsite, and listening to paranormal mountain stories from other trekkers.

The next day, we hiked from Dayara to Litham, ascending to an altitude of about 11,500 ft and covering roughly 6.5 km. The Litham campsite was beautifully set beside the Pabbar River. Its icy-cold, clear water originates from the Chandranahan glacier to the north of Litham. To the north of the campsite, I could see a waterfall near Chandranahan in the distance. We spent the afternoon unwinding by the river and playing a game of cricket with a bat fashioned from a tree branch! The big rocks near the river were perfect for relaxing and listening to the stream.

The next day, I woke up to the sound of an avalanche on the mountain peaks to the south of our campsite. It sounded like thunder. We had another day to spend at Litham and explore Chandranahan. We hiked about 5 km north of Litham to the origin of the river, the Chandranahan lake. I’ve heard that there are seven lakes that are part of Chandranahan. We hiked up to the fourth one, at an altitude of 13,000+ ft. However, the weather began to change, and we decided to descend back to Litham. We covered a total of about 10 km. The extra day at Litham gave us some more time to acclimatize.

Rhizocarpon geographicum (map lichen) grows abundantly on rocks throughout the valley, thriving in clean, unpolluted mountain air.

The next day, we hiked from Litham to the Dhunda campsite, ascending to about 13,500 ft and covering 8 km. We crossed a stream on the way. Fortunately, the water was low enough for us to hop across on scattered rocks. The final stretch to Dhunda was a steep and demanding ascent. The campsite was chilly, and next to a water stream that seemed to emerge from the earth. We went to bed early that day because we had to get up at 2 AM to begin our ascent to Buran Pass!

Getting a good night’s sleep at high altitude is never easy – at least not for me! I consider managing anything above 5 hours a luxury! We woke up early and left the campsite around 3 AM. The grass was slick with dew, and I slipped more than once. The climb to the 15,000 ft pass was steep and relentless. In the dark sky, we could see countless shooting stars and satellites. By the time the sun began to rise, we were at the top. It was an exhilarating feeling to finally see across the pass towards Kinnaur Valley. We spent some time clicking pictures. To our delight, we were getting cell reception. I called home to let my family know that I was alive and well!

At Buran Pass (15,000 ft). Helmets were required due to the risk of rockfall on parts of the descent.

The descent from there was very steep and rocky. After a few hours of hiking, we stopped next to a stream to have lunch and take a nap. We reached the campsite sometime in the afternoon after about 12 km of walking. Unlike the previous campsites, which lay above treeline, this one was amid dense greenery at an altitude of around 10,000 ft. A river flowed nearby, and its sound gave a soothing backdrop.

The next day, we began our descent towards the village of Barua. The trail wound through dense forests, and I really enjoyed the walk. The hill people who live in these mountains are simple and caring. Their lives move at a different rhythm from city dwellers. As we passed their settlements, they greeted us and gave us handfuls of apples and peaches. We reached Barua after about 8 km of walking. From Barua, we had a long drive back to Shimla.

After the trek, I returned to Bangalore for a few days before returning to Himachal again. This time, to unwind in Dharamshala. I’ll save that for my next post.

The SolderSmoke Direct-Conversion Receiver

June 20, 2025June 20, 2025Ashish Derhgawen 1 Comment

I have been on the on the lookout for a receiver to pair with my Nandi-40 transmitter. When I came across the SolderSmoke Direct-Conversion Receiver challenge, it seemed perfect. It checked all the right boxes: 40m band, fully analog, built entirely with discrete components, no ICs in sight. This project was originally intended for high-school students to build and learn about electronics and radio. This simple receiver has gained a following in the ham radio homebrewing community.

While browsing the project’s Discord channel, I “liked” one of the posts and Bill Meara (N2CQR) noticed it. If you’re not already familiar with Bill, he runs the popular SolderSmoke podcast and website. He emailed me and said I should also build the receiver! How could I say no?

Bill and I have known each other for a while. I first discovered his work through his book, Global Adventures in Wireless Electronics. If you appreciate the magic of radio and electronics, I would strongly recommend reading his book. The book is a memoir of Bill’s radio adventures, and it captures the thrill of the hobby. Bill has written about some of my projects on the SolderSmoke website.

The receiver consists of four sections: a band-pass filter, an oscillator, a mixer, and an audio amp. Usually, folks build each module on a separate board. I like this approach to building because it is perfect for experimentation, even though it may not be the most compact. You can test and modify individual sections easily.

I started off by building the mixer, the heart of this receiver. It is a double-balanced diode ring mixer. Alan Wolke (W2AEW) has some excellent videos on this topic. To test the mixer, I fed signals from a signal generator into the RF and LO (Local Oscillator) ports and terminated the output with a 50-ohm load.

To verify that mixing was happening, I tuned a nearby shortwave radio to the difference frequency of the two input signals. When both signals were applied, the radio went quiet—proof that the mixer was generating a signal at the expected intermediate frequency. I also used the math function on my oscilloscope to view the mixing products and calculate the conversion loss. It was fun to see theory meet practice.

Next, I added the audio diplexer circuit which extracts the audio signal and terminates the unwanted higher-frequency components. Before moving on to the other modules, I wanted to verify that the mixer could actually receive a signal. I hooked up a signal generator to act as the VFO and borrowed an audio amplifier from another project for testing. And sure enough, it worked! I could hear signals coming through, which was a great moment of validation.

However, it quickly became clear that a band-pass filter was essential. Without it, the receiver was swamped with strong AM broadcast stations bleeding in from all over the band. Front-end filtering was necessary for selectivity.

I built the band-pass filter next. I didn’t have NP0 capacitors, so I used regular ceramic capacitors that I had in my junk box along with some trimmer capacitors. With a NanoVNA, I measured the insertion loss. It was on the high side at around -2.8 dB. Not ideal. However, I decided to continue on and revisit the band-pass filter later when I had better capacitors.

The audio amp is a straightforward 3-stage amplifier. The only part I didn’t have was the output transformer, which I was able to order online. The amp had a tendency to oscillate, but it works better with a stiffer 9V supply using two batteries in parallel.

Next, I built the VFO, which is a simple Colpitts oscillator with a JFET buffer stage. What’s interesting is that the oscillator uses an inductor for tuning. It is based on the design by Ashhar Farhan (VU2ESE) from his Daylight Again radio. The tuning coil is wound on a 3D-printed former. The original designed called for stable mica capacitors in the oscillator, but I couldn’t find them anywhere in Bangalore. I spent an entire afternoon navigating the crowded lanes of SP Road in search of these elusive capacitor with no luck. So, I built it with regular ceramic capacitors and hoped for the best. Needless to say, the oscillator drifted all over the place.

Eventually, I ordered surface-mount NP0 capacitors online. They were easily available, much cheaper than mica capacitors, but a pain to solder. After a few fumbled attempts, I somehow managed to solder them in place. The oscillator is now rock stable and hardly drifts at all.

I also swapped out the fixed capacitors in the band-pass filter with surface-mount NP0 capacitors and the insertion loss dropped down to about -0.4 dB. These capacitors were definitely worth the soldering hassle.

In the end, I hooked all the modules together and the receiver worked perfectly.

My first modification was an essential one: adding a power switch and a status LED. After that, I added a connector for an external speaker. It sounds amazing with an enclosed speaker. The connector auto-disconnects the internal speaker when an external one is plugged in, eliminating the need for a separate speaker selector switch.

Here is Bill’s writeup on my project. It is officially in the SolderSmoke Hall of Fame.

Here are some videos of the receiver in action:

SSB:

CW:

Sloshing Electrons and Crossing Oceans

April 20, 2025April 21, 2025Ashish Derhgawen 1 Comment

On December 12, 1901, Guglielmo Marconi sent the first radio transmission across the Atlantic Ocean. This historic moment marked the dawn of a new era of wireless communication. The transmitter he used was designed by Sir John Ambrose Fleming. This demonstration proved that radio waves could travel beyond the line of sight, bending with the curvature of the Earth. Imagine how Fleming must have felt, knowing the transmitter he built had accomplished this remarkable feat. Even after more than 120 years, it still seems awe-inspiring. What better way to relive that excitement than by building your own transmitter and sending a signal across the vast open ocean?

Designing a Transmitter

I began working on a CW (Continuous Wave) transmitter inspired by the book “Crystal Sets to Sideband” written by my dear friend Frank Harris (K0IYE). In general, CW transmitters are simple to build and understand. You slosh electrons back and forth at just the right frequency, and you will generate a form of invisible light that can travel thousands of kilometers. I’m referring to electromagnetic waves, of course! CW doesn’t transmit voice; it is just an unmodulated carrier wave. Communication happens through Morse code.

Oil lamps for an authentic early 1900s feel — Oil lamps for an early 20th-century vibe

I built the oscillator and buffer amp exactly as described in Frank’s book. The oscillator is a Butler-type oscillator, which is a rather unusual choice. Most circuits use a Hartley or Colpitts oscillator, but Frank mentioned that the Butler oscillator has little start-up drift.

The buffer amp bumps the output of the oscillator to about 50 mW and adds a layer of separation between the oscillator and the subsequent stages. This helps keep the load on the oscillator as light as possible.

I didn’t have the same transistor Frank was using in the final PA (Power Amp), so I took it as an opportunity to design the rest of the circuit on my own. I decided to use the ubiquitous IRF510 MOSFET, which I had in abundance. After experimenting with the IRF510, I realized I would need at least 200-300 mW of drive to generate 5W+ of RF power, which is what I was aiming for. I realized my puny buffer amp didn’t have the juice to do it.

But aren’t MOSFETs voltage-driven devices with sky-high input impedance? Yes, that’s true when you look at them from a DC perspective. But the high impedance goes out the window once you start dealing with high-frequency AC signals. The gate is essentially like a capacitor. Depending on the frequency (and other factors which I’ll discuss), its impedance changes. So, I had to design another “driver” amp stage to drive the IRF510.

I experimented with several types of driver amp circuits. Some of the ones that I tried to design on my own didn’t work so well. I finally settled on a design I found in the book “Experimental Methods in RF Design” by Wes Hayward et al. It produces about 300-400 mW with an efficiency of about 70% (typical for Class C). This is more than enough to satisfy the IRF510.

What makes the IRF510 interesting is that it was made by International Rectifier in the 1970s for the automotive industry. It was used to replace clunky electromechanical relays being used in blinkers and dimmers. Who would’ve guessed that decades later, it would become a favorite among amateur radio builders, operating at frequencies far beyond what it was originally designed for!

The Full Circuit

The Gate Capacitance Paradox

Connecting the driver stage to the power amp was a challenge. In RF circuits, matching one stage to another often involves some sort of matching network. I’ve talked about this in my blog post about impedance matching and archery.

I tried to be methodical about designing the matching network. I started off by determining the output impedance of the driver stage by loading it with different resistors and recording the voltages. Here is the formula:

This formula is useful when driving the amp without a load is undesired. However, if the open circuit voltage can be measured, this formula could be used:

These equations are not magical; they could easily be derived using Ohm’s law and the basic voltage divider equation. I determined the input impedance of the IRF510 (about 220 ohms) by using a signal generator and measuring the voltage across the gate with an oscilloscope. The output impedance of the signal generator and the input impedance of the IRF510 form a voltage divider. The voltage divider formula can be used to determine the unknown input impedance.

I discovered that the input impedance of a MOSFET isn’t fixed. It is dynamic and changes with the drive voltage. If the drive voltage is below the IRF510’s gate threshold voltage V_gs(th), its impedance appears quite high. Once the voltage goes above the threshold, it starts dropping. After some speculation, I came to the conclusion that this is the result of the Miller effect. I’ve encountered this phenomenon in another experiment in the past.

“In electronics, the Miller effect accounts for the increase in the equivalent input capacitance of an inverting voltage amplifier due to amplification of the effect of capacitance between the input and output terminals.“

So, when the capacitance increases, the capacitive reactance decreases, since they are inversely related. This reduces the effective impedance of the gate. Well, I guess that explains the mystery.

I initially tried designing an LC L-match to match the output of the driver to the input of the MOSFET. But despite careful calculations, the PA didn’t perform well. I also experimented with transformer coupling using a 10:3 turns ratio, but that approach didn’t yield good results either.

Eventually, I settled on an L-match configuration with the shunt element placed on the driver side, which has lower impedance than the MOSFET’s input. This goes against convention, as the shunt element is typically placed on the higher impedance side. Yet, to my surprise, this unconventional setup worked significantly better. I think this is because this LC network forms a low-pass filter, which removes frequencies above a certain cut-off frequency. In my case, the values I’ve used would have a cut-off near 7 MHz. However, this filter could be tweaked to improve performance. It seems these values are not too critical. If you reduce the cut-off frequency, you would have a lower voltage at the MOSFET’s gate.

I’m not aiming for maximum power transfer with this matching network. Instead, I’m using a low-pass filter to ensure sufficient voltage swing—around 10V at the MOSFET gate—to drive it into saturation without exceeding its ±20V maximum gate rating.

The Gate Resistor’s Little Secret

I have a 470-ohm resistor connected to the gate of the MOSFET to provide a ground reference. I initially used a large 1 megaohm resistor to reduce resistive loading on the driver stage, but the transmitter performed poorly.

After some experimentation, I came to suspect that this resistor plays another important, and less obvious role: discharging the gate-to-source capacitance (Cgs) between RF pulses. This helps ensure the MOSFET remains off during the portions of the cycle outside its conduction angle. At 7 MHz, the period is just 142 nanoseconds. If the resistor is too large, Cgs doesn’t fully discharge between pulses. After experimentation, I settled on a 470-ohm resistor. A 1K resistor also worked well, but going much higher started to degrade the results.

Transmit, Receive, and Keying

The transmitter has a transmit/receive (Tx/Rx) switch. While it doesn’t have a built-in receiver, an external receiver can be connected and used alongside it. For Tx/Rx switching, I use a small DPDT relay. In receive mode, the relay completely cuts power to the transmitter to avoid interference.

I also added a “spot” button that can be pressed while in receive mode. This powers up just the oscillator, allowing me to hear its tone on the receiver. This lets me check if I’m in tune with another operator’s CW signal, kind of like tuning a guitar to match pitch.

In transmit mode, the oscillator runs continuously. Pressing the key activates the buffer amplifier stage using a PNP transistor. Once enabled, the buffer amp passes the oscillator signal to the subsequent stages. This approach ensures that the small PNP transistor isn’t burdened with the high current demands of the driver and power amplifier. Instead, it simply triggers the buffer amp.

I also added two LEDs for visual feedback. One LED indicates when the transmitter is in Tx mode. The second LED lights up when the transmitter is keyed, specifically, when an RF signal is being fed into the antenna. It helps verify that everything is working as expected. If this LED doesn’t light up, it means no RF is reaching the antenna, and something’s wrong. To minimize the current draw from the RF path, I used a transistor to drive this LED.

Cleaning the Output

Unlike Marconi, we can’t use a noisy spark transmitter on the air. The output from the PA was noisy and full of harmonics. I used a low-pass filter (LPF) to clean it up.

I tried designing a 5-element filter on my own. While it worked, its filtering wasn’t as strong as I would have liked. I did learn about filter design and Smith charts in the process.

In the end, I settled on a 7-element 40m LPF kit from QRPLabs that was cheaper than buying the components individually. It worked really well.

The Finished Transmitter

I named the finished transmitter “Nandi-40” because I live near Nandi Hills, and “40” refers to the 40m wavelength of the radio waves it produces. With a fully charged 12 V battery, it puts out about 10-12 watts of RF energy. With a 9V supply, it delivers about 4 watts. It definitely exceeded my design goals. Now it is time to see if it can cross the ocean like Fleming’s transmitter!

Transoceanic Transmission

I waited for good propagation on the 40-meter band. As the sun dipped below the horizon and conditions aligned, I tuned in to a web-SDR receiver in Mt. Barker, Australia. These Australian receivers have a very low noise floor. If the Nandi-40’s automotive MOSFET was indeed stirring the ether, there was a good chance I might hear it.

I keyed the letter “S” in Morse code, three simple dots, just like Marconi did. And to my amazement, those three little blips rose above the noise floor in Australia, nearly 6,700 kilometers away!

Sure, long-distance communication is nothing new in the world of amateur radio. But there’s something uniquely satisfying and almost magical about doing it with a transmitter you built yourself.

Photoelectric Effect and Neon Lamps

August 26, 2024Ashish Derhgawen Leave a comment

Neon lamps can do a lot more than just emit light. They can be used to build oscillators, logic gates, tone generators, and even sense light. Check out my YouTube video which explores the photoelectric effect using neon lamps.

Faraday’s Green Flame

July 20, 2024Ashish Derhgawen Leave a comment

I decided to step outside my usual comfort zone of electronics and explore the fascinating world of chemistry. My interest sparked after reading Michael Faraday’s book, The Chemical History of a Candle. One of the experiments in the book describes Faraday creating a green flame. A green flame seemed fascinating. I decided to replicate the experiment by synthesizing copper chloride. You can watch my video to see how it turned out!

I created my own spectroscope using a broken CD and a cardboard tube to study the spectral patterns of different light sources.

The Quantum Tunneling Transmitter

February 24, 2024July 20, 2024Ashish Derhgawen 3 Comments

This has to be one of the most bizarre projects I have worked on. It is a transmitter that works without any transistors or tubes. It utilizes a strange phenomenon known as quantum tunneling. Quantum tunneling is a phenomenon wherein a particle can disappear from one side of a barrier and reappear on the other side even if it doesn’t have sufficient energy to surmount the barrier. It seems as if the particle “tunnels through” the barrier, hence the name. Quantum tunneling is a consequence of the wave nature of matter. It is nothing less than magic.

The first semiconductor devices to utilize this phenomenon were invented in the 1950s and were called tunnel diodes. These devices exhibit negative resistance, which means that the current through the semiconductor becomes inversely proportional to the voltage across it. This only happens at certain voltages. Below and above that negative resistance region, the tunnel diode exhibits a normal current-voltage relationship.

Current-voltage characteristics of tunnel diode vs conventional diode (from: RCAs Tunnel Diode Manual)

This negative resistance anomaly allows tunnel diodes to be used in very high-frequency oscillators. Tunnel diodes are an exotic breed of semiconductor devices. Commercially made ones are rare and expensive. However, recently, a genius inventor by the name of Nyle Steiner discovered an easily made substitute. He found out that he could make tunnel diodes at home by heating galvanized sheet metal!

Anyone with a propane torch and a few scraps of galvanized sheet metal laying around can easily make a negative resistance device. With this device, it is possible to make very simple RF oscillators, audio oscillators and even amplifier circuits. It is almost like making your own transistor.
Nyle Steiner

Mr. Steiner has some of the most interesting experiments I have seen, so do check out his website – http://www.sparkbangbuzz.com/.

I recreated the experiment by heating a zinc metal plate with a torch. The back side of the zinc plate (opposite from the flame) had several active spots. These black spots are where you are most likely to find negative resistance. I used a “cat whisker” style detector similar to the ones used in crystal radios to probe the zinc plate. For the “whisker”, I used #30 AWG copper wire.

I’ve used a 4 MHz crystal to set the frequency of the oscillator. The 10K pot is used to adjust the biasing. The 1K resistor was initially added to reduce short-circuit current when the pot is at its lowest setting. This resistor may no longer be necessary after I added a 1K resistor in the keying circuit (which consists of a CW straight-key, 1K resistor, and an electrolytic capacitor). The keying circuit was a later addition, so I forgot to remove the 1K from the biasing circuit. The resistor and electrolytic capacitor on the keying circuit serve an important purpose. They smooth out the keying. The charge on the capacitor prevents the voltage on the cat-whisker from changing abruptly. For some reason, the homemade tunnel diode doesn’t like abrupt keying. Without the smoothing circuit, there are many cat-whisker settings that stop working after opening and closing the key. These oscillators are quite fussy.

Adjusting the cat-whisker is time-consuming task. Be prepared to spend several minutes trying to find good spots. The easiest way to know if the circuit is oscillating is by using a radio or an oscilloscope. I tuned my Icom transceiver to the frequency of the crystal (4 MHz) and kept it nearby in CW mode. You will hear clicks and beeps as you slide the cat-whisker. If you don’t, adjust the biasing. Finding a stable spot can take a while, but once you’ve found it, it can last for a long time if you don’t disturb it. In my experiments, I was able to use a spot for an entire day. It would have lasted longer if I hadn’t disturbed it.

Output of negative resistance oscillator

Here is the “quantum tunneling transmitter” in action:

I will try injecting an AM audio signal into the oscillator. I’m sure that would also work. If only the professor on Gilligan’s Island knew about it!

WSPR-ing Around the World

January 13, 2024July 20, 2024Ashish Derhgawen Leave a comment

I feel that one of the charms of amateur radio is its unpredictability. It feels like throwing a message in a bottle into the vast ocean, not knowing where the currents would take it, or who would read the message. Radio waves often take unpredictable paths when traveling. To a large extent, how these waves propagate is determined by the state of our planet’s ionosphere. The ionosphere consists of layers of charged particles that affect how RF signals travel. These layers move and shift and undergo cycles of strengthening and weakening, all under the influence of the Sun. Radio waves can bounce off the ionosphere, and essentially “skip” around the Earth.

To understand how waves propagate in different bands a protocol known as WSPR (pronounced “whisper”) was developed in 2008 by Joe Taylor (K1JT). It is an acronym for “Weak Signal Propagation Reporter”. It can tell us what is possible with low-power transmissions and see which radio bands have a path to which points on the globe.

A WSPR transmission conveys the sender’s call sign, station location, and power level using a compressed data format with strong forward error correction (FEC). The message is modulated using frequency-shift keying (FSK) at a very low bit rate. Sending a single WSPR message takes almost two full minutes! The WSPR protocol is effective at signal-to-noise ratios (SNR) as low as -28 dB in a 2500 Hz bandwidth, some 10 to 15 dB below the threshold of audibility.

My QCX transceiver contains an inbuilt WSPR mode. I set it up and sent a single WSPR message on the 40m band with about 4-5 watts of power. It took almost 2 minutes to send, and I was worried about heating the BS170 transistors in the power amplifier since I had never really stress-tested them in this manner. To my relief, no magic smoke was released. I monitored the transmission on a local WebSDR to make sure it was sending a decipherable message. After it was done sending, I checked the WSPRNet website to see if any stations received my feeble signal. I wasn’t expecting anything spectacular, but to my surprise, the signal had traveled much farther than I imagined!

I had reached Antarctica, Hawaii, New Zealand, Australia, the Canary Islands, and Norway (to name a few)! Well, that proves that the antenna I built is working. There are lots of tools to analyze the WSPR data. I liked the analysis tools available on WSPR Rocks. For example, you can view a SNR vs distance chart. It was interesting to see that some distant stations copied my signal better than stations which were nearby. On the website you can see the names of the stations and other details.

I also pulled all the data into a Google Sheet for analysis.

I reached 26 locations with a single WSPR transmission. Incredible! The results have encouraged me to try building a dedicated WSPR beacon using the Raspberry Pi for use on other bands. Stay tuned!

The Not-so-Random “Random Wire” Antenna

January 6, 2024July 20, 2024Ashish Derhgawen 1 Comment

I live in an apartment where installing an antenna for HF use is a challenge. After evaluating various antenna options, I chose to install the so-called ‘random wire’ antenna, stretching it between my balcony and my aunt’s next door. Random wire antennas aren’t very random at all! Certain lengths perform better than others. I decided to go with a 58-ft wire based on the recommendations on this website.

A random wire antenna has an unpredictable impedance that varies with frequency. Moreover, the impedance is usually so high that most antenna tuners need additional help from a 9:1 unun transformer to bring the impedance down to a workable range. The 9:1 unun is an autotransformer with a 3:1 turns ratio, which results in a 9:1 impedance transformation.

I had an untested unun that I built a long time ago. I connected it to the antenna and used 25-ft of coax to connect it to my Emtech ZM-2 antenna tuner. Despite all my efforts to tune it on the 40m band, I failed to bring the SWR to an acceptable range.

I tried adjusting the antenna length, but it didn’t make much of a difference. The SWR was too high when I checked with the NanoVNA. What was wrong in my setup? It was time to sit on my armchair, put on my detective hat, and light a cigar.

My hunch was that the untested 9:1 unun was the culprit. To confirm this, I removed the antenna wire from the unun and replaced it with a 470-ohm resistor on the output. If the transformer was doing what it was supposed to, I expected an impedance of approximately 50 ohms (470/9) on the output.

I used the NanoVNA to plot the transformer’s frequency response. The frequency plot revealed the problem. The 9:1 transformation was happening around 30 Mhz. At 7 MHz (40m band), it was nowhere close to 9:1. No wonder the tuner was struggling!

The 9:1 unun’s terrible frequency response

In the plot, the yellow trace is the impedance. In an ideal world, a transformer would exhibit consistent performance across all frequencies. However, in reality, a transformer’s bandwidth is influenced by its inductance and various parasitic elements. The VNA trace shows that the impedance increases between 1-30 MHz, where it is approximately 50 ohms. The 9:1 transformation occurs near 30 MHz, signaling that the low cut-off frequency is higher than optimal—a clear indicator of insufficient inductance! Time to light another cigar.

I wasn’t sure what toroid core I was using in the unun. I decided to replace it with an FT50-43 toroid that I had in my junk box. Mix #43 toroids can be used for wideband transformers between 3-60 MHz according to this website. The FT50-43 is a small toroid, so it was a bit difficult to fit 3 wires side by side. I used a twisted trifilar winding to save space and get about 8-9 turns on the core. The inductance increased from about 3 uH (old core) to 230 uH (new core). On the NanoVNA, the impedance response looked much more uniform across the HF band.

Uniform frequency response with the FT50-43 toroid

I connected the random wire antenna to the unun and measured the SWR across the HF band with the VNA. The impedance varies with frequency, with dips at specific frequencies. As you can see, there is a dip in the 40m band that allows the ZM-2 to easily tune the antenna.

I connected some of my QRP transmitters to the antenna, and was able to hear them on Bangalore’s webSDR!

There are still some unresolved mysteries to explore. Do I need a separate counterpoise? Would that make a difference? I haven’t observed any noticeable improvements from adding one. I believe the coax shield is the counterpoise in my setup. Does the position or orientation of the 25-ft coax cable make any difference? It does seem to affect the antenna’s impedance, so I think it does. How does the angle of my antenna affect propagation? So much to explore, so little time!