Is there any need for additional error correction?

Hi all,

I am at the point of fundamental design choices for my 2nd gen mods. Is there any need need for more advanced error correction for backing up tapes that have minor drop-outs on normal players?

I am not sure what this mean or how it would affect it.
The ideal situation is an exact copy/dump of the original file. How would a dropout be correctable?


The Read Amplifier is outputting a time multiplexed signal where each of the tracks is only forwarded 1/10 of the time. Legitimate cost saving measure because the DEQ2 only needed one instead of 9 filter channels with ADC. Thankfully the TDA1317 read amplifier has mostly unused (at least in Philips consumer machines) 9 seperate outputs for the tracks that could be used without having to care about the details of the head. Being able to analyze all data at every point in time would greatly improve error correction. Multiple passes and/or error correction based on the higher layers would also be possible.

My implementation will have to be a compromise of:

  • Implementation effort
  • Licensing: Basing on MATLAB Simulink would by far be the easiest but apart from the Student license everything required would cost about as much as all DCC Museum expenses ever combined so all collaboration would be dead
  • Pin accessibility: Using just two pins that are transferred between boards with a connector is by far easier and less risky than getting 9 pins from an IC without even the possibility of using test pads which is common practice for various mods.

So what do you think, should I take better error correction into consideration or is it just a nice idea without much practical use?

@Jac What do you think?

He plans to rely on player error correction, so that’s that.

At the moment do you have any tape with droputs of any value at the museum?

Next time I will come across one, I will save it for you.

The error detection and error correction should work for any data that’s encoded on the tape, PASC or otherwise. But it doesn’t make it infallible. For music, a dropout is an “acceptable” failure, but if you want real data stored with high integrity, so that one little dropout won’t make your 250MB file unrestorable, you’re going to have to add extra redundancy.

This is a bit out of scope for me so I don’t want to invest too much time in this at this time, sorry. You may want to look at solutions such as QuickPar which were designed to recover large files (such as movies and CD-ROM images) even if parts of the file aren’t available. It basically generates multiple sets of parity information. The more parity files you have, the more redundancy you get. For example, I made digital grabs of my mini DV video tapes, and put them on DVD recordables. Each tape takes up about 3 discs. Then I used QuickPar to generate about 60% redundancy data. That takes up 2 extra extra discs but it means that even if two of the discs of one set become unreadable, I will still be able to recover the video files.

Obviously on a tape that has slow access times, you may want to come up with something more sophisticated: parity information and real data should be close together on the tape so you don’t waste too much time. But on the other hand, it’s more likely that disasters happen in the same area of tape so you don’t want integrity information TOO close to the data that it’s supposed to protect. Besides, if you want to make it possible to record a tape in multiple sessions, you’re also going to have to take that into consideration.

As I said, this is a little out of scope for me. I’m not that interested in recording data to DCC tape. The capacity and speed are too low to be practical, and there are plenty other (and better) solutions for that (think DAT and DDS, for example). I’m more interested in storing audio in other formats on DCC. Or maybe video, just for the heck of it. And for those, the built-in error correction of DCC will be just fine.

=== Jac

1 Like

It is, but the hardware has limits as stated. I believe I could build improved decoding with additional downsides.

I am asking i. e. @drdcc if the museum has any unique, important tapes that suffer dropouts and @Jorn the same if anything unique important to him is on the 5% of his tapes from the 90s with droputs. In short, I am unsure if it would be a useful feature or just a tech demo to help me with my design decision. To be clear, I want to start with non-standard data after everything standardized for DCC is finished, I am not sure if I will ever start with a different codec on DCC as the practical use is close to zero.

Exactly this is my goal, I plan to go one layer deeper than @Jac, because he plans to extract the audio from I2S and the SYSINFO and AUX from the respective bus with error correction done by the player. I plan to do this part myself to keep the frames intact and am interested in your input regarding the error correction.

1 Like

Talked to @drdcc: Will try to implement it especially for the first known DCC recordings ever as they are on tapes in analog shells. For the sake of convenience I may implement something on the connector basis, too.

1 Like

I have a tape with short drop-outs in a few spots in an area that occur only some occasionally on the same spots. This fits my theory, and I strongly assume that capturing all tracks in parallel could avoid that to 100%. Another addition or alternative could be dumping raw data in multiple passes and merging it later.

1 Like