Cassette death by take-up torque

[Karen O]

I had a long conversation recently, with a friend who still uses cassettes extensively. I mentioned a recent topic on this forum - excessive take-up torque, and the damage this can do to a tape.

I've experienced this myself - a player that played too fast and too unevenly. I concluded that the take-up spool was pulling the tape past the capstan. I proved this to myself by unwinding some slack. For a moment at least it ran at the right speed.

The fix is to change that fibre disk in the clutch, which I did and restored the player to normal operation (a good clean of the capstan/roller helped too!) But how do I know that the torque is correct? I can stop the take-up hub fairly easily with my fingers - a good sign but hardly precise.

This issue concerns my friend so much that he's contemplating modifying a transport such that take-up torque is provided by a DC motor instead of a clutch that slowly fails. He can then adjust the torque by controlling the current through the motor. This strikes me as a good idea and one I might try myself.

[Panrock]

When I was at boarding school 50 years ago, one of my pipe dreams was to be able to record and view television programmes on my radio/cassette recorder.

I should point out that Karen, as a member of the NBTVA, has since achieved this using a standard audio cassette at normal speed, and the results are highly watchable!

This goes to prove any help we can give her will be in a good cause.

Steve

[stevehertz]

Wow?! How's that done (basically)?

[julie_m]

Basically is right!

An audio cassette has maybe 15 kHz of bandwidth. That means in one second, you can fit 15000 pixels. At 16.67 frames per second, which is close to the old hand-cranked ciné film rate, that gives 900 pixels per frame; call it 30 lines of 30 pixels each. (You could double that, if you used both stereo tracks and fiendish multiplexing; but then you would get only silent video.)

Mechanical scanning has the advantage of no flyback time, so all the bandwidth is usable.

[EMI BTR 3]

I've long thought this was also why tapes used in car stereos jammed themselves eventually. With the tape being wound under excessive tension, over time the tape very gradually stretches out of shape and no longer winds 'true'. Individual turns of the tape bunch and bulge outward causing the binding that leads to jamming (this is ignoring for the moment how much influence on the potential for jamming is due to a clean or gummed up/sticky pinch roller).

I've sometimes wondered if tapes longer than C-90 would have been more reliable if cassette machines would have been equipped with a means of reducing the takeup torque when using the longer, thinner tapes.

[ben]

The quality of the drive definitely helps. The worst are those Tanashin mechs, (commonly found in portable/radiocassettes/boomboxes or mini-systems and many cheaper car decks) which almost always have far too much TU torque. And no brakes. In addition, they have a plastic 'finger' rubbing against the tape which serves as a crude auto-stop. This will give 'railway lines' along the tape after only a few plays. The absence of any backtension also means that the tape is far more likely to get skewed.

The best decks are those with 3 motors and proper servo control, yet you don't need to spend hundreds on a Revox. The Tandberg TCD series are excellent, no unreliable clutches there. The more basic models can still be picked up for reasonable money.

[TIMTAPE]

A lot of better cassette decks had a "two motor" drive with one for capstan and the other for the reel hubs. The reel drive usually had a pivoting rubber tyred pulley which engaged the relevent take up or supply hub - one or t'other. (many VHS machines carried the same arrangement). The weakness in some was the built in felt type clutch which after a while either increased or decreased its torque. Sometimes take up torque in play/record got so weak that tape spilled. Some of the better models also reduced the motor's torque for play/record and motor torque was adjustable. Toshiba made some Aurex models like that.

With simple felt clutch type take up arrangements, I found the increased torque could often be returned to about normal by carefully dismantling the clutch and cleaning the felt and the surface it ran against. Often the clutch material is glued to a brass disc which is pressed onto a hardened steel shaft. Some of the better models carried a service adjustment to alter the pressure of the clutch disc.

Of course the torque is greatest at the start of a cassette "side" and slowly decreases to a minimum at the end of the "side", so the most likely stretching of tape is at the start and the most likely failure to take up reliably at the end.

Having a torque gauge really helped in normal servicing. Here is a picture of a standard torque gauge, as well as one I made up out of scrap parts including a piece of cassette leader tape and a Biro spring. The "home made" one was actually more rugged than the professional one because there was no fragile tape for a faulty machine to chew which was always a risk when testing a machine with torque that turned out to be far too high! I used it a lot when servicing high speed cassette duplicators which had the happy knack of destroying normal test cassettes in an instant!

[stevehertz]

How did you calibrate your home made one? Using another (linear) tension gauge?

[Karen O]

I think the world is ready for the first cassette vacuum column drive!

Wouldn't that look cool!

Slightly off topic: Almost all of the trouble with recording NBTV to cassette is the coupling capacitors in the deck. They cause horrendous phase shift - to the extent that everything below 50Hz (on my decks at least) is so far shifted that there would be less error if those components weren't there at all.

I managed to undo almost all of this low frequency phase nonsense by using an inverse all pass filter.

[TIMTAPE]

Originally Posted by stevehertz

How did you calibrate your home made one? Using another (linear) tension gauge?

It was some time ago Steve but I guess I calibrated it to the commercially made one shown. But it was only ever approximate as it measured tension with the hub stalled whereas the commercial one measured running at actual speed, which of course varies with tape pack diameter. Even so for a given machine I got to know what was the "normal" range. A problem with the standard commercial tool was that on a high speed tape duplicator the dial spun around so fast it was impossible to read it...

['LIVEWIRE?']

I still have one of the commercial torque gauges shown on picrure 1 of post#7. It must be nearly 30 years old by now, and, despite having been used in a wide variety of cassette machines, mostly car stereos in the early days, it's still useable, though if a new one were available at reasonable cost I might buy it. In many palyers & recorders the take-up torque is adjustable, although, as TimTape says, wear of the felt clutch pads has always been a problem. The normal takeup torque range in machines of which I have had experience is between 30 & 50p., measured with one of those gauges at normal running speed.

[TIMTAPE]

In my post #7 I made an error. On the 2 motor drive system there is usually a tiny felt clutch but it doesnt actually control the drive torque. Instead it is the force that initially moves the idler onto either reel table. If the clutch weakens, there may not be enough force to do this and there can be weak or no drive, usually worse on take up in play/record when the drive to the reel motor is weaker.

Nakamichi decks used a good example of this in their "classic" dual capstan decks. The clutch was generously sized and rarely failed in my experience. Unfortunately they put out some models with a revised smaller clutch/idler and that gave a lot more problems.

Ben's mentioned 3 motor cassette decks with separate reel table drive motors from Revox and Tandberg.
Akai also put out some such models in the 70's. I have a GXC-760D. The wind speed in rewind and fast forward is controlled by the speed of the non driven motor on the opposite side. The voltage is sensed and as this reel's tape pack decreases, the increased voltage reduces the speed of the driving motor. Only trouble is it's not completely stable unless tape tension is maintained. I feel the Tandberg system with true servo controlled reel motors via optical tachos was a more reliable design.

Because a cassette tape isnt normally withdrawn from the shell it's hard to design a proper servo tape tension sensor system for constant tension, so it has to be constant torque at best, but Tascam did design a servo back tension system which sensed or inferred the tape speed and regulated the back tension by (I think) a magnetic clutch system. They only used it on their expensive 122 3 head machines where back tension is more critical to reliable tape to head contact.

[McMurdo]

In the paper and textile industries, reel take-up is a very common problem, albeit on a bigger scale. DC Motors are the preferred option, at least on the less modern stuff. It's quite an art. The current is controlled/limited for torque control as you'd expect. It's not constant however but linked to a speed sensor on the take up motor. That way the circuitry can estimate both the torque and the diameter of the spool. As the spool fills, the speed decreases of course, but the current must increase to compensate for the 'moment' as the reel fills. As the reel approaches full capacity the winding tension must slowly decrease to prevent a layering/lapping error where the last few turns develop a tendency to go on the spool in an irregular way (also called telescoping).

The DC motor method overcomes the common problem of cassette clutches which is that they are very taught when the spool is empty but can slow and stop when the reel starts getting full.

[TIMTAPE]

Interesting comments McMurdo, thanks.

[Karen O]

I had a feeling that all these issues had been addressed before

It's amazing to think that something that was intended for dictation machines slowly morphed into something with such high performance.

There is software available now (at a price ) which can correct the wow and flutter e.g. as caused by stretched tapes. I've often wondered whether the bias frequency could be detected and used as a kind of audio clock. But that's another topic!

Thanks to all who replied to this thread. Much to think about!

[TIMTAPE]

There is software available now (at a price ) which can correct the wow and flutter e.g. as caused by stretched tapes. I've often wondered whether the bias frequency could be detected and used as a kind of audio clock. But that's another topic!

Yes for some time there's been a system to recover the bias signal and use it for time base corrections. The commercial name is Plagent Process. Unfortunately it normally would only work on higher tape speed reel to reel recordings. Normally on a cassette the bias signal is not retrievable.

The other more recent tool is " Celemony Capstan" which works by comparing the pitch deviations in different recorded musical tones/ instruments and only when those variations are the same does it assume it's wow or flutter. It cant work from speech unfortunately, but Plagent could of course.

Normally the cause is not stretched tapes though but a tape speed stability problem when recorded. I've heard people say "I threw out my old recordings as all the tapes had stretched". Probably they hadn't stretched but for whatever reason were not holding speed when played back years later.

[FrankB]

IIRC, ECG made the cassette torque test tape I used.
Not cheap at the time though, like +- $32.00 USD back in the 70's.
The one I used told the speed with a built in strobe disk, and the fwd & rewind torque.

You might try joining: http://antiqueradios.com/forums/index.php and place a "Wanted" ad there.
Its the US group I belong to, and for the most part, are good folks.

[ricard]

Originally Posted by TIMTAPE

Yes for some time there's been a system to recover the bias signal and use it for time base corrections. The commercial name is Plagent Process. Unfortunately it normally would only work on higher tape speed reel to reel recordings. Normally on a cassette the bias signal is not retrievable.

How does that actually work? I thought the bias signal essentially works as on an erase head but the zero level being offset due to the LF signal, thus actually leaving the LF signal on tape, but without any remnant of the bias signal itself.

[TIMTAPE]

I've never attempted to recover or even view the bias signal but apparently Plagent Process works. I think it's complicated by the fact that the bias frequency is well above the highest frequency recorded, and recovering it may require a special repro head more suited to such high frequencies. In pro machines the bias frequency might be 120kHz or more and that might only be recordable/resolvable at tape speeds of 15ips or 30ips minimum where the wavelength is within the abilities of the tape stock.
Werent ordinary video recordings basically the recording of a very high frequency bias signal and frequency modulating it?

[Ted Kendall]

Analogue video recording was FM, so the question of bias in that sense didn't arise except for the audio/control tracks.

Plangent works, if you have the budget for it and can work with a bureau system. I use Cedar Respeed to good effect, according to the reviews. It doesn't need a clock signal to work - it's one clever algorithm which I don't pretend to understand, but it works. No experience of Capstan, but judging by some ARSCLIST posts, it's a temperamental beast more akin to Autotune. You pays your money...