Hacker News Reader

Comment by bsder | Hacker News Reader

Is there way to read magnetic tapes like these in such a way as to get the raw magnetic flux at high resolution?

It seems like it would be easier to process old magnetic tapes by imaging them and then applying signal processing rather than finding working tape drives with functioning rollers. Most of the time, you're not worried about tape speed since you're just doing recovery read rather than read/write operations. So, a slow but accurate operation seems like it would be a boon for these kinds of things.

View on HN

For anybody who is into this this is a a good excuse to share a presentation from Vintage Computer Fest West 2020 re: magnetic tape restoration: https://www.youtube.com/watch?v=sKvwjYwvN2U

The presentation explores using software-defined signal processing analyze a digitized version of the analog signal generated from the flux transitions. It's basically moving the digital portion of the tape drive into software (a lot like software-defined radio). This is also very similar to efforts in floppy disk preservation. Floppies are amazingly like tape drives, just with tiny circular tapes.

OP here! Yes I'd highly recommend this video, I stumbled across it early on when trying to familiarize myself with what the options were-- and it's a good video!

At the very least, and the cost for this perhaps would be prohibitive, but some mechanism to duplicate the raw flux off the tape onto another tape in an identical format, a backup of the backup. This would allow for attempts to read the data that may be potentially destructive to the media (for example, breaking the tape accidentally) and not lose the original signal.

Sounds like at least in this case that ASIC in the drive was doing some (non trivial) signal processing. Would be interesting to know how hard it would be to get from the flux pattern back to zeros and ones. I guess with a working drive you can at least write as many test patterns as you want until you maybe figure it out.

At the very least the drive needs to be able to lock onto the signal. It's probably encoded in a helix on the drive and if the head isn't synchronized properly you won't get anything useful, even with a high sampling rate.

I would be surprised if it used helical recording. Data tape recorders rarely do because it's much more complex, increases tape wear, and the use cases don't usually demand that kind of linear bandwidth.

Wasn't the whole selling point of these drives that they were able to encode more on the tape than their competitors? I figure that must include some clever tricks.

The data format is documented at a high level here:

https://github.com/Kneesnap/onstream-data-recovery/blob/main...

It's a linear, 192-track format that's read 8 tracks at a time. The other files in that repo are also worth reading.

You still need to know where to look, the format, and using specialized equipment which cost wasn't driven down by mass manufacturing, so, in theory yes, in practice not.

(Completely guessing here with absolute no knowledge of the real state of things)

Yes. There’s some guy on YouTube who does stuff like that (he reverse engineered the audio recordings from a 747 tape array) but it can be quite complicated.

Would you have a link by any chance? Thanks!

https://youtu.be/MU02pQe3E5Q Is the one I’m thinking of, digital would require more work

Reply on news.ycombinator.com