We use PCIe to connect Intersil TW6865 chip for the surround view solution.
This is the connection of PCIe to iMX6Q SabreSD board.
This is the block diagram of the connection:
This is the 4 camera surround view:
Code base is L3.0.35_12.10.02 release. You can merge the patch file to the latest Freescale release.
Please check the attach file for the patch code.
Note: It is only a test version.
The last code for L3.0.35 BSP: L3.0.35_GA4.1.0 Patches.7z
The last code for L3.10.53 BSP: L3.10.53_TW686x_patch.7z
Patch for L4.1.15 1.1.0 GA BSP: TW6865 driver for Linux L4.1.15_1.1.0-ga.7z
I am doing the same thing! I use the tw6865 and I can read the picture from the cameras.what confuse me is that I can not change the resolution of the caption.
I use V4L2 lib to config the camera,when it is configed with 720*576, i will get a full picture,just like this:
when it is configed with other resolution,such as 360*288,640*480 and 320*240, i will get a Incomplete picture ,just like this:
I use a definition to define the resolution:
#define IMAGEWIDTH 720
#define IMAGEHEIGHT 576
here is the configuration code:
int Camera_type ::init_v4l2()
{if ((camera.fd = open(camera.device_add, O_RDWR| O_NONBLOCK,0)) == -1)
{
printf("Error opening V4L interface\n");
return (false);
}
//query cap
if (ioctl(camera.fd, VIDIOC_QUERYCAP, &camera.cap) == -1)/* 获取设备支持的操作 */
{
printf("Error opening device %s: unable to query device.\n",camera.device_add);
return (false);
}
else
{
printf("driver:\t\t%s\n",camera.cap.driver);
printf("card:\t\t%s\n",camera.cap.card);
printf("bus_info:\t%s\n",camera.cap.bus_info);
printf("version:\t%d\n",camera.cap.version);
printf("capabilities:\t%x\n",camera.cap.capabilities);
if ((camera.cap.capabilities & V4L2_CAP_VIDEO_CAPTURE) == V4L2_CAP_VIDEO_CAPTURE)
{
printf("Device %s: supports capture.\n",camera.device_add);
}
if ((camera.cap.capabilities & V4L2_CAP_STREAMING) == V4L2_CAP_STREAMING)
{
printf("Device %s: supports streaming.\n",camera.device_add);
}
}
//emu all support fmt
camera.fmtdesc.index=0;
camera.fmtdesc.type=V4L2_BUF_TYPE_VIDEO_CAPTURE;
printf("Support format:\n");
while(ioctl(camera.fd,VIDIOC_ENUM_FMT,&camera.fmtdesc)!=-1) /* 获取设置支持的视频格式 */
{
printf("\t%d.%s\n",camera.fmtdesc.index+1,camera.fmtdesc.description);
camera.fmtdesc.index++;
}
//set fmt
camera.fmt.type = V4L2_BUF_TYPE_VIDEO_CAPTURE;
camera.fmt.fmt.pix.pixelformat = V4L2_PIX_FMT_YUYV;
camera.fmt.fmt.pix.height = IMAGEHEIGHT;
camera.fmt.fmt.pix.width = IMAGEWIDTH;
camera.fmt.fmt.pix.field = V4L2_FIELD_INTERLACED;
if(ioctl(camera.fd, VIDIOC_S_FMT, &camera.fmt) == -1)/* 设置捕获视频的格式 */
{
printf("Unable to set format\n");
return false;
}
if(ioctl(camera.fd, VIDIOC_G_FMT, &camera.fmt) == -1)/* 获取设置的视频格式 */
{
printf("Unable to get format\n");
return false;
}
{
printf("fmt.type:\t\t%d\n",camera.fmt.type);
printf("pix.pixelformat:\t%c%c%c%c\n",camera.fmt.fmt.pix.pixelformat & 0xFF, (camera.fmt.fmt.pix.pixelformat >> 8) & 0xFF,(camera.fmt.fmt.pix.pixelformat >> 16) & 0xFF, (camera.fmt.fmt.pix.pixelformat >> 24) & 0xFF);
printf("pix.height:\t\t%d\n",camera.fmt.fmt.pix.height);
printf("pix.width:\t\t%d\n",camera.fmt.fmt.pix.width);
printf("pix.field:\t\t%d\n",camera.fmt.fmt.pix.field);
}
//set fps
printf("init %s \t[OK]\n",camera.device_add);
return true;
}could you please tell me where i am wrong?
Hi!@
I am doing the same thing! I use the tw6865 and I can read the picture from the cameras.what confuse me is that I can not change the resolution of the caption.
I use V4L2 lib to config the camera,when it is configed with 720*576, i will get a full picture,just like this:
when it is configed with other resolution,such as 360*288,640*480 and 320*240, i will get a Incomplete picture ,just like this:
I use a definition to define the resolution:
#define IMAGEWIDTH 720
#define IMAGEHEIGHT 576
here is the configuration code:
int Camera_type ::init_v4l2()
{if ((camera.fd = open(camera.device_add, O_RDWR| O_NONBLOCK,0)) == -1)
{
printf("Error opening V4L interface\n");
return (false);
}
//query cap
if (ioctl(camera.fd, VIDIOC_QUERYCAP, &camera.cap) == -1)/* 获取设备支持的操作 */
{
printf("Error opening device %s: unable to query device.\n",camera.device_add);
return (false);
}
else
{
printf("driver:\t\t%s\n",camera.cap.driver);
printf("card:\t\t%s\n",camera.cap.card);
printf("bus_info:\t%s\n",camera.cap.bus_info);
printf("version:\t%d\n",camera.cap.version);
printf("capabilities:\t%x\n",camera.cap.capabilities);
if ((camera.cap.capabilities & V4L2_CAP_VIDEO_CAPTURE) == V4L2_CAP_VIDEO_CAPTURE)
{
printf("Device %s: supports capture.\n",camera.device_add);
}
if ((camera.cap.capabilities & V4L2_CAP_STREAMING) == V4L2_CAP_STREAMING)
{
printf("Device %s: supports streaming.\n",camera.device_add);
}
}
//emu all support fmt
camera.fmtdesc.index=0;
camera.fmtdesc.type=V4L2_BUF_TYPE_VIDEO_CAPTURE;
printf("Support format:\n");
while(ioctl(camera.fd,VIDIOC_ENUM_FMT,&camera.fmtdesc)!=-1) /* 获取设置支持的视频格式 */
{
printf("\t%d.%s\n",camera.fmtdesc.index+1,camera.fmtdesc.description);
camera.fmtdesc.index++;
}
//set fmt
camera.fmt.type = V4L2_BUF_TYPE_VIDEO_CAPTURE;
camera.fmt.fmt.pix.pixelformat = V4L2_PIX_FMT_YUYV;
camera.fmt.fmt.pix.height = IMAGEHEIGHT;
camera.fmt.fmt.pix.width = IMAGEWIDTH;
camera.fmt.fmt.pix.field = V4L2_FIELD_INTERLACED;
if(ioctl(camera.fd, VIDIOC_S_FMT, &camera.fmt) == -1)/* 设置捕获视频的格式 */
{
printf("Unable to set format\n");
return false;
}
if(ioctl(camera.fd, VIDIOC_G_FMT, &camera.fmt) == -1)/* 获取设置的视频格式 */
{
printf("Unable to get format\n");
return false;
}
{
printf("fmt.type:\t\t%d\n",camera.fmt.type);
printf("pix.pixelformat:\t%c%c%c%c\n",camera.fmt.fmt.pix.pixelformat & 0xFF, (camera.fmt.fmt.pix.pixelformat >> 8) & 0xFF,(camera.fmt.fmt.pix.pixelformat >> 16) & 0xFF, (camera.fmt.fmt.pix.pixelformat >> 24) & 0xFF);
printf("pix.height:\t\t%d\n",camera.fmt.fmt.pix.height);
printf("pix.width:\t\t%d\n",camera.fmt.fmt.pix.width);
printf("pix.field:\t\t%d\n",camera.fmt.fmt.pix.field);
}
//set fps
printf("init %s \t[OK]\n",camera.device_add);
return true;
}could you please tell me where i am wrong?
Hi, Qingrong Jiang!
I'm a FAE from STC, one of the NXP distributors in Korea.
We made TW6865 test board, and we can see 4ch image on HDMI with MX6Q-SABRESDP.
But we can see afterimage when I move my hand fast in front of Camera. (Please see attached video clip.)
(BSP version : L3.0.35_4.1.0)
I met intersil FAE, and he said it is the problem of deinterlacing.
Do you have any guide or solution to solve this issue?
Dear Qingrong Jiang:
Hi
When I use this methord to get the data of images by the v4l2 capture.c,
I meet like this:
Dear JiangJustin ,
We have developed a camera board using Intersil TW6865 chip (As same as Freescale TW6865 miniPCIe Daughter Board) which can input two analog cameras to iMAX6Q Sabre Lite Development board using PCIe interface.
I need to clarify following >>
In iMAX6Q Sabre Lite for PCIe port there are 7 pin connections as follows. +3.3V, PCIE_RXM, PCIE_RXP, PCIE_TXM, PCIE_TXP, GND, GND.
Intersil TW6865 chip has REFCLK-, REFCLK+ pin connections for differential reference clock inputs.
Don't we need to connect this pins with iMAX6Q board if we want to interface with PCIe interface ?
Thank You,
Peter Amond.imx6_pcie_msi_handler
