• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 28 subscribers
  • Views 775 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.

DMD cooling scheme

Jian Tay
Intellectual 260 points

Hi,

We've been using a couple of DMDs in the lab - one is the 0.3" LightCrafter and the other is a 0.7" XGA device. We noticed that above a certain illumination power, the DMDs appear to develop dark spots in their images, which go away when the beam is removed. It seems like this is caused by heating on the DMD.

We were wondering if there were ways to overcome this issue, as we need a greater illumination power for our applications. We tried a simple 12V fan + heatsink, but don't get a large improvement. So we were wondering:

1. If better cooling would allow us to use, say, twice the amount of light we are now?

2. If changing the illumination wavelength would be better, for example using NIR? In this case, do the DMDs work well for these wavelengths?

Thanks!

  • 0
    TI__Guru 54685 points

    Hello Jian,

    I am sorry for the delay getting to your question.  Do you have any estimate of the power level (W/cm2) that you saw this effect?  I agree this sounds like a thermal effect.

    Regarding the IR, the DMD mirrors are slightly more reflective above about 900 nm, but the window is not optimized for the NIR band.

    Fizix

  • 0 in reply to Fizix
    Intellectual 260 points

    Hi Fizik,

    Thanks for the reply.

    We measure a power level of 270 mW using a laser repetition rate of 1 kHz, and a pulse duration of 5 ns. Based on this, I estimate that the average power density is 0.18 W/cm^2 - which I think is much less than the damage threshold (20 W/cm^2). However the peak power density is high - 37 MW/cm^2 by my calculation. Not sure if this is the cause?

    With regards to the window not being optimized, how much are the losses you would estimate on this?

    Thanks!

  • 0 in reply to Jian Tay
    Intellectual 260 points

    Just a follow up. I've tested the device now with an NIR laser, while the performance seems better, it still doesn't seem to be able to support more than double the power. So I'd like to know (if possible):

    1. Does the PEAK POWER matter here? The average power seems to be within safety limits, but the peak power is very high due to the 5 ns pulse duration. However, I don't seem to see anything about a peak power threshold in the manuals.

    2. Any explanation on how heat affects DMD operation. What I'm observing now is some mirrors seem to change state when the illumination gets too high. Mirrors that were on, turn off, and the state persists until a new image is written to the DMD. 

    3. Does the illumination angle matter? Is it possible I've aligned the DMD wrong and some light is getting in between the mirrors and ending up on the back substrate layer?

    Any information would be highly appreciated!

  • 0 in reply to Jian Tay
    TI__Guru 54685 points

    Hello Jian,

    1.  Yes PEAK POWER does matter.  You are correct the the time average power density is ok, but quite possibly the individual pulse energy may be pushing the peak mirror temperature too high.  5 ns duration at 37 MW/cm2 comes out to about 185 mJ/cm2 which is a very large pulse energy.  I would have to go through some modeling but I can tell you that at this pulse energy, you are pushing the peak temperatures a little high.

    You may want to take a look at the following white paper --> http://www.ti.com/litv/pdf/dlpa027.  However the charts do not show the region you are operating in.  Your pulse duration is at the very left of the charts, but the peak power density is energy is 1000+ times higher than shows in Figure 12.

    2.  It is possible with a high enough power density to make the memory cells below the mirrors lose their latched state.  You need to refresh the memory before each "Mirror Clocking Pulse" [MCP].  The mirrors should remain in their latched state until the next MCP regardless of the illumination, but should be refreshed with the data (even if it has not changed) while the illumiation is off before the next MCP occurs.

    3.  Illumination angle does matter, but only a little.  The typically expected illumination is at 45° azimuth (with 0° poiniting to the top of the DMD) and 24° down from the DMD normal (equivalent to 66° elevation from the surface).  See any of the DMD datasheets at the following link for illustrations --> http://www.ti.com/lsds/ti/analog/dlp/overview.page

    Fizix

  • 0 in reply to Fizix
    Intellectual 260 points

    Hi Fizix,

    Thanks very much for your helpful explanations. I have read the white paper you cited, and will look more into the modelling.

    But in the meantime, I guess this brings us back to one of my original questions, which is whether a better cooling system would help. However, if it's the peak energy/temperatures that is causing the issue, then it sounds like it won't since there isn't enough time to take the heat away before the memory cells are affected. Does this sound about right?

    Thanks!

  • 0 in reply to Jian Tay
    TI__Guru 54685 points

    Hello Jian,

    If the dark areas are only because the memory cells are losing their state (from photons booting the charges out) then you can try refreshing between pulses but before MCP's.

    If it is due to µ-mirror heating during the then a cooling solution will not help.  You are correct that the overall heat load (DMD bulk heating) is relatively small.  However,  I have very serious concerns about 185 mJ/cm2 pulse energy causing very high individual µ-mirror heating above the average array substrate temperature.

    Fizix