In fact, the picture above shows:

Green trace: HF data coming from the laser pickup.
Yellow trace: Recovered clock
Blue trace: Recovered data

The data is from a Datel disc (“ultimate cheatcodes”) and shows the “laser mark”; except that it’s not written with an actual laser, but simply embedded in the datastream!

To compare it with the picture from the last post, which showed a real, genuine GC disc. There are subtle differences (like the “noise” in the zeroed region).

So toxic is also right with his assumption that this will be the cliffhanger for the 3rd part of the story.

Just as toxic stated.

First, the information-density of the green trace is equal to that of the blue trace, as Tyler pointed out. Blue is in fact, as adam pointed out, the digitized, or “sliced”, version of the green one. Yellow is the recovered clock.

adam is also right when comparing the green signal to an overdriven opamp output; however, no opamp characteristics are involved here. But the waveform distortion can be described as a kind of low-pass filter. In this case, the low-pass effect is caused by the physical proportions of two elements in the system.

You will likely notice the sparseness of the blue signal compared to the clock; that indicates that a special line code has been used, in this case a code to limit the minimum and maximum number of consecutive ones or zeros; this is called an RLL-code, a run-length limited code. The upper bound is set to allow a successful clock recovery, the lower bound to limit the high-frequency components that would get lost due to the “low-pass filter” characteristics. Because of the used code, there is 50% redundancy on top of the payload data. Quite a lot, but that’s life, and also already a 1/17th improvement over the previous generation of this technology.

The long zero sequence, however, is a violation of the RLL rule. And while the data is NRZI encoded, so the data polarity doesn’t matter, the violation will always be low, and never high.

the analog signal seems to be suffering from phase inversion. notice each “pulse” has an M shape to it. i mean to say, if you take the “V” part of the “M” shape and folded it over the horizontal, you would have a wider, taller _upside_down_ “V.”

many op-amps tend to do this when you overdrive their inputs. but i dont think thats whats happening here, because the inversion seems to happen at different voltage levels. this leads me to think that this some type of phase encoded RF. the small amplitude also tends to make me think this is upconverted RF.

i also notice the frequency measurements. the mean frequency is ~25Mhz, with almost perfect STD deviation of 400kHz. this leads me to think that this is a baseband signal between 0-400kHz, modulated on a 25Mhz carrier.

the clock trace is shown for a reason, im guessing its a recovered clock. leading me to think this is some kind of NRZ/machester encoded data.

maybe the break through is that the clock has been recovered, and the data is being demodulated correctly, even though the dead-time bursts…

anyway, im driving myself nuts trying to figure this one out, hopefully the answer will show itself soon.

PS – i briefly looked thru the blog posts, nothing jumped out at me :(