[time-nuts] DMTD phase shifter

Bruce Griffiths bruce.griffiths at xtra.co.nz
Sat Jul 25 03:29:45 UTC 2009


EWKehren at aol.com wrote:
> On the chance that I may be totally wrong let me suggest the following. In  
> my opinion you are looking at the wrong information. When you mix a 10 MHz  
> signal with a signal offset by 1 Hz you are subtracting from a 10 E 7 
> signal 10  E 7 and as a result get 1 Hz representing 10 E 7. Counting it with a 
> counter  with 1 ns resolution, you in effect get at the last digit 10 E 16. 
> Obviously  some of the last digits are irrelevant due to noise and trigger 
> error. Using two  mixers with the common offset oscillator you want the two 
> phases to be as close  as possible to reduce the noise contribution of the 
> offset oscillator. That is  why you want the phase shift. Measuring the time 
> difference between the zero  crossings in my opinion is not the way to go.
> I am inspired by the simplicity of the NBS unit (thank you Corby for the  
> info) to lay out a PC Board; reading some of the comments I would appreciate 
> any  links to more info on that subject that may make the design more 
> effective. Any  help would be greatly appreciate. 
> Bert Kehren  WB5MZJ
> **************A Good Credit Score is 700 or Above. See yours in just 2 easy 
> steps! 
> (http://pr.atwola.com/promoclk/100126575x1222377105x1201454426/aol?redir=http://www.freecreditreport.com/pm/default.aspx?sc=668072&hmpgID=115&bcd
> =JulystepsfooterNO115)
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>
>   
Bert

With a classical dual mixer system, the only useful measurement is the
time difference between the zero crossings of the 2 beat frequencies.
The limiters used distort the signal and amplify the zero crossing slew
rate, thereby precluding meaningful measures of anything else but the
zero crossing times.
Time stamping the zero crossing times for each channel may be more
useful rather than just measuring the time differences.

The zero crossing circuit employed by NIST is far from optimum (Oliver
Collins paper on the design of low jitter hard limiters wasn't published
until May 1996), some of the JPL designs are better but are still some
way from optimum.
The timing jitter and phase drift of the ZCD output limits performance.
JPL achieved a zero crossing jitter of around 40ns in their 1Hz beat
frequency ZCDs, it should be possible to reduce this by a factor of 3 or so.
There later DMTD systems use a beat frequency of around 100Hz or so with
a 100MHz mixer input frequency.
The zero crossing detector filter component tempcos will limit the phase
shift stability for larger tau as will the mixer phase shift tempco (~
ps/C).
It is also essential to avoid low frequency ground loops that affect the
2 zero crossing detectors.
The mixer IF port grounds need to be low frequency isolated from the RF
grounds of the other 2 ports.
One needs to know the phase detector output noise in the flicker region
in order to optimise the zero crossing detector design.
Unfortunately phase detector flicker noise specifications don't appear
on the datasheet so one has to measure it.
One can do this using a low noise preamp for a sound card or a spectrum
analyser.
Using an apprpriate mixer IF port termination (a 50 ohm or even 500 ohm
resistive termination may be far from optimum) will affect the beat
frequency signal jitter.

If one eliminates the limiters (however the distortion probably needs to
be kept down) then, in principle a COSTAS receiver could be used to
measure the phase shift between the beat frequencies independently of
the individual beat frequencies and proximity to a beat frequency zero
crossing. However ADC noise will limit the performance unless one uses 2
or more ADCs per beat frequency and uses correlation techniques to
eliminate the ADC noise.

Bruce




More information about the time-nuts mailing list