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.

What is the frequency and duty cycle of a single pixel in a DLP LED projector displaying a primary colour?

Other Parts Discussed in Thread: DLP4501, DLPC6401

I am displaying only constant solid red at full intensity (R=255, G=0, B=0) over the whole image.
What is the pulse profile (frequency and duty cycle) of a single pixel in the image?
How does it change at lower light intensities, eg 1/2 red (R=127, G=0, B=0)?

I would appreciate links to any documents explaining the mirror switching timings, in relation to the multiplexing of different coloured LEDs and the PWM of different intensities.

Background
----------------
I am an engineer working in a biology lab where we are using a projector to stimulate opto-genetic neurons, meaning we can activate or disable a neuron by shining coloured light on it (cool huh?). Since neurons may behave differently to a pulsed light source compared to a constant light source, I need to know the characteristics of the emitted projector light.

Ideally I would like to know the exact pulse profile of the emitted red light, but if I have not provided enough information to get an exact answer, I would still like to know what order the pulse frequency is likely to be so that I can assess whether it is relevant in the timescale of neuronal activity.

Specifications
-------------------
Projector: Optima ML750 pico (LED)
DLP chipset: DLP4501, DLPC6401
Resolution used: 1280x800 (native)
Frame rate used: 60 Hz
Effective Pixel Clock Rate: up to 90 MHz
DMD interface: High speed DDR


Thanks,
James/.

  • Having found a little more information from TI documentation*, my understanding is as follows. Please let me know what I have got right or wrong.

    For 24-bit colour, per frame there are 24 time slots, i.e. 8 slots for each colour. The time slots are not even in duration; per colour there is one which is 1/6th of a frame, one which is 1/12th of a frame, etc. (1/6, 1/12, 1/24, 1/48, 1/96, 1/192, 1/384, 1/768). For the whole duration of one time slot the pixel mirror can be on or off. Which time slots are on and which are off depends on the required colour intensity. Any red intensity from 0-255 can be achieved by turning on a combination of the 8 red timeslots. For example at full red intensity, the mirror will be on during all of the red time slots. For half red intensity the mirror will be on only during the 1/6th red time slot. Similarly for green and blue.

    The documentation is unclear, but it seems like these time slots are arranged in the frame in a way that improves the perceived quality of the image. My guess is that the slot colours would be interlaced R,G,B,R,G,B,... etc. and that the slots are not arranged in order of duration but rather the durations are mixed up to spread the intensity more evenly across the frame. I would further guess that the pattern of time slots is different from pixel to pixel so that nearby pixels are out of phase with each other.

    So to answer my own question:
    At full red the frequency would be 60Hz x 8 on slots = 480Hz and the average duty cycle would be 1/3.
    At half red the frequency would be 60Hz x 1 on slot = 60Hz and the duty cycle would be 1/6.

    If my understanding is correct, I have another question:
    If the DLP projector is capable of 120Hz but I am using it at 60Hz, does it update each mirror state more often than 24 times per frame?

    *This is the document I used, page 55 onwards: www.ti.com/.../dlpu006e.pdf
  • Hello James,

    The frequency for the Red LED is depending on your firmware the EVM is programmed with. Can you please let me know where you bought this EVM from ?

    In general you are right. The frame will be split in different color segments and will be arranged to achieve the best image quality. The RGB color in Gamut space doesn't have anything to do with frequency or duty cycle.

    The change of the RGB in Gamut just brings the color points in a different location. This will be adjusted separately.

    Best regards,

    Nadine
  • Hi Nadine,

    Thanks for responding to my question.

    I am using an off-the-shelf pico projector (details in my first post) not EVM.

    Without knowing the firmware details, is it possible to say how many times per frame the mirror is pulsed on and off if I am displaying full red only? I assume 8 pulses per frame, is this correct? Or, for the given hardware, what is the maximum number of pulses per frame when displaying full red? Does it depend on frame rate?

    Sorry, this is not my domain, could you explain your remark "The change of the RGB in Gamut just brings the color points in a different location. This will be adjusted separately."

    Cheers,

    James/.

  • Hello James,

    This is all depending on the sequence and therefore on the firmware. Without having access to the firmware of the projector I am not able to answer your questions.

    It would be better if you would have purchased an EVM this would allow us to support you in a better way. Changing the RGB intensity doesn't impact the frequency or duty cycle for Red. It moves the color point in the Gamut into a different location (CIE Chart).

    Thanks,

    Nadine

  • Hi Nadine,

    I do appreciate you responding. Please help me to understand the physical output pattern of a DLP LED projector.

    Let's say, for example, I am using one of your EVM with factory firmware (or any other configuration you choose), and I am inputting an RGB image where all pixels are 255,0,0. What is the output pattern of one pixel over the duration of a single frame?
    In this case is only red light projected or might there be some blue and green used to 'mix' the desired colour?
    If I then project an RGB image where all the pixels are 127,0,0, how does the pattern change?

    I realise now that you can't give me exact answers but if you can give me a typical example that really help my understanding. Or do you have a link to a document that will help me?

    I also need to know what are the typical or recommended peak wavelengths of the LEDs used in a DLP LED projector.

    Thanks,

    James/.

  • Hello James,

    You can always buy or build your own photo detector to see the output of the projector this will make you independent from any firmware.

    In the firmware for the DLP LightCrafter Display 3010 EVM we specify three looks which means three different sequences.

    Look 0 is running 23% Red , 48% Green and 29% Blue.

    In this specific example we have 4 time slots for RED.

    1. 872.35us

    2. 928.81us

    3. 901.62us

    4.1129.34us

    By using an intensity of 127 the time slots will be pulsed to accommodate for the lower intensity.

    As mentioned earlier these timings are changing with each look and are different for each firmware.

    Best regards,

    Nadine

  • Thanks, this has helped me somewhat. But I am still struggling to understand how 255 different shades are generated with only 4 time slots. I thought there was at least 8 time slots per colour.

    In your example, what is the projected colour, is it white (255,255,255)? Can you give me an example that is not full saturation?
    What is the total frame duration? Is it 16.66ms (60 Hz)?
    Are these time slots fixed in duration, i.e. the same 4 time slots are used to produce all shades of red?

    " the time slots will be pulsed" - can you explain this please? Does this mean there is a pulse pattern within time slots? If so, what is that pattern?

    Cheers,
    James/.
  • Let me chime in and attempt to shed some light on this.

    First of all, most of the time the looks programmed into the firmware should not be mixing colors. I doubt that in your case this is an issue. As Nadine has stated, using a photodetector may help you verify this.

    Next, you can generally correspond the intensity of the color (0-255) to the amount of time that the particular color is active in a given sequence. That is to say, in a 60 Hz video output, you can expect each color to be on for its full duration (1/3 of the 16.66 ms frame time, in general). Using a photodetector, you will be able to observe that each color is produced individually in the case of pure white light.

    That being said, the timing within each of these windows of when the LED is on or off (if at, say, 127/255 intensity) is dependent on the firmware itself. If you need to capture more information about exactly when the particular LED is on or off, you will need to use a photodetector to measure it or test using an EVM with a known firmware configuration.

    Hope this helps.

    Regards,

    Philippe Dollo
  • Hi Philippe, thanks for your response. This helps to confirm my understanding up to a point.

    From Nadine's and your response, I accept that the only way for me to know the pulse profile of my specific projector is to measure it. I have an idea how I can do this with equipment in the lab so I will try it soon.

    Judging by the fact there is no simple explanation for how the pulse pattern is generated, I figure it must be more complicated than I first imagined with different vendors choosing different methods.
    However, I have now become interested to know how it all works, for my own satisfaction. Please can you point me to some documentation I can read?

    Cheers,
    James/.
  • James,

    I don't believe we have any documentation regarding this matter, as the underlying tech is proprietary to TI and not publicly available. I apologize for the inconvenience.

    Regards,

    Philippe Dollo

  • For anyone who might be interested in my conclusions...

    I have used a 10,000 FPS monochrome camera to get an indication of the pattern of output pulses on my projector, for individual pixels and groups of pixels. While 10,000 FPS is not fast enough to capture the briefest pulses, it does reveal many features of the pulse pattern.

    Here are some observations from my tests:
    • The 3 colours, RGB, cycle 4 times per frame (figure 1), giving a strong frequency signature at 4x the frame rate = 240 Hz
    • There are lesser power peaks at 60Hz intervals.
    • The patterns for each of the 4 sub-frame cycles are not the same (figure 2).
    • The pattern over a whole frame does repeat every frame.
    • Adjacent pixels of the same colour are not synchronised (figure 3).
    • Blue has 1/2 the mirror time (pulse length) compared to red, presumably because the blue LED is brighter or has more visual effect on hue. Green is on for slightly less time than red. (figure 1)
    • 'Pure' red (255,0,0) contains a small amount of green. So does blue. (figure 1)
    • There is a brief pulse whereby all pixels of any colour receive an output (figure 1). It's not possible to say what colour LED is on during this pulse (the high speed camera is monochrome) - it could even be all LEDs at once (white) to add brightness.

    I have some images to support these observations but I can't find how to attach images in this forum, surprisingly. Even the 'rich formatting' interface fails to import my images.

    I hope this helps other visitors.
    J/.