Hacker News Reader

Comment by abdullahkhalids | Hacker News Reader

The vinyl groove is about 25-100 micrometer, and feature sizes on it go down to about 22.5 micrometer [1]. Lasers are extremely cheap today compared to what they were back in the day, and doing laser interferometry to detect features this small is not that difficult. I imagine a hobbyist with the right skill set could hack together a reader for a few hundred bucks at most.

[1] http://www.gzvinyl.com/About-vinyl.aspx

View on HN

Not sure about hobbyists, and the price point.

Lasers are cheap, tracking for them is less so. To accurately read these grooves with sufficient resolution, you need 2 lasers (LPs are stereo), both looking at 45 degree inclination from different angles, and focused at the same point. That kind of accurate spatial + temporal sync ain’t trivial to accomplish. CDs/DVDs are much easier to implement. They don’t need similar resolution because digital, and they light their laser along the normal, only need to track 2 axes.

Even if you’ll manage to accurately focus both lasers, another problem is interference between left and right channel lasers.

> Even if you’ll manage to accurately focus both lasers, another problem is interference between left and right channel lasers.

Would aiming the lasers at slightly different spots and putting the channels back together in software work?

Would polarization angle have enough signal to be useful ?

Probably yes but it’s not trivial. If lasers are 10mm apart from each other, a radial track curves by 3.8° between them at outer radius, by 9.5° at inner radius. Probably too large difference to neglect. The sensors probably need to rotate around vertical axis too, to compensate for decreasing track radius. Technically doable, just all these precise mechanical parts add cost, and unlike electronics or software they eventually fail due to mechanical wear.

The technical problem is already solved, but the product costs $15k: https://diffuser.fm/laser-turntable/ As you see from the description, they did it with 5 lasers instead of 2.

Yep, modern optical computer nice have sensors north of 10000 dpi, so ostensibly a hacked up mouse/scanner could read all the data with plenty of room for error.

A colleague was doing this in a decade ago, I wonder if he ever got it working.

I'm not sure I can find the specific project I'm thinking of now (possibly [1]), but I recall that ~10-15 years ago someone scanned vinyl records with a consumer or "small business" grade scanner well enough that the scan could be played back with enough fidelity to be recognizable. Certainly far short of "all the data", but it's a data point for what can be accomplished with mass-market hardware.

[1] https://www.cse.huji.ac.il/~springer/DigitalNeedle/index.htm...

As I read that summary, feature sizes start at 22.5 µm and go down to 75 nm.

As they say, "significantly lower than the wavelength of visible light".

Why does a laser have to work in the visible light spectrum? Lithography lasers have been drawing at 175nm for over a decade.

And how much of that 75nm feature is getting picked up by a diamond stylus? The stylii used in metrology grade surface roughness and contour analyzers can't pick up features even remotely close to that size and those are far finer and more sensitive that a turntables stylus.

Reply on news.ycombinator.com