[time-nuts] DMTD using Active Mixers

[Bruce Griffiths]

Mario Sanchez wrote:
> Hi Bruce...
>
> Thanks for your answer...
>
> I have a couple of questions:
>
> 1. I am using a 1kHz beat frequency, and measure back2back period using a
> Pendulum CNT-91 high resolution frequency counter. Does the DC stability of
> the mixer still matter?
>
> 2. I have been comparing the AD831 with some passive mixer, but i can not
> see any great improvement. I am basically following this procedure:
> a. As I just want to compare the mixers, using only one of them I mix 10MHz
> from a Rb with 10MHz+1kHz generated by a R&S signal generator referenced to
> the same Rb.
> For passive mixer, I amplify before the mixer, I have also tested different
> input levels at both LO and RF signal.
> b. Filter the high component.
> c. ZCD - Hard limitter, using Collins-type amp.
> d. Calculate ADEV at 1s.
>
> Any approach on how to test the mixer?
>
>
> My goal is to reach an accuracy of 5E-14 @ 1s
>
>
> Regards,
> Mario
>
>
>    
Mario

1) dc drift (referred to the mixer inputs) does affect switching 
thresholds or equivalently the mixer phase shift.

2) To characterise the mixer phase noise simply drive its inputs (LO 
+RF) in phase quadrature with the same signal and measure the output 
noise (after the low pass filter) using a spectrum analyser (eg sound 
card plus high gain low noise preamp - I have some tested low noise 
amplifier designs that may be useful). Then measure the mixer phase 
sensitivity by driving one input at a frequency offset from the other 
input by 1kHz and measure the zero crossing slope of the beat frequency 
signal.
Then calculate the mixer phase noise spectrum from these results.
Typically for a 10534B or similar mixer the output noise is only a few 
nv/rtHz @ 100Hz.

If you don't know the mixer output noise spectral density then its not 
possible to optimise a Collins style limiter.
The higher the mixer output noise spectral density the lower the gain 
and bandwidth of the first stages in an optimised Collins style limiter.

Your tests may well be limited by the phase noise of the 2 sources you 
are using and not either mixer.

Its more important to have a low phase noise source driving the mixers 
in phase quadrature than a low drift one when testing.
A good crystal oscillator may be quieter than your rubidium source.

Bruce