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.

MBMC Request.

Hi,

I came across the application report on MBMC pattern sequence mode (see http://www.ti.com/lit/an/dlpa035/dlpa035.pdf), and would like to submit a request.  Here's the required info:

b)  For my application, I require an exposure time of 375 microseconds.  (Note that it may later be necessary to set this to 385 microseconds.  I intend to trigger the MBMC patterns with an external trigger of period 375 microseconds.  According to the user named PetroGelabert, the minimum trigger period to avoid buffer rotate overhead issues when using over 24 patterns is 385 microseconds: see  http://e2e.ti.com/support/dlp__mems_micro-electro-mechanical_systems/f/850/t/168690.aspx .  However, I have yet to observe any buffer rotate overhead issues when using an external trigger with a period of 375 microseconds.  Anything below that 375 microseconds does create issues though.)

c)  The table at the bottom of my post contains the MBMC pattern information sequence input (sorry about the length, but I'm trying to follow the instructions explicitly).  The table contains 96 rows that defines 96 1-bit patterns that cycle between red, green, and blue LEDs.

d)  The development time for this project is 2 weeks (mid-June).

e)  I'm making this request for two reasons.

First, I want to perform structured light scanning (projecting light and capturing photos) with white illumination at fast rates.  Currently, I'm forced to do structured light scanning using either red, green, or blue light.  Fulfilling this request would enable one to perform structured light scanning with 32 different white structured light patterns by capturing one photo for every three patterns (same pattern displayed three times, but using different LEDs).

Second, I want to perform structured light scanning with color illumination at fast rates.  So instead of displaying sets of three identical patterns using red, green, and blue illumination (as described in the previous paragraph), the three sequential patterns can be different.

Thanks,

Matt

Pattern bit-depth Number of patterns Pattern color
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue
1 1 Red
1 1 Green
1 1 Blue

  • Hello Matt,

    Thank you for sharing the details. I wanted to highlight important point about the trigger period wrt pattern 'Exposure Time'. The MBMC Pattern Sequence the trigger period is computed based on several other parameters. By design it will be higher than pattern exposure time so you will have to program the trigger period >= the value specified in the Readme.txt file. For instance, at 375uSec exposure time for the above MBMC pattern sequence  it will be 735uSec. 

    Please find the requested sequences.

    1bpp RGB MBMC Pattern Sequence with Exposure Time = 375uSec

    1bpp RGB MBMC Pattern Sequence with Exposure Time = 385uSec 

    http://software-dl.ti.com/dlp/embedded/catalog/dlp3000/dlp3000mbmc.html

    Regards,

    Sanjeev

  • Thanks Sanjeev.

    For my application, a short trigger period is very important.  So, I'd like to request one additional sequence to minimize this trigger period.  Presumably, this can be done by using the minimum 250uSec exposure.

    Also, is it possible to shed a little more light on how the minimum trigger period relates to other pattern sequence parameters (number of patterns, bit depth, exposure, ...)?  Specifically, is there a way to further reduce the minimum trigger period (i.e. by reordering the patterns to display 32 red patterns, then 32 green patterns, and finally 32 blue patterns)?

    Cheers,

    Matt

    EDIT: Also, can the trigger period be minimized even further by limiting the number of patterns from 96 to 24?

  • Hello Matt,

    Apology for delay in response to your email. 

    Trigger Period calculation in MBMC Sequence:

    Before answering, let me explain a bit about the how the 1bpp and non-1bpp (i.e., 2bpp, 3bpp, ...8bpp) pattern sequences are operating. Earlier for 1bpp pattern sequence the pattern 'exposure time' is controlled by FPGA. For the other bit-depth pattern sequences the 'exposure time' is directly controlled by sequences and the FPGA is configured in by-pass.

    Now, in case of MBMC Sequences you can have 1bpp and any other bit-depth patterns, because of this the FPGA has to be always in by-pass wrt to 'exposure time' signal generation.

    So for the 1bpp pattern display the 'trigger period' information is programmed as follows -

    1st 220uSec pattern data load on the DMD. Then generate the signals for the 'exposure time'. For > 24 patterns include buffer rotation over-head = 135uSec. 

    To display 1bpp pattern at 250uSec in MBMC sequence require a minimum of 220uSec + 250uSec + 5uSec (turn-off exposure signals) + 135uSec = 610uSec.

    For the case < 24 patterns you can exclude the 135uSec buffer rotation time. In this case the minim trigger period will be 220uSec + 250uSec + 5uSec (turn-off exposure signals) = 475uSec should work.

    The sequences will be put here http://software-dl.ti.com/dlp/embedded/catalog/dlp3000/dlp3000mbmc.html

    Regards,

    Sanjeev

  • Great, very useful information!  Thanks Sanjeev.

    I'll work with the sequences you sent me earlier.

    Cheers,
    Matt