# Video DAC's how much of a difference?

Discussion in 'Archived Threads 2001-2004' started by John-Miles, Sep 27, 2002.

1. ### John-Miles Screenwriter

Joined:
Nov 29, 2001
Messages:
1,216
3
Trophy Points:
110
Im not an electrical engineer.... but I do understand DAC's that being said I am wondering what kind of a noticable difference various DAC's provide in DVD players. for example I have seen some 54MHz 10 and 12 bit DAC's as well as some players with 108MHz 12-bit DAC's. Obviously the 12 bit 108MHz is preferable, but how mucha re you losing out with a 10-bit 54MHz?

2. ### Jagan Seshadri Supporting Actor

Joined:
Nov 5, 2001
Messages:
528
0
Trophy Points:
0
Short answer: 54MHz/10-bit DAC is probably all you need.
Any time that processing is applied to a digital signal, mathematical operations are performed on the digital data. If you represent the signal using 'bare minimum' requirements, then it is quite likely that digital signal processing will end up throwing away information that would otherwise be retained in the 9th or 10th bit if those bits were around.
Example: \$1.25 (3 significant digits...think of it as 3 'bits')
Digital Processing (Task: divide by two, multiply by two):
\$1.25 / 2 = \$0.62 (actually \$0.625, but we can't store the 5)
\$0.62 * 2 = \$1.24
So, now we have an error of \$0.01 thanks to doing digital processing using the bare minimum bit-depth. If we had allowed another couple of significant digits, the signal information would be retained and we'd end up with:
\$0.625 * 2 = \$1.25....in other words, no error.
Back to DVDs:
DVDs record the video signal component data [luminance, chrominance(blue), and chrominance(red)..in other words Y'Cb'Cr] with 8 bits per component. Furthermore, the luminance component is has the greatest signal frequencies and must be sampled at 13.5MHz (interlaced). Therefore, the bare minimum DAC requirement for interlaced DVD is 13.5MHz, 8-bits.
Progressive-scan DVDs have double the luminance frequency (double the amount of lines per frame). Therefore, the bare minimum DAC requirement for progressive DVD is 27 MHz, 8-bit.
By increasing the bit-depth of the DAC, you end up with less error in the resulting analog video signal, and by increasing the sampling frequency from 27MHz to 54MHz (2x oversampling) or even 108MHz (4x oversampling) then it is easier to design the video signal 'reconstruction filters'. This results in sharper picture.
The law of diminishing returns applies to these numbers. You probably won't be able to tell the difference between a 54MHz/10-bit DAC and a 108MHz/10-bit DAC. Depending on the digital video processing algorithms used, you may benefit from having a 54MHz/12-bit DAC instead of a 54MHz/10-bit DAC, but you'd have to see it for yourself.
Frankly, these DAC differences show up when you are viewing test patterns (like on the AVIA disc), but are less noticeable when you are watching a movie and are caught up in the action. Still, it is good to know what you are paying for in a DVD player.
-JNS

3. ### John-Miles Screenwriter

Joined:
Nov 29, 2001
Messages:
1,216
3
Trophy Points:
110
Thanks Jagan, as always you come up with the answers i need.

4. ### Rich G Stunt Coordinator

Joined:
Aug 22, 2000
Messages:
115
0
Trophy Points:
0
That was very well explained, Jagan. I've tried to ask the same question before but never got a usable answer.

Can you explain the differences in those DAC's using interlaced or proscan modes? In other words, do the faster DAC's help in interlace mode or just in proscan?

Thanks,

5. ### Phil Nichols Second Unit

Joined:
Sep 7, 2000
Messages:
345
0
Trophy Points:
0
Jagan,
You stated that "......by increasing the sampling frequency from 27MHz to 54MHz (2x oversampling) or even 108MHz (4x oversampling) then it is easier to design the video signal 'reconstruction filters'. This results in sharper picture.....".
I believe this may not be "exactly" the result? Doesn't the high over-sampling rates made possible by the higher DAC bandwidth mean that the natural DAC conversion noise can be moved out of the video band so that it can then be filtered out better and with less expensive and sharper filters? Therefore your "This results in sharper picture...." might be more accurately stated as "This results in a less noisey picture.......".
We are probably saying the same thing, but I always think of "sharper" as different than "less noise". I've read that it's the greater DAC bit depth that provides for more image detail and the greater DAC bandwidth that provides for more freedom from conversion noise.
Of course some guys would call all this "marketing hype"!

6. ### Jagan Seshadri Supporting Actor

Joined:
Nov 5, 2001
Messages:
528
0
Trophy Points:
0
Phil said:

7. ### Jagan Seshadri Supporting Actor

Joined:
Nov 5, 2001
Messages:
528
0
Trophy Points:
0

8. ### Ian Montgomerie Stunt Coordinator

Joined:
Feb 2, 2002
Messages:
112
0
Trophy Points:
0
The "stats" themselves in modern DACs, which usually start at 54/10 and go up from there, are basically meaningless in themselves. You don't really get visually noticable differences from increasing bit-depth/frequency from this to 108/12 and so on. However, there are differences in DAC quality that don't relate to these simple stats, and in general a more expensive DAC will be a little bit better even independent of its "stats". Diminishing returns still apply quite strongly, though. The typical modern DVD player ships with video DACs integrated into the decoder chip, and the entire decoder chip sells for 5-10 dollars. Buying a 500 dollar DAC might get you 10% better quality. Just barely noticable (although self-delusion makes a lot of people think it's HUGELY OBVIOUS when they know in advance which player is "supposed" to look better), not a night and day difference.

Diminishing returns occur very quickly with this kind of digital technology because the expensive stuff doesn't cost 10-100 times as much due to having 10-100 times more circuitry, more logic gates, more software development, etc. The actual hardware is usually not that much more impressive, but prices are driven up phenomenally by the low volume of sales, the fact that the part isn't integrated with another chip, etc.

Joined:
Aug 22, 2000
Messages:
115