I've finally got composite Video Input Board (LI-VI365), but i can't get video out from LeopardBoard365.I was trying using modified "encode" demo from DVSDK,with DVSDK 2.10.00.18, 3.10.00.12 (DM365 EVM), and RindgeRun844702 kernel, filesystem, and sample.The same program works perfectly on DM365 evaluation module.
Encode sample uses DMAI. Capture device TVP5146 is successfully created and configured without any errors.Program stops at Dmai function : Capture_get(hCapture,&hCapBuf) which is called, but never returns. This function should return captured video buffers in a loop.I looked at that DMAI Capture_get() function , and found out that program stops at : ioctl(hCapture->fd,VIDIOC_DQBUF,&v4l2buf),which is called, but does not return.
Has anyone any suggesstions ?
Good news. Now i can get Composite Video Input Board (LI-VI365) work.
I ask the same question as above to the LeopardImaging support team, and i got real quick answer.
Here it is:
"To use the LI-VI365, you'd better populate R196 and R197 on the Leopard Board 365 (which is located on the back side of the board, near the back of camera interface). These two resistors are of 0 Ohm value so you can also short them with a wire."
So i did as suggested, and after that video works fine,
just using kernel, filesytem end encode demo from TI DM365 Evaluation Module DVSDK 2.10.00.18.
Thanks Richard from LeopardImaging support team.
I received today the LI-VI365 board and I did the same steps as you but I got a failure from dmai function Capture_create().
Can you confirm you are using the sample code in ..../dvsdk_2_10_01_18/dvsdk_demos_2_10_00_17/dm365/encode?
Yes, i use the sdk you mentioned.
Right now i have to go out, i'll try to compare my code with the default later
and inform you.
You also have to change your bootargs to : davinci_capture.device_type=1
I'm trying with your suggestion, I'll inform you about result.
the Capture_create() function is now working fine after changing the colorSpace variable initialization value to Colorspace_UYVY both in videoThrFxn() and capturThrFxn(). Now I'm facing the same problem you described: the DQBUF ioctl never returns nevertheless R176 and R177 are in place. Tomorrow I will continue investigating.
They are R196 and R197, i hope you've just mistyped.
I use ColorSpace_YUV420PSEMI, and it also works.
There is one more thing i discoverd. After you populate R196 and R197, you must not use camera board anymore on that Leopard, unless you unsolder those two again. I broke one LeopardBoard that way.
I put camera board on a Leopard, after i solder those two, and started the program,
but from that time i couldn't get any data from LeopardBoard anymore, neither with camera board, nor with LI-VI365(DQBUF never returns).
So, with populated R196 and R197, don't put camera board on, only LI-VI365, otherwise you could breake LeopardBoard.
I have 2 DM365 Leopardboards. The old one didn't have the resistor mounted, the new one I' received together with LI-VI365 was already equiped with resistors. I' ve modified the old one soldering the missing resistors. Both boards work fine with MT9P031 CMOS sensor using the ipipe_aew_demo you can find in the ./myapps/ipipe_aew_demo in the RR sdk.
Can you check your Leopardboard Version and what camera board you are using? we will check what happens to figure out a solution. Thanks
Thanks for support.
I have LI-5M03 camera board.
First i tried to short R196, R197 on LeopardBoard365 ver 1.3 (with originaly unpopulated R196, R197),
after that analog video with LI-VI365 started to work.
I tried the same with LeopardBoard365 ver. 1.4 (with originaly unpopulated R196, R197),
and analog video also started to work. Then i exchanged input board on the same Leopard, and put the LI-5M03 on,
and after that i could't get any video anymore, neither from analog nor from CMOS sensor.
I have also just received 2 new LeopardBoards365 ver 1.4, but those already have R196, R197 populated with resistors.
However analog video with LI-VI365 doesn't work. I tried to unsolder those resistors, and insted just shorted R196, R197,
but still can't get analog video working with these new boards.
Hi Bill, Marko,
just to add a little bit of information,
I have 2 dm365 leopard both version 1.4, the older one was without resistor mounted the other was shipped with 0 Hom resistors already in place. Both boards works with LI-5M03 including the old one after soldering 0 Hom resistors. Both boards dont work with LI_VI365, no interrupt coming from CCDC controller. It seems to be a problem related to ITU-R BT 656 sync detection since there is no VD interrupt. Is the problem related to ver 1.4 board ? It seems the only board working with LI_VI365 is version 1.3.
Marko and Michele
Thank you very much for your detail testing, we did a total research, finally we found an issue on the compatibility between Leopardboard 355/Leopardboard 365 and LI-5M03 camera board, I will announce this issue in next message so that everybody can see it.
Again, thanks community for giving us the report, we will take care of it and let you know how we can fix it and have good solution.
I am using Leopardboard 355 and LI-5M03 camera board and I was wondering if there are any compatibility issues.
I am trying to use the endode/decode functions comming with the dvsdk_1_30 and they don't work. However the IPNC demo works well which means the sensor is functional. Also I have tried the ipipe_aew_demo but it says "davinci_capture driver not found".
I can give you bootargs, linux version and all my configuration in detail if you want to.
Any ideas would be really appreciate it.
Thanks a million
No news since last post by Bill PU last October 2nd. I'm wondering whether I've lost the information about solution.Michele