Patterns over Video Capture from G400-TV

[Phire]

While not even Matrox Tech's can figure out how to fix this, I was wondering if anyone else could figure it out. We think its currently interference causing the problem, but it'd be nice if someone could truely confirm it, someone who has expierence in interference problems.
http://forum.matrox.com/mgaforum/For...ML/000320.html

[Michel Carleer]

Hi Phire,
If you remember, I already suggested that it was some sort of interference. I further stated that it might be the colour subcarrier that would be the interfering frequency. Then Haig said that if the colour subcarrier was off, then you would have B/W pictures. This is not what I had in mind. What I had in mind was that for some reason (a filter not properly tuned or defective) part of the subcarrier was leaking in the luminance signal. The reason that I suggested interference with the colour subcarrier is that it would produce roughly the kind of patterns you showed us on your shots. But of course without any further proof, this is only speculation. What is sure however, and it is sure because you actually see a pattern and not random noise, is that the interfering signal is in one way or another frequency locked to the line (horizontal) frequency of the picture. In other word, I don't think it comes from radiation by i.e. the CPU because there is no reason that the CPU frequency be frequency locked to the line frequency of your video signal. This points me towards an interference by a signal generated on the capture card itself and derived from the video signal itself. Once again, the colour subcarrier is a possible candidate, because by design there is a relationship between the line frequency and the subcarrier frequency. In fact, the frequency of the colour subcarrier has been chosen such as to minimize cross-over between the chrominance and the luminance signals. The important word here is 'minimize'. If your source equipment is not properly adjusted and generates a subcarrier with too much amplitude, then it will be seen in the luminance of the picture in the form of the patterns you see. And of course it will be most noticeable on parts of the picture which have a very constant luminance value to start with. This is very often the case with cartoon pictures, but rather rare on footage of the real world. Cartoons usually show very strongly any problems in the generation, transmission and reception of video signals, because of these large 'a plat' (constant luminance or chrominance over large areas) that they are made of.
Other than that I am not a specialist in interferences nor in TV signals, so that I don't know how to proove or rule out the ideas above. But one thing for sure: if it is interference, then the frequency of the interfering signal is somehow connected to the frequencies of the video signal. Otherwise you would not see a regular pattern in the pictures.
Michka
[Edited portion]
Wait, wait: Can't you find (or generate) a purely B/W video source material, preferably of the cartoon type? Because in the case of B/W pictures, the amplitude of the coulour subcarrier is zero, and you should see no pattern in your pictures.

[This message has been edited by Michel Carleer (edited 19 April 2000).]

[alp]

I have this problem too, I am highly disappointed with Matrox about the pitfalls around the Marvel drivers.
At the end is like the video features are in a constant Beta testing, and the only thing that works is the O/S Display driver.
Probably is the Zoran chip to blame since all the products using this chips are surrounded with endless bugs.
If the so called car's lemmon law were applied to computer part manufactures, many of them would be out of buisness or consummers will have products that work as advertised.

[This_Idiot]

Phire,

What kind of environment are you working in ? A while back we had some instances of interference that were traced back to RF interference from outside sources. Especially bad are those low-voltage desk lamps that have a halogen bulb and a transformer in the base. I've got one here that I don't often use (I usually work either in natural daylight or in the screenglow), which really upsets my monitor.

There was another guy who worked in a room behind his kitchen, and the refrigerator was just the other side of the wall - that gave some interesting results. Do you work with your PC open, or with the case on ? Mobile phone or walkabout handset anywhere nearby ?

Just a shot in the dark, but RF interference can definitely give the kind of moire that you are seeing

Chris

[Phire]

Well, I have a few devices outside of the puter.. I even have an extention cord near the puter case, not too close but near. and my DSL modem near the case. With all that in mind.. Shouldn't my computer case block off all outside Interference?

Ok thanks for your input Michka, i will tell this to haig and see what he has to say.. Unfourtunatly i dont have any black and white sources to check from..

[DrMordrid]

Not all computer cases are that well shielded against RFI. Most cheap ones aren't, but some inexpensive ones like the Supercase are fully RFI certified.

Many of these cases have large openings behind the face panel or in the back. Many have poor grounding. Some leak like sieves.

And then there are those power supplies, cards and mainboards that give off RF like a 60's AM station themselves. Mostly these are also the cheapies.

On top of that every cable you have running in and out is a potential antenna. Unless you have audio/video cables the quality of the Monster or other major brands RF can get in through them.

Not to mention that cheap VCR's have shielding that might as well be made of air.

Shop carefully....

Dr. Mordrid

[Michel Carleer]

Addendum to my previous post, and kind of note added in proof:
Phire, I have looked carefully at the picture you published in your post to the matrox forum (the car above the platform, and the sky and see to the right). Have you noticed, and it is especially clear in the blues on the right side of the picture, that the more the colour is saturated, the stronger the pattern. This points definitely to my original idea of a cross-over with the colour sub-carrier (the amplitude of the subcarrier is proportional to the colour saturation, it's phase being proportional to the hue).
Unfortunately, it does not tell us if it is the video source equipment or the Matrox card that produces the pattern.
Michka

[Phire]

Is there any devices you can buy that detect RFI interference on a meter?

[DrMordrid]

Yes, but their readings and instructions only mean anything to an experienced technician. Also the good ones, those that can be tuned to a particular band, are $$.

The money would be better spent on good cables, preferrably with gold plated connectors, and the other things I mentioned.

Inside the CLOSED case it's important to have the best quality power supply you can afford, which will help not only RFI but stability.

Outside the box component placement can be very important. Bear in mind that most of the RF coming off monitors goes to the sides, back and top. These emissions mostly come out of the flyback (HV) transformer and image tube field coils. The bottoms are usually well shielded as there is usually a full coverage shield over the traces of the circuit board. The front has the benefit of the image tubes x-radiation shield.

Because of these emissions placing the VCR next to, or worse yet above, a monitor or TV can be a rather bad idea.

In my lab I use a homebrewed "bridge" under my systems monitor and place the main SVHS VCR under it, directly in line with the monitors internal shield. The video preview monitor is on the wall, above and to the side of the main monitor and on a metal mount, which itself acts as a shield.

The other decks are on shelves, well out of range, with well shielded cables connecting them to a distribution box & Elite BVP-4 video processor.

Dr. Mordrid


[This message has been edited by DrMordrid (edited 21 April 2000).]

[Fred H]

Heh...
RFI comes not only via the case. The keyboard, monitor itself, mouse and loudspeakers(!) captures a high amount of RFI.
VCRs contain oscillators, TVs radios too.
As an old radioamateur, I remember when I tried to send RTTY (RadioTeleTYpe) first time, with my old 8086 CPU. Pressing the letter key generated a different letter on the screen because the high RFI level.
It was extreme, anyway.
Grounding cases, ferite cores on loadspeaker wire, but better quality of the equipment and connections etc. helps.

[Phire]

I think I should mention that no patterns appear on RGB capture :P so I think the wires are passing through properly.. 640x480 RGB capture doesnt show a single pattern. So something is causing MJPEG chipset interference or its a defective card.. One thing that sticks in my mind is could it be the other pci cards above and below it.

[Michel Carleer]

Phire,
Did you try with the computer box open? Does it increase or decrease the intensity of the pattern?
And yes, you might also try swapping the PCI cards around.
Michka

[egov]

Ok, after weeks of misery we found the cause of the zigzag patterns -- power supply! Matrox G200 and G400 is very sensitive to the noise from the power supply. I tried my G200 TV on few different supplies with no improvement. Here's my original message:
http://forum.matrox.com/mgaforum/For...ML/002509.html
If you find a sollution, please, tell me.