What makes the 2D quality of Matrox better then other ?

[Slashhead]

Yes, it is indeed showing the cable. I was a little bit confused by some other pictures i have seen on the net.
It was not a RC filter but a lowpass filter they put on the circuit.

Let me show you the picture I was intend to show:


The reason for the LPF is the huge amount of harmonic signals caused by the square waves from the output DAC.
What they are doing is decrease the bandwidth of the signal to meet the CE mark.
As you can see, there is not much left from the original signal.
Seeing this, why bother to have a sharp edge ?

Picture is comming from here an also interesting article on image quality.

[This message has been edited by Slashhead (edited 19 April 2001).]

[az]

I wonder why some manufacturers (like matrox) who care about visual quality don't put a copper box around their output stage to avoid the filters and still don't emit high freq signals - i know this would rise the price of their boards a few $, but if it increases quality and if that is one of their key sales arguments?

AZ

[rubank]

Slashhead,
You´re saying the higher the frequency the lower the temp in the DAC?
I don´t think so.

A stereo amplifier (a very analogue device) always degrades the signal, to some extent. Does that mean you shouldn´t care about the quality of your records/CDs (the input) ??
Of course not. Same thing as your "bother to have a sharp edge", no?

I don´t see the point to your reasoning.

As for the article from "porotuner" regarding the MX-cards: been there, done that, doesn´t work.

rubank

[This message has been edited by rubank (edited 19 April 2001).]

[Slashhead]

<font face="Verdana, Arial, Helvetica" size="2">Originally posted by rubank:
You´re saying the higher the frequency the lower the temp in the DAC?
I don´t think so.

</font>

What do you mean ?
I have posted something a couple of hours ago, but i had removed it hours before you did your reply.
What I was stating with temperature is the changing in the DAC from a "0" to "1". It has nothing to do with frequency. The change from "0" to "1" takes the most energy. So to keep the energy as low as possible you need to switch very quickly. This has absolutely nothing to do with the frequency. Probably it has something to do with the sharp edges you have to remove in a later process. (LPF)



[This message has been edited by Slashhead (edited 19 April 2001).]

[rubank]

Slashhead,

changing 0 to 1 to 0 to 1... IS frequency.
The more rapid the changes, the higher the frequency.

Pls reread my former message, it´s been edited.

rubank

[rylan]

Slashhead: Some answers to your post.

Min required DAC speed is HRESxVRESxREFRESH
So, for 1600x1200@100Hz the DAC must be capable of at least 192MHz.

Not sure what you mean exactly by 'available bandwidth of your 2D circuit'. Your max resolution/refresh is limited by the rating of the DAC. However, this number is usually lower due to the filtering on the output stage or crosstalk at the high dac frequencies.

The filters on the output are for reducing EMI from the board. In addition, it helps with the problem of overshoot. This can occur with extremely fast rise times, where the signal voltage actually overshoots its mark and then comes back down to normal.
For example, a full white screen may be a 750mV dignal, however with a really fast rise time, each pulse may actually peak at 800mV or more before it drops back down to 750mV. This also introduces 'ringing' where the signal has to bound around a little bit before it stabilizes.

Hsync/Vsync are are controled by the drawing engine (GPU). Hync is a high frequency analog clock, and Vsync is low frequency (70-120Hz usually for most monitors).

RGB crosstalk usually occurs within the DAC itself as it is converting the digital signals to analog outputs. It can also occur at the amplifiers on the output stage, especially if isolated voltage rails for each line are not used.

Best way for fast rise/fall times are to use a fully shielded cable (that does NOT have beads on it) and get rid of some of the filters on the output. Basically, thats what we do were I work.. we designed a metal mesh wrapped BNC cable with metal jacketed connectors. Because of this, we didn't need any filtering on the output, and have really fast edge rates right to the monitor, with no emissions due to the shidled cable.

[Maggi]

<font face="Verdana, Arial, Helvetica" size="2">Originally posted by rylan:
Slashhead: Some answers to your post.

Min required DAC speed is HRESxVRESxREFRESH
So, for 1600x1200@100Hz the DAC must be capable of at least 192MHz.
...</font>

there must be something wrong in that calculation, because 1600x1200@85Hz already uses a RAMDAC frequency of 229MHz !

[rubank]

Hi Maggi,

there seems to something wrong with your calculator,

or is it to much beer

rubank

[Slashhead]

Rylan, the calculation for the RAMDAC is not correct.
Look at the table shown here

Also on the german site:

<font face="Verdana, Arial, Helvetica" size="2">
Bereits bei einer Auflösung von 800x600 Punkten und 85 Hertz Bildwiederholrate liegt die Pixelfrequenz bei 50 MHz.
</font>

Rubank, that's what i ment to say with hrowing with numbers without any explanation.


It is not just multiplying the resolution and the refreshrate.

Thanks for the other answers Rylan.


[This message has been edited by Slashhead (edited 19 April 2001).]

[rylan]

Sorry.. you're right. I forgot the conversion constant to take into account the time needed for the CRT electron beam to sweep and draw onto the screen.
This constant is ~1.32 timex my above formula. However, the actual ramdac speed is set to about ~1.4 for some safety margin, and to account for signal propogation in the VGA cable.

So, if you use H x V x Refresh x 1.4 you can get pretty close to your actual DAC speed.

The straight formula speed I posted earlier actually is technically correct for the raw -MIN- required DAC speed needed, but the dac is cranked up more to account for the display delay etc.

[Slashhead]

<font face="Verdana, Arial, Helvetica" size="2">Originally posted by rylan:
Best way for fast rise/fall times are to use a fully shielded cable (that does NOT have beads on it) and get rid of some of the filters on the output. Basically, thats what we do were I work.. we designed a metal mesh wrapped BNC cable with metal jacketed connectors. Because of this, we didn't need any filtering on the output, and have really fast edge rates right to the monitor, with no emissions due to the shidled cable. </font>

Sounds very interesting, on what website can we see some work of the company you're working for?

Those sharp edges from the DAC are they necessary to load as many energy in the filter as possible for a particular pixel, so that your effective voltage is almost the same as you want?
What kind of filter is recommended for that kind of signals, since you must use one. Are there special components for this kind of filtering ? Perhaps when you are using cheap components the filtering could be pretty poor.

[SteveC]

Bloody analogue circuity! I hate it! Give me ones and zeros any day.

I know what the answer is to all our 2D image quality problems - everyone should use a Digital Flat Panel screen with the digital interface.

------------------
Cheers,
Steve

"Life is what we make of it, yet most of us just fake"

[rylan]

Slashhead: The company I work for is Dome Imaging Systems.. website is www.dome.com
some of our display controler and stuff are on there. Its a new website, and I haven't looked around it yet, so I don't have any specific links to give you :P

SteveC: Exactly.. thats why we have a high resolution digital flat panel now heheheh

[rubank]

Slashhead,

I´m sorry if I´m a little slow to answer from time to time, for now I´m on a modem connection, so I just check now and again.

The answer to the quote you made from tecchannel.de (last post on side 1) is a couple of rows above your quote in the article:
the pixel frequency is a product of the horizontal resolution multiplied with the HSYNC frequency, so in this case the math would be 50 000 000 / 800 = 62 500, i.e. the horizontal frequency is 62,5 khz.

So, they don´t just throw numbers around after all

Steve, if only those bloody flatpanels were as good as CRT:s in every other respect besides sharpness (including price ).

When it comes to burning your forehead LCD:s are utterly useless

rubank



[This message has been edited by rubank (edited 19 April 2001).]

[Slashhead]

<font face="Verdana, Arial, Helvetica" size="2">Originally posted by az:
I wonder why some manufacturers (like matrox) who care about visual quality don't put a copper box around their output stage to avoid the filters and still don't emit high freq signals </font>

I think the problem also occurs in the VGA cable. It will probably working like an antenna with that kind of banwidth.
Perhaps you should do the DA conversion in the monitor itself. And make the monitor CE compatible.