Thanks to Leon Heller for the DDS PCB, which made soldering the tiny AD9851 much easier. I did not fit the MMIC amplifier, or the 125MHz canned oscillator: the reference oscillator is the 10.000MHz OCXO. I also did not use surface mount components (apart from the AD9851) as the board was designed for. To add insult to injury I also wired up the AD9851's internal comparator such that I have both a CMOS-logic level squarewave output, and 50-ohm analog output. I used a 14MHz 7-pole low pass filter.
I did not use a microprocessor, PIC or whatever. Instead I connected 24 DIP switches to the 24 most significant programming bits of the 32 available. The least significant switch has a resolution of 3.6Hz. These are in banks of 8 bits, OR'ed using diodes, and sequenced into the AD9851 by a 74HC4017 counter IC. The 6'th 74HC4017 output toggles the AD9851 frequency load pin and resets the 74HC4017 ready for the next load. A 74HC04 functions as RC oscillator on about 8KHz, which clocks the 74HC4017. This produces about 1,300 frequency updates per second. Practically as soon as you change a DIP switch setting, the frequency updates. The third IC is a 74HC08 which is used to gate 4 external input bits onto the AD9851 frequency control, giving an externally controllable amount of shift on the 0.9, 1.8, 3.6 and 7.2Hz resolution bits.
Here's the original handdrawn circuit diagram.
The OCXO has an adjustment, which allows about 280Hz range on the 10.000MHz output.
Here's a nice computer-drawn diagram prepared by Ingo DK3RED for publication in the German QRP magasine QRP Report.
In the foreground of the picture, to the right of the DDS board, is my 30m direct conversion receiver. I wired the mixer to the DDS instead of the receiver's VFO, and set the DDS for 9996KHz. The intention was to adjust the OCXO for zero beat against Moscow RWM's time signal on 9996KHz. Unfortunately the 10.000MHz OCXO signal itself was very loud in the receiver. At 4KHz shift, it was attenuated considerably by the low pass audio filters in my HF Receiver, shown in the background and functioning as the audio stages of the receiver. The audio response of the receiver below 50Hz is not good either (by design!). Therefore I was unable to hear RWM properly in the final few decades of Hz as I approached zero beat.
Here's a close-up of the DDS on top of the 30m beacon cobtroller, which is operational on 10,140,040Hz.
DDS by Armand ON4TL
Johan ON5EX sent me this wonderful photo and said: "Armand ON4TL (in his eighties) has succeeded in designing a 'no-processor' 9851 DDS VFO, inspired by your project on your website. He showed me a working DIP-switch sample last Friday. It is a charm."