Gamma correction. Do we need it?

Forum for discussion of narrow-bandwidth mechanical television

Moderators: Dave Moll, Andrew Davie, Steve Anderson

Postby Steve Anderson » Mon Jul 02, 2007 2:27 pm

Stephen wrote:Also, thermionic valves and FETs have a square-law characteristic and such amplifying devices have a highly non-linear effect that can do the same thing. That is why any high quality amplifiers that employ such devices have to have negative feedback circuitry. .


For reasons I cannot comprehend there are a lot of members of the 'audiophile' community who absolutely distain the use of feedback. This generally applies to the tube/valve lovers, if an amplifier uses feedback, they simply won't buy or build it.

So to get around the problems introduced by individual stages they often use 'complimentary distortion'. The output stage has a certain distortion characteristic, so they arrange the preceeding stage to have the compliment or inverse characteristic.

At first when I heard about this I scratched my head and thought, "Surely, if you distort an already distorted signal, you'll simply end up with more distortion".

So to convince myself about this or otherwise I rigged up two op-amps. Running off split-supply rails, the first one had a gain of 1.0 for positive signals, 0.5 for negative signals. As expected it sounded awful!

The following one had a gain of 1.0 for positive signals and a gain of 2.0 for negative signals the ouput of which sounded fine, no real audable difference from the source material. I was stunned.

Of course it seems correct and logical, but there is a catch. The op-amps I used (just general purpose ones, nothing special) have a bandwidth far more than our ears need. But, if you place a bandwidth limitation between the two it starts to fall apart.

In a simplistic way consider two audio signals at 15 and 16kHz, with distortion (deliberate or otherwise) these will generate additional signals at 1kHz and 31kHz plus thier fundamental harmonics of 30 and 32kHz and so on.

Recording this onto a CD or using a soundcard the 30kHz and above harmonics/sidebands are lost, so 'undoing' the distortion will not result in the original signal, it will be improved, but not perfectly recreated.

Now, of course a lot of research went into this and Gamma correction (a form of distortion) was adopted as its positives outweighed its negatives (excuse the pun).

However within the context of NBTV I can't see that it confers any advantages. Our detectors are linear, our display devices are linear* and with CDs and using soundcards, signal-to-noise ratio is not an issue. The only time we need to use Gamma correction is to undo that applied to the output of a FSTV camcorder or off-air signal etc. or if you are using a noisy circuit like HF.

As far as quantization noise goes I haven't noticed any artifacts when using 8-bit video as one would expect. Perhaps with something like 6-bit you might as I mentioned earlier.

Your thoughts gents?

Steve A

*Except those like me using CRT display devices, but there are ways around that which I'm quite happy to accept.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Previous

Return to Mechanical NBTV

Who is online

Users browsing this forum: No registered users and 26 guests

cron