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.

DLPDLCR2000EVM: Using the DLP for spectroscopy

Part Number: DLPDLCR2000EVM
Other Parts Discussed in Thread: DLPC2607,

I've been interested in applying TI digital micromirror devices for spectroscopic purposes (digital filtering, this has been an area of growing interest in spectroscopy for over a decade).  This application requires rapid cycling through a limited series of 1-bit (black and white) patterns, typically stored in a buffer, and removed the LEDs and projection optics. It's basically running the light path backwards, with light reflected off the DMD routed into two detectors, corresponding to pixels in "on" or "off". I'm aware that TI knows about this and makes an IR spectrometer, but that's not suited to my particular application.  I have been able to successfully remove the optical engine from my WVGA DLP (608 x 684 pixels, the original Lightcrafter), there's some help available on this topic on the internet.  Yes, I'm aware that trying to do this on this unit will void the warranty, but at 99 bucks, that's an easy risk to take, and this unit has the micromirror on the end of a ribbon cable, making it easy to integrate it optically. On my earlier Lightcrafter there were detailed specs about the maximum frame rate I could achieve, including in monochrome mode. Since the local Linux processor was part of that unit, and it had internal pattern capability, it's clear why this could be specified in detail.  I'm sure the max frame rate may be limited by what controller I'm using this with, so let's suppose I'm using a BeagleBone Black (ver C). Would you folks have any idea about the monochrome frame rates achievable here are, if in fact, monochrome can be done at all? Since I'm not planning on using this for DLP purposes, it's the monochrome (1-bit) performance that concerns me most, and I'd like to try something similar with this unit. I can find information about nHD line rates in the DLP2607 documentation (48 kHz in landscape mode), but I've found nothing about using this as a monochrome display. The power handling and wavelength capabilities of the mirrors are plenty adequate; the light is visible and level is extremely low compared to projection).  Am I simply too far out of the box in trying to do that with this unit?

  • Hello Timothy,

    Thanks for your interest in DLP technology...
    Let me check with my team and get back to you with some straight options. We have applications straight for monochrome purposes. But it depends on what are you specifically targeting as far as speed.

    Below is a similar E2E thread from another angle if interested:
    e2e.ti.com/.../677627

    I agree with you; your front end processor will limit your capability at the end of the day. One side question I noticed from your explanation: I'm trying to get some more information regarding your "performance" concern. Can you elaborate a little bit more into this?
  • Sure. The Lightcrafter 3000 had a high speed monochrome mode using the internal buffer, it could hold up to 96 frames, and run at 4 kHz. I am looking to do something similar here. I don’t know if an internal buffer can be set up on this unit.  What I’m trying to do is create a rather small series (<20)  of 1bit patterns in a buffer and simply cycle through them at rates  of 100 (should be easy) to > 1 kHz ( 4-10 kHz if possible) with a means to communicate to the host processor each time that the frame has toggled ( enabling me to synchronize the light detection system,  remember, I’m doing spectroscopy). I know the now obsolete 3000 could do this, and similar things have been done with the higher end evaluation boards.

    Since the number of frames is small, and they’re just one bit, the size of the dataset is also pretty small. If it can be loaded into a buffer on the 2607 or the mDDR (sorry, I’m not well acquainted with the architecture here yet) then transmission rates from the host become less important: it’s set up on initialization, then left to cycle until the cows come home. Not at all what one does in projection, but useful to us. So the performance areas I want to know more about is, can an internal buffer be created, how many frames can it hold ( yes, depends on frame size and bit depth, so  native 0.2 nHD image size, 1 bit if possible) and how fast can it cycle. The concept is very similar to the test patterns, it’s just my patterns will be chosen differently, e.g., Hadamard masks.

  • Timothy,

    From a high level I can see two problems that will interfere with your attempts to use the DLPC2607 chipset for this sort of application:

    > First, the DLPC2607 is a display chipset. In being so, there are algorithms that execute in hardware to improve display image quality. With these algorithms the DMD is no longer guaranteed to display a perfect 1:1 mapping of the content you push through the parallel port. Thus, this chipset is poorly suited for high-accuracy light control/pattern display applications.

    > The DLPC2607 does not have a pattern programming option, mainly because it is not designed for light control applications in the first place. If you are trying to run through pre-programmed patterns, you will probably need to use a proper light control chipset/EVM.

    For what you are trying to do, one of our more recent light control EVMs may be better suited to your application. The closest one to the DLPDLCR2000EVM is the DLP2010-LC EVM, which uses a 0.2" WVGA DMD:
    www.ti.com/.../DLP2010EVM-LC

    I hope this helps.

    Best Regards,
    Philippe Dollo