After testing the Alexa today for infrared issues it became clear that you should be carefull using plain ND filters.
We recomend to do your own testing but we found that ND filters above 0.6 (2 stops) already gave a clear magenta hue and color shifts. Additionally we found a big difference between the Panchro hotmirror filter, (working very well), and the Rosco True Color doing only half the job.
There where also big differences between what we measured with our light meter, what the camera recorded and what the density if the filters was.

Sometimes there was more than a stop difference between what we measured and what the filters were supposed to give. LIke the Panchro 1.5 hotmirror, is suppose to block 5 stops of light be it ony blocked three to four. When you are depending on your light meter as ARRI recommends, you might be in for a surprise.
We went around this by attaching a Transvideo monitor with a Waveform parade. We placed a big Greycard in the shot and set it to exactly 50% on the waveform. This way we were pretty sure we did the same exposure for all tests. Don't leave home without a proper waveform :-)
The infrared problem mainly comes up when shooting outdoors and we decided to test at the native 800 ASA but we were also curious how the lowest 160ASA setting would react.

Here are some test pictures, first we start with a comparison between Normal ND1.2, a Panchro HotMirror 1.2 and a Rosco True color with ND1.2:
(Click on the picture to enlarge and switch between them with the arrows on top, the pictures are 50% scaled down from the original HD size)

  
        
  

    

  
Panchro Hotmirror 1.2   
  


  
        
  

    

  
ND 1.2 - Alexa @160ASA  
  


  
        
  

    

  
Rosco TrueColor+ ND 1.2 - Alexa @160ASA  
  

Clearly visible is the magenta tone in de black fleece on the right. With ND 1.2 it is already strong, the Panchro takes it all away and the TrueColor sits somewhere in the middle. Also note the overal color shifts the filters give. THe only correction we did was to get an equal exposure on the chip.

Now lets look if it makes any difference using 800ASA or 160ASA setting. THe 160ASA setting holds very well. This should be extra tested because the chart is not able to show the full range the camera can handle but it comes close. No worries there to use the lower ASA setting before you put those filters on.

  
        
  

    

  
ND 1.5 - Alexa @800ASA  
  


  
        
  

    

  
ND 1.5 - Alexa @160ASA  
  


  
        
  

    

  
Panchro Hotmirror 1.5 - Alexa @800ASA  
  


  
        
  

    

  
Panchro Hotmirror 1.5 - Alexa @160ASA  
  

  
        
  

    

  
ND 1.5 - Alexa @160ASA + Face  
  


  
        
  

    

  
Rosco TrueColor + ND 1.2 - Alexa @160ASA + Face  
  


  
        
  

    

  
Panchro Hotmirror 1.5 - Alexa @800ASA+face  
  


  
        
  

    

  
Panchro Hotmirror 1.5 - Alexa @160ASA +Face  
  

Talking about post: the material is under investigation today, if any other results popup, I will update here.

Thanks to Niels Boon NSC and Cam-a-lot for helping with this test.