[time-nuts] DMTD Question
bruce.griffiths at xtra.co.nz
Mon Jan 25 21:41:19 UTC 2010
Bill Hawkins wrote:
> I'm not learning anything from this thread because I don't know the
> purpose of DMTD or how it would produce a beat note for a phase error.
> Now, DTMF I understand.
> Bill Hawkins
A dual mixer time difference (DMTD) system for 10MHz with a 10Hz beat
note frequency would use a pair of mixers with a common LO frequency of
say 9.99999 MHz.
The mixer RF input signal for each mixer would be nominally 10.000000MHz.
Each 10MHz frequency to be compared is connected to the RF input of one
The mixer If output is low pass filtered to eliminate the unwanted sum
frequency and other spurious products.
The resultant mixer output is nominally 10Hz.
The phase shift between the 2 10Hz beat frequency outputs is equal to
the phase shift between the 2 10MHz signals being compared.
Measuring the time interval between the zero crossing of one beat
frequency output signal and the corresponding zero crossing of the other
beat frequency output is to first order independent of the phase (and
phase fluctuations) of the common LO signal.
Its much easier using classical signal processing techniques to measure
such phase differences at 10Hz than at 10MHz.
However classical DMTD systems suffer from aliasing as the zero crossing
detectors only sample the phase once per cycle and the low pass filter
has a bandwidth greater than the Nyquist limit (otherwise the amplitude
of the beat frequency at the filter output would be very small). If the
noise is white phase noise then aliasing doesnt produce any artifacts.
If the phase noise is highly coloured the aliasing artifacts will also
More information about the time-nuts