Erik, When using a common clock into a DMTD it is fairly easy to make sure your DUT and REF are close to in phase at the start of a measurement. This allows the cancelling of noise from the offset oscillator. However if you use separate clocks of different frequencies you will lose that advantage. For measurements that don't need ADEV better than parts in the -11 or -12 range it would probably work, but a trial would tell the tale. Cheers, Corby

Corby, I'm struggling to understand. Suppose the single internal reference of a DMTD is 10MHz and 0.1Hz high due to phase noise Now take two inputs, one at 10MHz, 0.3Hz high and one at 20MHz, 0.4Hz high For simplicity assume the internal IF is zero Hz so we can have negative frequencies. The 10MHz input gets mixed with the internal reference producing negative 0.2 Hz The 20Mhz input gets mixed with two times the single internal reference, resulting in 2 * 10Mhz at 2 * 0.1Hz to high being 20MHz at 0.2Hz high and after mixing this results also in a negative 0.2Hz. The measured difference between the two input signals is now zero Hz. How can I remove the offset of the internal reference to get to the actual 0.1Hz difference when the phase noise is multiplied with different factors because the need to match the different frequencies of the two inputs? Erik. On 7-10-2022 18:00, Corby Dawson via time-nuts wrote: > Erik, > > When using a common clock into a DMTD it is fairly easy to make sure your > DUT and REF are close to in phase at the start of a measurement. > This allows the cancelling of noise from the offset oscillator. > However if you use separate clocks of different frequencies you will lose > that advantage. > For measurements that don't need ADEV better than parts in the -11 or -12 > range it would probably work, but a trial would tell the tale. > > Cheers, > > Corby > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe send an email to time-nuts-leave@lists.febo.com

Hi What’s being talked about: In a “normal” DMTD, the edges are what matter. You convert the sine signal out of the mixer into a square wave. You then use some form of counter to look at the delta between the edges. Typically the reference is set up to give you a beat note of 1 to 10 Hz depending on the design. With atomic standards both beat notes will be virtually identical in frequency. The “right” way to set it up is to phase shift one DUT input so that the edges are quite close to each other. With atomic clock sort of sources, you can hold that very small delta for quite a while with no further phase shifting. Why? As the edges move further apart, you are getting more and more phase on the common mode oscillator into one or the other DUT inputs. Since you can have a significant amount of time If the edges are not aligned, there can be a noticeable amount of noise. Yes, this is talking about a system that is trying to get to < 1x10^-13 at 1 second tau and all the fun things you play with to get there. Bob > On Oct 9, 2022, at 2:13 AM, Erik Kaashoek via time-nuts <time-nuts@lists.febo.com> wrote: > > Corby, > > I'm struggling to understand. > Suppose the single internal reference of a DMTD is 10MHz and 0.1Hz high due to phase noise > Now take two inputs, one at 10MHz, 0.3Hz high and one at 20MHz, 0.4Hz high > For simplicity assume the internal IF is zero Hz so we can have negative frequencies. > The 10MHz input gets mixed with the internal reference producing negative 0.2 Hz > The 20Mhz input gets mixed with two times the single internal reference, resulting in 2 * 10Mhz at 2 * 0.1Hz to high being 20MHz at 0.2Hz high and after mixing this results also in a negative 0.2Hz. > The measured difference between the two input signals is now zero Hz. > How can I remove the offset of the internal reference to get to the actual 0.1Hz difference when the phase noise is multiplied with different factors because the need to match the different frequencies of the two inputs? > Erik. > > On 7-10-2022 18:00, Corby Dawson via time-nuts wrote: >> Erik, >> >> When using a common clock into a DMTD it is fairly easy to make sure your >> DUT and REF are close to in phase at the start of a measurement. >> This allows the cancelling of noise from the offset oscillator. >> However if you use separate clocks of different frequencies you will lose >> that advantage. >> For measurements that don't need ADEV better than parts in the -11 or -12 >> range it would probably work, but a trial would tell the tale. >> >> Cheers, >> >> Corby >> _______________________________________________ >> time-nuts mailing list -- time-nuts@lists.febo.com >> To unsubscribe send an email to time-nuts-leave@lists.febo.com > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe send an email to time-nuts-leave@lists.febo.com

I've reached frequency resolution of just above 1e-13 at tau=1 and the fun is already starting. Any small temperature change causes a visible shift of the phase, also even a small motion in the cables becomes very visible. Erik. On 9-10-2022 13:59, Bob kb8tq wrote: > Yes, this is talking about a system that is trying to get to < 1x10^-13 at > 1 second tau and all the fun things you play with to get there.