TVP5158 4-Ch D1 Super-Frame Format

Please check table 3-11 and figure 3-15 in the TVP5158 datasheet available on the TI product folder web page. (http://www.ti.com/product/tvp5158)

It is also recommended to ensure you are using the latest available patch code for the video decoders, linked from the produce folder and available here...

http://software-dl.ti.com/dsps/dsps_public_sw/dsps_swops_houston/ANALOG_VIDEO/Analog_Video_Decoder_Versions.htm

BR,

Steve

Steve,

Thanks for the quick reply. I have gone through the Table 3-11 from the TVP5158 datasheet. Since we dont need audio right now, we have not gone through the 3-15 table yet.

I wanted to ensure that my understanding is correct on the Super-Frame Size in bytes.

As per Table 3-11, for 4-Channel D1 we should have the following byte configuration per horizontal line.

EAV [4 bytes] + EAV2SAV [280 bytes] + SAV [4 bytes] + Start Code [8 bytes] + SAV2EAV [1416 bytes] => 1722 bytes per Horizontal line and this repeated 4 times for each of the channel with the Start-code data changing to indicate seperate channels.

Is the above understanding correct.

Also, how and what is the break-up for the Vertical resolution?

Thanks,

Ravindra

Check FIGURE 3-15, not table 3-15.

We did notice one mistake in Figures 3-15 and 3-16 though.   The comment under the EAV2SAV period should not say “64 clock cycles (fixed)”.  It should instead say “Refer to Table 3-11 for the duration.”

Also, the TVP5158 does not guarantee the order of the video channels in the super frame.  The Channel ID provided in the Start Code must always be used to determine the channel number.

The embedded syncs for the super frame are defined to look like a progressive frame with a 2-line vertical blanking interval.  Don’t confuse the embedded syncs for the super-frame with the embedded syncs for the channel data that’s included in the start code.

BR,

Steve

Steve,

Thanks for pointing out the problem with the Figure 3-15.

Yes, I know that the TVP5158 does not guarantee the order of the video channels in the super-frame. In our product, we will need both i.e., demuxing of a Super-frame and also construction of a super-frame.

Since you have pointed to the latest TVP5158 driver, I think it will help in resolving any open issues on the demuxing of the super-frame. I wanted to understand the working of the TVp5158 from the construction aspect of the super-frame. Please let me know your inputs on what will the total length/width of the super-frame if we manually construct a Super-frame.

We will be getting 4 different decoded D1 video streams and we need to construct a Super-frame out of it. I wanted to be sure that we add the correct embedded sync when constructing the super-frame.

Thanks,

Ravindra

Hi Ravindra,

I drew graphical explanation of TVP5158 line multiplexed mode. Please see attached file.

There are four channels video data from 4 decoders. Four channels input videos are asynchronous each other. TVP5158 combines four channels video to one super frame and it also asynchronous to any video channels. One super frame is not guaranteeing to have full frame of each D1 video. However in this example Channel one is shown in full frame in this super frame. Other channels video does not include full frame video. The line multiplexed data is not evenly distributed in the super frame. It may show a random assignments between channels. If you can see first few lines in the super frame, the sequence of lines are Ch1-Ch2-Ch3-Ch4-Ch4-Ch2-Ch1-Ch3 and so on.

Davinci processor should demultiplex super frame data and sort each line to the frame buffer as shown in the figure. Channel 1 is only shown full frame. Other channels are incomplete frame and it'll filled in the next super frame.

I hope you to understand TVP5158 output formatter better. If you have more question, please let me know.

Regards,

Hyun

Dear Hyun,

Thanks a lot for sending the excel sheet. I have gone through the same and let me clarify one small part.

In our system, there will be a TVP5158 which can provide super-frame input into the DM8148 and in that part we can use the latest Video Decoder SW/Driver release available from TI to perform the de-mux of the Super-frame into discrete frames.

This is one part of the system. The other part of the system will receive network based H.264 encoded streams. There will be 4 such different streams of D1 Resolution. We will be configuring the HDVICP to run 4 different H.264 Decoders and once the decoded streams are available, we need to construct a super-frame out of the same.

For constructing the super-frame out of the 4 different YUV streams, I wanted to understand the horz and vertical sizes of the Super-frame and the contents to be added into the starting of the super-frame.

Thanks,
Ravindra

Hi,

That question is out of scope in this video converter forum and I'm not familiar with IP video. However I'll check who can answer it.

Regards,

Hyun

Hi,

for the 4 D1 multiplexing, the super frame consists of 2102 line (according to the TVP5158 DS). each source consists of 480 active lines and 45 inactive lines.

I am trying to construct a superframe to be used as input to the DaVinci.

I need to know what is the structure of the super frame. on what lines should the two artificial vsync appear ?

dose the super frame syncs are embeded inside the the SAV/EAV of the two artificial vsync lines ?

an illustration (similar to Figure 2-6 of sprugx8.pdf - Davinci Technical Reference Manual) will be helpfull.

Regards,

Boris

Hi,

Please look Table 3-7 and Table 3-9 from data manual of TVP5158 http://www.ti.com/lit/ds/symlink/tvp5158.pdf.

Regards,

Hyun

Hi Hyun,

Table 3-7 and 3-9 reffer to non-interleave mode and pixel-interleaved mode. I am using line interleaved mode.

I need this information to generate the superframe from an FPGA to the Davinci.

Thanks.

Boris

Hi,

FVH bits are same in both format. Vertical blanking is indicated by V bit in EAV/SAV code.

Regards,

Hyun

Hi,

Vblank is 2-lines and can be placed at the beginning of the super frame. These are inserted artificial lines.

Regards,

Hyun

Hi Hyun,

are those 2 Bblank lines should be consequent ?

according to my understanding, 4 interlaced D1 should look like this:

- 2 artificial vertical blanking lines

- 19 x 4 top vertical blanking of filed 1 (V and FlD equal zero)

- 240 x 4 active lines of filed 1  ( FlD equal zero)

- 3 x 4 bottom vertical blanking of filed 1  (V and FlD equal zero)

- 20 x 4 top vertical blanking of filed 2  (V and FlD equal zero)

- 240 x 4 active lines of filed 2 ( FlD equal zero)

- 3 x 4 bottom vertical blanking  of filed 2 (V and FlD equal zero)

please correct me if my understanding is wrong.

Best Regards,

Boris

Hi,

This is correct. One super frame consists of 2 lines (consequent) of VBLANK and 525x4 lines.

Regards,

Hyun

Hi,

I was able to detect only one source out of four (corresponding to source ID 3). what could be the reason ?

The data stream consists of two blanking lines with vertical sync in the EAV/SAV fields and 2100 lines with V=F=0 in the EAV/SAV fields.

the identified video source is identified as 720x240@60hz.

Thanks,

Boris