Bob, you are absolutely right. The preselctor filter is absolutely necessary.
--- In [email protected], Bob Camp <li...@...> wrote: > > Hi > > The only "gotcha" is that in the upper octave, the DDS can't produce falling > edges that are 90 degrees apart. It can indeed produce sampled waveforms that > will eventually re-construct to the needed signals. Filtering prior to > slicing is important. > > ------------------------------- > > With no filtering a DDS will indeed produce a *lot* of output frequencies. > With a 75 MHz clock, tuned to 30 MHz you also get (for no added charge) 45, > 105, 120 MHz and so on. I confess to having feed the first one I built into a > spectrum analyzer to check that out. Feed the full output, unfiltered into > something like a diode ring mixer and you will down convert them all pretty > much equally well. > > Bob > > > On Jan 24, 2010, at 8:26 PM, ni9n wrote: > > > > > You guys had me worried for a few minutes, there! Had to test it on 10 > > meters just to make sure. The two synched AD9834's do indeed produce > > signlas 90 degrees apart, even at frequencies above 1/4 of the master clock. > > > > I think some of you are having a little trouble understanding bandpass > > sampling. Nyquist doesn't really say the signal has to be bandlimited to > > less than half the sample rate, that's just the case for a "baseband" or > > lowpass signal, such as sampled audio. If you sample a bandpass signal, the > > signal must have BANDWIDTH less than the sample rate. > > > > If a baseband voice signal is limited to the usual 3 kHz (that is, no > > energy to speak of above 3 kHz) it has a two-sided spectrum from -3 kHz to > > +3 kHz. When it is translated up to a carrier (suppressed or not) > > frequency of, say, 10 MHz as double sideband, it has a bandwidth of 6 kHz > > because both sides of the two-sided spectrum are translated together. If > > you pass the RF through a brick-wall filter centered at 10 MHz and having a > > bandwidth of 6 kHz, you can use a sample rate of 10 kHz and theoretically > > lose nothing. This is because the bandpass antialiasing filter (that is, > > the brick wall filter) rejects the image responses at 9.990 and 10.010 MHz. > > As a matter of fact, without the bandpass filter you would have an image > > response every 10 kHz, but the bandpass filter rejects all the unwanted > > ones. > > > > The symmetry of the sampling waveforms (the I and Q LOs) does not matter, > > their duty cycle can be anything you like. All that matters is that the > > falling edges are 90 degrees apart. > > > > By the way, what does the "AW:AW:AW:..." mean? > > > > --- In [email protected], Bob Camp <lists@> wrote: > > > > > > Hi > > > > > > A full wave rectifier can also be used as a simple broad band doubler. > > > > > > The problem with the output of the DDS is that without filtering it is > > > far from symmetrical in the upper octave. Some use bandpass filtering to > > > get around the problem, but that usually is not an option on a wide band > > > radio. > > > > > > Bob > > > > > > > > > On Jan 24, 2010, at 5:08 PM, Rotten Robbie wrote: > > > > > > > For this application Nyquist does not matter. It is just used as a > > > > clock.And > > > > what you need for the detector is a square(rectangular) waveform. > > > > > > > > If the waveform is reasonably symetrical you can use both edges for > > > > frequency doubling. It just takes a few inverters for a time delay and > > > > an > > > > exclusive OR. > > > > > > > > Bob Macklin > > > > K5MYJ > > > > Seattle, Wa. > > > > "Real Radios Glow In The Dark" > > > > ----- Original Message ----- > > > > From: "Dave Wade" <g4ugm@> > > > > To: <[email protected]> > > > > Sent: Sunday, January 24, 2010 1:56 PM > > > > Subject: Re: AW: AW: AW: AW: [soft_radio] Re: LD-1 Discussion on > > > > Garage-shoppe.com Blog > > > > > > > > > > > > > > > > > > > k5nwa wrote: > > > > >> At 12:04 PM 1/24/2010, you wrote: > > > > >>> > > > > >>> Thanks for your explanations, Bob! > > > > >>> > > > > >>> I understand the principles of driving the mixer with a squarewave > > > > >>> signal and that you use a comparator to change sine to squarewave. > > > > >>> > > > > >>> Where I'm failing at is the trick how to receive 30 mhz with a DDS > > > > >>> that make 37 MHz max without getting problems with nyquist. Can you > > > > >>> use a comparator to double your VCO frequency without getting > > > > >>> problems at the duty cycle? > > > > >>> > > > > >>> > > > > >>> Stephan > > > > >>> > > > > >> > > > > >> This is a mixer, you are not trying to digitize the 30MHz signal > > > > >> where you would need a minimum of two samples, instead you are mixing > > > > >> it down to the base band. > > > > > > > > > > No but the DDS chip is synthesising a sine wave from samples, and > > > > > Nyquist bite both ways. However some DDS chips (I don't know about > > > > > this > > > > > one have on-board frequency multipliers... > > > > > > > > > > > > > > >> > > > > >> > > > > >> Cecil > > > > >> k5nwa > > > > >> www.softrockradio.org www.qrpradio.com > > > > >> < http://parts.softrockradio.org/ > > > > > >> > > > > >> Never take life seriously. Nobody gets out alive anyway. > > > > >> > > > > >> > > > > >> > > > > >> ------------------------------------ > > > > >> > > > > >> Yahoo! Groups Links > > > > >> > > > > >> > > > > >> > > > > >> > > > > > > > > > > > > > > > ------------------------------------ > > > > > > > > > > Yahoo! Groups Links > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >
