Precision optics are judged by their surface form and transmitted wavefront, and the tool that measures both is usually a phase-shifting interferometer. Lens makers, mirror shops, coating houses and the builders of the interferometers themselves all depend on it. Inside every one is a camera, and the camera sets limits on what the instrument can resolve and how far its numbers can be trusted.
This post is for engineers designing or upgrading an interferometer or other fringe-based metrology system. It covers the camera questions that matter most, with Finger Lakes Instrumentation (FLI) cooled CMOS cameras as worked examples. For a wider view of inspection work, see our post on cooled cameras for display, solar and optics inspection.
What the camera does in a phase-shifting interferometer
A phase-shifting interferometer combines a reference beam with light from the part under test. The camera records the resulting fringe pattern several times while the reference phase is stepped, often by a piezo actuator. Software then turns those frames into a phase map, and the phase map into surface height or wavefront error. Hariharan, Oreb and Eiju's five-frame algorithm (Applied Optics, 1987) is one well-known example of how the frames are combined.
Because the phase at each pixel is calculated from intensity ratios, the camera is a measuring element. Every pixel has to respond the same way to the same light in every frame of the sequence.
Pixels across the pupil
Lateral resolution comes from how many camera pixels cover the image of the test aperture. A round part imaged onto a rectangular sensor wastes the corners, so square sensors fit naturally.
The FLI RV533 uses a 3003 × 3003 sensor with 3.76 µm pixels, so a pupil image filling its 11.3 mm height spans about 3,000 pixels. For larger pupil images, the KL4040BI offers 4096 × 4096 pixels of 9 µm on a 36.9 mm square sensor, and the KL400BI offers 2048 × 2048 pixels of 11 µm. Rectangular formats such as the RV571 at 6244 × 4168 or the AR455 at 9568 × 6380 suit systems that image a pupil plus a test field, or several parts at once. Subarray readout lets you read only the region you need. Our post on pixel size and sampling covers the general matching math.
Fringe sampling
Phase can only be recovered where the fringes are resolved. The hard limit is two pixels per fringe period, and practical designs leave margin above that. That matters most on steep aspheres, parts with a lot of tilt, and systems that add carrier fringes on purpose.
More pixels across the pupil raises the fringe density the instrument can handle before data drops out. It also helps with mid-spatial-frequency errors, which need fine lateral sampling to show up at all.
Fringe contrast and low-reflectance parts
Fringe contrast is highest when the test and reference beams are about equally bright. An uncoated glass surface reflects only about 4% of the light at normal incidence, and anti-reflection-coated parts reflect far less. Measuring those surfaces means working with weak signals or poorly matched beams.
Low read noise and a large full well help here. Aurora and Revelation cameras list 1 e- read noise in rolling mode and a 50,000 e- full well, which gives room to keep fringes well above the noise floor without clipping the bright ones. Window choice matters too. The F116 window transmits farther into the near-IR and the F101 farther into the near-UV, so pick the one that matches your source wavelength. The window has to be chosen before the camera enters our clean-room build.
Linearity and phase error
Phase-shifting algorithms assume that recorded intensity follows the fringe pattern faithfully. When the detector response bends, the calculated phase picks up systematic errors that look like real surface features. Creath's review of phase-measurement techniques (Progress in Optics, 1988) lists detector nonlinearity alongside phase-shifter miscalibration as a known error source.
Check linearity at the gain and readout settings your software will actually use, with exposures that cover the full range of your fringes. Our post on acceptance testing for gain, noise and linearity lays out the tests. FLI provides a production test summary for each camera by serial number.
Timing the phase steps
Aurora and Revelation sensors use a rolling shutter, with a global-reset mode on models such as the AR455 and AR533. In plain rolling mode, rows start and finish exposing at slightly different times. If the phase shifter moves during a frame, different rows see different phases. The simple rule is to step the phase between frames and let it settle before the next exposure starts.
Global reset starts every row together. Paired with a pulsed or gated source that fires inside the exposure, it gives every row the same light. External trigger in and out is standard, so the camera can drive the phase shifter or follow it. In triggered global-reset mode, allow a row-reset delay of about 300 to 350 µs before the light event. Our posts on synchronizing cameras with triggers and automated fluorescence instruments cover similar timing.
Vibration and air turbulence change the phase during the sequence too, so isolation and enclosure design remain part of the camera decision.
Stability, cooling and darks
Interferometer exposures are usually short, so dark signal is small. At the published 0.002 e-/pixel/s at -20 °C, a 50 ms exposure collects about 0.0001 e- per pixel. Cooling earns its place another way: a regulated sensor temperature keeps response steady from one sequence to the next and through a long shift.
Aurora and Revelation cool to 50 °C below ambient with air, with liquid cooling as an option. That matters on a vibration-isolated table, where fan vibration can be unwelcome. Aurora is not offered with a mechanical shutter, so take dark and background frames with the source blocked. Revelation's thin body suits compact instrument heads, and every Revelation camera includes our software development kit (SDK). For optical layout and cooling choices, see our posts on back focus and flanges and air versus liquid cooling.
Questions to settle before you choose
- What aperture sizes will you measure, and how many pixels must span the pupil image?
- What is the highest fringe density you need to resolve, including carrier fringes?
- What reflectance range do your parts cover, from bare glass to coated surfaces?
- What source wavelength do you use, and which window suits it?
- How does your phase shifter step, and which device will be the timing master?
- How much vibration and heat can the instrument head tolerate?
Talk to our sales team
Send our sales team your aperture, fringe density, wavelength and timing scheme, and we will suggest the standard Revelation, Aurora or Kepler configuration that fits. If an instrument needs a modest change to a standard camera, we are glad to discuss it. Request a quote or email sales@flicamera.com.
FLI cameras are Made in USA, and our sensor chambers are assembled in our 1,000+ sq ft ISO Class 7 clean room. See how on our in-house manufacturing page.
