I analysed some virus samples (d~190nm) in the DLS with a protein (d~7nm) as stabilisator. As the protein is present in such abundance the z-average is decreased significantly to
Hello Friederike Eilts , there are a few things to consider. The "edit size classes" you found will change the bins for the size distribution analysis. So yes, you can force the fitting range to a different grid of potential size classes. However, this can have unintended consequences. The size classes have no relation to the z-average, so it is correct that the z-average would not change.
My recommendation would be to look at just the size peak and observe changes in that peak parameter (peak mean and area). You can also "eliminate" peaks by not displaying them, as described in https://www.materials-talks.com/blog/2016/06/15/solvent-peaks-how-to-get-rid-of-the-buffersaltadditive-dls-contribution/
If you are completely set on the idea that you require a parameter like the z-average but have to exclude everything below ~15nm : this can be done and was described in a previous discussion, see https://www.researchgate.net/post/How_to_remove_solvent_buffer_salt_additive_peaks_in_DLS_and_get_new_PDI_and_Z_average
In other words, if you want to change the z-average fitting parameters you have to goin to the 'researchmode' - and be aware of what you are modifying. Some details about this advanced set of features are also in https://www.materials-talks.com/blog/2015/08/19/advanced-research-software-features-for-the-zetasizer-nano/
PS: if you are looking at aggregation kinetics, you might find another post of interest at https://www.materials-talks.com/blog/2016/04/21/aggregation-kinetics-with-dls/
By definition, the z-average is calculated using the "cumulants" method which, in effect, is the best single exponential to fit the autocorrelation function. There is no way to limit the size range.
The instrument uses a different mathematical model to try to provide distributions with multiple peaks. So, even if you change the size range for that model, the reported z-average won't change since it is an independent calculation.
In principle, you could omit some of the data points in the autocorrelation at very short times but I don't think the Zetasizer software can do that.
EDIT: It looks like @Ulf Nobbmann was writing a reply at exactly the same time as me :) He has provided some useful links that expand on my answer and may indicate you can use a subset of the raw data.
Ulf Nobbmann and John Francis Miller thank you very much for your quick and detailed answers! I'll dive into your recommendations and come back to you if questions remain.