Raspberry encode camera and output IQ

[G4KLB]

Had better results with IQ Mode.

[f5oeo]

Great Colin...
For Tutioune, as the camera encode only video, it helps to set a PID audio by hand to a random value..1000 for example.
I think you have plug you IQ in your modulator as is. For fine tuning, there is surely some level to adapt to obtain a better MER.

I am curious to know why Ugly mode doesn't work well...Can you make a snaphot of Tutioune with Ugly mode ?

[G4KLB]

I am sure I can obtain a better MER. The modulator had Nyquist filters for 4000Ms/S anyway! as well as some other wrong components, but was fine for a quick test.
Need to find Nyquist filter values for 250 and 500Ms/S for next tests.

Will have another look at ugly mode, I'm sure I haven't set it up correctly, must have missed out a step somewhere.
Perhaps some step by step instructions for each mode would help others trying this for the first time, especially those new to Linux
Thanks for the tip about the audio in Tutioune.

[M0DTS]

I see up to 18dB MER using the DigiLite modulator and 250Ks (ish) IQ filter. With Ugly mode it's about 14dB max.

The reason i was not seeing anything on tutioune video is because i was impatient, it takes upto 20sec for it to 'see' the video appear sometimes on low SR.

Today i've been testing 250Ks mode some more, checked the modified script was correct and it looks ok, attached is a screen grab of live transmission in Ugly mode!
The video content effects the dts/pcr timing error quite a lot at low sr, the video data must be more variable....
Reducing the raspicam video bit rate to 250kb/s when the mux rate is 400kb/s stops it happening and works very well.

Rob
M0DTS

[f5oeo]

Excelent Rob...
Yes, H264 coming from camera is not constant bit rate right now. Reducing the bitrate of the video to get margin is the solution right now, but we loose video quality. I work right now on parameters of H264 encoder, and it seems now to be quiet CBR. However, there is some drop frame sometimes which is not well handle by the TS encapsulator. Still need some time to fix that. Target is also to reduce latency.

Just a question : on your screenshot you use 1.291.66 GHz : Normaly it is on 1.0625GHz.. Have you set a particular frequency on Ugly to do that ?

[M0DTS]

Yes as the frequency i put 1280 in UglyDATV commandline, i guess it just gives some nearby harmonic?

Rob

[G4KLB]

DATV from a Raspberry Pi, with no hardware, just a short wire on a GPIO pin!!!!!!!!!!!!!
Crazy but it works

[g8gtz]

Success

Ugly mode

437.5MHz (via SUP2400), 1 Msymbol, 7/8 FEC, 17 dB MER

RPI.png (364.49 KiB) Viewed 9255 times

[PA3CGN]

Hello Evariste.
I'm carefully following this topic, and i am curious if there is also the possibility in using an usb capure device for inputting video ,to be more flexible in using other camera's etc
I would like to test this also on 23 cm but i have to change the 62.5 Mhz. to have an odd harmonic in the 23 cm band isn't it ?

thanks in advance from PA3CGN Hans

[f5oeo]

Hi Hans,

Capturing a composite input through USB is in the pipe. I tested a Ezcap and have to work to use the video input to work with the hardware H264 encoder. This will be possible in a near future I guess.
For 23 CM, in the current version you could only have main frequency an integer divider of 62.5. So harmonic on 23 cm is about 1.0625GHz. With the future release you could choose a non integer divider which make a larger choice of frequencies, thus harmonics to suit in the 23cm band.
Stay tuned !

73's Evariste F5OEO