Timing and optimization of the system

Response time cameras The cameras were triggered with the timer box. After a camera receives the trigger signal there is some delay before the camera will react. This is often referred to as the response time of the camera. See figure 4.7 for a schematic representation of the response time. Even though the documentation on the Basler cameras is very detailed, no indication on the response time of the cameras is given. This is due to the fact that too many factors (e.g., operating temperature, production spread, ageing of opto-couplers) may influence this (User’s manuel for GigE Cameras, 2015). We thus have to find this response time experimentally.

A LED was connected to the timer box and one of the cameras is connected to the trigger signal. The camera can be triggered on the rising or falling edge of the trigger signal. The response time for the rising edge is shorter compared to triggering on the falling edge. Furthermore the spreading in response time for the falling edge is larger (User’s manuel for GigE Cameras, 2015). Hence for the first experiments the camera was triggered on the rising edge of the signal. The exposure time of the camera was set much longer than the pulse width of the light source. The exposure delay, i.e. the delay between the response to the trigger and the recording, in the camera was altered until photos were captured with the LED on. A sequence of 100 photos was captured every test. This resulted in a number of images with the LED turned on, and the rest of the images with the LED turned off.

After looking into the details of the system, it turned out that the oscillator in the timer box can have a jitter, i.e. variation in the frequency of the signal, of a few percent and can be temperature dependent. On the contrary the clock in the camera has a µs accuracy. After a while the clock of both devices may get out of phase. The frequency of the oscillator was set to 10Hz. For a jitter of e.g. 5% this can result in a time difference of ±2.5ms for half a period of 50ms. Triggering on the falling edge solved this problem. Due to a different reference point in the signal the jitter of the oscillator is less severe.

Minimizing differences between image pairs Working with 2 cameras, 2 different optics and 2 illumination sources induces differences in the image pairs. First of all, the intensity of both images was not equal. This can partly be con- tributed to different scatter properties due to different wave lengths of the light, the camera having a different spectral response for red and blue wavelengths, but also

Figure 4.7: Schematics of camera response timetPLH. Top panel depicts the trigger signal and the lower

panel depicts the response of the camera when triggering either on the rising or falling edge. Figure adopted fromUser’s manuel for GigE Cameras(2015)

.

the optics used had different aperture stops. The aperture was adjusted such that by visual inspection the intensities in both images seemed to be similar.

Secondly, the laser diodes used are designed to produce a certain amount of light, not a perfect beam profile. As a consequence the focus of both lasers is not optimal, resulting in a different thickness of the sheet. After optimizing the focus of the lasers, experiments were performed with slits in a very thin metal sheet, made by means of a laser cutting technique. As a result a slightly improved sheet thickness was obtained.

From the response time tests it became clear that the timing of the cameras could be arranged accurately enough for the signal range used during the experiments, that it was possible to time the system with the cameras. Therefore only 1 laser was sufficient, which was still pulsed with the timer box. Cameras were triggered with the timer box and exposure time and exposure delay were controlled with the camera settings. Reducing the number of lasers also minimized the differences in the image pairs. See figure 4.8 for an example of an image pair. In this figure it is visible that still some differences are present: Image intensity is still not perfectly equal and the 1st image is sharper than the 2nd image. This blurriness was already clear while focussing the cameras. Optics used with camera 2 were difficult to focus perfectly.

Data stream problems The cameras were connected with the computer via the university network, due to the availability of power over Ethernet. The data stream which resulted from two cameras with a frequency of 10Hz was too unstable for 1 computer to handle. This resulted in images with stripes where data informa- tion was missing. This problem was fixed by using 2 computers to capture the data. For future experiments it is recommended that camera power is not supplied over the network connection but directly to the camera. The cameras can then directly be connected to one computer.

(a) (b)

Figure 4.8: Example of image pair