Application of dual-mixer time-difference multiplication in accurate time-delay measurements

The optimized dual mixer time difference (DMTD) multiplier shows an outstanding ADEV background stability /spl ap/ 7/spl times/10/sup -15///spl tau/ in the short term at 5 MHz. For TDEV, the flicker floor is 2 fs (2/spl times/10/sup -15/s) for sampling times 1 s to 100s. May this DMTD system be used to measure time delays with an accuracy matching the femtosecond noise? It turns out that impedance mismatches severely limit the accuracy of the delay measurement, especially at low RF around 5 MHz. However, as this spurious effect varies with frequency according to a simple relation, a proper model should allow one to fit calculated and measured results in an appropriate frequency range (5 to 100 MHz) and so substantially reduce the effect of the reflections. In the first-try work reported here, we have operated the IREE DMTD multiplier without any change in either components or technology. We have observed that, in order to achieve the expected accuracy of 1 ps or better, different technical means must be implemented and various sources of errors must be suppressed and/or taken into account in the model, besides the signal reflection. Experiments have shown that the technology and the operating conditions of the components are critical and also that the electromagnetic compatibility rules must be strictly followed. As a conclusion, the DMTD multiplier optimized for the frequency stability measurements cannot be readily used for accurate time delay measurements. We give the guidelines which are to be followed for the realization of the next version of the DMTD multiplier, that will be specially designed for very accurate time delay measurements.