[time-nuts] Phase shifter circuit for DTMD

Magnus Danielson magnus at rubidium.dyndns.org
Fri Jul 24 14:42:39 UTC 2009


Lux, James P (337C) wrote:
> 
> 
> On 7/24/09 6:32 AM, "Stephan Sandenbergh" <stephan at rrsg.ee.uct.ac.za> wrote:
> 
>> Hi,
>> Many sources and papers refer to the actual DMTD mixer system (and
>> zero-crossing detector etc.). But, I've seen very little on the phase
>> shifter circuits, that are used in conjunction with this system.
>>
>> I was hoping some of you might have experience, and perhaps a circuit
>> example. If it is similar to what is used in audio applications, the single
>> opamp, 0 - 180deg, RC-all pass filter, it is simple. However, I assume there
>> are more to it than that. It seems such is the case with all precision time
>> related things :)
>>
> 
> For RF in coax, it's probably more like a mechanical line stretcher or
> trombone line. For 10 MHz there are some variable transformer sorts of
> schemes.
> 
> One can also do a 90 degree hybrid with a variable power combiner on the
> output, or, essentially the equivalent, a I/Q or vector modulator. There are
> also "linear phase modulators" based on diodes. With anything using an
> active device, one is going to worry about the usual things: noise, aging,
> etc.
> 
> 
> All of the "electronic"  phase shifters will require some form of
> calibration to get better than a few degrees accuracy. In some measurement
> schemes you don't really care what the phase shift is, just as long as you
> can get it to quadrature.
> 
> For the DMTD, aren't you really looking at timing of the beat notes, not the
> absolute phase? 

For DMTD you use a local oscillator which is detuned to produce a beat 
frequency of say 10 Hz. That means that you will scan through all phase 
relationships 10 times every second. For 10 MHz that means about 1 
milion different phase-relationships. Now, the phase relationships 
producing a beat voltage going through zero will choose themselfs. Thus, 
no need for a phase-shifter when doing DTMD as such. However, it can be 
wise to shift the second input such that it does not occur neat the same 
time, since that way time-correlation could be reduced. However, such 
phase-steps does not need to be made in linear steps, but rather crude 
steps approximating 0, 90, 180 and 270 degrees will suffice.. however 
other means to achieve the same thing can be used to create the separation.

Cheers,
Magnus




More information about the Time-nuts_lists.febo.com mailing list