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.

DM385: ISIF pushes to SDRAM all pixels in line instead of LNH

Part Number: DM385
Other Parts Discussed in Thread: SYSBIOS

Hello!

I'm trying to store raw Bayer video data coming from AR0330 videosensor to parallel port of DM385

So I'm using CAMERA_LINK_CAPMODE_DDR mode when data are stored to SDRAM fom ISIF

Resolution is 1920x1080

Here are IPIPIF/ISIF setings^

 [m3vpss ] ISS_CLKSTAT REG f0000013
 [m3vpss ] Capt addr b841c480
 [m3vpss ] IPIPEIF ENABLE 0,     CFG1 0
 [m3vpss ] IPIPEIF RSZ 10,       GAIN 200
 [m3vpss ] IPIPEIF DPCM 0,       CFG2 6
 [m3vpss ] IPIPEIF INIRSZ 0,     OCLIP fff

 [m3vpss ] ISIF SYNCEN 3,        CGAMMAWD 0
 [m3vpss ] ISIF HSIZE 78,        CADU 5c2
 [m3vpss ] ISIF MODESET c CCOLP 1b1b
 [m3vpss ] ISIF SPH e,   LNH 77f
 [m3vpss ] ISIF LNV 437,         VDINT0 433
 [m3vpss ] ISIF FMTCFG 0 FMTPLEN 0
 [m3vpss ] ISIF FMTSPH 0 FMTLNH 0
 [m3vpss ] ISIF FMTRLEN 0 FMTHCNT 0

But in my SDRAM data,  every valid 1920 pixel video line is followed by

another line with 480 pixels of zero data and 1440 pixels of garbage (0x8080).

It looks to me like ISIF after storing 1920 pixels in line doesn't stop and continues

till the end of HSYNC pulse (which makes 480 more zero pixels) and then upon reception of new sync front starts new

line and offsets to its beginning, leaving the rest of memory (1440 pix) to be filled with 0x8080 pattern

Isn't ISIF supposed to store in RAM  just amount of pixels programmed in LNH register and wait till beginning of new line?

Do I have to format data with input formatter?

Thanks a lot

German Baranov

  • Hi,

    Which IPNC version are you using?
    Can you tell us about your usecase?

    Regrads,
    Anuj
    Pathpartner Technology Pvt Ltd.
  • Hi,
    I'm not sure exactly about IPNC version, since eval board provider nSketch (from whom we got RDK) keeps their own rdk name and numbering. (carbb rdk - 1.5.6)
    I think its about 3.7
    (just couple of ti tool version numbers: cgt470 5_0_3, hdvpss 01_00_01_37, iss 03_80_00_70, sysbios 6_34_04_22,
    Linux 2.6.37-psp04.01.00.02)
    We are using our board with DM385, 512 Mbyte of SDRAM (single 16-bit chip), 512 Mbyte NAND,
    in development process I'm loading everything from SD.
    AR0330 sensor is connected to DM385 via parallel interface (MSB to MSB)
    We are running test program under Linux, it starts AR0330 and configures ISS for
    RAW input - RAW output mode fand
    storing RAW data from ISIF.
    Every 20th frame is stored from RAM to SD.
  • Hi,
    I've changed picture width to 2400 in hope to see (in addition to valid picture) just dark space at the position of HD pulse and no garbage data, but my expectations were wrong.
    I still have lines with valid data (but now thay also have garbage tail from 1920 to 2400) and lines with HD+garbage, though size of garbage
    changed.
    Now It seems to me that DM365 catches extra HD pulse at the end of sensor data line, probably real electrical spike.
    By the way does it have limitations to the length of sync pulses?