• State Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 2 answers
  • Subscribers 30 subscribers
  • Views 263 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.

Original question:

DLP9500: DLP9500BFLN

DLP9500: DLP9500

Park Sunghun
Prodigy 20 points
Part Number: DLP9500
Other Parts Discussed in Thread: DLPC410

Hello
First of all, thank you for your kind reply.
I'm asking you an additional question, so please check it out.
1. What is the actual skill level of Data Load Time 5.76 μs/2Block?
(Ex) 3.0-5.76 μs.
2. What is the actual skill level of 4.5 μs from Reset Start to Mirror Reaction Start?
(Ex) 2.0-4.5 μs.
3. What's the actual level of 8μs from Mirror Reaction Start to Mirror Reaction Complete?
(Ex) 2.5-8 μs.
This is an additional inquiry, so please check it.
Thank you.

  • +1
    TI__Expert 3425 points

    Hi Park,

    Welcome back to the E2E forum!

    The team is looking into these questions and will get back to you by next week. 

    Thank you for your patience.

    Regards,

    Akhil

  • 0
    TI__Mastermind 43225 points

    Hello again Park,

    Please help me to answer your inquiry by clarifying what is meant by "the actual skill level" 

    1. What is the actual skill level of Data Load Time 5.76 μs/2Block?
      (Ex) 3.0-5.76 μs.  To load all rows in the device actually takes 43.2 μs.  Therefore one block (72 rows) takes 2.88 μs so that two blocks will take the 5.76 μs as you have shown.  This is running at 400 MHz clock rate.  Each row is 16 clocks (16 * 2.5 ns = 40 ns), so the time to load 72 rows is fixed to 2.88 μs within about 0.25 ps.  You can insert an even number of extra clocks between row loads, but it cannot be faster.  If you do not need to (re)load all rows, then you can load it in a shorter period of time calculated [40 ns *  n] (where n is the number of rows).
    2. What is the actual skill level of 4.5 μs from Reset Start to Mirror Reaction Start?
      (Ex) 2.0-4.5 μs.  This time is fixed at 4.5 us, regardless of the number of pixels in a block(s).  This is required by the physics of the pixel and the Mirror Clocking Pulse logic in the DLPC410 controller.
    3. What's the actual [skill] level of 8 μs from Mirror Reaction Start to Mirror Reaction Complete?
      (Ex) 2.5-8 μs.  Again this is fixed by the physics of the pixel.  This is value is listed in the data sheet and operating faster than the value listed is not guaranteed.

    NOTE:  The 8 μs is for the mirrors to mechanically settle before addressing them again.  This is only for the blocks that have received a mirror Clocking Pulse.

    FIzix

  • 0 in reply to Fizix
    Prodigy 20 points

    Hello
    Thank you for your kind reply.
    Looking at your reply, it is understood that the mirror does not move during the 4.5 μs from Reset Start to Mirror Reaction Start time. If the mirror doesn't move, we're going to expose it by that time, so please check it.
    Thank you.

  • 0 in reply to Park Sunghun
    TI__Expert 3700 points

    Hello Sunghun,

    Yes, we will verify your assumption for this platform.

    Regards,

    Matt

  • 0 in reply to Park Sunghun
    TI__Mastermind 43225 points

    Hello Park,

    This is not correct.  The 4.5 μs is the time that the controller marks Reset Active.  The Mirrors start moving almost immediately.  For this platform you can easily look at the reset waveform on the two MBRST test points.  If you use a Global Reset, you can look at the light output during the operation.  

    In a typical optical system (like most consumer projection systems) the light moves out of the output aperture very soon after the reset waveform starts for pixels that are going from on to off.  For pixels that are going from off to on, it takes a little longer to cross over so that the light enters the pupil.

    Exactly how quickly is very dependent on the system optics.  We encourage you to make a simple measurement, using a high speed photodiode and an illumination bundle that is similar to your desired operation.  This can be set up on an optical bench.

    I hope this helps.

    Fizix

  • 0 in reply to Fizix
    Prodigy 20 points

    Hello Fizix

    Thank you for your kind reply.

    Q. When do micromirrors start transition to the next state?

    DLPC410 datasheet states that 4.5us (RST_ACTIVE high period) + 8us (micromirror settling time) is required for micromirrors to complete their transition. It seems like micromirror reset pulses (MBRST) are prepared during RST_ACTIVE high period and all micromirrors are found to be fully tilted to either on or off state after additional 8us is expired. But when do the micromirrors start to move? Can the motion occur　during RST_ACTIVE high period? Or is it guaranteed to take place only after RST_ACTIVE returns to low state? If the latter is true, is it OK to turn on a light source during RST_ACTIVE high period and turn off only during micromirror settling time?

  • 0 in reply to Park Sunghun
    TI__Mastermind 43225 points

    Hello Park,

    The actual crossover or stay movement happens during the 4.5 μs.  There is a brief period where the mirror voltage goes from V_bias to V_reset.  This is around 100 ns for this pixel.  As soon as the mirror voltage goes up to VCC2 (V_offset) the mirrors begin their flight to either the other side or to return to the same side if the pixel data has not changed.

    Please see this IEEE publication for more information:

    C. Gong and D. Mehrl, "Characterization of the Digital Micromirror Devices," in IEEE Transactions on Electron Devices, vol. 61, no. 12, pp. 4210-4215, Dec. 2014, doi: 10.1109/TED.2014.2361855.

    The Controller starts RST_ACTIVE and the Micromirror Clocking Pulse simultaneously, so that within 100 ns the mirrors begin moving.  The mirrors have completed their primary movement during the 4.5 μs but continue to "bounce" a little during the 8 μs that follow.  The result is that the data cannot be addressed to the memory (data) until the full 12.5 μs is complete.  Your optical setup will determine whether you can "see" the mirror settling. 

    So the answer to your question is that the primary micromirror motion occurs during the RST_ACTIVE high period.

    Fizix

  • 0 in reply to Fizix
    Prodigy 20 points

    Thanks for the detailed explanation about the behavior of micromirrors. I have read the IEEE paper (C. Gong et al.) and found Fig. 6 showing micromirrors' transition between on/off state is almost done within 4.5us. Is the 4.5us duration of RST_ACTIVE high period is designed to notify micromirrors are in motion? Or　is it just irrelevant? Also, is it safe to assume that bouncing of micromirrors during 8us of micromirror settling time is so small that the light source be turned on then? Wondering if there is any documented guidelines on when to turn on/off the light source, though it could just be dependent on application or user's optical system.

  • 0 in reply to Park Sunghun
    TI__Expert 3700 points

    Hello Sunghun,

    When we finalize our specifications in the datasheets, we have to account for as many manufacturing, chipset, and system variables as possible. There may be differences in DMD, or controller, or board design that impacts the timing of the devices. The 4.5us and 8us that you mention are given that way to allow as many users as possible to design with the TI technology.

    Hopefully you have enough info to make the best design choices for your system. Please let us know if you have other technology questions.

    Thank you,

    Matt

  • 0 in reply to Matt Soucek
    TI__Mastermind 43225 points

    Hello again Sunghun,

    Let me add to Matt's comments.  The system optics are the biggest factor in where in the micromirror "reset" cycle you can turn the illumination on.

    Specifically, if your illumination under-fills the output pupil by quite a bit, then you might be able to turn the illumination on sooner.  If you are working with a large F# and the illumination closely matches the output pupil, you can and will see the bounces as variations in the output brightness.  This is why it is very dependent on the system optics.  

    Fizix