1.1

Selected commercial digital micro-mirror device

Considering the targeted specifications for the SURPRISE demonstrator, the model DLP7000 from TI was selected in its evaluation kit version (DLPLCR70EVM) together with a controller (DLPLCRC410EVM) to facilitate performance tests. The main optical specifications of this DMD are summarized in Table.

Table 1-1.

DLP7000 optical specifications.

Each micro-mirror of this DMD can be actuated independently along binary positions, either deflected at +12° or -12° with respect to the horizontal plane.

2.1

Selection of materials

Sapphire was selected for replacing the protective window since this glass presents a high transmittance over a wavelength range from 400 nm to 5 μm, is characterized by a high hardness and its coefficient of thermal expansion (CTE) almost matches the metallic housing frame. The window was procured from Crystran (UK) with a diameter of 25 mm and 1 mm thickness. The aperture covers the full DMD size, and a 1 mm thickness offers a sufficient mechanical resistance to the shear stress resulting from the assembly process. The frame was manufactured in Kovar, with a CTE that matches both the sapphire window and DMD frame.

Figure 1:

Exploded view showing (bottom-up): unwindowed DMD, window frame in Kovar, and sapphire window.

2.2

Indium wire sealing

A CAD model of the DMD, replacing window and frame was designed, as shown in Figure 2.

Figure 2:

Exploded view showing (bottom-up): unwindowed DMD, indium wire (crushed shape), window frame, and window.

Prior finalizing the designed reworking procedure, we realized a preliminary validation test of indium sealing. This test was conducted onto an existing DMD that was sealed with its package into a mechanical housing. The resulting assembly was then tested with a helium leak detector. This test proved a lack of hermiticity of the assembly. After the leak detection test, the DMD was disassembled for further inspection and confirmed a poor adhesion of the frame onto the housing. Figure 3 shows pictures of the different parts after disassembly.

Figure 3:

Disassembled DMD and housing, together with the indium wire. It is noteworthy that the indium wire remains intact into the groove of the mechanical housing.

This first realization led to the conclusion that indium sealing was inappropriate for the following reasons:

2.3

Epoxy resin sealing

Considering the results obtained with the indium wire sealing implementation, we investigated a second approach based on epoxy resin sealing. The final reworking procedure consists of the following steps:

Initial tests followed a procedure in which steps 2 and 3 were combined and performed in a clean environment but at a normal atmosphere. These units were initially functional, but were progressively deteriorated, featuring more and more non-functional pixels. Further investigation showed that the presence of oxidants such as oxygen and water vapour inside the encapsulation were the main cause of damage of the micro-mirrors flexing structures.

The procedure had to be adapted to perform the encapsulation inside an inert or oxidant-free atmosphere such as nitrogen or any inert gas. Due to equipment availability, the aforementioned solution using a glovebox and argon was selected showing good results. Unfortunately, the stability of the solution was not sufficient, as eight days after encapsulation the micro-mirrors started showing the same problems. The final modification in the procedure to guarantee a longer-term hermicity of the encapsulation, the parylene deposition, proved to be sufficient, as at the time of writing, the reworked units are fully functional after 10 months.

This method allows obtaining a highly hermetic package and facilitates the development of a reliable and repeatable process that fulfils the application requirements.

Figure 4:

DMD on the milling machine, removal of the edges of the original DMD package.

Figure 5:

Left: DMD without its cover and epoxy resin sealing (grey ribbon along the edges). Right: DMD with the replacing window frame and sapphire window in place.

Figure 6:

Final assembly of the reworked DMD with Kovar frame and sapphire window before parylene deposition.

3.1

Visible spectrum transmission

The aim was to measure the spectral reflectivity of micro-mirrors in the wavelength range 400-2000 nm both with the new sapphire front window and without. The test was carried out using Agilent Technologies’ Cary 7000 UV-Vis-NIR Spectrophotometer, including Cary Universal Measurement Accessory (UMA) for reflectivity measurement. The characterization was performed with micro-mirrors in off-state. The resulting reflectivity in VNIR range of all both samples is depicted in Figure 7.

Figure 7:

VNIR reflectivity test results with (left) and without (right) sapphire window.

Interestingly, different incidence angles and polarization states do not greatly affect the reflectivity of both samples. No considerable impact in the reflectivity is seen between the samples with and without sapphire window.

3.3

Switching time

The objective of this test was to measure the switching time of a small set of micro-mirror pixels from ON to OFF positions and vice versa. Then, perform a statistical analysis over several random regions of the DMD. This test aimed at verifying the maximal achievable frame rate. Figure 9 presents the assembled test setup.

Figure 9:

Switching time test setup including light source in the middle, photodiodes in top and bottom and DMD on the right.

A focused light source illuminates a small area of the DMD where all pixels are switched from ON to OFF states and vice versa. Two optical lenses collect the reflected light onto two different photodiodes (PD), either when in ON state (PD1) or in OFF state (PD2). An oscilloscope records the signals arising from both photodiodes. Repeated several times, a statistical analysis of the switching time is performed. Since the DMD is mounted on a x-y translation stage, statistical analysis over several areas can be performed.

Results were obtained for 6 random positions on the active part of the DMD. For each position, 10 measurements were recorded during the switch from one state to the other. Figure 10 plots the results for a single position showing the mean and standard deviation of the switching time for all repetitions.

Figure 10:

Switching time test results for a single position.

It is first worth noting that as the photodiodes are not the same model, some differences are appreciable in the output data from each one. Mainly, PD2 features a lower output gain leading to a smaller signal-to-noise ratio. Nevertheless, this difference does play any role on the analysis. The signal from both PDs is simultaneously sampled at 0.5 GHz and triggered by either rising edge of PD1 or PD2. The logged data is then normalised to force low and high states at 0 and 1, respectively. The transition from one state to the other is defined at 0.5 threshold. Due to lower magnitude of PD2 signal, greater sensitivity to small misalignments is experienced, generating over and undershoots of the signal. As the shape and length of such characteristics are fully determined by the alignment, they are assumed to happen at high level, and not during transition.

To summarise, considering all measurements, the mean and standard deviation values of the switching time were measured as 2.99 and 0.05 μs. This short switching time ensures achieving the maximum frame rate of 32’552 Hz (30.72 μs) stated in the DLP7000 datasheet and presents a high stability.

3.4

Pointing stability

The objective of this test was to measure the pointing direction repeatability and deviation for a single micro-mirror pixel. Then, perform a statistical analysis over several random pixels of the DMD. This test aimed at assessing the necessary alignment precision and compatibility with the SURPRISE demonstrator design. Figure 11 presents the assembled test setup.

Figure 11:

Pointing direction stability test setup consisting of light source on the left, high-resolution CMOS camera in the middle and DMD on the right.

A collimated light source illuminates the DMD where all pixels are in OFF state. A single mirror on the optical axis of the detection system is then turned ON and OFF and one image is recorded to measure the reflection position on a camera, each time the mirror is turned ON, after stabilisation. By placing a camera (detector only, no lens) at a certain distance, computing the centroid of the recorded pattern enables resolving micro-mirror angle variations below 0.1°. The angular resolution will be given by the distance from the mirror to the camera, the pixel size and the precision in computing the centroid. The camera is located at approximately 45 mm from the DMD, considering a camera pixel size of 2.74 μm, an angular resolution of 0.0035° is achieved. Since the DMD is mounted on a XYZ translation stage, statistical analysis over several pixels can be performed.

Results were obtained for 6 random single micro-mirror on the active part of the DMD. For each pixel, 6 repetitions were recorded. For each image, as shown for a single pixel and two repetitions in Figure 12, a gaussian distribution is fitted for both X and Y positions, giving the centre (mean μ) and width (standard deviation σ) of the light intensity.

Figure 12:

Pointing direction stability test results for a single pixel and two repetitions showing the normal distribution fit for X and Y pixels, as well as their mean and standard deviations.

To summarise, considering all measurements, the standard deviation of the angle was measured to be inferior to 0.025°.