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.

DLP4710: RTIR to projection lens distance and spill between states

Part Number: DLP4710

1) Can you guys please confirm that there has to be 0% overspill at the projection lens between On, Off and Flat states of TRP DMDs and

2) What is the typical distance between RTIR prism and projection lens for 1.7 r. index RTIR prism and 17 degree illumination beam convergence.

Because there seems to be something wrong in our simulation, overspill is too high at 30mm distance from RTIR to projection lens.

  • Hi Mike,
    I will discuss with my colleagues to what could be the cause for such result and get back to you by tomorrow.
    regards,
    Vivek
  • TRP-Pupil.pdfHi Mike,

    1. Please see the attached  diagram, for a 34 degree illumination angle, the ON state is separated by 34 degree from Flat state. The separation between flat sate and OFF state is 34 degree. The ON state and Flat state are separated by 48 degree.

    One of the common technique used  to increase  contrast is  marginally increase illumination angle by 0.5 to 1 degree. This will shift all those pupil slight up and reducing the stray light in to the projection lens (projection lens is not moved). The contrast gain is at the cost of marginal loss in brightness.

    2. The projection lens can be placed as close as mechanically possible to the prism. The prism separates the ON light from the rest.  There may some stray light due OFF state scattering from edge of the prism. May want to check those too.

    regards,

    Vivek

  • Thank you.

    "This will shift all those pupil slight up and reducing the stray light in to the projection lens"

    What?

    Do you by any chance mean decrease, not increase? (less converging and less diverging)

    And could you please tell me if Flat state light during such operation is possible and if yes how so we know how to reduce that as well?

  • Hi Vivek,

    Attack the ray trace simulation for DLP470TP on and off stage. I'm place a dummy detector on the top of RTIR to check the light profile when DMD on and off state. I'm notice some area of detector is overlap when DMD is On and Off. I'm also found this happen in TI reference design 2500 –d3-720p (last 2 slides), after I remove the projection lens and the image plane is in front the RTIR. Would like to get some suggestion from TI how to solve this issue to get better contrast.

    DLP470TP DMD On and Off issue.pptx

  • Hi ChengWoei,
    I will get it reviewed by our optical expert and get back to you.

    regards,
    Vivek
  • Thank you again, Vivek.
  • Hi ChengWoei,
    Your observation is right and as expected. The On state and OFF state bundle will completely separate after certain distance. Before that point , you will see overlap int those two light bundles. Same is true for TI reference design. There is an aperture in the projection lens to block stray lights. You may consider adding an aperture or masking part of the lens to block such light.

    There are few specific feedback from our team on your illumination design:

    1. Illumination is not telecentric - this may result in coupling problem/issues with eyepiece /wave-guide. It may also cause non-uniformity.
    2. Eye pupil may cover only few cell of the Fly's eye lens array, this may result in uniformity problem.
    3. TIR prism capability is under utilized. You may want to check that you are able o meet your FOV and pupil size requirement down the design. However, this will not be applicable if your system is using a diffuser.

    regards,
    Vivek
  • Hi Vivek, 

    Thanks for the On and Off state explanation. On the feedback, telecentric illumination, mean parallel light source right?

    Regards, 

    Cherng Woei

     

  • Hi Vivek,

    Thanks for the On and Off state explanation. On the feedback, telecentric illumination, mean parallel light source right?

    In previous discussion, you inform that, DMD need to have acceptance cone angle 17 degree. If the illumination light is parallel, there is no cone angle on DMD, right?

    Regards,

    Cherng Woei
  • Thank you, Vivek.

    To add to Cherng's comment ,

    1. Why should the illumination be telecentric? And not sure we can even do that without laser-like point emitter (assuming by telecentric you mean the illumination beam (and therefore projection beam after DMD) is parallel)).
    With OLED microdisplays each pixel doesn't provide a parallel beam but rather a beam angle of a LED.
    So we probably don't understand this.

    2. Why? I assume you mean the human eye pupil.

    3. By diffuser you are referring to the projection screen, right?

    Thanks again for the great support.
  • Hi Cherng and Mike,

    I have attached a PDF drawing explaining telecentric illumination.  This should help. By telecetric illumination, I mean

      - Each pixel “sees” rays from the same cone angle and direction

    1.  Telecetric illumination ensure that TRI prism reflects light uniformly for every pixel.

    2.Yes, human eye- Typically human eye pupil could be ~2-4 mm depending  environment.  A rough estimate that eye pupil need to see around 3x5 cell for uniform image.

    3. Diffuser  - Yes  a projection screen, could be transmissive.

    regards,

    Vivek

    TelecentricIllumination.pdf

  • Thank you again. I'm pretty sure by telecentric illumination you mean collimated (parallel) beam illumination, yes? I don't think this is what you have suggested earlier that the beam can converge/diverge up to 17 degrees based on contrast/brightness requirements. Are you perhaps also referring to case when DMD is viewed via an eyepiece lens rather than projection?
  • Hi MIke,
    Please refer the picture I attached for telecentric illumination. All pixels are illuminated a cone of light at same angle (angle of cone). The cone by itself is not parallel. However, the center ray of the cones are parallel.
    Telecentric illumination is critical when TIR prism is used, independent of eye piece or projection.

    regards,
    Vivek
  •  "Telecentric illumination" is a parallel beam illumination, from the single company I can find the phrase mentioned by www.youtube.com/watch ,but I think I understand what you mean.I will ask Cherng to check this topic again. He already uses a flyeye lens in his design to make the beam profile tophat (uniform rather than gaussian) but maybe you probably mean the converging cones going onto the DMD have to be parallel which is not the case if you add a lens after the flyeye lens. Right?

    Cherng's current design which you say is not telecentric:

    It isn't?

  • Hi Vivek,

    Would like to know, how to confirm the illumination design is correct cone angle for the DMD? when the light is parrallel and still have some cone angle on the DMD pixel. Any suggestion?

    I have done another revision design, in this revision, the illumination light is almost collimated before enter RTIR. The Uniformity of the DMD look goods. The illumination designed was taking count the LED performance and space constrain.  

    However, I have done some modification on design (page 20-23) by making the light become convergence before enter RTIR by following the TI reference design. Can you please  check either this is telecentric design with correct cone angle? However, I'm getting the Guassian beam profile on the DMD active area.

    Maybe, there is some confusion on the collimated light and the cone angle for the DMD pixel.

    We would like to confirm the design is correct before move to next step.Illumination Optics Design for DLP470TP with SBM-40-LC LED rev3.pptx

  • Hi Cherng,
    Thank you for your patience. Here are feedback from our team:

    1. Checking telecetrnic illumination - You have  to check array angle at multiple sample pixels to ensure that they are with-in range define by the F/#. ex +/- 12 degree for F/2.4, +/-17 degree for F/1.7. In addition, please check the center rays at these pixels are in parallel at the best efforts.

    2. Improvement in your design - Please consider increasing distance between Fly eyes and lens after. Please see the TI design, there is significant space between those two. To get a better performance, you may want to consider adding another lens.

    Are you using 4-in-one LED package mentioned in one of the previous thread by Mike? if so, you should not consider as point source. Consider it as surface source and multiple color i.e RGB color from different LED die.

    regards,
    VIvek

  • Thank you, Vivek. I'm sure Cherng will respond himself but let me clear some things up for him before he does.
    Responding to your comment, please respond back when you can.

    1) > You have check array angle at multiple sample pixels ensure that they with-in range define by the F/#. ex +/- 12 degree for F/2.4, +/-17 degree for F/1.7

    Let me rephrase this slightly to make it maybe clearer. Please correct me if the below is wrong interpretation of what you meant:

    1. By array angle you mean individual "light cone" (diverging/converging beam) angle right? Not the angle of the whole array of beams, right?

    2. When you say within range of +/-17 degree you mean "up to +/-17 (0-17)", yes?

    > please check the center rays at these pixels are in parallel at the best efforts.

    In other words, make sure the individual "light cones" (diverging beams) are parallel to each other, using their center rays, yes?
    This is interesting that you say this as it seems like in some reference designs the array (center rays) is converging slightly and not parallel. We should avoid that, right? Illustration of what I mean, exaggerated, where center rays are highlighted in red:


    parallel, center rays are parallel:

    This is what you recommend, yes?

    Converging, center rays not parallel:


    Some reference designs seem to use this:

    I think Cherng got confused due to this and assumed the whole "array" (light cone bundle) has to converge to be considered telecentric. I think this is where the main confusion comes.

    2) In our current design there is one flyeye lens, not two. Is that an issue by itself? Off the shelf double sided flyeye lenses weren't found and we didn't want to make an injection mold for a prototype device.

    > Are you using 4-in-one LED package mentioned in one of the previous thread by Mike? if so, you should not consider as point source. Consider it as surface source

    Yes, it is the 4-LED chip, similar to the 2-LED (RB) chip in Lightcrafter2010.

    When you say "consider as point source" do you mean to start off each beam (each color channel emitter) as a surface source (each emitter is 1mm diameter) in the simulation or do you simply mean to not simulate each color channel/emitter as if originating from the same point?



    And finally, can we take from your post that Cherng's current design is telecentric, if it meets the criteria/recommended parameters mention in point 1 of your post?

  • Hi Vivek, 

    Thanks for the reply. 

    If I'm understand correctly, the illumination design need to follow the DMD tilted angle. The light cone angle need to  same as DMD etendue. I'm attach the DMD etendue document that provided by you here. 

    Besides that, I have done a calculation on the F/#, to understand how you get F/1.7 for 17degree and F/2.4 for 12degree. Please help to check either it is correct. 

    Because there is some space constrain for the product design and the RGB LED die is offset from LED chip, the design may not go for fast optics, like F/1.7. If I'm go for F/5, the light cone angle is will be 5.71 degree.Based on the DMD etendue document page 4, when light cone angle is smaller than the DMD etendue, the contrast is high but suffer on brightness is that correct?

    About the question on the LED, yes, surface source with LED angular profile is considered in the design. That's why there is some difficulty on the design, as the RGB color of LED die unable to align to the Lens center.  

    1447.DMDEtendue.pdfFnumber define from DMD Etendue.docx

  • Hi Mike,
    Your observation is correct.
    1 By array angle you mean individual "light cone" (diverging/converging beam) angle right? Not the angle of the whole array of beams, right?
    Vivek> Yes.
    2. When you say within range of +/-17 degree you mean "up to +/-17 (0-17)", yes?
    Vivek> Yes, cone of 17 degree

    Comment about TI reference design -
    The rays are telecentric at the point you pointed. However, when you consider light across DMD, the overall beam will look like converging at a angle or smaller than of the individual light cone.

    Hi Cherng,
    You angle calculation for F/# are correct
    F/1.7 - 17 degree
    F/2.4 - 12 degree
    F/5 - 5.71 degree

    You may use following formula to directly calculate angle
    = Sin Inverse of (1/(2*F))

    regards,
    Vivek
  • Thank you for the great support once more, Vivek.

    Cherng,
    Your design is not telecentric because your individual light cones are not parallel from each other like in the below illustrations but diverging from each other.

    The TI reference design is telecentric because the individual light cones are parallel. The beam in the TI reference design is not converging, the individual light cones are converging, but they are parallel to each other, so in the edges of the beam bundle (all of the beam cones combined, or the "whole" illumination beam) you still have converging individual beams/cones so the whole bundle seems converging to you, but it isn't.

    I have prepared an animated illustration for you to illustrate it very clearly.
    Here you have light cone starting at 0 degrees (parallel, a single line) and slowly turning into a typical DLP converging light cone.
    There are 3 of them in the first illustration:

    https://imgur.com/HSJu8dJ

    Now here are 7 of them. Since they overlap so much their combined beam appears as converging but it isn't:

    https://imgur.com/bmTGnSI

    Hope this clears the confusion.
  • Hi Cherng and Mike,
    Thank you for kind words. Please feel free to reach out if you have any additional questions.

    regards,
    Vivek