The personal website of Scott W Harden

First Homebrew QSO Ever!

Today is a very special day, as it’s the day I first made a contact with a radio transmitter I built completely on my own! The plans were copied from no where (although the concepts were obviously learned elsewhere), so it’s somewhat of a unique design (likely because it’s not very good!). I’ll be the first to admit there is MUCH room for improvement, but my goal was to design and build a multi-band transmitter which would produce RF (not necessarily efficiently) at multiple bands by dropping in crystals of different frequencies.

My first QSO was with Bob, KC8MFF in West Virginia at 5pm today on 7MHz. He heard me calling CQ and replied! He gave me a 559 which made my happy. I was sending about 8 watts at the time into a Mosley Pro 67 Yagi at 180FT and receiving from a 40m dipole at 150FT at the W4DFU Gator Amateur Radio Club station in Gainesville, FL. Although he’s was about 650 miles away, I hope to make a more significant contact as the band opens up later tonight. It’s such an exciting feeling! The aluminum plate gets very hot (even with the fan) and there’s a slight smell of smoke whenever I transmit, but it adds to the fun I guess! Here’s some information about the build, though I’m confident it’s less than optimal.

I’ll preface this by stating that my goal was to produce an experimental platform which I could use to investigate construction techniques of small moderate-power transmitters. This is by no means a finished product! Much work (and some math) must be done to calculate the best number of turns on each coil for each band, including the RF choke on the power (resulting in class C amplifier behavior), the RF transformer, and the inductor/capacitor values of the low pass filter - all of which were determined empirically (watching output on an oscilloscope while adding/removing turns on a toroid). At 10W, it’s not QRP, but it’s easy to tone down to QRP (5W levels).

One of my desires was to create a transmitter which could be built at minimal cost (total value of this is probably about $10). The microcontroller (ATTiny2313) was what I had on hand ($2), the buffer chip acts as a small amplifier ($0.50), and the power amplifiers are IRF510 MOSFETs ($1). The rest of the components are junkbox, and their values aren’t really significant! The power supply is a 19V 3.6A power supply from an old laptop - small, convenient, awesome! Hopefully with some tweaking I’ll have a nice transmitter which I’m proud to share and have replicated…

The overall schematic represents a crystal clocking a microcontroller at the transmit frequency, where the CKOUT fuse has been set, producing 5PPV square waves. These trigger an inverting buffer which (a) amplifies the current of the signal and (b) provides an easy source of inverted signal. The two (inverse) signals then fire a pair of IRF510s in tandem, each acting as a Class C amplifier producing about 60PPV waves (not quite as square-ish). The output is low-pass filtered with a Pi filter (3 pole Chebyschev), then sent to an antenna. Nothing special has been done to match the output to the antenna, so SWR with a 50ohm load is currently a bit high, but I imagine a variable capacitor on the output LPF would give me something to adjust to improve this. I should probably go back to square 1 and re-do the math from start to finish and follow my impedance values more closely.

Future work will be invested into adding an iambic keyer property to the microcontroller, as well as a button to send CQ at various speeds. It may be interesting to clock this from a Si570 digital synthesizer, allowing me to transmit on any frequency and no longer be crystal-bound. Additionally, using the same oscillator source to power a direct conversion receiver would yield obvious benefit, allowing transmit/receive from a home-brew device at minimal cost. Currently, I’m locked into using a commercial rig as a receiver. We’ll see how it goes…

Anyhow, that’s that. I wanted to document this because I know I’ll look back in the future and laugh at how poorly designed this project is. I’m just amazed it works, and for now this represents a gigantic step step in my learning and growth as an engineer. As poorly designed as it may be, it’s something I’m very, very proud of!

Great inspiration has come from Wes Hayward’s Experimental Methods in RF Design text. I’ve been checking it out from the library every few weeks (Interlibrary Loan, from Vanderbilt University to the University of Florida) but I finally got my own copy for Christmas. It’s such a great resource! The IRF510 push-pull idea came from figure 2.101.

PS: The image below is of a MOSFET I exploded in the development process. Too much current… oops!

40m Junkbox QRP Transmitter

I decided to sit down and build something last night, and I’m surprised by how functional it is! Nothing about it is extraordinarily complex, and it’s extremely flexible, accommodating almost any crystal you want to drop in. Although I doubt I’ll use this exact design for a permanent transmitter, it was fun to build and I’ll post photos hoping to inspire others to tinker with RF circuitry as well! The final device worked on 7.000MHz and had 3 components: power supply, oscillator/amplifier (making 20mW), and amplifier (making 1.5W).

First, I needed an oscillator. I had an easy source of one because I had a pile of ATTiny25 microcontrollers. Often I run a microcontroller at my transmit frequency with a crystal (applied to XTAL1 and XTAL2 pins) and collect the convenient 5V square wave on the CKOUT pin (after the appropriate fuse setting is applied). However, although the ATTiny25 has both XTAL and CKOUT pins, they overlap! This means that CKOUT cannot be obtained when using a crystal. This complicates things slightly…

I ended-up getting a nice sine wave from the XTAL1 pin, although it was less than 1PPV. I tried having this signal directly switch an N-channel MOSFET as an amplifier, but it didn’t work that well (a transformer might help increase PPV, but that complicates things). I instead used a 74HC240 (8 inverting buffers on one chip) to help boost the signal. However, 1PPV wasn’t enough to get the buffer oscillating. I therefore added a 2 resisters and a capacitor to the first inverting output, such that a persistent low would slowly raise the voltage of a wire, and I attached that wire to the input of the buffer chip. This way, although 1ppv wasn’t enough to start oscillations, a few milliseconds of time allowed the inverting output (high when the input is low) to raise voltage of the input until it was enough to fire the buffers. Once it starts, it starts! I’m trilled, because a voltage divider or a potentiometer would have been a pain, and required specific parts.

The result is about 20mW of power with no tuned circuit! This means it will work on pretty much any crystal you can pop in the micro-controller. This may be suitable for a QRSS transmitter, and since we’re not pushing any of the components very little heat is produced, should it should be thermostable and easy to regulate. Modulation is achieved by a reverse-biased LED varactor diode varying crystal capacitance to ground, discussed elsewhere on my site so I won’t go there again.

Power supply is one I built a while back and had available. 5V for the microcontroller, and 12V for the amplifier. Simple!

Amplifying the signal was pretty easy as well. The 5V signal output of the buffer goes from 0V to 5V, which was enough to trigger an IRF510 N-channel MOSFET with a convenient packaging that I screwed into a huge heatsink. I push the MOSFET a lot, and a lot of heat is produced, but as long as I keep it separate from the oscillator the heat shouldn’t affect frequency too much. Although on my workbench I use exposed wires connecting components, this is prone to getting RFI so obviously use shielded cable of some sort, or use extremely short leads. The MOSFET is arranged as a class C amplifier, with a RFC inductor at the drain.

In retrospect I’m doubting that 5V is enough to fully activate the IRF510. I should probably use some method to bring voltage just below firing threshold, so the 5V can more fully open the gate. I’ll try that later! The output is filtered with a PI lowpass filter. I use two 1nF capacitors and a coil which I wind until the output on the scope looks acceptable. I know there are more exacting ways. Anyhow, I had fun, so I thought I’d post. I’m just tinkering at this point!

It’s putting out about a watt and a half into 50 ohms. How cool? Adding a code key is trivial, as the 74hc240 has “gate enable” pins for easy on/off control - even from a microcontroller! Food for thought… 73!

UPDATE - I decided to slap a 10.140MHz (QRSS window) crystal in there and see what happened. I saw my signal locally (AJ4VD/W4DFU grabber), but not elsewhere, so I left it up for about a day. Vince Adams, N9VN spotted it in IL (about 1,000 miles away) and made a post on a mailing list asking who it was. Awesome! Note that for QRSS I used a lower-current power supply, so I don’t actually know what power output was, but I’d estimate it to be about 500mW.

(It’s the “V-shape” at the bottom)

Full-Auto Rapidfire Mouse Modification

I did this purely for the fun of it, and am aware there are many ways to accomplish the same thing. I was playing Counter Strike Source (you should buy it and play with me, name “swharden”) and my fingers are really cold from the winter weather, and wondered if I could have a button help with the rapid firing of pistols. I mentioned it on the microphone, and one of the players ("{Ẋpli¢it} shadow") said I should go for it. Because it was a fun little project, I documented it so I could share it. Check out the cool photos and video!

There’s a summary of the project in video form. Some details of the project are below…

__Here you can see the original circuit board in the mouse. __The microchip on the bottom right of the image seems to do the data processing, so I investigated it a bit and found the pin that the left-click button goes to.

Here’s the underside. It helped me identify good locations to grab +5V and GND solder points.

This is the microcontroller I decided to use for the project. It’s an ATTiny25, $1.33 USD (10+ quantity from Mouser), and has a built in 8MHz oscillator (which can run at 1MHz thanks to the DIV/8 clock prescaler.

I slap the chip in the homebrew development board (a glorified AVR-ISP mkII) and it’s ready for programming. Code and schematics are at the bottom.

After programming, I glued the microchip upside-down in the mouse case and soldered wires directly to the pins. I used small (about 28AWG) magnet wire because it’s a lot easier than stripping wires. Just heat the tip with a soldering iron, the coating melts away, and you can stick it wherever you need to with a dab of solder. Not too many people use this method, but I recommend you try it at least once! It can be very useful at times, and is about as cheap as you can get. (eBay has good prices)

BIG PROBLEM! It didn’t work AT ALL. Why? Didn’t know… I checked the o-scope and saw everything seemed to be working fine. It turns out that 50 clicks per second was too fast to register, and when I reduced the speed to 25 clicks per second it worked fine. Unfortunately I had to add extra wires to allow myself to program the chip while it was in the mouse - a major pain that complicated the project more than I wished!

__Here’s a good view of the transistor. __ Simply put, when the microcontroller sends power to the “base” pin of the 2n2222 transistor, the “collector” is drained through the “emitter”, and the transistor acts like a switch. It’s shorting the pin, just like would happen if you physically pressed the left click mouse button. When the mouse microchip is positive (+5V), it’s “no click”, but when it goes to ground (shorted by the click button), a click is detected. I biased the “base” pin toward ground by connecting it to GND through a high value resistor. This makes sure it doesn’t accidentally fire when it’s not supposed to.

Here you can clearly see the programmer pins I added. This lets me quickly access the chip and reprogram it if I decide to add/modify functionality.

__When it’s all said and done, it’s surprisingly slick and functional. __ I’m using it right now to write my blog, and the button isn’t really in the way. I think it’s one of those el-cheapo buttons you get in a pack of 10 from RadioShack, but I would highly recommend eBay as RadioShack is ridiculously overpriced on components.

There’s the schematic. Grabbing 5v and GND from a usb mouse is trivial. Heck, most of the circuity/code is trivial! Now that I think about it, this represents are really great starter project for anyone interested in microcontrollers.

Use this diagram of the pin functions for reference.

Remember there’s more than one way to skin a cat! For example, if you don’t want to program a microcontroller, a 555 timer is a simple method and there are tutorials out there demonstrating how to do this. I chose a microcontroller because I can precisely control the rate of firing and the duration. If you decide to do something similar, send me photos and I’d be happy to share them on the site! I love doing tangible projects, however silly they are.

And finally, the code!

#define F_CPU 1000000UL // frequency (20MHz)
#include <avr/io.h>
#include <util/delay.h>

void on()
    PORTB |= 1 << PB3; //led
    PORTB |= 1 << PB2; //heater
void off()
    PORTB &= ~(1 << PB3); //led
    PORTB &= ~(1 << PB2); //heater

void main()
    DDRB |= (1 << PB3) | (1 << PB2);
    int ticks;

    for (;;)
    { //FOREVER
        while ((PINB & _BV(PB4)) == 0)
        }                                     // NOT PRESSED, DO NOTHING
        for (ticks = 0; ticks < 125; ticks++) // CLICK FOR 5 seconds
        } // CLICK TAKES 1/50'th second

My Simple Balcony Antenna

I’ve been using a new HF antenna recently with surprisingly good results. Hopefully this page will be encouraging to those in apartments with severe antenna restrictions. I used to operate an indoor dipole mounted on my ceiling which was virtually invisible, but ever since solar panels were added to my apartment roof this antenna is picking up a huge amount of noise. In the past I played with a base-loaded vertical antenna made from copper pipe and it worked okay, especially when on my balcony, but it was bulkier than needed and awkward to store. My most recent antenna is made from 24AWG wire helically wrapped around the top element of a 3-element cane pole. My dad found a 15ft cane pole for $4 and it’s working pretty well for me. I guess the best description of this antenna is a “fully-loaded vertical” similar to a DIY hamstick. Here are some photos.

Notice that I only wrapped the highest element with wire. (The arrows in the above image depict where the helical element begins and finishes.) My logic is geared toward trying to get as much of the functional antenna above my apartment roof as possible. While it might not be a high-gain antenna, the level of noise reduction I experienced by raising the majority of the antenna above the roof is astounding.

I can hear stations nearly full quieting that I cannot even detect with my indoor dipole. Also, I hate reports like this, but I’ve only made a few SSB contacts ever with my indoor setup, and always local US stations. The very first contact I made with this vertical antenna was Slovenia! He was calling CQ, I responded, and it picked me up on the first try.

THIS ANTENNA IS UGLY and a certain violation of my lease agreement which specifically states no outdoor antennas are allowed. Therefore, this is something I can only set up at night. Notice the PVC fitting at the base of the antenna - it makes it easy to set up and break down. Maximum setup / breakdown time is 30 seconds. On the floor of my balcony I have wire running up and down the wooden boards which forms a makeshift mesh ground plane. It’s not optimal, but I’m limited and it’s what I came up with. When I feel ambitious, I have quarter-wavelength radials that I toss off the balcony and in the bushes to improve grounding. Although I’m sure I could have tap points and gator clips to select the antenna’s resonant frequency, currently I run the antenna right into a MAC-200 antenna tuner. I’ve used it on 17m, 20m, 30m, 40m, and 80m. Again, this antenna is far from optimal, it should represent the last resort for extreme cases, but when you’re faced with not being able to operate at all this little quick and dirty setup has been a godsend!

I’m sure a lot of people will read this and be angry or argue as to why this doesn’t make a good antenna. I’m not claiming it’s awesome, but for me it’s the best I could come up with in my limited situation. That’s my $0.02!

Crystal Oven Experiments

Now that I’ve finished my 6-channel data logger (previous post), it’s time to put it to the test! I’m using a handful of LM335 temperature sensors to measure temperature, and a 20 Ohm resistor to act as a heater. When 1A of current passes through it, it gets quite toasty! First, I’ll make some temperature probes…

UPDATE: Those photos show a partially completed sensor. Obviously the third wire is required between the resistor and the LM335 to allow for measurement! Here’s a more completed sensor before the shrink tube was massaged over the electrical elements:

Then I mounted the sensors on a block of steel with the heater on one side. This way I can use one temperature to measure the heater temperature, and the other to measure the temperature of the metal chassis. I then put the whole thing in a small Styrofoam box.

When I fire the heater, that sucker gets pretty darn hot. In 40 minutes it got almost 250F (!) at which time I pulled the plug on the heater and watched the whole thing cool. Notice how the metal chassis lags behind the temperature of the heater. I guess it’s a bit of a “thermal low-pass filter”. Also, yes, I’m aware I spelled chassis incorrectly in the graphs.

But how do we use this to build a thermo-stable crystal oven for a MEPT (radio transmitter)? I tried a lot of code, simply “if it’s too cold, turn heater on / if it’s too hot, turn heater off” but because the chassis always swung behind the heater, and even the heater itself had a bit of a delay in heating up, the results were always slowly oscillating temperatures around 10F every 20 min. That’s worse than no heater! My best luck was a program to hold temperature stable at 100F with the following rules:

What a great job! That thing is practically stable in 20 minutes. The advantage of this over an analog method is that I can set the temperature in software (or provide an interface to change temperature) and my readings are analytical, such that they can be conveyed in a radio message. Again, my best results came when I implemented rule 3 in the code above. More experiments to come!