Erik, Coherence is not the issue. Sampling at the same time (coincidence) is! For instance with a 1Hz offset if you have 5Mhz Ref and 5.00000000001 DUT Your TIC will read close to zero if they are near coincidence. It they are at the limit just before wrapping you will read close to full scale .999999999. The offset oscillator noise only gets cancelled at near to zero coincidence. The further away the more offset oscillator noise affects the measurement. You can use a really good oscillator for the offset oscillator and that will help. I use an FTS 1200. To see these effects take a look at this post. Go down to the 16th and 17th from the top for some realtime results showing this effect. https://www.eevblog.com/forum/metrology/dmtd-tutorial/ In my first DMTD system https://nvlpubs.nist.gov/nistpubs/Legacy/IR/nbsir75-827.pdf There is a variable phase shifter in one channel to allow adjusting to coincidence. However it's easier for me to adjust the REF or DUT slightly until coincidence is achieved and the frequencies are close enough so that the measurement counts are stable or drifting very slowly. Cheers, Corby

Corby, Thanks for the links. The tinyDMTD I'm using does not use a TIC after the mixer to measure zero crossing but samples the whole output of the mixers and measures the phase by I/Q down converting to zero Hz and calculating the phase of the complex signal. Previously (earlier version of the tinyDMTD) I use two signals at 10MHz and 0.01Hz apart, measured the phase over a 310 s period to check if the phase difference of the input signals had an impact see: http://athome.kaashoek.com/time-nuts/DMTD/Phase_difference.png and calculated the unwrapped linear residue see: http://athome.kaashoek.com/time-nuts/DMTD/Phase_difference_residue.png As you can see the maximum phase error due to pulling is +/- 60 ps The ADEV confirms the offset oscillator noise is canceled out independent of phase difference as the ADEV of this oscillator at tau = 1 s is very bad (around 1e-10) and the measured ADEV at tau = 1 s is 1e-12 see: http://athome.kaashoek.com/time-nuts/DMTD/Phase_difference_ADEV.png Erik On 9-10-2022 18:01, Corby Dawson via time-nuts wrote: > Erik, > > Coherence is not the issue. > > Sampling at the same time (coincidence) is! > > ... snip ... > The offset oscillator noise only gets cancelled at near to zero > coincidence. The further away the more offset oscillator noise affects > the measurement.