Color burst problem?

[tanaka]

Hi,
I use sample code posted on this forum and got picture displayed in color, but some TV monitor not(It displayed in monochrome).
I had probe a signal with oscilloscope from video output pin, and compared to RS-170 specifications on webpage(written in japanese... according to that, the color burst signal must be 40 IRE level = near 300mV).

It seems amplitude(peak-to-peak voltage) of color burst signal are not correct(not enough).

I tried changing the burst-level value but can't change amplitude of burst-signal...

Is there any solution?
I will appreciate your help with this situation.


RS-170 specification
http://elm-chan.org/docs/rs170a/spec_j.html

Code Posted(VS23S010 Breakout Board with Arduino)
viewtopic.php?f=14&t=2164

[Panu]

Hi!

Hmm, I looked at the code, and the BURST_LEVEL in that example, in file ntsc.h seems to be 0x0d00 + black level. If you change 0x0d00 to 0x0c00 or 0x0b00, does the amplitude rise sufficiently?

Hmm again... when you change the define, check that the compiled code REALLY changes - there's something strange going on, because BLACK_LEVEL and BLANK_LEVEL are both the same (0x66) BUT the DC level of the color burst is clearly not at the same level - or is that due to your changes? The high byte of BURST_LEVEL determines the AC colorburst amplitude and phase and the low byte determines the DC level, e.g. brightness (which should be at blank level during the burst.

Please see the palette editor for more information.

-Panu

[tanaka]

Hi, Panu!
Thank you for your quick reply!

Sorry for the lack of some information in my last post.
There are some difference between original(posted) and my code ...

BLACK_LEVEL changed to 0 IRE (Equal to BLANK_LEVEL) from 7.5 IRE.
BURST_LEVEL's lower byte changed to grater than BLACK_LEVEL.

that's all(maybe not...)

According to the RS-170 specification, it says black-level has two options.
One is the U.S. standard value and another is Japanese(defined as NTSC-J ?).
So, I changed it to 0 IRE on my code.
Changing of BURST_LEVEL's lower byte was from my misunderstanding...

After recieving your repray, I tried another values you adviced me.
Finally, I got a color picture displayed! (0x0B66)
Thank you Panu!

However, still insufficient amplitude(about 130mVp-p).
Change to 0x0866 got better(about 200mVp-p), even though continue trying with smaller values got worse.

The initial objective(one of the tv monitor not display color picture) was completed.
but if there are another solution to enlarge amplitude, please advice!

[Panu]

Hi!

Hmm, GREAT that you got the color on all of your monitors, but strange that you should have problems... from what I recall from how the design was made, the maximum colorburst swing that the modulator can generate should be 2*63/345*140 IRE = 51.1 IRE. But I've never given it much thought as I've never had any problems with it. So I'm really not sure. But I do know that stronger colorburst amplitude gives relatively more weak colors. So usually I've probably kept the colorburst value quite low to get strong colors. 0x88 for the burst would give the largest amplitude, but the colors are probably quite weak at that point.

I must chat with the design engineer about what level the burst should be able to generate. If it proves to be too low for your needs, you could use an external amplifier and boost the entire signal, or if you need a very high end solution, it would be possible to use VS23S040 and take luma and chroma from different video output pins and sum them on your PCB with different gain factors to the external amplifier to get the best quality signal.

[Panu]

Dear Sir,

After checking the design, it seems that your findings are correct, and the levels you have measured are correct. The amplitude of the color signal is less than what I have thought it to be, I am embarrassed to find out. It seems you are the first one to notice this, probably because televisions are very adaptible and can accept even very weak color signals.

Thank you for bringing this to our attention. We are going to investigate this further and try different software and hardware fixes here at VLSI. We will publish new example schematics and PCB versions as soon as we have found the best solutions to this problem.

The question right now is how can we help you best achieve your goal. Can you tell us a little what you are aiming to achieve?

I have a few alternative suggestions of what you can do, in increasing order of complexity.

Solutions requiring only software:

1) Accept the hardware as is, and find a suitable level for the color burst. You find that at 130mV all your displays work. The maximum you can get without hardware modifications is about 200mV.

2) Add a selection of different burst levels to the software, to allow user to fine tune color intensity to match the display used.

Solutions requiring hardware modification:

3) Add a video op-amp to the PCB to amplify the signal. This will amplify the entire signal: both the color and luminance signals, so it may be best to not use the highest levels of white in the picture. For lower power consumption, you can set the VS23S010 into the low level signal mode and amplify with the op-amp. Or you can keep the VS23S010 in the same mode it is now, and drive a 150 ohm resistor from the video out to ground. Then buffer (gain=1) the signal with op-amp to output. This will double the signal level in millivolts.

4) For improved resolution and color reproduction, replace the VS23S010 with VS23S040. Software needs to be modified so that luminance is taken from one output pin and chrominance from another output pin, just like S-Video. Add an external op-amp buffer to the PCB and sum the two video output pins to get a composite video signal. Different gains are used for the two pins: luminance and chrominance. This can give double resolution because 2 x VS23S010 is used. Also full scale signal for chrominance and luminance can be used, giving the best picture. The bad side is that color information and black-and-white information needs to be written to different memory locations in the softare.

Whatever you decide, we are here to help you complete the design!

Sincerely,
Panu

[tanaka]

Hi, Panu!

I'm sorry for the late reply.
I am interested in VL23S040 and want to try one of the solution(#4) you advised me if I could.
If it is possible, Would you please send me some samples for testing?

[Panu]

Hi!

Hmm, ok. We need to write the software to support option #4 and also make a board to design and test the video circuit. This will take some time, how much time do you have available?

To proceed, we need to know which video format you are targetting, what is the resolution and number of visible lines you are targetting, and how are the colors chosen in your application, e.g. is it computer generated graphics such as on-screen menu etc or do you plan to show pictures such as photographs or live video on the screen.

Sincerely,
Panu

[tanaka]

Hi,

This is one of the my DIY project and not for commercial product.
So, if there are no simple solution to fix this problem, it is ok.

but after hearing your advice,I interested in solution that using VS23S040 chip.
so, if it is ok to send me a sample(not completely set of hardware or software, only the chip), I will try to writing a code by myself.

I'm not good at writing (or speaking) in english, so may be confuse you... sorry!

[Panu]

Hi!

Ok, no problem. Sure, we can give you a couple of VS23S040 chips. Just write to sales@vlsi.fi and give your information.

We will try the circuit and software anyway, as soon as we have some time. The idea is to couple video output 2 to video output 1 via a DC blocking capacitor, choose the component values to adjust the relative levels of luminance and chrominance and then use an op-amp to amplify the signal to the correct level.

-Panu