Composite colour aberration on NMS 8245

Page 1/3
| 2 | 3

By Bondy

Resident (54)

Bondy's picture

05-01-2016, 09:03

I usually use the RGB provided by the scart connector on my 8245 but recently I tried the composite using a scart splitter(that just derives the composite signal). On Space Manbow and Grauw's Synthe6 program (haven't noticed on any other programs) There is a split or something in the reds so that they appear as red and green stripes. Here is a photo: This does not occur when using RGB. I'm wondering if anyone is familiar with this problem and can help me out :-?

Login or register to post comments

By Metalion

Paragon (1149)

Metalion's picture

05-01-2016, 11:28

Nothing really new there.

It's just the visible effect of the loss of video signal quality when switching from RGB (high quality) to Composite (low quality). In the composite signal, the colors are mixed whereas in the RGB signal they are separate, thus achieving higher quality. Another result is a lower capacity of scanlines: around 240 lines with composite signal, up to 576 lines with RGB signal (for PAL).

By Bondy

Resident (54)

Bondy's picture

05-01-2016, 12:14

Metalion wrote:

Nothing really new there.

It's just the visible effect of the loss of video signal quality when switching from RGB (high quality) to Composite (low quality). In the composite signal, the colors are mixed whereas in the RGB signal they are separate, thus achieving higher quality. Another result is a lower capacity of scanlines: around 240 lines with composite signal, up to 576 lines with RGB signal (for PAL).

I don't think so. Even though Composite is of low quality (colour bleeding etc) colours don't jump scanlines! If the guns were misaligned this problem would show up in RGB (it would be more obvious). I assure you the alignment on this monitor is pretty bang on. Also the TV res doesn't magically go up when you use RGB for the same output it's just a clearer and purer signal; The monitor is always showing 240 lines (or the PAL equivalent- 288 or whatever it is) at one time whichever you use. This leads me to think it is a hardware problem or this particular system doesn't do composite very well at all.

Edit: Unless the MSX has some kind of interlacing mode and this is a symptom of that?

Edit Edit: If this was a signal quality issue the picture would be smudged, not defined and demarcated by the scanlines (there is no red in the green and vice versa)

Edit Edit Edit: For instance the white in that photo wouldn't be white at all if the allignment was off.

By Metalion

Paragon (1149)

Metalion's picture

05-01-2016, 15:40

Look at all the white pixels on your picture. Some of them are indeed white, but a lot of of them are blue on one line and purple on the other. So the white color is also "split".

By sd_snatcher

Prophet (3296)

sd_snatcher's picture

05-01-2016, 21:19

To me, they look like Hanover bars, an undesired side effect of the PAL encoding when the color carrier is misaligned.

On a computer, this means that the color crystal frequency is drifting. Unfortunately, the entire video section of the NMS-8245 service manual from the Hans Otten website seems to be missing. So I can't tell you if there's a trimpot to adjust that or not.

It would be good if someone could scan and add the missing pages to that service manual.

By l_oliveira

Hero (530)

l_oliveira's picture

06-01-2016, 01:20

NMS-8245 use the usual RGB to Y/R-Y/B-Y transistor based converter with then a LV-510 Y/R-Y/B-Y to CVBS converter chip which is used on most Philips computers with V9938 VDP.

That board is driven with a free running 4.43Mhz oscillator crystal so yes, it is not running synchronized with the horizontal sync and there is your Hanover bars source.

By Bondy

Resident (54)

Bondy's picture

06-01-2016, 16:12

Thanks, can anyone scan those pages for me? Or should I try finding a replacement for the oscillator that l_oliveira mentions?

By l_oliveira

Hero (530)

l_oliveira's picture

06-01-2016, 16:56

The only fix for the problem pictured is use RGB as video connection.

The effect is inherent of CVBS connection system. Because it saves wiring connections by compressing the video into a single wire there's loss of video signal quality.

By flyguille

Prophet (3029)

flyguille's picture

06-01-2016, 17:37

Bondy wrote:

I usually use the RGB provided by the scart connector on my 8245 but recently I tried the composite using a scart splitter(that just derives the composite signal). On Space Manbow and Grauw's Synthe6 program (haven't noticed on any other programs) There is a split or something in the reds so that they appear as red and green stripes. Here is a photo: This does not occur when using RGB. I'm wondering if anyone is familiar with this problem and can help me out :-?

that is a problem in the PAL encoding, the encoding shift 180 degree the HUE on odd scanlines, it is a responsability of the TV SET to merge both scanlines color angles (HUE) (the ones in the upper even scanline + the next odd scanline) into the average one, that way the the standard avoid color shifting due to CLOUDS and anothers climate events.

But that feature also means that the color resolution will be HALF in the vertical than the direct RGB, this doesn't happens with the LUMINANCE signal that is full resolution.

Color HUE angle shifting (see the chart)

the white in the middle is the XY axis imagine a cartesian coordinates, and select the color and trace a vector to it, that is the color hue angle, I don't remember if the NTSC has it, but in PAL, the color angle is shifted 180 degree on EVEN scanlines, or maybe is the inverse.

The TV set has a acoustic delay line (it works converting signal into sound waves, and then pickup it again, as the sound is slower it delays one scan line), that way the TV set can has the information of both scanlines at the same time, mix that, and the color resulted is the average angle. If I thinks correctly, it shift 90º clockwise in even scanlines, and shift 90º in odd scanlines but in the other direction, when merging results in the original color.

(a side note: As the color angle shifting is limited by the limited chroma bandwidth, the angle can be shifted as fast as a certain speed, that is, the angle can be shift to get a complete opposite angle color like in 1 and half MSX PIXEL width, if not two msx pixels, that is why when thin vertical lines with completely different color are next together, the color BLURs, the angle can't be shifter faster as it necessary and ir shows all the inbetween color in ugly oscilation mixing all togeter result in grays oscilations and blurred colors).

Continuing with the topic.
If the chroma oscilator is out of synchronism (transmitter vs receiver), the decoding of the angles is wrong. And the merge result in a floating angle that is shifted 180º degree between odd / even scanlines, that way you see that it change to a completely different color for each next scanlines.

Why the TV set has its own chroma oscilator... because the chroma information is transmitter without the modulation carrier for saving bandwidth and don't affect in excess the luminance (both chroma & luminance are transmitted in the same bandwidth space), the guy that invented that was a genius, to saved all preexistent tv sets, they found a bandwidth hole at the rythm of the horizontal frequency, and they fits all the chroma in that tiny hole (it was the gloryhole of the TV norms!)!.

In the mix, TV set circuitry is not only the delay line, but also a intensity preset next to the delay line, for balance the mix, if that is out of adjust, one of the scanline information (say EVEN or ODD) is stronger, and the average angle result in a biases angle that shifts from odd/even lines because it is out of balance. Check if the TV do the same in AIR channels, if it does, you needs to adjust that preset, if not, it is an unbalance in the transmitter circuitry.

But, if the TV set is ok, receiving channels., it is most probably the transmitter is transmitting in the wrong PAL norm, or NTSC instead PAL or reversed. Now, TV sets since the 70's has a stage called "chroma killer", that is, if it detects that can't sync with the chroma frequency that is transmitted, it KILLs the color, and present a nice white and black image.

But sometimes, that stage can't detect that it is not in sync, because by example the sync., does within a subharmonic of the frequency.

And that is maybe what is happening.

First, check if the composite encoder is outputing the right TV NORM, and check if your TV set has MANUAL norm control, instead of AUTO NORM, try to set manually what you thinks the msx is transmitting.

If that doesn't works, most probable the msx circuitry is out of calibration, needs repair or adjust.

Check the cristal frequency, check if that match with the NORM you thinks the TV set expect.

By l_oliveira

Hero (530)

l_oliveira's picture

06-01-2016, 17:34

I believe the problem here is actually the fact that the encoder chips were made with NTSC in mind. (Japanese made chips) And while they operate in PAL satisfactorily, they're actually optimum for NTSC video encoding.

I remember seeing similar green tint leaks like that on NTSC SEGA Mega Drive consoles which were modified to operate on PAL standard. It only happened when RED saturated (the console outputted very strong red colors). Restoring them to operate in NTSC made the problem go away. The 50Hz/60Hz thing didn't affect the problem at all so operating at NTSC with 50Hz output I didn't see the green artifacts too.

By flyguille

Prophet (3029)

flyguille's picture

06-01-2016, 17:40

no, chips normally are dual norm, and the timming is not done inside the chip, it is in the discrete components that are around the chip.

Again, check the tv set seeing if you see the same effect in TV cable or air transmissions.

If it do, it is just a preset adjustment, you can't fail in doing that.

Page 1/3
| 2 | 3