What is the timing of the DLP image display, and camera trigger output

I transfer 3 8 bit gray scale images, to the DLP, and are cycling through the images, with the H/W Trigger.

I don't have any problems running, when I have the Camera Trigger the DLP. Only with the DLP triggering the camera.

For the DLP triggering the camera, I must insert a 5 mS delay, or I get multiple projections in the same image. Even once I do that, I still don't get a stable image. (The grayscale values are constantly changing, as I grab images.)

For my setup, of the Camera triggering the DLP, the image is very stable, and repeatable. I might get ~5% variation (not measured,) in gray level. Very minimal change, but the image is very solid.

Previously, I was using S/W trigger, with a 20 mS delay, and have decent results with that version assuming I always grab the image in a multiple of 20 mS. (I don't understand 20 mS, since the camera is running at 60 FPS, and the capture time to get a full image, should be 16.67 mS. But empirically, for the most stable image, I must grab from a multiple of 20 mS.) My understanding is that it takes 16.67 mS to for all the mirrors to move, and project all the bits of the grayscale image.

I'm now moving to a H/W trigger to reduce the image grab times, since I must grab 3 consecutive images for the 3D imaging.

I plan to update the DLP to Version 5.0, but have not updated this DLP yet. (Hopefully, that it NOT the cause of my non-repeatable image.)

Paul

The attached image, is my working Camera-DLP triggering

Here are my settings in the DLP Lightcrafter for the DLP-Camera Triggering.

I used past from word. Hopefully that worked.

In case if failed again. attach a jpg

Only problems with the DLP triggering the Camera. I have no problems with the Camera triggering the DLP, or with S/W triggering of the camera. (S/W trigger display the next image in the list of pre-loaded images, waits 20 mS, then does a capture.)

No online information about the camera. The "frame grabber board" is not the traditional board, but a combination of Certis device, for handling the serial stream from the camera head, and placing this information into memory.

No time shifting. I'm passing the trigger directly to the H/W on the "frame grabber board", and this H/W is using this to grab the image.

Basically, I programmed the DLP to generate H/W triggers without any delay, and the Camera H/W can look at the rising or falling edge of this trigger. Neither rising or falling generates a valid capture. The capture is a combination of the previous image, and the newly projected image. The only way to get valid image, is to delay 5 mS after the trigger signal, before grabbing.

So my guess, is the trigger is generated, before the image is stably projected.

I also find with S/W triggers, that I must collect a multiple of 20 mS to get a stable image. This is unrelated to this problem, but a question about the working of the DLP.

Paul.

Hi Paul,

I believe there is likely an issue with the pattern exposure and trigger period matching up with your camera specs. I created a test setup with a LightCrafter triggering a PointGrey Flea3 (FL3-U3-13S2C-CS) connected over USB 2.0 (although this is a 3.0 capable camera). I used the three 8-bit patterns found at the directory listed below.

<install-dir>\Texas Instruments-DLP\DLPLCRSoftwareBundle-3.0\API-SampleCode-v2.0-Windows\Images\PatSeqImages

• pattern_8_00.bmp
• pattern_8_01.bmp
• pattern_8_02.bmp

When I initially used your settings, I was seeing a combination of all three patterns. Given the specs of my camera, it could not capture at this speed. I have listed a few of the limiting specs of this camera that may be a source of your issue.

1. Camera integration time
   1. Using the external trigger mode and running at about 30 fps, my camera was taking about 40 ms to integrate
   2. Given this, I set the pattern exposure time to 2.5x the time (100 ms) with a trigger period of 1 s (auto triggering). The external trigger pulse width was 1 ms with a delay of 0 s
   3. The captured images appeared to match up with the projection
   4. A good image illustrating the camera integration time relation can be found in the Technical Reference Manual (TRM). I include a screen shot below

1. External trigger timing
   1. Similarly, your camera may have some propagation delay between the trigger and exposure time
   2. I've included another screen shot from the Flea3 TRM 

1. Maximum frame rate in external trigger mode
   1. I also found out that the rolling shutter caused the maximum frame rate to be halved for my particular camera

Let me know if this information is helpful.

To address your other question, can you explain what you mean by SW trigger? Do you mean you have a trigger connected to the LightCrafter and camera generated by SW on your PC?

Thanks,

Clinton

My camera is an on-demand global shutter. So no rolling shutter or delay from the trigger.

From my testing it appears that the trigger from the DLP to the camera comes, before the DLP image is displayed completely due to some unknown latency.

When I delay the camera integration by ~5 mS, I always get a stable image, with some slight grayscale differences.

For S/W trigger, I'm using the USB to display the desired image, by telling it to display a specific image index. Then I start a delayed trigger on the camera, with a 20 mS delay, so I will get static image.

Another question is the following: if my DLP is setup for 60 FPS, or 16.67 mS display, for each image, why must I wait multiples of 20 mS capture time to get a static image without varying gray values. (If I'm NOT at a multiple of 20 mS, I see a beat frequency in the gray levels.)

Since the camera is 60 FPS, I would expect this time to a multiple of 16.67 mS for image capture. Somewhere I saw something about 3 - 5 mS RGB displays, and 5 mS setup time, which would correspond to my 20 mS. But since I running in single LED color mode, I would not expect this behavior. My images are 8 bit grayscale images.