Sensor Coverglass Information

[marknorton]

Advertisement (gone after registration)

It's thought the "magenta problem" is caused by IR light reflecting off or emitted by objects in the frame and this is down to inadequate filtering by the sensor coverglass before it reaches the sensor.

 

Here's what I've found out about the coverglass used on the Kodak sensor. It may be of course that the version of the sensor supplied to Leica is different.

 

First, here is the spectral response of the standard Kodak sensor which provides sharp IR cut above 700 nm:

 

[ATTACH]14283[/ATTACH]

 

The M8 sensor coverglass appears to be made by Kyocera, here's a link to a product page which seems relevant:

 

http://global.kyocera.com/prdct/electro/pdf/has_suirezo/184_e.pdf

 

This coverglass is different from that used on the DMR sensor. That is 0.5mm thick, has an anti-abrasive coating on one side and an IR cut coating on the other and uases a glass type D-263 made by Schott but has a similar spectral response.

 

The specification of the glass indicates that it has a refractive index of about 1.5. This is what causes light to bend at air/glass interfaces and I believe it was a major objective of the M8 cover glass that this should not happen, all tied up with the angle of incidence problem.

 

Looks like the Kyocera is a sandwich of two glasses with the IR cut glass in the middle and it's possible this improves the optical performance compared to a single piece of glass. This Kyocera sandwich provides IR cut of 5% above 700 nm (top right graph) and you can see the spectral response of the Kyocera glass matches that of the sensor/glass combination.

 

What we don't know of course is what Leica selected for its version of the sensor which is letting so much IR light through.

 

[ATTACH]14284[/ATTACH]

[jlm]

so maybe instead of sending our lenses in for coding, we send them in for coating...with an IR filter layer!

[marknorton]

so maybe instead of sending our lenses in for coding, we send them in for coating...with an IR filter layer!

 

Based on what we know about the coding, it might be more useful!