Monochrome UV Camera Fact

[Andrea B.]

I was not sure whether planing off the Bayer filter of a typical digital camera to create a monochrome conversion would increase the "reach" into the shorter UV wavelengths. In to answer that, I just read the following about monochrome conversion on the MaxMax website (owner Dan Llewellyn). The bullets and bolding are mine.

 

UV-VIS-IR (Ultraviolet-Visible-Infrared) cameras have their ICF/AA (IR Cut Filter/Anti-Aliasing) removed and replaced with a clear optical window with the same thickness as the original ICF in order to maintain the correct focal plane.

• The clear optical window is made of Schott WG280 glass which is clear from 280nm past 1200nm.
  
• For a color camera, the sensor microlenses and CFA (Color Filter Array) will pass light from 360nm to 1100nm.
  
• For a monochrome camera, sensitivity will be from about 300nm to 1100nm, but UV sensitivity will increase by a factor of 6x.
  

 

Now I am personally a little unsure about the remark that microlenses and CFA only pass light between 360-1100nm because my use of a steep-slope, narrowband 340/10 filter on my converted Nikon has shown that it records quite well around 340nm. (Do recall, however, that that 340/10 filter has problems with reflection due to its mirrored surface.)

 

Remember one of the primary problems shooting between 300-340nm is making sure you have some UV illumination in that area. :D

 

 

{Disclaimer: UltravioletPhotography.com has no financial affiliation with any vendors. Any reference such as in this post is for your information only. Always investigate on your own before purchase.}

[Cadmium]

"For a color camera, the sensor microlenses and CFA (Color Filter Array) will pass light from 360nm to 1100nm."

 

Clearly, my Sparticle band pass filter array tests show sensitivity as low as 325nm, limited by the transmission of the lens + UV lens filter.

A 340BP10 (335nm-345nm bandpass) doesn't pass 360nm, 325BP10 also... set up the camera with 'no lens', just a body cap and a 325BP10, you will see more sensitivity than with a lens.

So the "color camera" is sensitive to UV below 360nm.

[Andy Perrin]

Probably it is a question of what arbitrary threshold or cutoff for "transmission" that the MaxMax site is using?

[Andrea B.]

Could be.

It is a bit mysterious.

[OlDoinyo]

If one could replace the microlens/filter array with a phosphor screen, that might be interesting...

[Alex H]

My question is - how Dan measured the transmission properties of CFA and microlenses, as well as monochrome camera "reach" into UV.

[Cadmium]

Maybe with an Arbitraryometer?

[JMC]

In the David Prutchi book "Exploring UV photography", there is a figure on page 26, which gives "Bayer Filter Transmission" as a percentage as a function of wavelength. Looking at that, it suggests that at around 360nm the transmission of the filter drops below 10%, perhaps this is where the 'the CFA will pass light from 360nm' comes from. If so it is a very arbitrary definition of 'passing light'. Unfortunately there is no reference for where this came from originally. As an aside my MaxMax monochrome EOS 5dsr conversion (which does have a Schott WG280 filter) seems to be able 4x more sensitive to UV, when used with a Baader U, vs my ACS converted EOS 7D (which uses the ACS own design filter over the sensor, and keeps the Bayer filter). Not quite a true like for like comparison, but there is obviously a significant improvement in overall UV sensitivity when removing the Bayer filter. What I do not know though is whether this is even more enhanced at the shorter wavelengths. Time and resources pending, I would love to look more into this in a systematic way, but I think I'll need that lottery win first.....

[Andrea B.]

Dear Santa Claus -- I haven't written you since I asked for that Tea Set when I was 4. (Thanks, by the way. Loved that!) But I have a serious request now. Please oh please leave me a really nice set of spectro gear under the Winter Solstice Tree on or about 25 December, will you?? We really do need to measure some stuff here at UVP. You know, like filter transmissions, lens transmissions, sensor transmissions - that kind of thing. I promise to continue being mostly good. Thank you so so much !!!!! Love & Hugs from Your Friend - Andrea B.

[Andrea B.]

Bayer trans diagram here:

https://micro.magnet...agesensors.html

Scroll down.

 

From the looks of these diagrams it is a wonder we are able to photograph anything in UV.

Oh well.

[Cadmium]

In the David Prutchi book "Exploring UV photography", there is a figure on page 26, which gives "Bayer Filter Transmission" as a percentage as a function of wavelength. Looking at that, it suggests that at around 360nm the transmission of the filter drops below 10%, perhaps this is where the 'the CFA will pass light from 360nm' comes from. If so it is a very arbitrary definition of 'passing light'. Unfortunately there is no reference for where this came from originally. As an aside my MaxMax monochrome EOS 5dsr conversion (which does have a Schott WG280 filter) seems to be able 4x more sensitive to UV, when used with a Baader U, vs my ACS converted EOS 7D (which uses the ACS own design filter over the sensor, and keeps the Bayer filter). Not quite a true like for like comparison, but there is obviously a significant improvement in overall UV sensitivity when removing the Bayer filter. What I do not know though is whether this is even more enhanced at the shorter wavelengths. Time and resources pending, I would love to look more into this in a systematic way, but I think I'll need that lottery win first.....

 

Yes, here is the conversation about that p. 26 graph before.

http://www.ultravioletphotography.com/content/index.php/topic/2102-flowers-with-two-false-colours/page__st__80

 

I don't trust that graph. There is no credit or source, and it might be 'arbitrary' or made with an Arbitraryometer, so... I trust my Sparticle, it is not arbitrary.

[JMC]

Andrea, that Bayer transmission diagram in the link you shared doesn't make any sense to me at all. The title for the graph is Bayer Filter Transmission Spectral Profiles, and yet the Y axis on the graph is Quantum Efficiency. Surely this is a graph of response of different pixels on the sensor through the Bayer filter, which will be a sum of transmission through the filter and response of the pixels (both a a function of wavelength). Running them both together makes it impossible to determine the wavelength response of the sensor (or the transmission through the Bayer filter).

 

Is it possible to buy a Bayer filter anywhere (ideally on a UV transparent substrate), or are they always just produced on sensors?

[Andy Perrin]

JMC, the Bayers seem to be made of photoresist -- they are directly produced on the sensors I think.

[JMC]

I've found information on the Fuji website about the dyes used to make RGB CMOS and CCD sensors, which gives a transmission curve (albeit only from 400-700nm);

 

http://www.fujifilmusa.com/products/semiconductor_materials/image-sensor-color-mosaic/rgb/index.html#applications

 

I've reached out to them to see if they have data for <400nm, so we shall see (although as often with big companies I wont be holding my breath). If I find anything I'll share it.