So where does thing sit right now? The next CC-20 board revision is just about ready to be implemented. I've had to move to a DDS with a higher master clock frequency and change out the product detector from a dual-gate MOSFET to a diode-ring mixer. One advantage of the new DDS is that I can greatly simplify the transmitter circuitry, but this will require the trade-off of a fairly significant revision of the PCB.

I have been getting my PCBs manufactured in China, and right now many of the manufacturing firms (my board house included) are shutting down for two weeks to observe the Spring Festival (Chinese New Year). So even if I do send my Gerber files to the board house, they probably won't be back for at least a month. In the meantime, I've decided to work on a side project that's been rattling around in my head for a while: a QRSS/CW/Feld Hell/Etc. beacon. Also, in response to a lot of positive response that I have received from my simple Twin-T code practice oscillator, I also spent a few days revising the circuit to make the output a bit more robust and then created a PCB for the circuit in Kicad so I could transition my EDA to an actively developed software package (I was using TinyCAD/FreePCB previously, which seems to be pretty much a dead end).

OpenBeacon Prototype

So allow me to tell you a bit more about the beacon project. For now, I've decided to dub it OpenBeacon (I know, so very original). But there is a decent reason for the name. Much like the CC-Series, I intend for this project to fill a niche in the market that is very empty right now. The list of notable open source/open hardware kits out in the market is very small. The only one I think of off the top of my head is OpenQRP. As far as QRSS kits, I'm only aware of the one from the talented Hans Summers. My goal for this project is to provide a kit that is open, extensible, relatively inexpensive and simple, and ripe for user modification. Let me tell you a bit more about the project specs and how they fit into this goal.

Let's start with the bare hardware. The transmitter is a standard, vanilla Colpitts oscillator followed by an emitter follower buffer, which feeds a class A PA with fully adjustable output power (provided by a very cheap and cheerful part, the BD139). At full-bore with 13.8 V_CC, the transmitter can put out about 300 mW into 50 Ω. The brains of the operation is an Atmel ATtiny85 microcontroller. The way that it interacts with the transmitter is via its PWM output, which can generate a voltage from 0 V to 5 V after proper filtering. This control voltage is fed to a reversed-biased LED which acts as a varactor to tune the oscillator in very tiny amounts (< 10 Hz). The PWM output is essentially an 8-bit DAC, so not only can the varactor be flipped between 0 V and 5 V, but it can be set to many intermediate values, which allows for things like Feld Hell and just about any kind of graphic or glyph you can think of to be transmitted. The transmitter PA is also keyed with a PNP transistor which is controlled by the ATtiny85, which allows the OpenBeacon to operate in standard CW beacon mode.

The main way in which this project will meet the goals I stated above is in its user interface. There is a handy open source project called V-USB which gives USB interface capability to AVR microcontrollers that do not have USB built-in. This allows me to wire a USB port to the ATtiny85 and have the V-USB firmware take care of all the ugly business behind the scenes so that I can focus on interfacing the OpenBeacon to a PC. With a simple command line program, the user will have the ability to switch between the many operating modes available, set his own callsign and beacon message without having to have the microcontroller programmed for him, upload custom glyphs to be transmitter, and monitor the status of the beacon. No need to mess with jumpers or in-circuit programmers (although the ISP port will be available for those who want to hack their OpenBeacon). The client program is written in C and should be able to be compiled for Linux, Windows, and OS X machines.

KI6FEN Grabber Capture

Right now, the prototype is pretty much complete save a few minor tweaks. Yesterday, I got the code for the CW modes completed and put the beacon on the air in DFCW 6 second dit mode just above 10.140010 MHz. Conditions weren't great, but I did manage to get a few weak captures on the KL7UK grabber and one from KI6FEN via Twitter. The signal was way too wide and extremely drifty, but I've solved those problems by changing the coupling capacitor between the LED varactor and the oscillator and by creating a rudimentary thermal chamber for the beacon out of pink antistatic foam. I'll be leaving the beacon on for the next few days when I'm not working on the project (which will be most of the day). Any reception reports would be greatly appreciated!

So the plan is to get the CC-Series PCB revisions hopefully done by next weekend so that they can be sent off to the board house before their vacation is over. In my little bits of downtime, I'll continue work on the code for the OpenBeacon. The plan for this project is to get the PCBs cranked out very quickly. Now that I'm familiar with Kicad, I think it won't be too difficult or take too long to design the boards. I'm also going to be trying out a new PCB vendor which promises much cheaper prices and faster turnaround times on smaller boards such as this. With any luck, I can fast-track OpenBeacon testing and production and have it out while the CC-Series is in it's final beta test. Stay tuned, this is make-or-break time!

IF Amp & Product Detector from CC-20 Beta 1

I've done a lot of reading in Experimental Methods in RF Design (EMRFD) about microphonics in DC receivers (read chapter 8!), and the number one cause of it is poor LO-RF port isolation in the mixer. The CC-Series uses a venerable old circuit which hasn't seen much use in a while. A dual-gate MOSFET is pressed into double-duty as a product detector and BFO (see above). Since the dual-gate MOSFET product detector is in a single-ended configuration, it inherently has bad LO-RF isolation. This allows VFO (or BFO in this case) signal to leak out the product detector input, and have a good portion of that signal reflect back into the product detector. So naturally, the CC-20 could be experiencing the microphonics because of this phenomena. One of the solutions mentioned in EMRFD is to put an amp in front of the mixer which has excellent reverse isolation (signals coming into the amp output don't tend to get out of the input, and therefore can't reflect back in again).

I had the suspicion that the common-source JFET amp in front of the product detector might be the culprit. So what's the best type of amp to place in front of a single-ended mixer? The common-gate JFET amp is a good and popular choice. However, VE7BPO notes on a recently published web page that the best commonly found amp configuration for this particular parameter appears to be the cascode (see the bottom of the page).

In order to test this theory, I went to work on a project that I had set aside earier: a direct conversion receiver based on the CC-Series product detector. When there was no preamp in front of it, the microphonics were unbearable. I figured that a good way to test my theory would be to put a cascode amp in front of this mixer and see how much it helped. I decided to put a dual-gate MOSFET preamp in front of it, as this is essentially a cascode amp and it fits with the dual-gate MOSFET product detector. Once the new preamp was added, the change was dramatic. The microphonics were gone.

Next, I decided to be a bit more rigorous in my study and quantify the exact difference between the common-source JFET amp and the dual-gate MOSFET amp. First I breadboarded the common-source JFET amp and ran it through the test procedure in the page linked above (at 18 MHz). The results were atrocious. Only 30 dB of reverse isolation, which is worse than the worst amp listed there (the feedback amp). Next, I dug out an old dual-gate MOSFET amp I had breadboarded for my 2008 investigations and ran it through the same test. As expected, the results were vastly superior: 68 dB of reverse isolation. This lines up nicely with Todd's measured results of >64 dB for the hybrid cascode (I used a spectrum analyzer while he used an oscilloscope, so I was able to get a pretty good measurement down to low signal levels).

So this appears to be strong evidence that the IF amp is the problem. It seems certain that the next version of the CC-Series is going to scrap those awful common-source amps for a much nicer dual-gate MOSFET amp. The lesson to take away from this is that if you are going to use a single-ended mixer for any but the most simplistic applications, it must be fronted with an amplifier with an excellent reverse isolation. While the typical common-gate JFET amp will work OK, for best results it looks like a cascode or dual-gate MOSFET amp is the way to go.

Dave AA7EE purchased and built a VRX-1 kit a while ago but was never fully satisfied with the performance due to an annoying 60 Hz hum. He and I had briefly traded comments on the topic via Twitter, but I never really seriously took the time to think about it until just recently. Dave had built and placed a peaked lowpass audio filter into the receiver thinking that would help with the hum, but unfortunately it did virtually nothing to help with it.

I was a bit surprised to hear of the hum problem, since I had never encountered any significant amount of hum, nor had I had other complaints of hum. The eureka moment came when Dave had mentioned that the hum went away when he disconnected the antenna, or it decreased in signal strength when he moved away from his home. I had assumed that the hum was a glitch in his audio circuitry, but this reminded me of the problem known as common mode hum. The best description of this phenomena is found on pages 8.8 - 8.9 of Experimental Methods in RF Design, but I can provide a brief overview. Common mode hum is the result of the LO leakage getting out of the antenna port, modulated by a mains power supply (like an old-fashioned model with rectifiers), and then re-received by the radio.

Due to the simple, single-ended mixer design in the VRX-1, I knew that LO-RF isolation was very poor. So the first suggestion to pop in my mind was to tell Dave to try a common-gate JFET preamp on the front end. Although these type of mixers have modest gain, they have a low noise figure, and even more importantly for us, excellent reverse isolation (on the order of 30 dB). This should be enough to kill any significant amounts of LO leakage.

Dave built a circuit from master homebrew experimenter, Todd VE7BPO. It's the last circuit on this page, and it looks rock-solid. A double-tuned circuit on the front and a single-tuned circuit on the output. Sure enough, that ended up doing the trick. Rather than trying to reinterpret Dave's thoughts, go visit that last link, then watch his YouTube video so you can hear the results for yourself:

I'm really pleased to hear that Dave's annoying problem is finally fixed. This makes me wonder, in retrospect, whether I should have just designed in a preamp to the VRX-1. It certainly isn't needed for noise figure purposes, but as you can see it can make a huge difference with those who might have problems with hum. There's also a well-documented problem of a loud impulse generated when the antenna is connected or disconnected during operation. I suspect at the reverse isolation of the preamp would also help this. Hindsight is certainly 20/20. If there is ever a VRX-2, then you can bet that it will get a stock common-gate preamp.

G3UUR Crystal Checker

I decided to make my initial Project X prototype PCBs at home using the old tried-and-true method of toner transfer (via Pulsar Professional paper and foil). Since I'm a novice at PCB layout, I didn't feel comfortable paying the money for a few proto PCBs from a board house, then finding out that I did something wrong and flushing that money down the drain. Instead of buying Pulsar's starter kit, I just purchased a pack of the transfer paper and a roll of the green foil. I also got the required GBC laminator from Amazon instead of paying significantly more for it from Pulsar.

Last night, it was time to give the process a whirl, so I decided to make a PCB of the G3UUR crystal checker circuit that was printed in the Fall 2010 QRP Quarterly (an excellent article, by the way). The instructions seemed clear enough, but I had about five failures before I finally figured out how to make the process work correctly. I was just about ready to chuck the whole thing in the trash bin, but I managed to keep my wits and persevere through it. For those who might be new to the process, let me help you to avoid some of the problems that I had:

• Give yourself at least 0.5 inches of copper clad clearance on each margin of the final board edge in order to give the laminator good purchase on the board and transfer paper. Putting the toner traces too close to the edge will result in those edges failing to adhere to the board. There's just not enough heat and pressure to do the job properly at the edge.
• When passing the copper clad plus transfer paper through the GBC laminator, I found that it worked better with four passes. Pass the board once, turn it 90°, pass it again, etc., until all four edges have been the leading edge through the laminator.
• I didn't see this mentioned in the instructions, but after you apply the foil on top of the toner traces, you must let everything cool down to room temperature before attempting to peel the foil off the board. Failing to do so will rip most of your toner off of the board!

Once I got the bugs worked out, I was quite happy with the end result. I also decided to try out a new etching method. Instead of using ferric chloride, I used the hydrogen peroxide/hydrochloric acid recipe that I've seen touted on the Internet. Let me just say that it worked out beautifully and is waaaaay cheaper than buying ferric chloride. It only took about 3 minutes to etch my small board in a Ziploc baggie. No need to mess with expensive equipment or chemicals!

The etched board turned out very well. There are a few places with very close traces, as you can see in the photo above. These etched out perfectly, no problems at all. You might notice some bad copper on the bounding rectangle on my board, but that was because of the close clearance between that trace and the board edges. I know how to avoid that in the future.

Tonight, I got the board all soldered together and it worked perfectly on first power-up! That's always an extremely satisfying feeling. Now that I've got a handle on the process, I feel comfortable using it on the Project X prototype. Stay tuned for more progress on the new radio!

G3UUR Crystal Checker - Bottom

G3UUR Crystal Checker - Top

The most common form of JFET biasing that you seem to find in homebrew QRP projects is self-biasing; where a resistor is placed between the source and ground, and the gate is tied to DC ground. This is a convenient way to bias a N-channel FET when using a single-polarity voltage supply, but it's definitely not the most stable form of biasing. According to Application Note 102 from Siliconix, using a constant current source to bias the FET will ensure a constant I_d, but that seems like a bit of overkill in a simple QRP rig.

(a) Constant Voltage Bias (b) Constant Current Bias (c) Self-Bias (d) Combo-Bias

The image above shows the transfer characteristic curves for a 2N4339 biased with four different methods: constant voltage, constant current, self-bias, and combo-bias. The two I_d-V_gs curves in each graph represent the normal production range of I_dss. As you can see in graph (c), the load line QA-QB has a fair amount of slope, which indicates that I_d will vary quite a bit as I_dss varies. Graph (a) shows constant voltage bias, which provides the worst variation of I_d by far. The source resistor provides the slope of the load line, so increasing R_s will flatten the slope and reduce the I_d variation, but the problem is that it also chokes off I_d and reduces the amplifier transconductance.

Combo-biasing (presented in graph (d)) helps to flatten the slope without losing all of your drain current. The load line in the self-bias graph has an intercept at the origin, which represents the gate at a DC voltage of 0 V. If you leave the source resistor in, and apply a positive voltage on the gate, the load line slope will flatten as seen in graph (d).

In order to see this effect for myself, I conducted a quick experiment with a batch of ten J211s that I picked from random from a bag of parts procured from Mouser. The first thing that I did was measure I_dss by configuring the circuit on the left below and measuring the current through the source. Next, to measure the self-bias configuration, I placed a source resistor of 1 kΩ in the circuit and again measured the drain current. Finally, the combo-bias configuration was tested by applying 2.9 V to the gate through a 100 kΩ resistor, for a target drain current of 4 mA (the procedure for choosing the proper gate voltage is detailed in the application note).

As you can see in the table above, self-biasing controls the drain current to a smaller range than I_dss, but if you compare the standard deviation to the sample mean, you can see that it's roughly the same. The values of drain current in the self-bias circuit are roughly proportional to I_dss. On the other hand, in the combo-bias circuit it's obvious that the drain currents are even more tightly controlled even though the mean is about double what it is in the self-bias circuit. The standard deviation as a percentage of the mean is approximately half of the self-bias circuit.

Note: it has been a while since I've taken Statistics, so I believe that I've made a valid comparison, but if I haven't I'm sure someone will let me know.

Now I understand one of the reasons that the string of source diodes is used in the Hycas IF amplifier. This provides the gate with a few volts of bias, which gives combo-biasing and drain current stability. The Siliconix application note shows voltage divider bias as a way of achieving this, but I've always been a fan of using diode bias where possible. This method of biasing should provide a good way of controlling for manufacturing variations when using JFETs in bulk.

The receiver is pretty close to its final configuration (I hope), so I took it in to work so I could measure the performance using the very nice calibrated test equipment at my bench. This is my first time making this entire range of receiver measurements, and I have to say that it was quite interesting. I wish it was something that I had started taking more seriously a while ago. I have a wish list for test equipment a mile long, but I've realized that I really need to get my hands on a pair of good signal generators, hopefully ones that can give me an output down to -140 dBm.

I followed the procedures in the ARRL Test Procedures manual as closely as I could (do yourself a favor and save a copy of this highly useful document). The numbers came out pretty close to what I would expect, so I'm reasonably sure they are legitimate. I'm just hoping that I can win the challenge so that my rig gets taken to HQ for the real battery of tests! Below is the results of my testing, followed by a very brief commentary on the numbers:

IF Bandwidth: 462 Hz
MDS: -126 dBm
3rd Order DR (20 kHz): 80.5 dB (S5 signal level per ARRL Lab Procedures)
IIP3 (20 kHz): -5.2 dBm
Blocking DR (20 kHz): 102.6 dB
IF Rejection: 23 dB
Image Rejection: 48 dB

Clackamas IF Response

I believe that the MDS, dynamic range, and IIP3 measurements are all acceptable for a receiver using 40 parts and a 7-part VFO. I know why the IF and image rejection is so horrible (it's because of a design trade-off), but I can't get into the details of that yet. The filter response is a bit funky, but I'm sure that my impedance matching isn't the greatest. On the air, it's my opinion that the rig sounds decent. Maybe I can record some audio this weekend and post it. Let me know what you think; does this sound reasonable to you for a compromise superhet? I'd love to hear your comments on this.

VU2PEP has a lesser-known SSB design on his website, that's a dual-band transceiver. Besides having 20 and 40 meter capability, it also has a different topology than the BITX series. Instead of reversing the flow of the signal to generate a SSB signal, this design sends the RX and TX signal in the same direction through the IF. Take a look at the schematics to get an idea of what I'm talking about.

I decided to make a "remix" of this design. The basic topology is the same, but most of the circuits are revised. The IF was moved to 4.9152 MHz, and the VFO is heterodyne-style to provide a ~19.12 MHz LO signal. My version is only for 20 meters. The front end has a preamp added and uses a cascode JFET mixer instead of a single JFET. So far, the RX strip and VFO is complete (although I might change the VFO because of some birdie problems), but the transmit amplifiers haven't been built yet. I got a good chance to work out the RX during Sweepstakes. Check out my YouTube video below to hear me describe the circuit so far and listen to the receiver on SS.

VRX-1 in 4SQRP Clear Top Tin

While I was away on my honeymoon, I noticed that the upcoming kit that I've been hinting about for months has finally been released. The Four State QRP Group announced availability of the VRX-1 direct conversion receiver. The VRX-1 is a simple 40 meter VXO-tuned receiver (crystal on 7.030 MHz), but it's not your typical NE602/LM386 combo. The product detector consists of only a 2N7000 MOSFET, a capacitor, and an inductor. The audio amplifier is a TDA7052 from NXP. This little 8-pin DIP can output 1 watt of clean audio into low impedance headphones or a small speaker. Current consumption is only about 40 mA, which makes the VRX-1 easy on your batteries if you take it out for portable use. The construction of the receiver is done Manhattan-style, but don't let that put you off if you've never built this way before. I provide a precise, detailed layout diagram to show you exactly where each part is placed and how it is oriented. There's also some very detailed build documentation to walk you through the build, which you can preview at the VRX-1 web page. Even the novice builder can construct this radio!

The VRX-1 was designed to be a companion to the NS-40, or other similar rock-bound 40 meter QRP transmitters. I also include instructions on how to use some of your own parts to modify the VRX-1 for operation on any HF band, so don't feel like you are stuck on 40 meters if you would like to try to experiment a little. In a future blog post, I'll walk you through the process of integrating the VRX-1 with a standalone QRP transmitter to make a complete station.

Proceeds from the kit sales go to fund OzarkCon 2010; I don't make a dime off of it (just the glory, LOL!). So please support the QRP community and try your hand at a new kind of kit. It's only $25 postage paid in the States, $28 for DX.

20 Watt Broadband Linear Amplifier

QRP is tons of fun on CW, but it gets a bit rough trying to work other stations on SSB with 5W, especially when you are using antennas that are low to the ground. I had been eyeballing the nice RF MOSFETs from Mitsubishi for a while, and since I got a hankering to get a bit more active on SSB, I took the plunge and ordered five of the RD15HVF1 devices. At a current price of $5.25 at RF Parts, they are a bit more expensive than the IRF510 that you see in a lot of 20-40 watt range linears, but these devices have a few advantages over the IRF5xx series. One of the biggest, in my opinion, is that these RF transistors are designed to run off of a 12 volt drain voltage, unlike the IRF510 amps which don't really work well until they get around 24 volts on the drain. These things can also take quite a beating from poor mismatches, and have the convienice of having the source connected to the metal tab on the case, making for a nice solid ground connection.

20 Watt Broadband Linear Amp - Inside

Having the appropriate parts in hand and some designs on the internet to steal from, I set out to build my own linear. There isn't a ton of creativity to be used when designing a linear of this class (Push-pull Class-AB). Every design that I've seen looks nearly the same. Not surprisingly, the real focus of the design is in optimising the input and output networks. Feeling lazy and anxious to just get on the air, I pretty much did "cut and paste" from some different circuits to find out what works best. I know, not the best method, but sometimes the desire to just put out some RF trumps proper procedure. I don't have a scehematic to post at the moment, but if you click through on the photo to the right, you can see a close-up with descriptions of major circuit blocks. Below, I've posted links to the two circuit resources that I used the most for this design. I'll have more details about the designs to comment on at a later date, when I can pull some proper notes together.

• HPSDR Pennywhistle
• TF3LJ SoftRock Amplifier

One of my weakest homebrewing areas is in the mechanical engineering, but now that I have a bit of a real "shop" in my garage, things have been getting better. A bit of scrounging at the surplus stores around town led me to some cheap heat sinks that looked like they might be suitable for this project. After attacking them with an angle grinder to get a lip off of the bottom side, I was able to bolt two of them to the lid of an aluminum Hammond enclosure. I nibbled a nice square area right out of the middle of the copper clad I used to build on, soldered the RD15HVF1 devices to some pads etched out with a Dremel, then bolted the MOSFETs and copper clad directly to the lid of the enclosure. Drilling the holes for the BNCs and the LED was a piece of cake with the aluminium box material.

Without getting into too many details at this point, I was able to get the amplifier working right off the bat. I didn't get quite as much output power as I initially liked (only got about 10 watts), but the amp was working correctly. More troubling was the fact that output on 6 meters was only 2.8 watts. Not too great when you are putting in 2.5 watts. I figured it had to be something with the input or output network. The input return loss measured quite good; -15 to -20 dB across all the bands. So I figured that left the output network. My initial iteration of the amp used a transformer similar to the one in the Pennywhistle amp (this is a configuration that I've also successfully used before in a push-pull class-C CW amp). Without doing any actual measurements and calculations, I dropped in the broadband transformer pair used in the TF3LJ amp, and immediately improved my output power by a few watts. But I was still a bit low on 6 meters. A bit more searching showed that I might need another compensation cap on the output, so I experimented a bit more until I found that a 1200 pF silver mica in series with the drain transformer outputs worked wonders and boosted my power on 6 meters to nearly 15 watts CW. I haven't done any analysis to see why this helped. I know, sloppy...but sometimes expedience wins.

Since there's no output filtering built into the amplifier enclosure, I had to assemble some outboard filters in order to get this thing on the air. I was going to use 7-pole low-pass filters until I realized that everybody else uses 5-pole filters because push-pull amps already suppress the even-ordered harmonics by at least -30 dBc. A bit of work with the new LADPAC software in EMRFD enabled me to crank out a table of filters for all of the bands (160 m - 6 m) using the silver mica caps in my junkbox plus T68-6 toroids. If you click through the photo below, you can get a glimpse of the copper clad enclosure sticking off the output of the amp.

Backyard Linear Test

Last Monday, after a bit of checking of the signal purity with my dummy load and scope, I was satisfied that everything was working OK and took the amp out for a spin on the back porch. I set up the Buddipole in Versatee Vertical configuration with the Low Band Coil. It tuned right up on the upper end of 75 meters, and I had no problems at all checking into the Oregon Emergency Net. One watt out of the 817 gave me about 25 watts out of the linear on 75 meters. I was too busy to do much else with the amplifier until today (the following Sunday), but I was excited to give the amp a try on 6 meters, since that was one of my biggest motiviations for building the thing. The Buddipole was set up in a simple 6 meter dipole configuration about 10 feet above the ground and I parked the 817 on 50.125 MHz. It didn't take long before I heard N6OR booming into Beaverton from Southern California (grid DM12). I snagged him on the first call, getting a 57 signal report in return and a report of good, clean audio when asked. He was running 100 watts into a quad, which you can see on his QRZ page. I was really thrilled since this was not only a victory for my mad homebrewing skillz, but was also my first 6 meter QSO!

I've been parked on 50.125 for most of the afternoon here at the NT7S shack and have picked up a few more QSOs. So far, all reports of the audio quality of the linear have been FB, so I'm satisfied that it (and the LPF) are working as they should be. I think I've about worn out my keys on this post, so I'll wrap it up for today (I always start with modest ambitions on these posts, they they grow exponentially). I'm having way more fun than I should be, and I'm very pleased to be back out of my ham radio funk.

Spring boarding off of this project, I intend on extending some of the circuit elements that came about during the development of the secret project and using them to create something new. I don't want to say too much at this point, since I don't want to give any clues which could spoil the QRP club's kit announcement. At this point I'll just say that if I can pull this off, it will be a new take on a QRP classic. But I don't want to hype anything too much. I would rather underpromise and overdeliver. Plus, this thing hasn't even gotten off of the engineer's notepad yet. It should be a fun one to try to accomplish.