Talk:NT (cassette)

Tracking mechanism

The article currently states,

NT stands for Non-Tracking, meaning the head moves at a shallower slope to that of the tracks on the tape, crossing several during each pass, albeit only reading partial data from each one. By making several passes it is possible to reassemble the complete data for each track, in memory. This considerably reduced the complexity and size of the head, and, therefore, the recorder.

The wording sounds dubious, perhaps a bit misleading. As worded, it implies Sony engineers deliberately made the machine to move the tape faster during playback, which, though not impossible, seems unlikely, as it would have required a ridiculously huge buffer memory, able to store many minutes of 32 kHz stereo audio data so that the sound could be played back in real time, slower than it was coming off the tape.

Helical scan recording systems typically use a separate linear control track to synchronize the rotary heads with the helical tracks; tape speed is precisely controlled with a capstan roller driven by a synchronous motor or servo motor. The NT system evidently lacked room for a separate linear recording head to lay down and read a control track, so the rotary heads couldn't be precisely synchronized with the previously recorded tracks. Another explanation of how it worked might be that during playback the heads in the drum traversed the tape at an angle identical to, or very close to the angle at which the tracks were recorded, with allowance for manufacturing variations from one recorder to another and tape expansion and contraction due to temperature and humidity. Since the recording is digital, most likely with embedded error-correcting code (ECC), two or three passes of a track on playback would yield sufficient information to reconstruct a complete, error-free track in memory before converting the data to an analog signal. The head might move at a shallower or steeper slope than the recorded track. This could be accomplished within a very moderate buffer memory size, just large enough to store the amount of data needed to mask the fraction of a second when the cassette reversed direction. The system also had to contend with decoding tracks recorded at two different angles, depending on which direction the tape was moving during recording. Having "forward" and "reverse" tracks on the same tape creates the potential for one set of tracks interfering with the other. Was this solved by using two different types of heads in the drum, with the gaps slanted or biased in opposite directions to minimize magnetic domain interference on the tape? This would have been a difficult problem, miniaturizing the mechanisms to that degree.

This is speculation on my part, based on having worked in the helical scan recording business years ago. Perhaps someone familiar with the technical details of the Sony NT system can eventually fill in the gaps in the article. — Quicksilver^{T @} 18:29, 20 November 2014 (UTC)

A paper describing the system is here: http://onlinelibrary.wiley.com/doi/10.1002/ecjb.4420760510/abstract . Definitely worth a read. Seems like a lot of your speculation is spot on. alex.forencich (talk) 01:16, 29 April 2017 (UTC)