• State TI Thinks Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 4 answers
  • Subscribers 30 subscribers
  • Views 1834 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

DLP 7000 0.7 XGA DMD triggering problem

Siqi Li
Prodigy 120 points
Other Parts Discussed in Thread: DLP7000, DLP7000UV

Hello,

We have a DLP700 0.7 XGA VIS DMD module with Vialux controller suit. We send in an external trigger signal at 120Hz. We set the trigger pulse widht to 6ms and delay to 0 (in another case we set the delay to 1ms). Then we start to see dark pixels under microscope. Attached is the microscope image focused on the pixel plane. It looks like these pixels are stuck in either ON or OFF position. When powered, we could still load images onto the DMD but these marks will not recover. From the microscope images, it looks like damage occurs along vertical lines, which is parallel to the longer dimension of the DMD. When we set the delay to 0, it shows rows of pixels flipping ON/OFF and the row scrolling across the chip.

Could you help us understand what is going on? What is the mechanism to do the triggering and timing? Does it do it row by row scrolling through the chip? Why are we seeing this kind of damages, and is there a way to recover it?

Thanks.

Siqi

  • 0
    TI__Expert 6370 points

    Hello Siqi,

    Can you give me a better idea of the application this DMD has been used for?

    For more information on micromirror reset, take a look at the Device Functional Modes section of the DLP7000 Datasheet.

    Best regards,

    Trevor

  • 0 in reply to Trevor Dowd
    Prodigy 120 points

    Hi Trevor,

    This is used for laser modulation. We have requested to put UV window on this DMD chip from a third party company. We have tested it can sustain a laser peak fluence of >1mJ/cm^2. The laser is at repetition rate of 120Hz and pulse length 2ps. We use the timing system of the laser to trigger the DMD as well. During these tests when the dark pixels showed up, we only used <50% of the peak fluence to illuminate the DMD. We do not think it's thermal damage.

    Siqi

  • 0 in reply to Siqi Li
    TI__Expert 6370 points
    Hi Siqi,

    Just to clarify, are you saying that a third party company removed the window and replaced it with a UV window? DMDs a sealed with a lubrication environment in them which is necessary for operation. If the window was removed, this environment would escape.

    Also, if this is a visible DLP7000 part (as opposed to the DLP7000UV DMD), you must make sure that your illumination is within the given datasheet specifications. For your application, it sounds like the DLP7000UV DMD would be the best choice.

    Best regards,
    Trevor
  • 0 in reply to Trevor Dowd
    Prodigy 120 points
    Hi Trevor,

    Yes, the third party company replaced the window with a new one that transmits UV. They repackaged it in a dry environment and they claim it works for other customers.

    We also saw that when we put in high fluence of UV the mirrors flip where the beam hits but this can be recovered by loading another image on the DMD. Is the lubrication layer organic? Could it be that the UV destroys the organic layer? How does the DLP7000UV work if the UV light destroys the organic layer?

    I understand that we are operating out of specs but it would be really helpful if you could help us understand what is going on so that we can avoid damage if we order new chips.

    Thanks!

    Siqi
  • 0 in reply to Siqi Li
    TI__Expert 6370 points
    Hi Siqi,

    Unfortunately, I cannot share much about the proprietary material contained within the DMD. To give you a better idea about DMD behavior in UV applications, take a look at the Reflectivity Characteristics graph in the Application and Implementation section of the DLP7000UV datasheet (note that this DMD is designed to be optimized for different wavelengths than yours).

    My colleague will respond to the question about operation with very high fluence.

    Best regards,
    Trevor
  • 0 in reply to Trevor Dowd
    TI__Guru 54635 points

    Hello Siqi,

    I work with Trevor and have some understanding of using lasers with DMDs.  As I read your post the dark areas do not recover when the DMD is turned off and reloaded with an all on pattern.  If this is true it would indicate thermal damage to the µ-mirrors.  Please confirm that the µ-mirrors do not recover.

    The pulse energy you list (>1 mJ/cm^2) over the pulse duration you indicate 2 ps gives a peak power density of > 500 MW/cm^2.  That is very high fluence indeed.  Even at 200 MW/cm^2 this is still a great deal of power.  Damage can occur to the µ-mirrors at this level.

    If the µ-mirrors do recover by reloading the data and then sending a µ-mirror clocking pulse, then there is a different explanation.  In this case the photons are liberating the charge in the memory cells under the µ-mirrors leaving these memory cells set to a random value.  The µ-mirrors will not lose their landed state unless another µ-mirror clocking is received after losing their set value.

    Looking at the image you show, the linear groupings of µ-mirrors suggest thermal damage, not memory cells losing their values.

    Fizix

  • 0 in reply to Fizix
    Prodigy 120 points

    Hi Fizix,

    Thanks for your response.

    The micromirrors recover but not completely. We could see the damage pattern changes over time. May I ask why do you think the vertical lines suggest thermal damage? The transverse shape of the beam is round, not vertical lines.

    Another observation is that the damage expands like a ring outwards if we keep the chip on WITHOUT laser illumination.

    I'm curious about the memory cell explanation. Does the discharging of charges affect neighboring chips which may cause the damage to expand outwards?

    We made some summary slides on the various observations that we've seen, with images which could tell you more what's happening and what we've done. Is there a way I can share those with you?

    Siqi

  • 0 in reply to Siqi Li
    TI__Guru 54635 points

    Hello Siqi,

    Memory loss due to charge liberation in the CMOS cells is non-destructive and completely recoverable.  It does not expand in time if there is no active illumination.T

    The vertical and horizontal line groups (4 - 8 pixels long) within the damage area suggest that there is physical damage that is transmitted to adjacent pixels, which do not tend to transmit diagonally.  I do not have an specific explanation other than we have seen occasionally this kind of permanent damage in very high energy density operation by customers.

    Please send me a Friend invite to share the pictures.

    Fizix

  • 0 in reply to Siqi Li
    Prodigy 190 points
    Hi Siqi,

    The Vialux controller starts reloading data immediately after the mirror clocking pulse (12.5µs), independent of frame rate.
    So maybe you should review trigger timing. The laser pulse must happen about 12µs after image display is triggered. If later, then it could disturb data of the next frame, because it is already sent to the DMD.

    (Please note that this does not answer the question about persisting/expanding damage without laser illumination.)

    Best Regards
    Frank