@david.kwon

I think you have the right idea conceptually, but unfortunately I don’t think your approach will work in non-sequential OpticStudio.  Let me explain.  First, speckle is created via a superposition of randomly phased waves, e.g., when reflecting a collimated laser beam from a scattering surface, and viewing the light on an observation screen placed some distance away from the scattering surface.  In this case, each point on the scattering surface can be thought of as a randomly-phased point source.  Duplicating this with rays in non-sequential OpticStudio requires a random phase surface (say using a custom Grid Sag Surface).  That’s the easy part. 

If you then apply a non-zero Coherence Length to the source, all you will be doing is creating more randomness to the phase of the rays.  The Detector Rectangle will still calculate a coherent superposition of all the rays landing on any given pixel, and the speckle pattern will remain.  In fact, the contrast of the speckle pattern may actually increase.  This is opposite of what you want.  To model temporal coherence correctly, the rays would have to be combined differently, in a much more complicated fashion that transitions from fully coherent to fully incoherent as the relative path differences between rays increases.  The way OpticStudio implements temporal coherence is based on a very simple (non-physical) approach that only provides a crude approximation in certain cases, such as a reduction in fringe pattern visibility that is accompanied by the presence of a speckle pattern.     

Here is a very simple example to illustrate.  It’s just a collimated beam falling on a Detector Rectangle.  With a coherence length set to a value that is larger than the propagation distance, the coherent irradiance is uniform as expected:

However, upon reducing the coherence length, speckle arises simply because the random phasing that OpticStudio applies is becoming more significant.  The contrast of the speckle pattern therefore increases as the coherence length decreases -- again, this is simply a result of the non-physical way that OpticStudio implements coherence.  In reality, the speckle should only diminish as the temporal coherence goes down.

Hope this makes sense.  For your problem, I think if you calculate the modal time dispersion of your fiber (i.e., the temporal width of the impulse response) and make sure that it exceeds one over the bandwidth of your source (i.e., the coherence time of your source) by at least of factor of 2 or more, then you may be able to mitigate speckle.  See Goodman, Speckle Phenomena in Optics, 2nd ed., Fig. 8.8 which essentially shows how the speckle contrast falls off as a function of the product of source bandwidth and modal delay time for a step-index fiber. 

Regards,

Jeff