Digital fix for wow and flutter

...which is to say, this probably can't begin to work on cassettes that lack either a bias-frequency recording or sprocket holes (16/35mm mag).  I take it to mean that without an externally applicable time reference, the Plangent approach is not possible.

It certainly won't be practical with cassettes, but not because they lack recorded bias - the bias is there, just as it is with a reel to reel recording. The problem is the amount you'd have to slow down a cassette, that's already only running at 1 7/8 ips, to reproduce it!

OK so a standard audio cassette doesn't have sprocket holes but surely it will have HF bias on it. It couldn't make a decent recording if it didn't.

I was just assuming that the bias frequency wouldn't be preserved because its wavelength on the tape would be too small.  If self-erasure is a problem at high audible frequencies, wouldn't the bias signal pretty much immolate itself?  How small ARE those little particles, anyway?  If I'm wrong, no doubt somebody could dream up a solution.  Didn't some of the Nakamichis have a 15/16ips second speed?

I was just assuming that the bias frequency wouldn't be preserved because its wavelength on the tape would be too small.  If self-erasure is a problem at high audible frequencies, wouldn't the bias signal pretty much immolate itself?  How small ARE those little particles, anyway?  If I'm wrong, no doubt somebody could dream up a solution.  Didn't some of the Nakamichis have a 15/16ips second speed?

For bias to work at all, it has to be able to affect the magnetic particles on the tape. I don't think that self-erasure is much more of a problem at slower speeds than at higher ones - that's not really the mechanism. As to how small the particles are - well on good tapes they are about 100 nanometers, but on specialist tapes they can go down to about 20 nanometers. That means that under the head gap of., let's say, 1micron (which is pushing it!) at any one time, there could easily be in excess of 10 particles - more than enough to record bias frequencies. Chances are with most cassette heads, there would be rather more particles, I suspect.

With cassette formulations they deliberately developed materials that have higher coercivity, as this would normally fall with shrinking particle size. The driving force for all of this was not actually audio cassettes, but DAT computer backup tape systems. The technology isn't exactly dead now either - it's the same principles at work in a hard disk, after all. But linearity doesn't matter there!

Anyway, the higher the coercivity, the more bias you need to get the modulation into the relatively linear range of the B-H curve, and to fit in with what was being manufactured by machine manufacturers, there were distinct limits to what you could do with tape formulations. So-called 'metal' tape performed best - in fact some of it was remarkably good. Getting a relatively flat response out to 20kHz was routinely possible on Naks, after all.

Even playing a tape back at 15/16 ips you've only halved the speed. Normally to get bias to play and be reproduced, you have to use 1/4-speed playback, and even then you can only see it on a scope - if you're lucky and the replay head is in good condition. The problem with this is that getting a tape mechanism to remain even vaguely speed stable at these speeds is quite a tall order, simply because of the friction losses and the reduction in the effectiveness of the flywheel. To get the flywheel as stable as it would be at 4 times the speed would, I think, require 4 times the mass... ain't gonna happen!

thanks, SteveG
...still a beginner to DSP, and have much to learn

Great stuff, thanks for making some sense of that, Steve.  So if the low-speed option is unattractive, what would it take to build a head and playback circuit flat to 100Khz (for the sake of speculation)?

So if the low-speed option is unattractive, what would it take to build a head and playback circuit flat to 100Khz (for the sake of speculation)?

Fortunately it wouldn't need to have a flat response - you only need the frequency variations in the bias signal to be detected. The problem is quite considerable though. It's been found to be true over a long period that trying to get much more than 9 octaves of linear amplitude response from a tape head simply isn't a realistic proposition. So if, for instance, you needed a response out to 100kHz, the LF response would drop away dramatically below 200Hz, so bye bye all the bass...

This was the problem that faced early video recorders, of course. The technical problem of recording a wide bandwidth was resolved by not making linear frequency recordings at all, but by using FM modulation techniques, which trade a wide frequency response against worse noise performance. And in order to record video even with FM, the required video frequencies required the writing speed to be much faster, which is why we have really fragile spinning ferrite heads on a head drum recording relatively long thin tracks very fast compared to the linear tape speed. The head gaps require a reasonably powerful microscope to see them!

Anyway, I'm pretty firmly convinced that until or unless somebody comes up with a really novel way around all of these little difficulties caused by the Laws of Physics, that the tricks you could do with a faster original tape speed and bias are not going to be possible with slow cassette recordings.