Magnetophon Operation page 2

An extension of the previous page Magnetophon operation


Our previous Magnetophon-operation page was becoming too much loaded.


Page initiated: 13 April 2015

Status: 13-18 May 2015

Page modification: 2a + 3b


Jaap Keijzer phoned me this morning (13th), telling me that he did fix the various problem concerning the capstan-motor (Tonmotor) of our Magnetophon apparatus. And that he is ready for additional tests.

I could not wait for the results and asked him whether he could come over that very day?



The capstan motor being re-mounted and its behaviour investigated


For it we did two different tests. First recording from our CD-player a 'piano concerto' as a piano is a good instrument testing tape-speed-inconstancies.

The results were poor, as the previous nuisances were more or less equal. One aspect was clearly solved and that was the one caused by the two motor bearings. Which had been replaced by two self-lining sintered-bronze-plain-bearings.

The kind of rumble proved to be a function of the voltage supplied onto motor. Jaap Keijzer discovered that the motor runs best at about 110 V instead of 220 where a severe hum was mechanically caused. It was found that in some way the motor responded best at a mains voltage of 250 V +, but becoming soon too warm.

After our experiments were finished I did send Manfred Firnhaber a brief report of what problems we had encountered this day.

He replied that very day, with a long e-mail on various aspects of capstan motors in the old days.

Die Ströme in der Hauptwicklung und in der Hilfswicklung müssen symmetrisch sein. AEG/Telefunken hat das mit zwei gleichen sehr niederohmigen Strommessinstrumenten durchgeführt und dabei mit einem passenden Widerstand und dem Phasenschieberkondensator in der Hilfsphase die Ströme symmetriert.
Die Spannungen in der Haupt- und in der Hilfsphase können dabei unterschiedlich gewesen sein. Aus dem Kopf würde ich sagen ca. 220 V in der Hauptphase und ca. 160 V in der Hilfsphase. Ich bin aber in Mecklenburg und kann nicht in meinen Unterlagen aus dem Braunbuch nachsehen.

This e-mail paragraph might give an opening how we can approach our current problems. Whether it can be solved this way stays open, but at least we should commence experiments into this direction.


Let us rely on audio/visual recordings made available on our YouTube channel.

I have to apologise myself for not always using the right (correct) technical terms, as these may differ from a straight one-to-one translation between Dutch - German and English.


Film 176:    Listening to our experimental recording, and encountering quite some 'wow and flutter'. Although, Jaap Keijzer did an excellent job in getting the rotor and both shafts in-line again; hence eliminating the failure being caused by bearings. The 'variac' on the floor in front is meant for supplying different voltages onto the capstan motor. It was found, that 110 V does work favourable in respect to 220 V mains.  

Film 177:    Viewing the performance of the capstan motor (Tonmotor) again, made visible by painting a first order Lissajous. Bearings running fine now, and all the problems most likely being caused by disturbances (interactions) between the rotor and the 'slip' between the rotor-speed and the rotating magnetic field. We neglect the residue of the HF bias (on the CRT screen) which should be filtered later.  


Quite some experiments laying ahead.  



On 21 April 2015


Yesterday, Monday 20th, Jaap Keijzer came along again.



Jaap is today demonstrating his new discoveries

For it we record, on our Magnetophon T8, a piano concerto again, because a piano is a delicate instrument and small variation of speed being quasi audible magnified.



The fuzzy meter pointer is indicating the hunting effect of the capstan motor, and will be dealt with in the following contribution


His first proposal, a few days ago, was damping the sometimes hunting phenomenon of the flee-wheel, by a mechanical means.

After severe considerations with a friend, they measured the rotor section by means of a coil (from a Hammond organ). An oscilloscope was showing clearly the oscillating effects of 'hunting'.


They finally came up with a theory, which proved to be quite valid.

They also investigated the modifications once accomplished; likely in Belguim, because they used insulations originating after, say, the late 1960s.

They did very very strange things!

He and his friend also discovered that the hunting effect can be nullified by counter shocking the hunting irregularities of the rotating system. What will be shown, is that by counter moving the motor-body against the motor housing the rotating system, can be brought to locking upon the typical rotor slip.

Hence, the rotor slip is getting a quasi 'low impedance'.

Electrical asynchronous motors can only operate with a particular amount of rotor slip versus the theoretical rotation speed caused by the supplied mains frequency and the number of 'motor poles' involved (coil pairs).

I once calculated, that, when we consider the diameter of the capstan shaft versus the theoretical rotation of 750 rpm (four coil-poles, 3000 : 4) the capstan is having an actual speed of 742 rpm. This difference being the so-called slipping factor; normally given in a percentage. My calculation might be inaccurate, because I considered that the tape speed being 38 cm/s. Whatever it actually is, there exists a 'slip'. An asynchronous motor otherwise cannot rotate by its own means. This so-called 'slip' is, in some way or another, a critical factor. This critical aspect is causing some form of motor-speed hunting. This rather low frequency (I guess, 1 -2 Hz) hunting aspect can be overcome by locking the rotor rotation onto the motor-specific slip percentage.

This 'locking upon' phenomenon will be demonstrated in two ways. By increasing the driving motor energy, in casu, supply voltage, and by speeding up the rotor rotation versus the rotating field of the capstan motor (or inverse); by means of shock-wise rotating the motor housing against the rotating rotor (whilst the 'hunting rotor' is running).

All the time Jaap is measuring the voltage across the 'starting up capacitor'; hunting is being made visible by means of a moving coil meter pointer. When locking occurs, the meter instantly is showing this by means of a typical 'discriminator' like tuning behaviour. Typical for locking is the phenomenon, that when locking is apparent the motor driving voltage can be varied over a wide range from, say, 100 V up to any other voltage, without getting out-off-lock again. Hence, the motor rotation stays constant. Actually, determined by the mains frequency minus the rotor-slip percentage (750 - Δ).

Repeating our hypothesis: Locking can be accomplished by rotating the motor body against the running rotor, or by means of the full motor driving voltage. This might mean, that the motor rotation speed being too low, so that locking never can occur. Feeding the motor with maximal power the rotor can reach the critical 'slip speed' and locking does occur. The other way around, rotating the motor-body against the rotating rotor in such a way that virtually the rotor speeds up against the (static) rotating magnetic field, as if the rotor runs faster (in the magnetic field) then it actually did; even when this pulse like movement is lasting for a very short interval only. From my former experience with the Tonschreiber b, I know, that when the rotor speed runs from a higher speed to a lower speed, crossing the critical 'slip speed', that locking will occur. Maybe the system inertia is helping us. 

Today I phoned Jaap again, and we discussed my visions, we both concluded that it is not yet certain, whether the rotor speed is above or just below the critical capture slip. He also pointed, that he rotates the motor body clockwise as is running the rotor.


We never expected that the slip of an asynchronous motor is causing quite low frequency hunting. And that when it finally gets into a state of locking, that the driving motor voltage can be varied over quite a wide range, without getting out of lock again. 


Jaap also discovered, that there hardly is any cooling airstream inside the capstan motor existing. There even is hardly space for an air flow possible.

Also the way the top fan being constructed is curious.

Normally, a fan is having some shape as to lead (forcing) air into a certain direction. The top fan has only radial Al-cast strips. The same, maybe even more critical is the way the lower flee wheel being constructed, having four rectangular teeth. There is only a few mm space left at the bottom end of the motor section. Though, no airflow possible as there does not exist a means of 'canalling' air into the lower section of the motor.

I remember that the previous owner told me once, that cooling had been a problem. Likely owing to this nuisance, they had to rewind the coil bobbins. Quite understandable, when there is no sufficient cooling airflow possible.


YouTube films


Film 178:    Jaap does currently test the motor. Reading off the voltage across starting-up capacitor. The hunting phenomenon clearly visible on the moving coil instrument. Interesting is, that when the rotor speed locks upon the necessary (critical) asynchronous rotor speed, that the supplying voltage can be varied over a rather wide voltage range.

Film 179:    First Jaap Keijzer explains what he is doing in Dutch language. Thereafter, he shows that locking upon the typical rotor slip versus the theoretical field rotation speed (in our case 750 rpm versus the critical slip rotation of, say, 742 rpm).

Film 180:    Jaap is demonstrating what principally is wrong in the current state of our capstan motor. The straight radial motor-fan-blades hardly will capture air and therefore cannot be used for internal motor cooling. 

Film 181:    Jaap is demounting the lower side flee wheel. Again, the four slits hardly will move an airflow (lacking the necessary shape). The three coaxial bottom-slits for airflow hardly can be effective, as the flee wheel is allowing only two or three millimetre space, and there is no means of generating an airflow whatsoever.

Film 182:    Viewing into the motor where the rotor being removed. Also the inadequate shape or construction of the top fan. Jaap doubt that the current state is genuine.


However, after today's experiments, he is convinced that he can get the capstan motor right again. Albeit, after some changes to the current state of affairs. Such as improving the airflow for cooling. Also by changing experimentally the bottom flee wheel. When possible also the top cooling fan.

During making the recent YouTube films, some rumbling sound is coming from the motor. We don't know yet the reason. It might originate from the lower bearing which also will be replaced by 'sintered-bronze-plain-bearings'.


Jaap is very excited, as he is going into the real phenomenon of electrical asynchronous motors.



On 13-18 May 2015

I proceeded with implementing the capstan motor (Tonmotor) which Bernd Fischer so kindly donated to us. It concerns a motor once used within an AEG Magnetophon type AW 1. This latter machine was meant for home use and was fit with a single AEG motor.

After having dismantled it from the metal section accompanying Al plate, it proved that the size equals the one - once used in the genuine T 8 Magnetophon machines. Even the screw-holes are being placed identically.


Why waiting any longer?



The AW 1 capstan motor mounted temporarily

The black tubes provide oiling; the top one is apparently here not being fed by means of gravity; and oiling should be commenced (in this current stage) by turning the recorder on its backside.



Viewing it from a different perspective

Maybe not directly visible, is that the shape of our AW 1 motor is similar to both genuine winding motors (Wickelmotor), albeit, the AW 1 motor being fit with a special oiling-facility, which is lacking elsewhere entirely. What I have removed, is the break-arm meant for stopping rotation of the capstan flee-wheel abruptly. Maybe once used for quick stopping, and I was told that it was used in professional machines as to facilitate a quick cutting stop.


It was soon mounted and being wired electrically. Although, wired according the AW 1 schematic, it ran anti-clockwise. After having inverted the wiring of the 'support coil' (?) (aanloopwikkeling) it runs now clockwise as does the previous capstan motor.

The next move, was checking what its response is doing.

Sadly, it does behave the same way as did the previous capstan motor, albeit, that this capstan motor does not show signs of 'hunting' and other nuisance.

What next?

Dick Zijlmans and myself did check the spindle deviation of the left-hand tape winding disk. It shows some amount of deviation, but can this result in a wobbling recurrence of 1 à 2 Hz? No, of course not, because the deviation should be then widely dependant upon the size of the tape loaded at the left-hand reel. When the tape bobbin being full, the wobbling recurrence should then be less than a Hz. Which apparently, is not the case.

There only remains a single trouble causing option - and that is the rubber-wheel-roller, which is pressing the tape against the capstan-shaft.   



The original rubber-roller-wheel


Let us remember what I have measured once, and did put on the web previously:   


 ... I, therefore, estimate, that nowhere in these values we find a dimension that correlates with 2 Hz deviations. Please notice, that my 2 Hz is my rough estimation, only derived from what the YouTube films do show us.

A second size was measured, the diameter of the rubber-roll which presses the tape against the capstan.

I measured: 45.51 mm multiplied by π (pi) this gives a perimeter of 14.29 cm.

Tape-speed being 38.1 cm/s. Calculating it all in mm we get 380 / 142.9 = 2.65 rounds per second (the 0.1 is skipped as its influence is negligible).

Which implies, that this figure has something in common to the tape-speed deviations. At least we should try to replace it, albeit, temporarily. It does not yet tell us whether we have found the origin of our technical problems.


I measured the diameter of the AW 1 capstan motor which is 9.71 mm or 0.971 cm. Let is now consider that the tape speed in conjunction to the the 38.1 cm/s standard.

We get a capstan shaft perimeter of: π d = 3.14 x 0.971 = 3.05 cm

Knowing that at 50 Hz such electrical asynchronous motor runs, theoretically, 750 rpm

we get 750 : 60 = 12.5 rotations/s

Hence, 12.5 x 3.05 = 38.125 cm/s

Standard it should be 38.1 cm/s. We may regard this value being within the 'noise of the system'.

This would implicate, that our motor does not possess an asynchronous motor-slip. However, when this should be a synchronous motor, then rotating manually we should notice that rotation of the capstan shaft should rotate with some bump-like movements; caused by the magnetic fields (various N-S poles) of the rotor versus the stator. This phenomenon is not encountered.


On Friday 15th, I approached the problem again and focussed my attention upon the rubber-wheel roller.


I measured, with some difficulties the deviations of the rubber-wheel-roller perimeter. The encountered difficulty was to mount the micro-meter stable at the tape deck? There are no holes in that particular deck-sector.  




Hand-writing by direct means of a computer mouse isn't easy. Therefore, this curve is only meant as to indicate that two different deviation peaks being measured within a rotation of 360°

The consequence of such behaviour, is, that the torque of the capstan shaft being modulated by the erratic load changes caused by the diameter deviations. On the other hand, my measurement provides only very limited information, as the micrometer-finger touches only at not yet a mm of the (circular) roller perimeter.


Another phenomenon that was encountered, unusual tape flutter; which is dealt with in the next two line drawings.



This simplified line drawing shows briefly the trajectory of the tape during recording- or in the play mode

Within the red sector the tape vibrates clearly; the exacerbated situation is shown below.



The tape-width being exacerbated as to emphasise on what happens


Afterwards, I detached the head-assembly and did view what might occur. I believe, that the spindle height does not fit, or the alignment against the other spindles should be readjusted. For it I would like to rely upon Jaap Keijzer's superior mechanical expertise, and have to have patience until he is back from his Hungary journey, about the end of this month.      

Reconsidering all, I did replace the current rubber-wheel-roller by another one which we once, most kindly, got from Bernd Fischer.

Its diameter differed about 2 mm, seemingly not much, but during brief experiments it proved being far too much difference. After running for some seconds, the tape speed apparently started to fluctuate. This could be countered by pressing the rubber-wheel-arm  against the capstan shaft.   Interesting is, that when the pressure put on the rubber-wheel-roller is virtually lacking, that the tape movement stops. Hence, the rewinding and the winding-up torque is in balance. Consequently, the capstan torque is only loaded by the resistive pressure of the rubber-wheel-roller.

What it also proved, is that the overall system performance increased quite much; albeit, the system is still performing imperfect.


For it I made two YouTube films.

Film 183:    Viewing another rubber-wheel-roller. Its circular perimeter is also imperfect though, less than the one originally with our T8 Magnetophon.

Film 184:    We are viewing our Lissajous-measurement setup. Seemingly the tape speed starts varying, but this proved not to be due to the capstan or rubber-wheel but it is being caused by the actual too small diameter of the rubber wheel. The mechanism functions such, that a curved disk is pushing an arm where upon the rubber wheel is mounted against the capstan shaft. There is, so far as I know, now particular arrangement with which the wheel or roller pressure can be adjusted. It does fit or not. A too big wheel diameter will influence the capstan torque. Please bear in mind, that the painted Lissajous is also showing residues of the HF erasing- and/or recorder-head bias.        

Very important, is, in my perception, is that the two ball-bearings should be new, having the least 'play' possible.


How can we optimise the performance of an existing rubber-roller-wheel?

Does some of you have a suggestion?

What technology should be approached?

What degree of flexibility should the rubber-roller perimeter have, versus the tape running between the rubber-surface and the capstan shaft?

How can we adapt modern materials?

How is it possible to get a 'sound flat surface' after having used machine-tools? 

Additionally, how can we adjust the arm-pressure-mechanism? To what we found, there is no such provision existing. Some say there is, but we cannot find one. The arm-swing and the accompanied roller pressure is determined by an eccentric disk; which allows only: a free arm and an arm at pressure. The pressure by its own right is determined by the diameter of the rubber-wheel-roller versus the mechanical dimension of the curved disk; nothing else in between. Hence, it is only an on or off mode. And, of course, by the accompanied diameter of the capstan shaft.


'Experts' please come forward, and send us an e-mail:

Please type-in what you read


To be continued in due course.  


By Arthur O. Bauer