John Langridge KB5NJD / WG2XIQ (Originally written in July 2012)
UPDATE October 6, 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: Always check the source site when building these projects for updated schematics and information. GW3UEP’s designs can be found at http://www.gw3uep.ukfsn.org/472kHz.htm
This month I am going to talk a little bit about my experiences with the development of a viable CW transmitter for 630 meters. As this is truly an experimenter’s band, the prospect of an off-the-shelf solution was neither likely to be found nor desired. Fortunately CW transmitters can be very easy to design and implement and the skills learned in their development are useful in other areas of amateur radio. In short, this IS your father’s amateur radio…
After the announcement of the 630-meter band on Valentine’s Day 2012, a number of kit developers began marketing to US amateurs. Up until this point, most of the market was in Australia and select European countries and most are in the $300-$400 price range. One was an all mode software defined radio developed by a VK and marketed stateside by a ham in New Hampshire under the name of “Genesis”. While reported to be a fairly easy kit, the board comes partially populated and requires band pass filters to be wound for the desired frequencies in addition to a handful of through-hole components. The transmitter puts out about 10 watts and while it is all-mode, I really needed about 60 watts to achieve the 1-watt EIRP that is likely to be approved as legal limit for the band.
The next model was developed as a full SMT kit by a group of Finn’s. Known as the JUMA TX-500, the kit touted a 65 watts output and while I did not need it, a receive down converter to convert signals at 472 up to the 80m band, allowing the use of a conventional ham rig which usually has been sensitivity on the 80m band than below the broadcast band. The price tag was impressive as well – nearly $500 plus a European value added tax. This was a bit of a turn off for me and the idea of SMT was not appealing as well as the unit had hundreds of parts. Also, after a few emails asking questions, the Finn’s stopped responding. I guess they were not used to having someone take such an interest in their product and they may not again 🙂
I had also considered using my newly modified SWR analyzer as an RF source and “rolling my own” high power amp. As much as I wanted to be able to say that the analyzer RF source was a workable solution, the unit has stability issues and tends to drift. You can observe this by looking at the frequency counter.
After a little more searching on the Internet, I somehow stumbled onto an inconspicuous Welsh website of Rog, GW3UEP, which I had missed on previous searches. Looking around his site, it was obvious that Rog was very knowledgeable and experienced in the operating at 500kc and below, having undertaken a number of mobile operations near the ocean and making good contacts with a number of hams on the continent as well as Ireland. Rog had just what I needed – a schematic for a simple but stable VFO/driver on 500kc that could be easily modified for 472kc as well as power amps for 25 watts and 100 watts on 472kc. After several emails, Rog proved to be very helpful and before I knew it, I had ordered parts for the VFO/driver and both models of the PA deck – less than $100 and I had enough parts for about 10 transmitters! I should go into business selling these things! (Authors note: most parts houses like Mouser follow the same model of selling bulk parts for cheaper than singles. If you were to buy parts individually for this project, you would have paid about $50 for a single transmitter – why not spend $50 more and have enough parts for 10? Good if something breaks!)
I started with the VFO/driver, using a 3 X 5 copper clad board and “dead bug” construction style.
Divide by 8 frequency divider for 472kc with 12v p-p driver – cheap and simple
The design is simple, very simple in fact, utilizing a parallel resonance network (literally a capacitor and coil in parallel) on 3.8 MHz. 80m is a popular place to build a frequency divider on the 630m band as divide by 8 semiconductors are very common and CHEAP!. The signal from 3.8 MHz is fed through a series of buffers on a hex inverter chip and then into a divide by 8 CMOS device. The output of this VFO is 12V peak-to-peak and right in the middle of the 630m band. If additional waveform shaping is not implemented, this driver produces a very clean waveform with click-free shaping. The only thing you have to remember about this VFO is that you need to find the signal with this highest amplitude – don’t just pick a carrier and call that 472kc… check it with a frequency counter. Earlier experiments showed that my fundamental signal was down near 200kc and I was using a harmonic that happened to fall in the part of the band I was using. CHECK THE FREQUENCY!
Next was the 25-watt power amplifier. I have to say, I learned a tremendous amount from this part of the project and really set the stage for building the 100-watt power amplifier that I will discuss later. The 25-watt model uses an IRF-510 MOSFET as the final amplifier operating in class-D with waveform shaping on the drain power feeder that controls the rise and fall time on keying. The other plus is that the IRF-510 is very forgiving of mistreatment and they are cheap. When they blow up, though, they sound like a shotgun going off!
The IRF-510 MOSFET is located on the left hand side, just right and down from the BNC connectors. Its enshrouded in a homebrew copper heat sink. Key shaping and powering of the drain on the MOSFET is accomplished by a transistor on the terminal strip at the top left of the board.
As stated earlier, this was the trial by fire part of the project. Having worked with tubes in the past, the MOSFET was a logical choice, not only because of their operating efficiency but because MOSFETs behave much like a vacuum tube – they are either completely on or completely off. Class-D is accomplished on this amp by way of a shunt diode with a leakage resistance across the input from the driver circuit. Based on some of the other variations on accomplishing class-D, this was a very novel approach and one that I appreciate. It also ensures efficiencies greater than 80%. One of the most valuable lessons that I learned in this phase of the project is that variable coils would have been very helpful. Simply winding a coil to spec and checking the inductance with an RF bridge is usually not the best way to do it and you have really have to tweek the coils for waveform and power output. A scope and power meter calibrated for 472 kc is very valuable.
Probably the most significant discovery that I learned or re-learned is that SWR meters that work well between 1.8 and 28 MHz rarely work well below 1.8 MHz and can give all sorts of erroneous values on 472kc. I cannot tell you how much time I spent trying to adjust the output network to match a 50-ohm load with an SWR of 1:1 only to be able to plateau at 2:1! It was maddening but I did learn a lot and feel very confident that when my SWR meter says 2:1, it really means 1:1. In the future I will manage antenna matching through use of the SWR analyzer as well as antenna system current monitoring (more about this in future articles).
After working out the operational bugs of the low power amplifier, I started looking for a platform on which to build the 100-watt power amplifier. I was exhausted after working out the 25-watt model so my goal was to build the 100-watt version once and with minimal stomach acid. Fortunately, this was not an unreasonable request.
The 100-watt PA utilizes similar waveform shaping with a beefed up power transistor, more capacity on the RF capacitors and an IRF-540 MOSFET that had a higher drive requirement that the 25-watt version. Higher drive power is accomplished with a transistor-based current boost circuit, driving the gate of the IRF-540. Class-D is still accomplished via a shunt diode and leak resistor.
In the development of my transmitter, I chose to use slug tuned RF quality inductors. I was not going to make the mistake of winding my own again, which lacked the robustness of inductance variability.
The PA was built in an “island” configuration on top of a large heat sink, utilizing a variety of conductive pads in dead bug configuration.
Parts are layered on top of each other creating a “busy” appearance. The IRF-540 is seen in the center-left of the heat sink.
This configuration has offered good performance and like the IRF-510, the IRF-540 is very forgiving. Output waveforms look great on a scope and are more than permissible for on air use. Harmonic suppression seems to be on the order or 70-80 db and this is evident in the heat generated in the filter network. It’s a well-known fact that at these frequencies, it’s not uncommon to have as much if not more power in the harmonics as the fundamental. A nice-sized fan blowing across the output section solves this problem. Slug tuned inductors are the way to go and I have headroom on my power requirements to spare so losses incurred are not of concern as long as the PA does not catch on fire.
Note the nice rounded leading edge and fall-time of the waveform. This is optimized for click-free (harmonic free) CW
The ripple is from variations in the tank circuit and was resolved after these pictures were taken.
It might be a little triangular but it is a pretty nice looking sine wave and I will be happy to put I on the air.
So what’s next as far as transmitters are concerned? I dunno. I had visions of making modifications to a sideband generator so that the signal could be mixed appropriately for computer generated digital modes. The complication is that those modes really need to be operated in a linear mode and class-E, which is the mode this PA operates under, would generate all sorts of hash in and around the band. Maybe there is a class A or class AB in the cards next. Implementation of the mixing would be easier, in my honest opinion, than getting two MOSFETS to operate in a balanced manner for push-pull but maybe I am over thinking that right now. I will revisit this after the entire system is implemented and ready for air.
This project has been amazingly rewarding and I’ve learned a tremendous amount in the process. Thanks to Rog, GW3UEP, for his development and consultation time as well as the use of his schematics throughout this article. I could not have accomplished this without his guidance.
Next month I will detail the construction of the ATU and the implementation of the variometer/loading coil at the antenna feed point. I have a few parts to look for at Hamcom so I can finish up!
Its been a wild, awesome ride so far…
73 and see you in the pileup!