Philips prisms are used to separate an incoming image into several spectral, intensity or polarization channels. The general layout of the Philips prism when used for color imaging is shown in Figure 1.
Figure 1 - Three-channel Philips prism layout
The Philips prism consists of three glass wedges cut and assembled to provide three output images with important characteristics.
- All three images are erect - that is, none of the images are inverted either top-bottom or left-right relative to the others because the light for all channels passes either zero or two reflections.
- The optical path lengths for all three channels are identical
- The difference in S versus P polarization is minimized
In the configuration shown in Figure 1, the blue portion of the entering light is reflected from a dichroic filter and the remaining red and green (yellow) portion is transmitted. The blue light returns to the input surface where it is directed by total internal reflection to the first output. It is not necessary for this first dichroic filter to be a longpass filter between the blue and green as it is in this example. It can be either longpass or shortpass at any wavelength that the prism material will transmit.
The yellow light is split by a second dichroic filter into reflected red and transmitted green. The red light is directed to the second output by total internal reflection (this is the reason for the air gap) and the green light passes straight through to the third output. The second dichroic filter can also be either longpass or shortpass with a transition wavelength in the band transmitted by the first dichroic filter.
If the light to be imaged is narrowband or additional spectral isolation is required between channels, then trim filters can be added at the outputs for spectral shaping. In color cameras, an infrared absorbing filter is often added at the input face to assure that stray infrared never reaches the sensors.
Other Splitting Types
The splitting filters inside the prism do not necessarily all have to be spectral separators. if the first coating is a dichroic neutral splitter with a 33:67 ratio and the second coating is a neutral splitter with a 50:50 ratio, then each of the outputs will receive one-third of the incoming light. This configuration is useful when the required channel spectral bands are not known. The spectral bands can be set by putting different bandpass filters at the outputs. When the right set is identified, prisms can be nuilt with the splitting inside to improve optical efficiency. Neutral splitters can also be used to increase scene dynamic range.
While the prism angles are designed to minimize polarization effects, the coatings can be designed to produce particular desired polarization splits when polarization images are needed. Generally, polarizing trim filters are required in these instances to produce sufficient extinction ratios.
Other Numbers of Channels
While the Philips prism was designed for use in broadcast color TV, is has many other uses, some of which require something other than three channels. Making a two-channel prism is straightforward, requiring only that a single glass block replace the second and third blocks. For four and five channel prisms, one prism is essentially stacked on top of another. The top one is a normal three-channel unit and the bottom is either a two channel or three channel unit. In either case, the straight-through channel on the bottom part simply feeds light to the top part. Figure 2 shows the configurations that result.
Figure 2 - 2, 3, 4 and 5 channel Philips prisms
In all cases, the image sensors are positioned so that all images are erect. The coatings can be selected for the properties needed and any or all outputs can include trim filters.
Lenses for Prisms
While the geometry is constant for all configurations, the size of the priems can vary widely depending on the sensor size and the lens f-number required. Increasing the sensor size or reducing the f-number requires bigger prisms to assure that all of the imaging rays stay inside the glass. Prisms have been made to accommodate sensors as small as 1/4" and as large as 55 x 60 mm. In all cases, optimum optical performance requires the use of lenses that include the prism glass thickness in their optical formulas. For most prisms, there is some common lens type that can focus through the glass but generally, the chromatic aberration is not as good with common lenses as with lenses designed for prism use. The best example of this is television broadcast, which uses almost exclusively lens series designed for prism use, including incorporation of image-side telecentricity to minimize color shading from ray angle variations acrosss the image.
Often common lenses are good enough for particular applications but users should at least examine the effects of these lenses relative to their performance requirements to decide if special or custom prism-optimized lenses are justified.
Additional Technical Data
Philips prism white paper - PDF here