Multiple Tau Digital Correlation ensures optimum statistical accuracy ...

Multiple Tau Digital Correlation is a unique way of obtaining spectral information on stochastic or deterministic signals using a temporal correlation function with variable sampling/lag time developed by Dr. K. Schätzel and ALV-GmbH in 1984. Different from standard correlation function measurements using a single (or some) sampling times, the Multiple Tau Correlation technique uses blocks of 8 linear channels (16 channels for the fastest sampling time) with a doubling of the sampling time from block to block. Real-world implementations use more than 40 parallel sampling times within a single measurement, covering a spectral (or temporal) range of more than 1 : 10 ¹³ and using up to 336 correlation channels.

Since Multiple Tau Digital Correlation is a fixed scheme (which simply does not  require any setting of sampling times, since they are optimum anyway), the burdon on the user to find a good fitting set of sampling times for the actual experiment completely vanishes !

Increasing the sampling time with the lag time ensures optimum statistical accuracy for measurements on stochastic signals with decaying correlation functions (as is the case for DLS, FCS, DWS ...). And this is a proven fact, published in several journals and books (see literature section).

... but only, if a 16/8 channel layout is used, Individual Monitor Channels are implemented as well and Symmetric Normalisation Schemes are used !!!

However, the Multiple Tau Digital Correlation must be coupled with individual monitor channels for each correlation channel, since the  statistical advantage of using longer and longer sampling times can only be obtained, if the fact that with that the number of samples (thus the number of samples taken per s, or minute or hour) decreases  accordingly. Not using Symmetric Normalisation along with Multiple Tau Digital Correlation (and for this, the use of individual channel monitors is simply required), a gain in statistical accuracy can not be  obtained - even worse, for very long lag times and very large sampling times, the statistical accuracy could even decrease ! Likewise, the implementation of “high-resolution” Multiple Tau Digital Correlation, such as using 32 linear channels at the initial sampling time and doubling the sampling time every 16 channels will not at all increase the “resolution”. The contrary is true, at least for decaying correlation functions : the statistical accuracy suffer, and it suffers even that much, that the increased number of correlation channels can not outweigh this effect. In summary, even less accurate results must be expected. 

... this guarantees optimum baseline accuracy !!!

Such high statistical accuracy is very well important - the accuracy of the baseline of the correlation function is solely determined by the accuracy of the long lag time correlation regime. Although, the correlation function baseline necessarily must be zero for all “energy limited” statistical processes (and thus for every real world process), it might not be zero “already”, since the measurement duration was not yet long enough to ensure all long time fluctuation of the process was measured and did decay to zero. A typical example is a slow laser intensity fluctuation, which can very well reach the 10 ... 100 second regime for the fluctuation time and thus can lead to nonzero baselines for duration significantly shorter than say 10 x the fluctuation time. For all such cases, optimum baseline accuracy is a must, since the precise knowledge of the actual baseline value is the most important key to high precision data analysis. Again, only Multiple Tau Correlation coupled with symmetric normalisation schemes will ensures this !
 

Multiple Tau Implementation on the different ALV-Digital Correlators

*(for the ALV-5000/E plus the ALV-5000/FAST Tau Extension the temporal dynamics is 4.0 x 10¹²)