The Philips ToUcam (Model P01-VC740K) is becoming popular for high resolution imaging of bright astronomical targets, since it can take short exposure images that can be aligned and summed to increase the signal-to-noise ratio. While several websites explain many features of this particular CCD or webcams in general, I haven't found some important data that describes its optical performance. I spent some time measuring the webcam's exposure time, linearity, and spectral sensitivity, all before the Mars apparition of 2003, so I would be ready to exploit the camera to its maximum capability. Here are some summaries of the measurements on my particular camera. I used the Philips drivers that came with the camera, but I can't guarantee that my results will apply to other camera or software versions. I was satisfied with the performance, enough to only use this camera for all Martian imaging, and probably for future imaging of Jupiter and Saturn.
The camera control software is apparently not quite accurate, when in manual mode, in giving the actual exposure time. For slow frame rates, the exposure time indicated by the slider is not correct. I measured the actual exposure time by taking images of a spinning bar connected to a small steping motor with a known stepping rate. The following table summarizes the actual exposure times:
For Mars, I typically use 10 frames per second at the "1/25 second" setting. This gives me lots of data to process and tends to fill up a hard drive. Using the longest exposures (200 msec), Uranus can be imaged as a disk at F/33 by only turning up the gain a little bit! If a driver exists somewhere that gives the actual exposure time for these settings on this camera, please let me know!
The linearity was measured by imaging a red LED that was pulse modulated at a high frequency at a constant current. The exposure time was long enough to contain many pulses, so the LED brightness is linear with an easily measured pulse frequency. At a camera control gamma setting of 0, the response is linear, as shown in the graph. The black line is the best fit to the data points, shown as red circles. The maximum level of 255 counts was not measured here; other data sets show linearity up to the maximum 255 counts. The main problem is the negative intercept for no input signal. Instead of a small noise level, all signals less than about 5% of the maximum level is clipped to zero. Normally, this will only impact the edges of a planet. This data is for a gain of zero; if the gain is increased to about 50%, the black level is above zero, but then the entire frame has more noise. More experiments are required to understand this tradeoff.
The data sheet for the Sony ICX098 CCD used in the Philips ToUcam shows that there is a lot of infrared sensitivity, but the curves don't show the response past 700 nm. To get a rough idea of the entire curve, I set up a simple prism spectroscope using a tungsten-halogen lamp as the source and a 45 degree BK-7 prism to disperse the light from a small pinhole. A HeNe laser was used to get a single point wavelength calibration, then the dispersion curve was calculated using a ray-tracing program. The resolution was not very high, but my confidence in the measurements is. The IR filter used here, and on my telescope, is taken from the ToUcam lens. (To remove the IR filter from the back of the ToUcam lens, I soaked the entire lens assembly in acetone. The black plastic dissolved, leaving the glass IR filter behind.)
The images are shown below; I forgot to record the CCD controller's white balance setting, but it was likely set on "Auto". The top spectrum shows the entire range of wavelengths on the bare CCD. Lots of IR signal is recorded on all three color pixels. The bottom spectrum shows the same image with an IR filter placed in the optical path; no trace of IR signal is seen.
Slices of these spectral images were exported to Mathcad for more analysis, producing the spectral response curves shown below for two different gain settings. The color response was corrected by applying a wavelength-dependent factor to convert from the estimated 3300 K color temperature of the halogen bulb to a solar value of 5000 K.
In the left-hand figure, all the gains were set at the lowest value. The curves with the solid lines are for the bare CCD, while the response with the IR filter is shown with dotted lines. The gain setting and exposures here are the same as used on Mars, so the results can be applied directly. Since the gain is set at the lowest value, the low-level leakage is not apparent. The right-hand figure shows the difference between the low gain curve (dotted lines) and the response with the gain set at 25%, to bring up the lowest levels. The peak heights were scaled to give the same peak height, and the scale only goes to 700 nm. A different IR-cutoff filter was used for the two spectra, so the red response is not the same.
For the higher gain, this set of curves and the Sony data sheet are similar. The red pixels do pick up some response to blue light, but not very much. This camera shows lots of sensitivity in the IR spectral region that might be exploited with new external filters. The near-IR methane band, for example, is well-matched to the blue pixels.
Return to the Image Gallery index page
Return to Stellar Products home page
About Stellar Products
Adaptive optics tutorials