A little background: I was always fascinated by radio, audio, and electronics in general. When I was a 13 year old kid I got my novice license, call sign WV2CRJ. I put together my first station, a Heathkit DX-20 and Heathkit AR-3, built them up probably not perfectly following the instructions, strung a marginal dipole and set out on 40 meter CW. It was frustrating. I had 3 crystals, the frequencies I could operate were always busy, and few people could hear me anyway.
I plugged away at the code so I could get my general class license and be done with CW, finally managed that, became WA2CRJ, graduated to a DX-40 with a VFO, and was able to check into local nets. The station was still zero-budget, the antenna poor, and I usually got poor signal and audio reports. The receiver was deaf and drifted all over the place. So did the VFO. I lost interest, let my license lapse and was done with amateur radio for decades. I’m sad that I don’t seem to have any photos of that station. The pictures would have been better than the experience ;-)
In college I became involved in the carrier current AM campus radio station. I designed and built a bunch of low-power transmitters to feed the multitude of dorms, learned a bunch of things not to do with radio, and flunked out.
In 1963 I got my first-class radiotelephone license and went to work at WHAM, the 50kW blowtorch in Rochester for a Summer job. That was a real radio station. They had a real transmitter. A 1943 vintage Westinghouse 50-HG1. 11,000 Volts at 6 Amps on the finals, high level modulated. The antenna was a 440 foot tall vertical with a bunch of radials. I got to do maintenance on that stuff and loved it! Even so, I was always bothered by our audio quality and did a lot of work trying to improve it.
These experiences left me with a fascination for good quality AM. Several times I thought about getting my amateur ticket back but I couldn’t bring myself to grind away at getting my code skills (the lack thereof) back When the FCC finally dropped the code requirement I jumped on the opportunity. I found the written exams easy and passed the three elements for extra-class with flying colors. So now I’m a new-old amateur with a new call, KC2RLQ. I checked and found that I could get the old call back as a "vanity" but I like the new one. It also makes me seem younger and what "vanity" is there in making yourself seem old? ;-)
"KC2 Radio Low Quality" :-) a good goof!
When I first put KC2RLQ on the air in 2007 I started with a Johnson Viking-II transmitter. I quickly changed from using its own modulation to an external setup based on an Ashly MOS-FET power amplifier and this sounded good.
After a while I wanted some more power but really didn't want to get into huge transformers, big expensive tubes, and dangerous voltages. Most of my time was spent on 75 meters and some of the class-E mos-fet transmitters I'd read about seemed promising. When I heard these rigs on the air they generally sounded excellent. So I started building and after a couple of years of trial and error and letting the smoke out of a few parts I had a working transmitter.
This project would have been much more difficult if Steve Cloutier, WA1QIX had not already blazed many of the trails. Steve is obviously proud of his work and is very generous in sharing it. They say you can always tell the pioneers, they're the ones with the arrows in the back. Steve saved me from a lot of arrows and I'm very grateful. Thank you!
In designing the KC2RLQ class-E transmitter I struck out in my own direction in several areas:
The pre-driver uses a pair of comparators driven by the 3 Volt sine wave from my digital VFO. One comparator switches off before the other switches on so the two comparators give about a 40% duty cycle square wave to feed the FET drivers. With no VFO signal the pre-drivers are both always off. (they better be!) The duty cycle of one of the comparators is slightly adjustable, when you look at the transmitter on a spectrum analyzer you can adjust this to null the even harmonics. The comparators feed a pair of 74HC365 high current hex CMOS buffers with all 6 gates in parallel to feed the FET drivers. Its a very simple scheme and its very stable.
I like to use an individual shunt capacitor for each of the FET's and mount it right at the FET. This minimizes effects of lead length which are very significant in a low voltage-high current design such as this. This costs a little more but I like it.
There are no relays. The power supply is always on and the T-R switching is done by the pre-driver enable circuit. I have a delay circuit using a basic PIC microcontroller chip to allow my antenna relay to settle before applying RF power to it. Some day I hope to also use the PIC to include some fault protection.
Most class-E transmitters use either a class-H analog or a PWM modulator. Either of these gives good results but being lazy I do it the old fashioned way. I used a big inductor (actually a 1500W surplus Ashly power transformer) to feed the DC to the RF deck and just capacitor couple the same Ashly power amplifier into it that I was using for the Viking. This sounds fine and given the relatively low average power of speech its really quite efficient if a little low-tech. Its pretty standard high-level modulation like broadcast transmitters have used for years but because the audio amplifier is a good match to the RF deck there's no modulation transformer needed!
The power supply is a pair of 24V 10A switching supplies in series. With this configuration I can switch between 24V for tuneup and 48V for full power operation. An additional 12V 4A switcher is used to power the drivers. These supplies weren't expensive but they don't seem to generate any significant RF noise.
Update: 12/10/2015 - The RF deck has been updated to use (4) ST Micro STW15N80K5 MosFets instead of the former (8) Fairchild FQA11N90 parts. The ST's are much easier to drive and have lower on resistance. So far, so good with these parts. One of these days I hope to put the rest of the case around it. In the mean time, please *DON'T* touch the left end of the copper inductor! You'd get a nasty RF burn!
The transmitter currently puts out over 400 Watts if I want it to, I'm running it at about 350 Watts which gives me 1500 Watt peak power. (legal limit)
I'm currently using a home-brew software audio processor based on "Reaper," a great audio workstation with a built-in DSP language. This does a bunch of things; overall eq, an 8th order 50Hz highpass filter, a brickwall 5kHz lowpass filter, and a 3-band RMS compressor. The bandsplitter for the 3-band compressor adds 720 degrees of phase rotation which greatly improves the symmetry of voice signals. The symmetrical signal is much easier on some detectors that don't do well with big positive peaks. The processor is followed by Reaper's built-in "ReaLimit" peak limiter plugin. This is a very transparent look-ahead limiter which absolutely limits peaks to make overmodulation impossible. Now the audio hardware is simply an old IBM laptop running Windows XP, an old Firewire interface retired from recording work, and the Ashly power amplifier.
The original Viking II modulation did rather poorly on both counts. The RF deck was not linear. A trapezoid test showed a noticeable kink in the slope at around 20% carrier. The modulator itself had poor frequency response and high distortion and both of these problems had several causes.
One problem with the RF deck was that Johnson derived the screen voltage for the 6146’s from the output of the modulation transformer fed through a 20k power resistor. A Canadian amateur, Paul, VE7KHZ (gotta love the call!) had made the same observation and moved the input to the 20k resistor to the B+ so it had no modulation on it. He noted the greatly improved linearity but also noted that he had to run the finals at a bit lower plate current to get good positive modulation. I found a big improvement could be made by using two 40k power resistors to feed the screens, one from the original connection to the modulation transformer output and the other to the B+ supply. This gives some modulation to the screens but keeps out of the non-linear region below about 100V. The result is a nice linear trapezoid while retaining good positive modulation. Paul took some nice scope pictures of the trapezoid patterns. The pattern from his modification looks just like mine. His photos are reproduced here with his permission. Please check out his interesting website http://www.qsl.net/ve7khz/.
Fig 1. Screen fed from modulator (note the distortion at about 15% carrier)
Fig 2. Screen fed 50% from B+ and 50% from modulator (much more linear!)
Fig 3. Schematic of modified screen supply
The performance of the original Viking modulator itself was pretty bad. Initially it wouldn’t even make 100% modulation. This was due to a .01uF ceramic capacitor across the 807 modulator tube plates which was breaking down. Removing it let the modulator go to 100% but it still had terrible frequency response (500-2.5kHz+/-6dB) and about 8% distortion, not exactly hi-fi.
Looking at the multitude of design problems with the modulator and in view of the fact that I have lots of audio equipment including commercial power amplifiers around I decided to bypass the internal modulator entirely and use an Ashly power amplifier combined with some Ashly processing gear instead. (When you own an audio manufacturer you get a good deal on left-over stuff ;-) )
The Ashly CFT-1800 amplifier is a stereo MOS-FET design which is very rugged and produces great audio at about 175W/channel into 8 Ohms. (more than we need for a 150W plate power input transmitter) If you operate it in bridged mode (the channels in opposite polarity with the output taken from the 2 hot terminals) you can get about 350W at 70 Volts of audio from it into 16 Ohms. A step-up transformer is necessary to provide the necessary modulation voltage for the Viking.
I tried a bunch of different transformers I had lying around and got great results from toroidal power transformers for a small Ashly amplifier. These transformers have a 120-240V primary and a 52VCT secondary. Feeding 3 of these transformers in parallel at the 52V winding and then hooking the 3 240V windings in series gave me about a 1:14 Voltage stepup with good frequency response from 40Hz-10kHz and very low distortion. It also allowed enough headroom to allow about 150% positive modulation. I know from experience that you can’t have DC current in a continuous core toroid so I capacitor coupled the toroid into the Viking using the original modulation transformer as a reactor. Although power transformers are designed to operate at 50-60Hz many of them have plenty of high frequency response for AM audio.
Now the audio was much nicer but I still had a couple of issues. The bypass capacitors (about .02uf total) on the RF deck caused the modulation transformer and reactor to resonate at about 8kHz causing about an 8dB peak at 8kHz. I changed those caps to something much smaller for a total of about .004uF and the response was much flatter. There was still a little high frequency boost so I added a passive RC filter at the power amp input to counteract that.
At 10% modulation the transmitter was now flat within 1dB from 40Hz-10kHz. However if I tried to run 100% modulation at 40Hz the reactor (original modulation transformer) made awful noises, the Ashly amplifier clipped, and eventually blew an output fuse! The old modulation transformer was being saturated by the DC current to the RF deck and just didn’t have enough core to handle low frequencies. It was pretty much a dead short at 40Hz and although the amplifier tried valiantly to force a low frequency AC voltage on it, at higher levels the audio current was just too high. I ultimately put a 10Hy 250mA choke in to replace the original modulation transformer and now 100% modulation at 40Hz was easy. At 90% modulation and 1kHz THD was below 1%.. Some commercial broadcast stations do worse ;-)
One thing that makes this all work so well is that the power amplifier is a "voltage source." This means that it has an effective output impedance of essentially zero until it clips. The small signal impedance on the secondary of the toroids is about 120 Ohms at 1kHz. and this is feeding a 3000 Ohm load. (the RF deck) Essentially the amplifier forces the output to track its input even with a non-linear load like a saturating choke.
Fig 4. - 3 toroid power xfmrs used as a step-up modulation transformer
Fig. 5 - KC2RLQ Audio chain block diagram
Fig.6 - Full-wave limiter detector with adjustable symmetry
Fig. 7 - the (2) 40k 25W power resistors replacing the original 20k screen resistor and replacement screen bypass cap. The AM/CW switch made it very convenient to mount these.
Fig 8 - (6) 68uF 400V electrolytics glued onto the chassis. 4 of these replace 2 very dried out original dual filter caps. The pair closest to the choke in the top of the photo are series connected to pass the audio from the modulation transformer to the reactor.
Fig. 9 - The 10Hy 250mA choke (modulation reactor) replacing the original modulation transformer
Fig. 10 - Closeup of the audio gear. From top to bottom:
1. DDS VFO
2. Mic preamp/mixer
3. Crossover used as tunable high and low pass filter
4. Parametric eq-Limiter combo (The limiter detector was modified for asymetry
(100% negative, 150% positive modulation))
5. Antique Simpson VU meter adding to the "broadcast look" ;-)
6. The CFT-1800 power amplifier
All the audio gear is standard Ashly Audio product. The DDS VFO is from a kit. It was designed by Craig, AA0ZZ, we made the panel for it at Ashly but it is a one-off, not an Ashly product. I really like that VFO, it lets me keep my transmit frequency spot-on! The plywood rack is left over from some rock-n-roll sound system and it sits on its rolling road case.
Info on the Ashly products: (Adobe PDF
format) 
MX-206 mixer - (the unit in the rack is its predecessor, the MM-106 - the 206 is stereo)
XR-1001 stereo 2-way/mono 3-way crossover
DPX-200 parametric eq / limiter
FTX-1501 stereo mos-fet power amplifier (actually I'm using the CFT-1800, same thing, convection cooled)
Fig. 11 - The audio/VFO rack from the rear
Fig. 12 - closeup of the modulation transformer made from 3 toroidal power transformers (pardon the plywood ;-) )
Fig. 13 - KC2RLQ headquarters (Feb, 2008)
Audio: This recording was made from a diode detector on the dummy load. When I'm actually on the air I restrict the bandwidth to about 4kHz. However for this recording I opened the bandwidth to a full 20kHz and turned off the high frequency preemphasis so you can hear the natural and unrestricted high-end of the transmitter.
Fig. 14 - Display of audio from above recording (Note about 4dB greater positive peaks relative to negative due to the asymmetry of my voice - that's about 160% positive modulation! - polarity is important on AM)
Epilog:
This station is somewhat a work in progress. I would like to build up a class-E RF deck sometime and play with different ways of modulating that. However, I've really enjoyed the tube technology in the Viking and may build up a higher power tube RF deck. If so, I'll probably modulate that the same way I'm doing with the Viking now. Another project on the list is a decent 160 meter antenna. For the time being I'm having a great time on 75 meter AM and this station has exceeded my expectations.
The Drake R8A receiver is in the corner of the bench. It feeds an Ashly SRA-120 amplifier driving an original Advent loudspeaker. You can tell good audio from bad with this setup. The only non-American component is the Hitachi oscilloscope which monitors both transmitted and received audio. The IBM laptop lets me check the internet and record and play back audio. This setup is giving me very nice audio, it measures good and sounds good. I'll update this page as I update the station.
-73 and see you on 75 meter AM!
--Bill KC2RLQ