Didier has my 5650A notes on his Wiki... note in paticular the info on properly setting the oscillator frequency. I do have some DOS/WIN98 software that implements this.
Notes on Didier's WIKI
Mark's schematic sketch
23 April 2009
I forgot to mention that the parts on my oh so artfully drawn schematic are approximately in the same relative locations as the parts on the FEI DDS board.
Also, the values of the caps you use to bridge the coupling caps to extend the low end of the freq range determine how well the unit will handle low freqs. 100 uF work well and give good signals down to 10 Hz, usable to 1 Hz, but are probably too big to get under the cover. It is fairly easy to find some 10uF units that will allow the cover to be used, but limit the performance on the low end. I think there are some 35uF units out there that would fit. Since I was going to install it in a box with a micro, I don't worry about using the cover.
24 April 2009
I oriented all my added caps as shown on the drawing... The '+' side going to the DDS chip IOUT signals (+ side of all the added caps are physically nearest to the DDS chip, - side nearest the output cable).
The cap with the 1.2K resistor across it is the most important one for getting a low freq output. For small values of caps (10 uF?), the 1.2K resistor is probably not needed. For larger values (100 uF) it removes a lot of distortion from the signal. Did you bridge the correct cap? It is the one centered between the (not-installed) crystal leads. Did you use my polarity? Are your caps good?
You should be able to get a good signal down to at least 10 Hz. I suspect your 20 MHz signal looks like crap on a scope... mine does... I think 15 MHz is a practical upper limit for a clean waveform. You might be able to do better if you don't want a low freq output.
Also did you remove the termination resistor from the back of the DDS board? For this app you should be terminating at the end of your signal cable. You do need termination somewhere, otherwise the bias resistors to the comparator chip will raise the center of the signal above ground.
I check my modified unit. The drawing is wrong! The cap to bridge is the 0.01 uF one centered between the crystal filter legs. The 1.2K resistor is shown properly (across the .047 uF cap in the row of six components).
Yes, the cap to bridge is the 0.01uF at the crystal filter option point. You can't get much of an audio signal through 0.01uF. The crystal pins are shorted together with a jumper on the back of the board.
When set up properly, my audio distortion analyzer shows less than 0.1% (-60dB) distortion. I have not looked at the RF output on a spectrum analyzer (I just got one in a few weeks ago).