Hi Dirk,

I think there's an inherent assumption in the coherent sum that causes this discrepancy, namely in how energy is normalized on the detector to ensure conservation of energy. Unfortunately I can't think of a workaround, aside from performing your own detector coherent sum (i.e. avoiding the inherent coherent power scaling in the Detector Viewer). Please review the section of the help files 'Comments on Coherent Data Computations', in the Detector Viewer page. Importantly, coherent energy is always normalized to be equal to or less than the incoherent, either on a detector-wide scale or on a pixel-by-pixel basis.

When you use 'Waveplate Mode', OpticStudio treats both the ordinary and extraordinary energy as a single ray; the output from the crystal is a single, Y-polarized ray which carries all of the initial energy (assuming no Fresnel losses, etc.). Nothing is lost as we propagate through the final analyzing polarizer. The incoherent power (sum((A^2))) is the same as the coherent ((sum(A))^2). On the other hand, when we split the ray into ordinary/extraordinary, the output of the crystal is 2 rays, each of which carrying 1/2 the initial power, and with (unnormalized) polarization states i.e. [x,y]=[-1,1], [1,1]. But here's the problem: OpticStudio doesn't actually sum the rays until the data post-processing step on the Detector Viewer, and in doing so, it scales the result to the incoherent power. Said another way, the coherent power isn't allowed to be greater than the incoherent power. 

Because 'Waveplate Mode' effectively re-combines the o and e rays immediately (or, never separates them is probably more accurate), we don't have to worry about this. So, depending on the setup and how the rays interact with subsequent objects, it might be more accurate to use this method. Alternatively, some additional manual post-processing of data (i.e. to scale the detector results) might help work around this issue, at least in this simple case.

Also Dirk, I'm curious why you don't want to use waveplate mode when simulating a waveplate...

Hello Zach, 

thanks for the detailed answer.
What if the crystal axis has a z-component (e.g. 1/1/1) and incident angle on the front surface of the waveplate (xy-plane)  is 0 Deg?
Then you will have two perpendicular polarized beams (ordenary and extraordenary) which have a walkoff and are
not coincident. So on a detector behind the waveplate these two rays will hit different pixels.

In this case: How can Zemax calculate the following information of the light behind the waveplate:

(i) coherent irradiance I (x,y) = [ E_ord (x,y) + E_extraord (x,y) ] ^2 on a single detector pixel A (x,y)
(ii) the phasedifference between the resulting two perpendicular polarization states of the light on a single detector pixel A (x,y)

Best regards
Dirk

 

Hi Dirk,

That case sounds like the beamsplitter mode, in which the ray is split into E and O rays. We really have a choice in modelling birefringence with rays: either split the ray into E and O, which is beamsplitter mode, or to treat the difference as a phase delay, which is waveplate mode. If the rays don't follow the same path, they can have an optical path length difference, but they don't have a phase relationship.

I don't think this is a limitation of the code, but of the ray model. If the two rays separate, use beamsplitter mode, if they recombine use waveplate mode. I think waveplate mode should be fine for what you want. It gives you I(x,y) (but not the individual Ex and Ey terms) and the phase delay.