Hi Maybe this is where the disconnect is happening:
The DDS output is essentially an analog signal rather than a digital signal. You have to take the output of the DDS D/A and run it through a low pass filter to make it look anything like what you want. Once you have the sine wave from the output of the lowpass filter you put that into a comparator. The digital output of the comparator drives the digital mixers. Put another way: If you look at a DDS output with a scope, it looks like a sine wave when the sample clock is very high compared to the output frequency. When you do the same thing and the output is > sample clock / 6, the output looks like garbage. To be useful, it needs to look like a sine wave. The low pass filter is very important in this respect. Phase shift of the signal is limited by the size of the accumulator. Because of the way things step with a DDS, it is actually putting out a phase staircase that averages to the phase you want. Again, the lowpass filter is your friend. It does not take care of all the spurs, but it at least makes the output usable. Just like with the other stuff, you do get a lot better performance at 1 MHz than at 30 with a 75 MHz clock. All that said, the radio will work a lot better at < 10 MHz than it will at > 30 MHz. That's not a knock on this design, it's just the nature of the beast. Hope that helps. Bob On Jan 24, 2010, at 11:37 AM, Stephan Schaa wrote: > > Hi Bob! > > I'm not a native english speaker, thats why I maybe not understand what you > ment?! > > > From what I know is: > > - you need (normally) double sampling rate from that frequency that you want > to sample (Nyquist, 60 MHz sampling rate for 30 mhz frequency) > > - in addition to the above said for a I/Q mixer you need two VCO frequencys > of the same frequency with a 90 degree shift. > (this is sometimes created by taking a frequency 4 times high than the rx > frequency and put it into a johnson counter to get, but there are sure > different ways of doing this, 2 DDS outputs with the shift , ... ) > > - you can do a trick in order if you haven't got a that high going frequency > clock that fits the nyquist frequency: take a single balanced mixer which > doen't attenuate lower harmonics that much and use the rx for that > frequencies. the problem is that in this case you are loosing some > information from the real signal because you are sampling only every second > time of the sine wave. > > > Stephan > >
