Accurate Mapping of MD 9-bit to 24-bit Colour-Codes

For anything related to VDP (plane, color, sprite, tiles)

Moderators: BigEvilCorporation, Mask of Destiny

Chilly Willy
Very interested
Posts: 2984
Joined: Fri Aug 17, 2007 9:33 pm

Post by Chilly Willy » Thu Apr 18, 2013 1:49 am

The "standard" tolerance band colors are none, silver, and gold, meaning 20%, 10%, and 5% respectively. They have since added even more colors for tighter tolerances... you can find those on wikipedia:

http://en.wikipedia.org/wiki/Electronic ... lor-coding

TmEE co.(TM)
Very interested
Posts: 2440
Joined: Tue Dec 05, 2006 1:37 pm
Location: Estonia, Rapla City
Contact:

Post by TmEE co.(TM) » Thu Apr 18, 2013 3:16 am

I was not talking about tolerance, but the article did remind me what I was talking about :

http://en.wikipedia.org/wiki/E12_series#E_series
Mida sa loed ? Nagunii aru ei saa ;)
http://www.tmeeco.eu
Files of all broken links and images of mine are found here : http://www.tmeeco.eu/FileDen

Chilly Willy
Very interested
Posts: 2984
Joined: Fri Aug 17, 2007 9:33 pm

Post by Chilly Willy » Thu Apr 18, 2013 5:42 pm

TmEE co.(TM) wrote:I was not talking about tolerance, but the article did remind me what I was talking about :

http://en.wikipedia.org/wiki/E12_series#E_series
That's pretty interesting - they didn't have that when I went through electronics. :lol:

r57shell
Very interested
Posts: 478
Joined: Sun Dec 23, 2012 1:30 pm
Location: Russia
Contact:

Post by r57shell » Sun Jun 09, 2013 11:53 am

Chilly Willy, how much deviation of your table?
I suppose +-0.05, then in RGB it will be:
0.05/4.7*255 = 27,128
So, it is +- 27 in RGB. :S
To make +-1 in RGB, needed 1/255*4.7 = 0,0184 deviation.
Image

Chilly Willy
Very interested
Posts: 2984
Joined: Fri Aug 17, 2007 9:33 pm

Post by Chilly Willy » Sun Jun 09, 2013 5:55 pm

The steps between values are different by the quantization amount, i.e., 1/256, or 0.0039. That means if you want your output to be accurate to within +/- 1 bit (very commonly wanted), you need better than 0.008, or closer than 1% (0.8% in this case). So from the table I linked, you want a tolerance grade C part (+/- 0.25%).

That kind of (ridiculous) accuracy shows why resistor ladders are not common over more than 5 or 6 bits. What you do instead is use the resistor ladder for 4 bits, then use PWM for the other 4 bits. A two stage DAC can use much more sloppy tolerances, and hence be cheaper.

r57shell
Very interested
Posts: 478
Joined: Sun Dec 23, 2012 1:30 pm
Location: Russia
Contact:

Post by r57shell » Sun Jun 09, 2013 6:01 pm

:S
I just want to know precision of this measurements, and therefore range of variations. I don't want to modify anything... yet :]

EDIT: Oh, it's my mistake in calculations: 0.05/4,7*255 = 2,713. It is much better. From HardwareMan's image: voltage varies 4.7/50 = 0.094, but very rapidly. So, range about +-5 in RGB 24bpp.

EDIT 2: There is symmetry in this relation, at point (3.66;2.4973) ((0..7);(0..47)). And, if you assume that it is halftone, then maximum is 5v. :]
Image

Post Reply