John Langridge KB5NJD / WG2XIQ (Originally written in September 2012)
UPDATE August 9, 2015: If you are building one of these transmitters, note that in most cases the low pass filter is not good enough for the 50 db harmonic suppression that is required by the FCC in the US. I hope to do a write up on this site in the future but if you are building a system now, send me a message on the contacts page for details on improving the low pass filtering so that its legal in the United States!
NOTE: This article was written in 2012 and since then numerous modifications and developments have occurred with this converter. Always check G3XBM’s site for the most updated information and schematics.
A few months ago I talked about a way of generating a CW signal on 630-meters using an 80-meter LC oscillator and a buffered CMOS-based divide by 8 device to develop enough signal to drive a much larger power amplifier with a high quality keying waveform. This class-D device is great for CW but what about other narrow band, low baud rate digital modes like OPERA, JT-65 and WSPR? Fortunately these are all FSK modes that don’t have more than one carrier frequency on the air at a given time so this means that I don’t have to deal with the linearity issues associated with modes like SSB even though these modes are often generated using USB modulation from a conventional HF rig. While SSB transmitters can be complicated by beat frequency oscillators, IF mixing and other scary (and conceivably costly) things, it’s a very simple affair to construct a device that take a signal from any HF rig on 80-meters and mixes that signal with another fixed frequency signal which ultimately produces the sum and difference of these frequencies. A little filtering allows some selectivity of the signal you are looking for and that can be output to a higher power amplifier to generate a usable signal level. It’s really shockingly simple.
So why 80-meters for a starting frequency? There is a lot of data to suggest that the lower the frequency of the local oscillator (LO) and the RF source that provides the frequency variability, the higher the frequency stability of the output signal. There are a lot of other factors that go into this but it’s a pretty good generalization. My original system was going to use a 28 MHz LO and the RF from the radio at 28.472-.479 MHz to generate a difference frequency between 472 and 479 KHz. Examining the stability of this system compared to the same arrangement using a 3.2 MHz LO with rig RF between 3.672 and 3.679 yielded many orders of magnitude more stability with the lower frequency system. It also helps that 3.2 MHz crystals are easy to come by today.
What makes this system unique is that mixing occurs not in a pricey and sometimes difficulty to obtain and utilize doubly balanced mixer from someplace like minicircuits.com, but rather utilizing a $.39 2N3904 bipolar junction transistor. What’s really nice is that if you burn up the transistor mixer because you injected too much 80-meter power, they are simple and cheap to replace.
What do mixers do and how do they do it? Well, there is a lot of black magic there but in the simplest form, a mixer is a transformer on the input driving a diode bridge kinda like a rectifier and then an output transformer. Yes, I am oversimplifying this in big way (primarily because I can’t really rationalize how it works entirely, it just does!) but the signals are introduced to that diode ring at different locations and the signals circulate through the mix and match the frequencies. At the end you get the sum and difference of the signals. There are all types of mixers out there. Some pass the original signals though (never understood where that would be handy) and some pass one of the signals through along with the sum and difference. It really depends on what you are looking for. With the 2N3904 transistor, the signals are just thrown onto the base of the transistor at similar power levels and the signals duke it out until they are amplified as sum and difference. I’m sure some of the LO and RF drive frequency is present as well but the low pass filter on the output takes everything out except for the lowest frequency, 472 KHz band. I imagine the purists are cringing to hear me say that but it works and it works well.
I was not the first person to come up with this idea in amateur electronics. Roger, G3XBM, was developing a similar system in July of this year so that he could operate OPERA and WSPR on 600-meters, right around 500 KHz, which is where many of the experimental licenses currently reside today. Roger is an accomplished electrical engineer and did a very good job developing his system and tweeking component values for optimum RF decoupling and maximum signal with inexpensive parts, many of which reside in the builder ham’s parts box.
My device differs every so slightly. As the radio I am using for 630-meters has a minimum output of 5 watts and mixers of any flavor have a tendency to be finicky – they don’t want too much power or too little power, they want “just enough” whatever that is – I opted for about 30 db of attenuation which resulted in about 5 mW of power to the mixer. If I needed more, I could always turn the power up just a bit because I scaled the attenuator for 10 watts of dissipation (multiple 2 watts resistors in parallel). As it turns out, 5 mW was plenty of power and judging the amplitude of the 3.2 MHz crystal oscillator waveform, the input was close to equal.
Additional differences in the system include the lack of receive loop back. It should be noted that in my system, I do not listen on the transmit antennas, instead utilizing separate directional receive systems. As a result the loop back is not necessary. It bears being said that this is NOT a true transverter. It is only a transmit downconverter and to receive on the target band, you must operate in split VFO mode, transmitting on 80m and receiving in the 472 KHz band on the other VFO. Modern radios accommodate this switching very easily. It may be necessary to utilize a receive preamp since most amateur transceivers do not offer very good sensitivity at these frequencies out of the box. A system that I am currently designing for Russ, KX5G, will have to address this problem.
Since I am utilizing an outboard 100-watt power amplifier, I do not need the 10 watt PA on board the downconverter that utilizes an IRF510. In its place, I use a few stages of what Roger, G3XBM, refers to as a “squarer”. This is nothing more than intermediate stages of amplifier. Since I did not really know how much boost I would need to achieve the 12v peak-to-peak signal I needed to drive my big PA, I simply started adding stages until I saw useful power. That works well as long as you are prepared to do good quality filtering on the output. In future versions I am considering utilizing the IRF510 at a lower supply voltage and then utilize an output attenuator to achieve the necessary drive power for the larger PA. This will make the efficiency purist cringe but I like to think that the heat generated in attenuators within the box ultimately contributes to frequency stability of the crystal oscillator after a few transmit cycles. It’s like an unintended crystal oven.
So far I have built and tested a number of these units over the past two weeks and they seem to have great frequency stability, drifting only a few hertz after initial warm up. Testing was accomplished using separate radios to transmit and receive within the ham shack and transmitting into a dummy load utilizing WSPR at 50% transmit cycle and 10 watts. Below is the WSPR interface showing the received signal on 475 KHz while the adjacent radio is transmitting on 3675 KHz into the down-converter:
There are a couple of things to note from this image. First, disregard the reported RF frequency. The WSPR system had to be “dummied up” for the test. Reception is, in fact, occurring on 475 KHz utilizing the Yaesu FT1000 Mark V Field. Secondly, note that the receiving waterfall shows a number of signals that are 60 Hz apart and that in the receive window there are two simultaneous received signals from the same transmit cycle. This is the result of power harmonics and is easily resolved by placing 1000 uF electrolytic capacitors in parallel with the existing decoupling capacitors on the mixer, LO, and the squarer. Note that I have notified Roger of this change but the above schematic does not incorporate these additions. That means you would add a capacitor in parallel to C7 and 8, C14 and 15, and C16 and 18 for a total of 3 additional capacitors. Additional testing has resulted in almost total attenuation of the harmonics. Depending on the quality of your power, you might need more or less.
Much of the work that Roger and I have incorporated into our respective systems utilizes RF PA design concepts that were developed by Rog, GW3UEP, who has been referenced in past articles regarding PA design. So many of these concepts are open source ideas and the level of development gets better and better through experimentation.
My long term goal is to build up a number of systems for use by select club members that I feel like would use the system either by setting up beacons or operating when the band officially opens. This includes setting up a beacon on the behalf of the club at someone’s house utilizing a minimalist system with a short vertical and loading coil design. If you have an HF system that you are not utilizing on a regular basis or have multiple radios, consider setting up a receive-only system to provide reports to other WSPR or OPERA systems on the air. I will talk about how to do this in subsequent articles because I think it is a very valuable thing to have a large number of receivers to listen for band openings. It may likely only require that you setup the free software as if you were going to run a digital mode and set the frequency on your radio.
Odds and Ends
I had intended on talking about the math associated with EIRP calculations and the measurements associated with it. Since the ARRL executive committee and counsel are still drafting the proposal for the FCC we don’t yet know what the max EIRP will be. It is presumed to be 1 watts but it might end up being 5 or even 20 watts. Until we know for sure, I will hold off on that.
A few other items that I have forgotten in previous months that I thought were important:
Filtering and the ATU: One of the features of a high Q antenna system and subsequent narrow SWR bandwidth is the filtering capability offered by the ATU. While the ATU’s primary purpose is to match the antenna to the feed line and for maximum power transfer with a minimum of conductor loses, because the match is narrow, signals that are out of band will generally be rejected because the resistance, at least in my system is a dead short, that is 0 ohms resistance. For example, when I am tune up for operation at 475 Khz, if I sweep upward in frequency, the resistance drops to 0 ohms close to 490 Khz and remains this way well into the AM broadcast band with an ever increasing value of reactance. In short, even if the filtering in the transmitter were missing, its my belief that there would be absolutely nothing left in terms of harmonics at the second harmonic since the SWR bandwidth is so narrow. This is a nice, additional layer or spectrum protection.
Beware of SWR meters for HF at low frequency: When I was developing my first signal source and PA’s I was utilizing an MFJ 949 tuner that I had as a spare around the shack. It has a dummy load and a cross needle meter and was great for the purpose of seeing the RF I was generating as well as terminating that RF during development. The problem was with how power is sampled. Almost every through line power meter on the market uses a ferrite core transformer of type-43 material to sample RF on its way to the load and utilizes a diode rectifier and voltage divider to ultimately display a voltage output on the meter that corresponds to an SWR value and power. The problem here is that type-43 material is VERY lossy at lower MW frequencies. In fact, it is so lossy, that Pat, W5THT, who operates in the ARRL 600-meter research experiment as WD2XSH/6 in Gulf Port, MS, had a core get so hot that it virtually exploded. Pat tried to modify the core material to utilize type-77 core material, which is more compatible with these frequencies but was unsuccessful. In short, it was just easier to start from scratch. The take home message here is that you should not believe the SWR readings on your meter at these frequencies. In fact, SWR values will always read high. In my case, into a 50 ohm load, my meter still reads over 2:1 SWR. It took me a long time to figure out what was going on. Those meters are only speced to 1.8 MHz! Even Bird Thruline Wattmeters read erroneously at these frequencies without the properly matched slug. What’s more and less obvious is that using the wrong core material sinks a tremendous amount of power and for some time I left this meter in line with my high power system which resulted in a significant amount of power burned up as heat in this core. Fortunately mine did not blow up but it looks like it’s been hot. That meter is now out of line and I am seeing an appropriate level of system current into a dummy load. When selecting a ferrite for use as a current transformer to indicate the antenna system health, be sure you use Type-77 at these frequencies lest you have lots of loss. Power line ferrites also seem to work well.
Check your HF rigs for the MARS/CAP mod: You might be surprised! Many radios that have the capability to be modified for extended transmit will make power at 630-meters. What these systems often lack is filtering. A simple low pass filter will clean up the waveform and result in a very good quality signal. Russ, KX5G, determined that his radio would make 25 watts at 475 KHz and an examination of the waveform indicates that a little filtering will make it suitable for air. That will simplify his installation tremendously. In my case, the Yaesu’s FT1000 and 920 are difficult mods that are compromises at best and a transmit down-converter is a more logical solution.
That’s a good stopping point for this month.
73 and see you in the pile up!