I Digital Micromirror Device based optical setup

[Peter_44332]

Hi everyone,

I have a question regarding the design of an optical setup based on a Digital Micromirror Device (DMD) and would love to hear your thoughts.

I am using a multimode fiber-coupled laser (fiber core diameter = 50 µm) to illuminate my DMD. The setup is intended for an optical test stand that can accurately modulate light with a projected pixel size of around 7 x 7 µm. Since the micromirror pitch is 10.8 µm, the DMD projection needs to be scaled down.

My approach to designing the system in Zemax (non-sequential mode) is as follows (see attached sketch):

1. I simulated the fiber end as a Source Two Angle, using the corresponding diameter and numerical aperture (NA).
2. An aspheric lens is placed one focal length away from the fiber end to collimate the light.
3. The DMD is rotated 24° relative to the incoming collimated beam and 45° relative to its own axis.
4. In the ON-state path, I placed another pair of aspheric lenses (f = 50 mm and f = 30 mm) to scale down the DMD projection, resulting in a pixel size of approximately 6 x 6 µm.
5. An aperture is placed between the two lenses to filter out multiple diffraction orders from the DMD.

According to my simulation in Zemax (after optimizing the distances between the first aspheric lens in the ON-state and the DMD, as well as the position of my sample) the setup can create a precise projection of the individual DMD pixels with the necessary resolution. Do you think this approach is effective, or do you have any suggestions for improvement?

Thank you for your input!

 

[Andy Resnick]

Hi everyone,

I have a question regarding the design of an optical setup based on a Digital Micromirror Device (DMD) and would love to hear your thoughts.

Off the top of my head, using a multi-mode fiber could create unintended effects due to speckle.

 

[Gleb1964]

5. An aperture is placed between the two lenses to filter out multiple diffraction orders from the DMD.

The aperture acts as a lowpass filter, destroying high frequencies. I doubt that your projection would resolve pixels taking into account wave nature of light. The geometrical raytracing may show that pixels are resolved, but diffraction would vanish the sharp details.

 

[Gleb1964]

To emphasize what I mentioned earlier about aperture filtering:

When we reimage a periodic object (such as a grating), it produces multiple diffraction orders. If we filter and recombine only the 0th and 1st orders, we obtain a sinusoidal carrier frequency of the periodic object. To achieve a more rectangular shape of the object, we need to recombine higher orders.
Truncating all diffraction orders with an intermediate aperture effectively produces spatial filtering, where none of the periodic structures of the object pass through the aperture.

 

[Gleb1964]

4. In the ON-state path, I placed another pair of aspheric lenses (f = 50 mm and f = 30 mm) to scale down the DMD projection, resulting in a pixel size of approximately 6 x 6 µm.

Regarding this item: your system looks like it not focused, it would not deliver any image.
If your aim to reimage DMD pixels you need an optical relay with linear magnification -0.6x, presumably double telecentrical in object and image space.
Something like this:

The performance of optical relay depends on the aperture opening and the type of illumination, incoherent or coherent. Because your are using laser illumination, it can be the case that you deal with the coherent or partially coherent object, which change a game i bit.
Here is simulation of performance of the optical relay from the picture above, working with your DMD pixels 7x7 um and 10.8 um pitch, using incoherent and coherent illumination:

 

[Peter_44332]

Thanks a lot for your replies. I will try to recombine higher diffraction orders too.

Regarding this item: your system looks like it not focused, it would not deliver any image.
If your aim to reimage DMD pixels you need an optical relay with linear magnification -0.6x, presumably double telecentrical in object and image space.

Thanks a lot for the input and the simulation results!
My idea was to collimate the light before the DMD and to scale the projection of the individual mirrors (10.8 x 10.8 µm) down a bit using the telescope in the ON-path. After the last aspherical lens, the light would then be collimated again. Therefore, I did not focus the light.
Since I am relatively new to optics, I am not entirely sure what you mean that it would not deliver any image . I was assuming this would be sufficient to get a projection of the pixels (since the light is collimated) that are in the ON-state, e.g. a square of light from a pixel in the ON-state.

The performance of optical relay depends on the aperture opening and the type of illumination, incoherent or coherent. Because your are using laser illumination, it can be the case that you deal with the coherent or partially coherent object, which change a game i bit.
Here is simulation of performance of the optical relay from the picture above, working with your DMD pixels 7x7 um and 10.8 um pitch, using incoherent and coherent illumination:

Regarding the light source: I was also trying to simulate interference effects, and will probably use a broadband source combined with a bandpass filter to mitigate these effects.

 

[Gleb1964]

After the last aspherical lens, the light would then be collimated again. Therefore, I did not focus the light.

In that case you would not have pixels projection at your sample plane because diffraction would vanish details.
In the initial post you have mentioned that you want project DMD pixels:

the setup can create a precise projection of the individual DMD pixels with the necessary resolution

 

[Gleb1964]

The process of parallel beam propagation does not need any optics. But the interaction of the beam with DMD mirror would result some diffraction light out of parallel beam direction. That diffracted light contains all the information about interaction. If the optical relay can intercept that diffracted light and focus at sample that would create an image of DMD pixels at sample. If let that diffracted light to spread out of the parallel beam (or filter it by spatial filter) the parallel beam would be clearned as if no interaction with mirror have been.