Because of the Thanksgiving holiday in the U.S., TI E2E™ design support forum responses may be delayed from November 25 through December 2. Thank you for your patience.

  • State Resolved
  • Locked Locked
  • Replies 15 replies
  • Answers 5 answers
  • Subscribers 30 subscribers
  • Views 2025 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DLP3010: smaller optical engines for VR/AR usage?

Armen Tmsn
Intellectual 900 points
Part Number: DLP3010
Other Parts Discussed in Thread: DLP2010

I have notice that the optical engine providers are all making optical engines ready to be used for pico projectors. None for home theatre or office video projectors, and especially for VR/AR.

Issue is, they seem to be making *all* of their optical engines to be as bright as the particular DMD can handle before reaching its damage threshold caused by overheating. So the optical engine used in Lightcrafter 2010 is much less bright and uses much smaller LEDs than the one used by Lightcrafter 3010.

That said, for things like VR/AR, you do need as high resolution as possible, at least 720p or 1080p but you can't afford the size of the weight and size of the OEM optical engines and this seems solvable as you don't need anywhere near 300 lumens, 50 or less should be sufficient.

So one of my questions would be, is it possible to use smaller LEDs and therefore smaller optical engine compartment (where the LED collimator lens, dichroic mirrors and flyeye lenses go) before the larger DMD and just have a beam expanding lenses before the flyeye lenses? For example use the tiny LEDS from Lightcrafter 2010 for DLP3010?

For comparison here is the size difference of a DLP2010 optical engine vs a DLP3010 optical engine:

DLP2010: 

DLP3010: 

It is about twice as big while DMD size difference is only 0.1 inches.

Second question would be, is it possible to use smaller projection lenses after the DMD? The lens size difference between lightcrafter 3010 and lightcrafter 2010 is huge and a big deal for VR. If yes, what would it require and why isn't it used by all OEM optical engines since smaller projection lenses I would assume would be more compact and cheaper to manufacture?

My final question would be, are there 3rd party partners you guys know that have some DLP3010 optical engines meant for VR/AR with the benefits I mentioned (smaller leds, led collimators and everything related to the LEDs before the DMD) or something they can work upon rather than starting from scratch? Because I would love to save money and time on a premade or not completely custom design built from the ground up.

I haven't found anything of the sorts from the optical engine providers , they are all focused on pico projectors exclusively.

  • 0
    TI__Mastermind 28005 points

    Armen,

    Have you tried contacting Vuzix about this? I believe they offer AR kits based on the 2010EVM with a modified light engine. I do not know if they have a 3010EVM variety available, however.
    https://www.vuzix.com/


    I hope this helps.

    Best Regards,
    Philippe Dollo

  • 0 in reply to Philippe Dollo
    Intellectual 900 points

    Are they product manufacturers or also OEM optical engine providers and custom optical engine design service?

    Doesn't look like they provide OEM optical engines.

    Also, this doesn't answer my first two questions.

  • 0 in reply to Armen Tmsn
    TI__Mastermind 28005 points
    Armen,

    I don't believe they provide custom optical engines. As for your other questions:

    "So one of my questions would be, is it possible to use smaller LEDs and therefore smaller optical engine compartment (where the LED collimator lens, dichroic mirrors and flyeye lenses go) before the larger DMD and just have a beam expanding lenses before the flyeye lenses? For example use the tiny LEDS from Lightcrafter 2010 for DLP3010?"

    Yes, it is certainly possible to use a smaller design with a larger DMD. Of course, the design will then be unable to make full use of the DMD's maximum possible brightness.

    "Second question would be, is it possible to use smaller projection lenses after the DMD? The lens size difference between lightcrafter 3010 and lightcrafter 2010 is huge and a big deal for VR. If yes, what would it require and why isn't it used by all OEM optical engines since smaller projection lenses I would assume would be more compact and cheaper to manufacture?"

    You can use a smaller lens, with the caveat that you must ensure that your etendue is large enough to capture all the light coming off of the DMD. If your lens is too small, you will lose brightness.

    I hope this helps.

    Best Regards,
    Philippe Dollo
  • 0 in reply to Philippe Dollo
    Intellectual 900 points

    Thank you for the informative response.
    As my optics knowledge is limited, I have to ask, when you say the projection lenses can be smaller in cost of brightness, how smaller? Can they be as small as the DMD itself, or even smaller? It is about the size acceptable for VR headsets (not AR) and fortunately here brightness requirements are very forgiving as the whole projection screen is the size of a mobile phone screen and if it were used in pitch darkness.

    I do not believe anyone has developed a VR (not AR) DLP based head mounted display yet so forgive me asking so many question, we just don't have anything to use as reference. Avegant Glyph and similar products are not true head mounted displays but "personal cinemas" not using projection screens and large fields of view.

    A great deal of space (and weight) is taken by the prisms that direct the illuminating beam to the DMD and then from it to the projection lens. Are there alternative reference designs that do not involve large prisms?

  • 0 in reply to Armen Tmsn
    TI__Expert 4610 points
    Hello Armen,

    This is a very difficult question to answer without actually doing a projection lens design to optimize for size.

    DLP3010 EVM has the projection lens approx 50 mm in length and 21 mm in diameter. You can see it here : www.ti.com/.../technicaldocuments

    The lens for this EVM is optimized for max brightness, the largest aperture the DMD can support without compromising contrast and for 100% image offset. Most likely the lens consists of 6-7 lenses which is typical for a the 3010 DMD with offset(which in VR is not really necessary). If you do not require offset, you could probably reduce the lens count down to 5 or 6 lenses. By reducing the lens count, this would reduce the length of the lens. The diameter would also decrease significantly. From our design team experience, a lens designed for the 3010 DMD with 0% image offset (on axis projection) can be 10-12mm in diameter. Now if you also slow down the F/# it may be possible to reduce down to 4 lenses, further reducing the length and slightly the diameter.

    The lens cannot be as small as the DMD unfortunately. You do have other contributing factors such as a prism or airspace in front of the DMD that increase the size of the projection lens. This is required to separate the illumination bundle of light rays from the projection bundle so that the system can have good contrast.

    I hope this gives a little background on the contributing factors for projection lens size. Unfortunately without an actual design for the required specs, it is difficult to say exactly how small it can actually get.

    Thanks,
    Navneet
  • 0 in reply to Navneet Singh
    Intellectual 900 points
    Thank you.

    First I would like to mention that while I do not have the 3010 EVM I do have the 2010 EVM.
    The 0.2 inch DMD in millimeters would be 5.08 mm and that seems about right from the datasheet width and height. www.ti.com/.../dlp2010.pdf (page 17)

    This is the lens of 2010EVM I measured the diameter of: i.imgur.com/IUO0nJo.jpg
    It is at least the same size as the DMD, unless I have the wrong info on DMD size.
    Looking inside it appears while the first lens elements are wider (but don’t appear bigger in height) the final lens(es) are not bigger than the DMD. i.imgur.com/1dpV1BT.jpg

    I do not need the image offset to be anything other than 0, although the 2010EVM has 100% offset and doesn’t seem to need big lenses. How can you explain this?
    And I don’t need max brightness so I probably need smallest aperture, right?

    “The lens cannot be as small as the DMD unfortunately. You do have other contributing factors such as a prism or airspace in front of the DMD that increase the size of the projection lens.”

    Why do the prisms and airspace matter, as long as the illumination beam and resulting projection beam after the DMD are kept collimated?

    I would like to clarify, my goal is not to have all of the lenses the same size as the DMD, something like the 2010 EVM lenses will do (wider but not higher in size).

    My goal is to have the beam parallel (as much as possible) and small (as much as possible) before they reach relay lenses for each eye screen of the VR device. Just think of it as there being a beamsplitter cube in front of the initial projection lenses and same final lenses for each screen after few cm and polarization and LCD shutters determining which frame each eye sees.

    I think the best idea is to have a parallel (collimated) projection beam before it reaches the projection lenses for each eye. Is that possible?

    From page 15 of this document on your website it appears it should be but I would like a confirmation www.ti.com/.../dlpa044.pdf :

    “It is possible to project an image of the DMD which is smaller than the actual DMD. This is used in
    lithography and other applications when it is desired to reduce the size of the pixels and cover a very
    small image field. Examination of the lens equation (Equation 2) shows that if the distance to the DMD, d1,
    is increased to exactly twice the focal length of the lens, the image formed at d2 will be exactly the size of
    the DMD. That is, the magnification will be unity (1×), or actual size. If the DMD is moved to a greater
    distance, d1 >> f, the image formed at d2 will be smaller than the DMD. This means that demagnification is
    possible. In practice, the amount of demagnification is limited by the physical distance that the DMD can
    be separated from the lens, by numerical aperture (NA) considerations, and by the wave nature of light
    (diffraction).”

    From the feedback I got from previous employee:
    “You can use a smaller lens, with the caveat that you must ensure that your etendue is large enough to capture all the light coming off of the DMD. If your lens is too small, you will lose brightness.”

    About how much small is too ”small”?

    Thank you very much.
  • 0 in reply to Armen Tmsn
    TI__Mastermind 23070 points
    Hi Armen,
    This thread has continues for to long and context is lost. Could you please summarize your questions? I will try to get you information.
    regards,
    Vivek
  • 0 in reply to Vivek Thakur
    Intellectual 900 points
    The questions are in the post right above yours.
  • 0 in reply to Armen Tmsn
    TI__Mastermind 23070 points
    Hi Armen,
    Answers to your questions will depend on the actual optical engine design and trade-offs. It is not feasible to provide answers to all these question in forum. You need to engage with the optical designer and make those trade-off based on your application requirement. I have tried to capture responses to questions. For further detail and result you need to go through design process.

    1. Smaller LED – Yes, you can use smaller LED for larger panel if it meets your brightness requirement. You can additionally leverage smaller LEDs :
    - If you decide to use multiple Led in one package (RGB in one) subject to that combined package size matches with Etendue of DMD. It will simplify illumination optics in the system eliminating dichroic and reducing number of lens.
    - You can also consider using slower F#, this not only reduces the size but also improves contrast ratio of the of the system.
    2. Your comment -“I think the best idea is to have a parallel (collimated) projection beam before it reaches the projection lenses for each eye. Is that possible?”
    It is not practical because LEDs are not point source and we have to image LED on DMD array. The light beam would be expanding after reflection from DMD.

    3. Smaller lens of DMD – It is part of optical design. It will also depend on optical configuration telecentric or non telecentric. The distance between DMD and lens is also critical factor in size. Please note that light after DMD is expanding and as distance from DMD increases the light cone is bigger. In case of telecentric system, you have prism between DMD and lens. Whereas for non-telecentric system, the lens can be placed after incoming light cone has separated from projection.

    The size of the lens has direct trade-off on how much light is collected after DMD. This will not only impact the brightness but also efficiency because you are not collecting all light. The collection efficiency will also depend on the F# of the optics. As you can see there is no one number or answer, it depends on actual design.
    Hope this helps.

    Regards,
    Vivek
  • 0 in reply to Vivek Thakur
    Intellectual 900 points

    Of course, I just want to make sure I am going to a discussion with an optical engineer knowing that what I need is possible to begin with.

    1. My only concern is the DMD will not be illuminated evenly by each channel as the red , green and blue LEDs on the chip are offset from each other by 0.5mm and are 1mm each. That said, I have the Lightcrafter 2010 which has the red and blue (but not green) LEDs on one chip and the projection seems evenly illuminated. I'm not sure which optical element helps with that, perhaps the flyeye lens? i.imgur.com/1dpV1BT.jpg

    2. Do you mean there is an inherent characteristic of how the DMD works that any beam reflected by it becomes expanding or you think it would be hard to perfectly collimate the LED?

    What about using laser diodes?

    3. "Please note that light after DMD is expanding" . Please check my response to 2. What causes the expansion of the beam after the DMD?

    "Whereas for non-telecentric system, the lens can be placed after incoming light cone has separated from projection."

    Are there any articles, reference designs or at least diagrams/illustrations for such a design? All the DLP otpical modulates I have evaluated use a TIR prism. I wasn't aware there is another method.

    "The collection efficiency will also depend on the F# of the optics. As you can see there is no one number or answer, it depends on actual design."

    Sure, but as you see I am able to provide the design requirements.

    "Hope this helps."

    Yes, very much. Thank you.

  • 0 in reply to Armen Tmsn
    TI__Mastermind 23070 points
    Hi Armen,
    Please see my response to your questions:

    Armen Wrote….
    "1. My only concern is the DMD will not be illuminated evenly by each channel as the red , green and blue LEDs on the chip are offset from each other by 0.5mm and are 1mm each. That said, I have the Lightcrafter 2010 which has the red and blue (but not green) LEDs on one chip and the projection seems evenly illuminated. I'm not sure which optical element helps with that, perhaps the flyeye lens? i.imgur.com/1dpV1BT.jpg"

    - Yes , it is critical that DMD is uniformly illuminated, This is achieved by the illumination optics which could be a combination of len(s) and Flyeys or len(s) with light tunnel.

    Armen Wrote….

    "2. Do you mean there is an inherent characteristic of how the DMD works that any beam reflected by it becomes expanding or you think it would be hard to perfectly collimate the LED?
    What about using laser diodes?"

    - No, DMD is simple mirror and it reflects the light. “Perfectly collimate the LED” is little ambiguous. Ray coming of every point of LED may be perfectly collimated but those collimated rays from multiple point may not be going in same direction.
    The illumination optics is used to image the LED on DMD. DMD is in object plane which projected on the screen. Often aspect ratio of LED is different than that of DMD, as a result uniformity and efficient capture of LED lights becomes challenging.
    Laser diode will be easier to collimate because they are point source but the challenge is to uniformly illuminate DMD.

    Armen Wrote….

    "3. Please note that light after DMD is expanding . Please check my response to 2. What causes the expansion of the beam after the DMD?
    "Whereas for non-telecentric system, the lens can be placed after incoming light cone has separated from projection."
    Are there any articles, reference designs or at least diagrams/illustrations for such a design? All the DLP otpical modulates I have evaluated use a TIR prism. I wasn't aware there is another method."


    - You can refer to “DLP™ System Optics” application note. It will give you an overview about various design aspect.
    www.ti.com/.../dlpa022.pdf

    Please also refer to section 4.2, figure 2 of “Optical Engine Reference Design for DLP2010 Digital Micromirror Device” application note. This section lists design option for optical engines.
    www.ti.com/.../dlpa074a.pdf

    Hope this answers your questions.

    Regards,
    Vivek
  • 0 in reply to Vivek Thakur
    Intellectual 900 points

    Thank you very much for your detailed answer, however I believe there is still one detail left out or unclear to me.

    You say LED rays may be collimated but they may not be going in the same direction. I'm sorry but isn't going in the same direction or being parallel the definition of collimated? The same way it is still unclear to me why you said "It (skipping this step of needing to create an extra focal plane with extra lenses for the beamsplitter by having the illumination LED beam going to the DMD collimated) is not practical because LEDs are not point source and we have to image LED on DMD array. The light beam would be expanding after reflection from DMD." Why does the light source have to be a point source, and then, why won't lasers work for this? And in your second sentence you say the beam will be expanding after reflection from the DMD and by the wording it is suggested that the DMD will be the cause of the expansion because you mention the expansion will happen after the DMD but now you say DMD is just a mirror so it may be just poor wording. But then, is it expanding before reaching the DMD as well? Why?

    I am simply struggling to understand the reasoning behind having a non collimating beam before the DMD and collimating it after it passes the DMD with projection lenses when needed. If, as you explained, the DMD doesn't cause divergence by itself and is like any other mirror then for cases where you need the beam to be as small as possible or needing to mask or split the beam with a clean cut without needing to create an extra focal plane with extra lenses, you would want it collimated as close to the LED as possible (before the DMD), so having it expanding after the DMD and dealing with beam collimation with bigger lenses after the DMD where the optical path has needlessly increased seems like a bad design. Right?

    Having an expanding/diverging beam exiting the DMD seems only a good idea when you are certain the projection lenses are going to be bigger than the DMD and want the design of the illumination engine to be independent from the projection lens by having a diverging beam and determining size of the projection lens based on the distance from the DMD to the first element of the projection lens (since the beam is expanding, the farther it is the bigger the beam). Is this a correct understanding behind the reason of not collimating before the DMD?

    If yes, then for my specific use case which I admit is probably less than 5% of use cases where I want the projection and projection beam to be as small as possible, I should rather have the beam collimation done before the DMD, similar to how laser collimator lenses are better to be put right in front of the laser dye rather than after other lenses, because with VR you need about 10 lumen brightness or less but need lenses as small as possible as these are wearable devices and weight is one of the most important requirements to be met.

    Do I have the correct idea? Because all the articles seem to focus on pico projector engines where brightness and high FOV projection beam (not long throw) is of most importance rather than the other way around so this idea of why beam is not made parallel/collimated before the DMD doesn't appear to be addressed.

    To give you an idea on how different VR/AR illumination/brightness requirements are different from pico projectors, I can say that it should be acceptable to simply crop the LED beam to account for the different aspect ratios between the LEDs and the DMD.

  • 0 in reply to Armen Tmsn
    TI__Mastermind 23070 points

    Hi Armin,

    Thank you for your patience.

    I have attached a PDF file to  explain " LED ray may be collimated but they are not going in the same direction."

    If you try to try to use such collimated beam, the efficiency will be very low. For this reason, illumination optics is used to image LED on DMD and light uniformity across DMD is critical.

    I hope this explains.

    Regards,

    Vivek

    CollimatingLED.pdf

  • 0 in reply to Vivek Thakur
    Intellectual 900 points

    Thank you. Have you kept the beam coming out of the lens still very divergent for the sake of illustration?

    Does this also apply to laser light source?

  • 0 in reply to Armen Tmsn
    TI__Mastermind 23070 points
    Hi Armen,
    Those are the actual ray traces for that combination. Divergence will be higher as you move away from center of LED & optics.

    Even for laser , we want to image image laser to DMD and eventually DMD to the projection screen or pupil for eyepiece. The DMD has to be in image plane because it has image .

    regards,
    Vivek