One of my motivations for building and Arduino controlled DDS-60 was to use it as the VFO for a QRP transmitter. I have had success with this by using the RF stages of the Universal QRP Transmitter Mk-II by Wes Hayward W7ZOI as described in QST April 2006. Click here to download the article for free from the ARRL.
I got an 80 meter version going first using the crystal controlled oscillator before I had the DDS. I bought the "bag of parts" from Kanga US. That was an easy way to get started although I didn't use it all because I haven't built the key shaping or sidetone circuitry. More about this later. It worked well. Note that Wes has posted an errata document on his web site with some important modifications to the oscillator especially for 80 and 40 meters.
Unfortunately the 2SC5739 transistor has been discontinued by Panasonic. A good substitute is the BD139. This was a suggestion from Wes in response to a question I posted on the EMRFD Yahoo Group / mailing list. I can confirm that the BD139 works very well for both Q2 and Q3, at least on 80 meters, and seems to be in plentiful supply at Mouser for 30 cents. It's in a TO-126 plastic package so no insulation is required when it's mounted. The Philips data sheet shows a typical fT of 190 MHz so it should be good for most of the HF spectrum. It seems like a perfect choice. Note that if you mount it in the usual way with the writing and orientation dot upwards, the BD139 pin connections are the reverse of the 2SC5759. With the wires pointing down, left to right, the BD139 is emitter, collector, base.
Wes commented that bipolar switching transistors suitable for RF amplifiers seem to be slowly but surely disappearing from manufacturers' catalogs and being replaced by FETs. Let's hope the BD139 stays around for a while. I think a few suppliers such as and Kits and Parts still have some 2SC5739 and Kanga is still selling the bag of parts so I guess Bill has some too but I didn't feel comfortable building something new and experimenting with (and potentially destroying) a device that is no longer being made.
Replacing the crystal oscillator with the DDS is very simple. I've thrown away the oscillator and T1, i.e, everything to the left of point X on the schematic and simply connected the DDS output to point X at the top of the R17 trimpot. My R17 is 500 ohm and I seem to have lost the recommended 270 ohm resistor in parallel but it works well. The output level trimpot on the DDS-60 is set to a resistance of about 100 ohms. This gives an output of about 3 v p-p which is enough to comfortably drive the transmitter. R17 adjusts the power output smoothly up and down. The final result is a nice simple, proven circuit with no tuned circuits to adjust.
The BD139 final is bolted to the PC board with no heatsink. It gets a little hot if I key down for a while but it seems to stay at a reasonable temperature during normal CW contacts. A small heatsink would probably be a good idea.
My operation is rather simplistic at this point. I wanted to prove the RF side of it first so I only have the circuitry for Q2 (driver) and Q3 (final) and Q4 (keying) but that's enough for me to operate quite easily. I use my Kenwood TS430S for receiving. The T/R relay is a double pole double throw as in Wes' design but I use both parts of it for antenna switching. The other half of the relay is wired so that the receiver input is grounded on transmit. It didn't want any risk of blowing up the front end of the TS430S. A front panel toggle switch acts as a T/R switch by switching power to the transmitter and relay on and off. It's not quite as convenient as semi or full break-in keying but I find it easy to use for casual QSOs. When on receive I hit button 2 on the DDS to put it in Quiet mode so that I don't hear it in my receiver. I didn't build the sidetone generator either. On transmit I simply listen to my own signal on the receiver. With the antenna input grounded, its just a good strong signal. I hear some backwave between dits and dahs but I don't find it a problem.
If I crank up the R17 drive control, I can get the power output up to just over 6 watts into a 50 ohm dummy load. I leave it at about 4 watts. I measure the power with an oscilloscope setting it to about 40 v p-p which is about 4 watts. This corresponds quite closely to Wes' observed 9.4 volts DC on the diode rectifier / RF probe at the output so that gives me some confidence that my newly acquired second hand oscilloscope is reasonably accurate.
It's built very roughly and was mostly intended as a proof of concept but it's fun to operate and I've made quite a few contacts. My next project is to build another one nicely in the same sort of metal box as the DDS. I think I'll try 40 meters this time. I will do something to enable semi break-in keying. I'll eventually have to do something about sidetone because my next project after that might be a matching direct conversion receiver. I'm not sure if I'll do the sidetone with an oscillator like the original design or try to get the Arduino in the DDS to produce sidetone.
The RF circuitry is so simple with only two transistors that it should be reasonably easy to duplicate it several times in one box to make a multiband transmitter. Lots of possibilities.