Magnetophon restoration

attempt 5


This is my fifths approach getting our Magnetophon tape machine operating in a good order.


Page initiated on 16 February 2016

Status:  28 June 2016

Extension: A + 5B + 5C + 5D + 5E

New approach called: 6D + 6F + 6G + 6H + 6K


Previously we encountered, for what ever reasons, ongoing problems with various capstan motors.

This time I took an AW 1 motor meant for 76 cm/s tape speed. Not very convenient, as 1000 m tape allows only about 22 minutes recording time.

However, the aim this time is double edged, as on the one hand I would like to get momentum in the Magnetophon project again, as well as allowing Dick Zijlmans to start up with a new project. Being the attached audio preamps combined with genuine faders and Siemens 'Lichtzeiger' signal level-indicators.

My very first attempt didn't run the way I expected, but the nuisance was caused by a bad contact in the 'Hilfswicklung' circuit of the capstan motor circuit.


Starting up again restoration attempt number 5

My first move was fitting a quite new capstan roller, which in last August Günter Hütter kindly gave me. For it I need bearings because these are lacking.


A superb address is:

Lager means bearing and boer stands for farmer, thus bearing-farmer.

It was Jaap Keijzer who brought my attention onto this exceptional firm.

According Jaap, they buy bearings of all sorts from everywhere in the world. They offer very good bargains, and a wide range of bearing types of good brands on stock; like SKF and that quality like.


Preparations for fitting a new set bearings inside the capstan roller

The shown bearing are previously taken from another capstan roller, but of used quality.


The new capstan roller in more detail

Please notice the slightly increased diameter just there where the tape is running along. By this means the tape is soundly pressed onto the capstan and also reduces unnecessary roller/capstan frictions. My guess, once there existed a patent to it.


The capstan motor, before mounting

On top of the capstan shaft we notice a replicable shaft section.

There exist two types of AW 1 capstan motors. One for 38 cm/s and one type for 76 cm/s. From outside you cannot see the difference; or you must be acquainted to AEG motor serial-number successions. 


Viewing the painted first order Lissajous, which is the result of the interaction between a tone signal of about 1000 Hz and the output of the separate play-head

That the trace line is broader than usual, is being caused by the residue of the HF-recording-bias which signal is in the order of 40 kHz. The erasing bias is 80 kHz.

This residue must be filtered out first, as it otherwise can interfere with the tape audio-output.

I first adopted the same 'Anlauf C' configuration, being a capacitance of 1 µF in series with a resistor of 300 Ω (used for the 38 cm/s capstan motor). It proved however, that the capstan motor speed does 'hunt'. This stopped after the series resistor being shortened. All further experiments should be commenced this way; except further notice.

After running for the second session, I measured an outside motor temperature of 26.2° C. I suppose that this is a jolly good value.


YouTube Films

Film 0217     Viewing during restoration attempt 5 at the painted Lissajous figure, where we see the interaction between the sending signal of 1000 Hz towards the recording input versus the according play-head signal. Every irregularity will be noticed instantly. The rotation of the ellipsoid axis points into speed irregularities and the small measure of the ellipsoid indicating flutter or tape amplitude variations.

Film 0218    Continuing with the second approach today. The current tone frequency is about 1000.75 Hz. The broader than usual scope trace is due to residue of the HF recording bias, which is at 40 kHz. This signal component should be removed (filtered) before the future preamps being operated.   



On 27 February 2016


I have filmed what has been encountered some days ago.


For it I especially made a YouTube film


Film 219:    We are viewing to day what a few days ago was found. We discovered that a great deal of jitter can be eliminated when the left-hand side tape roller being hold. Then the T8 Magnetophon performs rather well. Albeit, that we operate currently at a tape speed of 76 cm/s. From foregoing experiments I learned already that it might be originating from the left-hand side winding motor; maybe a side effect of asynchonism.




On 25/29 March 2016


I would like to continue with working on our Magnetophon system again.

Our aim this time, is, to start with implementing line-signal amplifiers. For compactness and simplicity we have decided to implement standard printed circuits.


 The circuitry is quite basic

One may ask me why using this odd type module?

Simply, because Dick Zijlmans does have some boards on hand.

However, I have to modify the input circuit as to match it onto the output impedance of the W24a fader device.


This basic concept of what I would like to experiment with, first

Profil-Regler W 24a is an historic German sound-fader. Already designed and operated before the war on behalf of the RRG (Reichsrundfunkgesellschaft)


Its schematic, copied from a Braun Buch page, the 'holy grail' for German audiophiles

(courtesy Mr. Gerd Kuper)

The purpose is: that whatever the fader setting - the in- and output impedance remains more or less constant; in German practice being 200 Ω.


Also derived from the Braun Buch (courtesy Mr. Gerd Kuper)


These devices are of rather complex design

(shown is our device, which we with kind support of Bernd Fischer could obtain. (Screening-cover taken off)

All small bobbins being bifilarly wound with resistor-wire each having a particular value.

Bifilarly winding was accomplished - as to cancel the otherwise inductive-coil-components, which aspect would make such device highly frequency depending.




Our current experimental PCB; providing two separate amplifier channels

It regularly constituting a 1 x amplification; however, provision is taken to increase amplification, by means of changing two resistor values.


Before continuing I would like to know whether the PCB functions well, which was the case.


Today Wouter Elzinga assisted me with holding matters - or was directly supporting me otherwise

Everybody knows, that in these circumstances one need quite often more than two hands!


It proved, however, that when this Profil-Regler W24a being interconnected  with the experimental line-amplifier that it responded faulty.

It was discovered, after removing the screening case, that a resistor wire hung 'in the air'. But onto what point should this particular wire be interconnect with?

Of great help was that we, luckily, possess three W24a type faders. 


I do not have to explain the problems one is facing when somewhere in the centre a single thin wire 'hangs open in the air'

(Photo courtesy Gerd Kuper)

After the problems have been cured, the Profil-Regler type W24a was implemented into the amplification chain again.


As to make life easy, I loaded the amplifier output with a 1 kΩ resistor and observed at the same time whether this will influence the overall signal output; instead of using a second (output) line transformer. It operated soundly. Albeit, that it was noticed that the sliding-fader-contacts need to be moved several times before responding smoothly. Let us image that these devices haven't been operated for more than, at least, half a century!


However, the moment came that the amplifier being supplied with 24 V as well as supplied from 'tone-generator'. A sound sine wave signal appeared on the oscilloscope screen

In the background we notice another type W24a fader.

After some experimental settings the next move could be undertaken.

First looking whether the two Siemens Pegel-Zeiger instruments do work appropriately; these I got last summer from Günter Hütter as a present.

One unit wasn't visually responding; the other one with lacking some screws, did work instantly.

Please notice. That these types of moving-coil systems do not indicate anything as long as not a minimal current being supplied (the same happens when the instrument current surpasses a certain limit) (the light image still being projected - but outside the visible db scale). The according 'Aussteuer-Verstärker' unit is not a simple signal rectifier and meter driver, but a quite ingenious circuit. Responding logarithmic and signal levels upward responding virtually without delay, but falling backwards goes slowly. By this means it becomes possible to adjust at quickly changing sound-effects. There cannot be noticed any overshoot effects, so often a nuisance when operating fast moving moving-coil-systems.

My next move was first operating our more modern 'Lichtzeiger' instrument; we possess already since January 2015. 


Siemens & Halske (S&H) Type Rel Bv 663K3010. Siemens Serial number F: Nr. 5/504579


Such device can only be operated in conjunction, for example, with an 'Austeurverstäker' type U 70 or U 70a

The difference between them, is - that the first type uses valves and the latter type, in contrast, is fully transistorised.

However, the mechanical sizes, as well as their connector layout, are equal. Luckily, because it proved impossible to find the schematic of the U 70a module. I therefore could rely upon the 1950s schematic of type U 70.



Three photos have been taken as to show the interaction between W24a fader setting and the indication on the 'Lichtzeiger' instrument (sound level meter calibrated in db (dB)

What is being indicated is not yet calibrated or adjusted correctly. Today my main aim is to show the particular fader settings providing an according 'Lichtzeiger' deflection.


 The W42a fader being set at another (lower) sound level

Please notice that the early types W24a being calibrated in Neper - a logarithmic scale based upon the (mathematical) natural figure e; later every body adopted the db scale based on the figure 10. Also adopted by the Germans somewhere during the war.


Example three


To understand the interaction between both Neper and db's - notice the next comparing scale.


It is clear that 1 Np equals a signal strength difference (ratio) of 8.68 dB



Finally, showing how today our experimental setup was accomplished

A small note. Why is the unit type U 70a standing so typically on the table? The reason is quite pragmatic; because its housing is having a special asymmetric shape. I assume, that this was done as to prevent that units not fitting to the system can be plugged-in.



On 29 March 2016

Today we will succeed with the measurement results undertaken on 27 March.

My aim is plotting the output levels versus frequency and settings of the W24a Profil-Regler also known generally as (sound) fader.


Shown is in front an experimental amplifier print and right of it the vintage W24a Profil-Regler (fader). On the table, in the background, an analogue meter provided with a convenient db scale

What also was measured was the frequency response of the Lichtzeiger instrument in the background.


My first measurement today starts with measuring the overall frequency response whilst the W24a fader setting is - 1.4 Neper (≈ - 12 db). This value was chosen as a good equilibrium versus meters readings.

Please notice hereafter also that the amplitude scales, but a bit less for the frequency scales, do not represent pure logarithmic nor linear scaling. But all being more accomplished by a practical point of view. Also originating from the fact, that I currently do not possess a sound logarithmic graphic table.


The next series is using as indication instrument the Siemens Lichtzeiger type Rel. Bv. 663

For this occasion we leave the fader adjusted at about - 1,4 Neper but reading virtually* off now 0 db from the Lichtzeiger (light-slit meter)

* Please notice the considerations below


We have adjusted the overall signal driving level, originating from our 'tone-generator', such - that the Lichtzeiger light-slit (=Zeiger) is set on the Rel. B.663 meter scale at just - 5 db. This being considered our current 0 db reference, and the drawn values above do refer to a value differing from this virtual 0 db level


Because, I did then not yet know whether levels could reach higher than the current real zero db level. The Bv 663, but also comparable types, do entre then within a read marked sector, where reading off is a bit delicate. This is the great advantage of using db references!

What we may confront with later, is that the indication at the right-hand of the Lichtzeiger is reliable and maybe more or less correct, but at the left-hand side of the scale, thus at low signal levels, the read-off accuracy is not correct and reliable. Should be particularly studied in due course.

This might be originating from within the U70a Aussteuerverstärker.

However, I have opened it once in early 2015 and became quite afraid of the very many potentiometers (trimmers). Without a solid manual - I dare to touch anything! I therefore closed the box - and was very much relieved that it still responds!

I assume, that we should look out for the more simple version type U70. 


 The W24a fader adjusted at 0 Neper, thus at a straight forward signal transfer

Please notice again, that the used scaling is not logarithmic nor analogue fully; but in my perception provides, nevertheless, a quite good impression.

What might be noticed:- that a matching error or other problem prevails above the rest of the circuitry.

The next graph shows a different fader setting.


W24a fader being adjusted at - 3Neper

It is quite clear that the meter driving 'Ansteuerverstärker' does not allow responses much higher than about 12 ą 13 kHz

Please notice also, that the responses between 15 Hz and 100 Hz is suffering from too little measuring references (frequency spots).

It is quite evident, that this response compared to the previous plot is showing a rather good overall frequency response.

A preliminary conclusion:- there apparently is a problem being caused when the W24a fader is adjust at 0 Neper. This will cause a direct signal transfer through the W24a fader device, whereas with inserted attenuation the Aussteurverstärker is responding, in respect to frequency response, quite good. What not is coming out of this test series yet, is the Lichtzeiger scale correct operating (indicating)? 


Finally, we conclude today with:

W24a fader set at - 3Neper using as indicator the Philips PM 2401 analogue meter instrument. 


W24a fader adjusted at - 3Neper (≈ - 26 db), reading off from the Philips PM 2401 analogue meter

Please notice again:- graph scale is neither fully logarithmic nor analogue. Second, the curves between 10 Hz and 100 Hz, as well as 1 kHz and 10 kHz apparently suffering from too few measuring spots. I have, however, tried to smooth it a little bit, but still some kind of uncertainty is prevailing.   

The difference between what is shown in this graph (drawing) compared with the previous measurement, is that some might originate from mismatch originating from the way we do currently measure. Albeit, it isn't much difference.



On 4 April 2016


Last Saturday I encountered quite some problems when continuing experiments with the experimental dual channel amplifier PCB.


My current experimental setup

The problems encountered were, that the channel started with some time ago did work quite fine (I thought). But became apparent - it did not at all!

Let us first go back to the schematic of the concerned circuit.


As already just mentioned, the first channel (1 out of 2) seemingly operated fine

The second channel implemented on the same PCB was brought to act. A terrible distorted signal appeared on the output capacitor.


My first action, was checking everything. I did not find any incorrectness.

My second step was, to approach the first amplifier chain again and checking whether something might have been wrong.

Just it was!

The first implemented transistor TR 1 wasn't of a NPN type but a  PNP!

After having replaced it by the quite universal BC 547 this stage behaved as is doing the second channel. Also the dc levels being quite in line.

But amplification being far too high. In the schematic a provision being noticed converting the circuit into a so-called line-amplifier. Necessary was changing some resistors.

This option did not work appropriately at all.

As so often, in these cases I cannot sleep well and went through the principles of its schematic.

It just popped up in my mind, that it should be possible to manipulate TR 1's amplification by influencing the negative feed back factor. For it I replaced the 12 kΩ emitter resistance R 4 by a small potentiometer trimmer, where the fixed carbon layer constituting the emitter resistor and the variable arm being connected onto elco C3. It becomes now possible to get some variable feed back.

However, I must admit that I had expected a bit more amplification swing. However, 3.4 db isn't on the other hand too bad.

Practically the signal-output with a constant input of 1.7 Veff gave an output swing of ca. 3.4 db.

Another nuisance encountered was still overloading of the first amplifier transistor. Please bear in mind, that studio praxis was for decades to orientate signal levels onto 0 db. (what nowadays the sound reference is I don't know)

What was, for practical reasons, being neglected because of mall functioning of the first amplifier stage (using an PNP instead of an NPN transistor), I had fed directly the line levels onto the input capacitor C 1. To omit overloading I went back to the input resistor of 47 kΩ and the resistor R 1 of 100 kΩ. Briefly giving a signal level reduction of a third or, say, 9 db. The output loading resistor of the fader section remained 1 kΩ. Might be later replaced by means of a real line transformer (likely parallel onto the 1 kΩ resistor).


The potentiometer with attached elco C 3 is good visible

Visible is an experimental additional 2.2 kΩ resistor. Where I would like to see whether the 10 k potentiometer or its 10 kΩ + 2.2 kΩ making 12.2 kΩ and does this make a difference? Which it did not (hardly measurable).

However, the advantage - by this means the output level can be manipulated over 3.4 db.


Accurate level measurements have been accomplished by means of the Philips PM 2401 meter

Please notice also what is written down on the meter scale:

0 dB 1 mW 600 Ω 0.775 V

This what I meant with professional studio 0 db level.

I suppose, that the RRG used this level related to an 600 line impedance, or they took 200 which was their line impedance before and mostly during the war.


It was also measured whether the W24a fader output correlated with the analogue meter scale. It did quite well.

Also astonishing, was, the reproducibility of the W24a fader settings.

The full range of the W24a fader attenuation being measured between 0 Np and 6 Np, the latter value taken just before the signal being short-circuit by means of a special internal provision.


Finally a proper sine-wave being display at the scope CRT screen

Shown is a sensitivity of 2 V/cm



On 6/11 April 2016

Extended below on 13 April

In the meantime new approaches have been commenced.

During a telephone conversation some days ago with Jaap Keijzer, Jaap suggested - why not using TL072 operation amplifiers?

After some considerations, I decided that it is indeed worth doing, at least, some experiments.


My first move was purchasing new components, like experimental boards en ICs, transformer and some other devices.

My first approach was to build a dual voltage power supply providing + 12 V and - 12 V

Because the latter concept is not yet known fully, I have built it in such a way, that it stays versatile.


I don't like 230 V mains quite next to the audio circuits during my experiments. I therefore decided first to construct a dual voltage power supply mounted on a separate experimental board


Simplified supply concept

During the beginning of my experiments, I encountered some strange interference, which was found within an oscillating -12 V regulator. After some additional electrolytic blocking capacitors all functions well.


     My next move was to built on a second board an experimental amplifier, leaving room for extensions

The wiring towards us constitutes the input line

The TL 072 opamp not yet been mounted, because I would like to check first whether the power supply lines have been wired correctly. 


My favoured concept for decades is the application of an inverter circuit, as is shown here

Its simplicity of adjusting the amplification gain, stability of the output off-set is quite striking.

This layout has been chosen as to match onto the foregoing transistor PCB experiments.


The principle circuit equals the foregoing loads

It proved being very convenient to adjust the amplification gain by means of the small 10kΩ potentiometer (feedback gain).


During the course of the experiments I referred continuously to a 0 db output level. The usual signal reference in studio techniques

It has to be said though, that I am not acquainted to nowadays studio (sound level) standards.

The board hold in position by one of the foots of the level-meter is the power module. Which is, for obvious reasons, being kept away from my current experiments. 


Giving you an impression of its experimental set-up


My last approach was checking the overall frequency response. Meter set at 0 db signal level (0.775 V @ 600Ω). My aim is to measure the overall frequency response at 0 db signal level output between 10 Hz and 100 kHz

Left of the module U70a we notice the former PCB


The measured overall response of the newly implemented TL 072 opamp circuitry; still in place the auxiliary circuit around the W24a fader

In my perception, the overall response measured between 10 Hz and 100 kHz is quite impressing.

Please notice: that the scale isn't logarithmic for both axis, nor it is linear. Its only purpose is to express  the sound frequency response.

Level deviations between - 0.4 db (10 Hz) and +0.1 db between 10 kHz and 50 kHz.

This time W24a fader being set at 0 Np (signal passes straight through it, maybe a parallel resistance remaining)


My preliminary conclusions today:

It is very useful to change our pre-amp concept from transistor controlled PCBs to the full application of integrated circuits.

It has proved possible to omit all capacitors in series with the audio signal paths. The off-set is for our purpose negligible. Signals being interconnected (matched), at least at the input and output level by means of high quality audio transformers, once designed by the former Dutch NRU broadcast organisation. They developed these superb devices themselves and these were once manufactured somewhere in Holland. Our signal (levels) today have passed through one of these superb transformer devices.  


Today 12/13  April 2016

I did a small test series.

Until now we have relied upon the application of the W24a Profil-Regler (fader). But we also possess a more economised version type W 44. I guess, it was introduced that very year. It is of a far simpler design and therefore consuming less labour-time and materials.


For it I have pulled the W24a fader a bit out of its housing, herewith pulling it out off the circuit; the W44 being used instead

It was already previously encountered that the W 44 type does give an additional 8.6 db signal damping (loss). This value equals 1 Np.

I measured also its frequency response W 44 at 0 db and the controlling Philips PM 2401 meter reading off 0 db. This was within the amplification range of the experimental TL 072 amplifier circuit.

Going through it from 10 Hz up to 25 kHz we did encounter a maximum deviation of - 0.1 up to + 0.1 db

We may thus draw the conclusion, that the entire circuitry responds more or less flat.

The same being repeated W 44 set at -12 db.

The maximum deviation between 30 Hz and 20 kHz -0.2 db.

50 Hz = - 0.1 db at 100 Hz - 0.1 db

and between 300 Hz up to 20 kHz entirely flat at 0 db level.

Isn't it impressive?



On 16/17 April 2016

On Thursday 14th, I have started with extending, on the same board recently created, a microphone preamplifier section. Not in a direct experimental manner, but just when it is performing well, it can stay this way.



The circuit concept is quite straight forward, but performs exceptionally good

As to open it in PDF just click on this drawing

It proved to be possible:

with an input of 1 mVrms  to obtain an output level of 0db or even beyond.

The first opamp section provides a fixed amplification of 21, hence providing a gain of, say, 26 db The second opamp-circuit is capable of a variable amplification gain of up to 25 times (34 db), but can also act as a kind of attenuator (buffer stage), when the potentiometer being set that it forms a short-circuit between output and input; then the input resistor of 1000 Ω is acting as an attenuator (against virtual ground). The maximum obtainable gain being 26 + 34 = 60 db (= 1000 x).


The microphone preamplifier DUT


My next move is to feed the microphone signal through the W24a fader assembly (drawn below) onto the already existing amplifier stage (right-hand side)

Here I encountered my first setback, because all started well with an output level of 0 db. However, the output signal showed irregularities on the CRT screen, it was soon discovered, that this was due to induced hum originating from the not (yet) grounded cable-screen. Because, for this experimental occasion, the TG10 generator output being galvanic wired onto the microphone amplifier input (left-hand side).


After having connected the screen experimentally onto the ground of the signal source, hum vanished; leaving the cable-screen free from contacting the DUT*

* Basically, in symmetrical systems, grounding of a cable-screen should always be maintained at the side where a signal originates from! With the exception of, for example, a (non pre-amplified) pick-up or microphone signal source, in these cases grounding of the cable-screen should take place at the entrance of the first amplifier stage. It should always be prevented: that hum currents can flow through the cable-screening towards the next stage. Hence, whatever a situation - grounding of cable screens should only be maintained at a single side. When still hum problems being encountered - then the sections should be separately interconnected onto a common ground-potential-point or rail.


However, it was encountered thereafter, that the microphone pre-amp showed considerable gain instability.

I measured the output off-set which was  about 1.9 mV. But its value varied with the apparent gain drift. Sometimes even the entire signal disappeared; but after a while it came back again.

Then I disconnected the line between the input transformer of the W24a fader assembly. The dc output off-set increased considerable.

Then I changed the linking transformer by a better type TLX1-B, similar to the series I am intending to operate anyway.



System insertion damping is less, and for the time being due to less amplification the microphone output did perform a bit better.

But after a while the nuisance repeated again.

Might it be - that the current load of the second opamp output is too heavy?

I therefore disconnected the green wire in front and the off-set voltage rose up to about 130.9 mV!

Apparently the off-set current value have to be counted with, seriously! My move was to implement in series a 1 µF capacitor, not an electrolytic type, of course. All the nuisance vanished instantly!

Everything performs well since.

What should be checked: might the considerable off-set originate from an off-set caused by the TG10 generator output?

Checking this must be simple, as only the TG10 generator BNC connector has to be disconnected.


It is clear that the value of 1 µF is actually a bit too low, but I have currently no access to a sufficient (small in size) capacitance type. I did calculate, on the back of an envelope, that 10 µF will do well. But have to look whether our distributer has these on stock. Its value becomes only critical at lower frequencies (@ 100 Hz signal -5 db down, versus 0 db @ 1000 Hz valid for the rest of the higher audio spectrum).

My next move was to 'load' the IC output-circuit with the input of the genuine recording amplifier type V47b. Its loading did not have any measurable influence on the parallel wired 0 db meter (PM 2401).


Between the microphone amplifier and the output IC we would like to operate a W24a fader assembly.  


This constitutes our current (basic) fader concept

The transformer on the left-hand side is currently type TLX1-B

However, no great deal technically but very efficient, as long as the interconnections between its output circuit and the next stage will not induce additional hum or other interference. However, never forget the complexity of such fader device!

The W44 fader basically acts similar, but I suppose - that its internal circuit might differ a bit. Also should be taken into account, that the W44 type generally is causing an additional  8.65 db insertion loss (= 1 Np). However, it will also be separated, at least by a driving (input) transformer. Which provision will always be compensated for by an additional amplifier stage.


I also was able getting ultimately a brief idea how the amplifier-interface housing might be constructed.


The two boarding transformer strips are of equal design, once used by the NRU (Nederlandse Radio Unie), the forerunner of the Dutch radio broadcast organisation (1950s and /60s)

Their performance is of an exceptional high order, and I don't like to change its PCB concept; providing totally 8 x 1 : 1 separation possibilities.

I guess, the box might get a height of ca. 8 cm, and will fit well within the additional  created space at the attached frame-bottom.

The flat Siemens Tuchel connector (30 pins), will be used for interconnecting the fader assemblies. The second Siemens Tuchel connector (12 pins) should supply level signals onto the two U70(a) Lichtzeiger-Instrument driving units. One for the input- and one for watching the output-levels.

I got, some time ago, from Dick Lucas, a multi-cable section having 10? screened symmetrical audio cables. I suppose perfect for linking our four fader-assemblies onto the future amplifier box.



On 24 April 2016


My first aim was checking whether my brief calculation being valid. Yes, indeed, 10 µF as coupling capacitors between the last opamp and the transformer BLX1-B is performing well below 100 Hz. With quite minor level drop of  1 db we can run the system easily from below 30 Hz onwards.

My second move on 22 April was building a CD opamp interface, where at the input circuit the two channels L and R being mixed together, as we only operate a 'mono' Magnetophon system. 


The third opamp from the left being build today

The bright blue led is indicating the existence of - 12 V supply, the red led for + 12 V has not yet been changed for a brighter shining type.

Photo taken during testing it.


For it I drive the circuit straight from our CD-player


The circuit concept is about similar for all opamps involved.


Rather straight forward technology, but it is working sound this way

For those interested in a more nice reproduction, please click at it as to open it in PDF

The only change today , is, that we know since that in the output line, when being loaded with a transformer we should implement a series capacitor of 10 µF; because the offset of the circuitry cannot be neglected. It might also be possible to implement also an input capacitor of the same value, which will lower the off-set voltage, but even then an output capacitor cannot be omitted.

This knowledge will cause a slight concept change, because there is no space left on the main amplifier board. But I would like to add an extra board only carrying a series of 10 µF capacitors.



On 23 May 2016

My next move was building a second buffer stage. This stage should follow-up the output of the faders (Profil-Regler). Their outputs finally have to be supplied onto a three channel mixer.

2 microphone channels

1 channel for a CD player

Actually I have wired only a single microphone channel yet.




Therefore, we currently have to deal with two channels. For this reason, I have provided two additional buffer stages and one mixer stage. Its output should be fed onto V47/2 (V47b) recording module, genuinely meant for this application. Up the mixer stage were yet only a single buffer-output being interconnected with; originating from the CD channel


The system being under test

This time apparently also the second buffer-stage-output being linked onto the mixer circuit too.


A bit more close up view of the board layout

The blue led indicates the availability of -12 V the red signals + 12 V

Front left the microphone channel 1 pre-amp, next, still empty, meant for microphone channel 2

The third IC constitute the CD preamp channel, its output have to be fed, via a special transformer onto the according fader arrangement. Its output signal being supplied onto the fourth buffer-stage

In front, most on the right-hand side we notice another buffer-stage, which will currently handle the microphone fader signal. The outputs of the latter two buffer-stages being fed onto the 3 channel mixer stage (IC top right).

The potentiometers do constitute the adjustable feed-back settings. By this means it is possible to adjust each stage amplification factor separately; in my perception a very convenient provision. 

Some days later, I have built the buffer-stage for driving the Lichtzeiger (LZ) instument. Which is indicating ly means of a light-slit the actual audio-level fed onto the recording unit V47b (V 47/2); genuinely the recording head driver stage combined with the erasing-bias generator.


V 47/2 (V 47b)


My basic concept, followed through-out the entire module


LZ stands for: Lichtzeiger instrument 


Top row, testing currently the buffer-interface driving the Lichtzeiger driver module U 70a


Board layout

The empty places being meant for future modification.


Testing the performance of the LZ output, feeding the Aussteurverstärker U 70a

All quite straight forward technique, by the way.



On 11 June 2016


Quite some time have been passed since my last contribution on 23rd May.

Why, too lazy?

No. But I have been heavily engaged in a huge (intriguing) research project together with my friend Phil Judkins.

We are working comprehensively on a subject, this time related to German espionage networks where British S.I.S. could never lay their finger fully on it.

Why espionage?

Because what British services knew, and that was quite much throughout most of the duration of WW II, relied mainly upon W/T interceptions. Following, for example, Paul Georg Fidrmuc's (cover-name Ostro) whereabouts living in Lisbon, almost all communications went through W/T channels. British R.S.S. (RSS ..v.W...) reported on 24/7 basis from mid January 1942 until 3rd May 1945. One gets a tremendous inside vision in their daily occupations. Although, Ostro was a commercial business man, his informations were in Berlin regarded as most important; albeit, that some (Britain's) considered it being purely invented nonsense. One cannot imagine the S.I.S. efforts put into these subjects. It was even discussed to kill him in spring 1944, because he had predicted the area- and about the days for the likely Allied invasion on the European continent (Operation Overlord). It was finally decided in the XX Committee not to eliminate him, as the implications (danger) might have been that Britain's 'most secret and secure source' would have come to the understanding of the Germans (Ultra).

There existed generally two types of secret communication services ISOS and ISK. ISOS dealt with manual codes, whilst machine codes had been managed under the designation ISK. In a post war summary, providing all intercepted messages, every W/T telegram had been accompanied with an according ISOS- or ISK running number. In Ostro's case (3rd May 1945) at the final end their successive numbering had reached >140,000! (this should be regarded altogether what once had been decoded via other network channels as well)

Also interesting, is, that sometimes there exist irregularities in the succession of received dates. Here the ISK numbering is providing the solution - they might have encountered difficulties with decoding, as it happened that a message of, say, mid December 1944 carried an ISK number of about 140,000 (and popped up between February or even March 1945). The only reason for this, in my perception, being the fact that processing for some reason had been delayed. In a smooth running system it might have been likely that difficulties occur sometimes.

By all means, without 'German Wireless Communications' the whole British endeavour could never have been accomplished, and Bletchley Park would never have done such a superb and utmost decisive job!

Consequently, that is why our new commitments aren't far off the aims of our Foundation; because it concerns aspects of 'German Communications'!


Our preliminary focus started with the KLATT complex. Frustrating (upsetting) British services even up to the beginning of our new century.

The experiences are striking! Digesting sometimes formerly 'Top Secret' materials and the great many efforts put in post war decades as to bring ultimately the proof but with annoying results.

What to think of an important player like:- Andor Gross alias* Gyorgy, who was born on 15 April 1905 and possessed the Hungarian nationality.

* I also learned that some British wartime typing machine were equipped with an @ key! Therefore in such cases we find instead: Andor Gross @ Gyory; by the way, he also acted on behalf of British S.I.M.E being named Gyorgy!

  However, rarely - on the front page of his main file wearing his photo taken in front of the Union Jack. Next, encountering a white paper page:- ..... this personal sheet is for registry use and to facilitate amendments. ....

.... File status RED

Date? 04.10.1989

He must have been 84 years old by then.

But why?

This date correlates to the days of the 'opening (lifting) of the Iron Curtain' in Hungary. But, this date was more than 44 years after the Germans surrendered in 1945? Some pages still being withheld, visible by blank pages, marked by means of a printed stamp referring onto a law reference of 1958.

What to think of: KLATT, heading a most mysterious German intelligence (Abwehr) organisation, first stationed in Sofia (Bulgaria) and early 1943 had moved, for obvious reasons, from Sofia to Budapest**. He ran until early 1943 a Wehrmacht intelligence (Abwehr) Dienstelle, being a private person, and he was a Jew!

** In Budapest he continued working for Ast Vienna, albeit under the cover of Hungarian Intelligence Services.

British S.I.S. could read his communications starting on 15 October 1941 up to 12 February 1945, (almost) entirely.

Klatt died sadly due to medical implications, early 1953.

What to think of - that his main file was being lifted (opened) in 2003 still?

We may consider that Richard Kauder @ Klatt was frustrating (upsetting) British intelligence institutions for very many decades since!

One cannot imagine, in how many cases, in post war decades, the word Klatt being refereed onto!

I sometimes get the strong impressing - being British's biggest frustration!


Deo volente, Phil and I would like to give a comprehensive presentation on these many unknown aspects, in autumn 2017.

In the third week of May Phil came over for several days as to discuss our progress. During his stay we logged onto the British National Archive, and we downloaded more than 5000 document pages. Which had to be printed - digested and being stored in my dbase system (using mainly keywords). Phil left on May 20th and yesterday, 10th June, I could finish this endeavour!

That is why our website could not be upgraded in the meantime. It is astonishing to notice, what quantity of knowledge can be stored in an alert human brain; Deo volente, of course!


So far this interlude.


You might remember, that we faced since we got our T 8 Magnetophon apparatus in March 2013 malfunctioning capstan motors.

Thanks to Bernd Fischer of Germany we, generously, got an AW2 tape recorder motor. Originating from an AEG consumer machine type from, say, mid 1950s. About 1963 Mr van Thijn of 'Arti Sound'  showed me proudly such machine. He also mentioned, that it was a quite good performing tape recorder; of course, for those days standards; albeit, that he was using commercially Studer machines. I never could obtain one, until quite recently. But now half a century have been  passed since!


My first fear, it's housing diameter might be too large for our application.

Jaap Keijzer prepared for us especially and experimental 'nylon' disk which prevents from hampering mechanical and electrical annoyances. 



Please notice the white coloured disk between the top of the motor and the recorder-deck frame

To be discussed hereafter, the concept of this motor allows two synchronous rotation speeds; being provided with separate field-coil sections - including according support coils.

Divided in a 4 fields configuration or an eight fields application.

My current calculation considers that 4 main fields implies 3000 rpm : 4 is providing 750 rpm rotor speed.

And, 3000 : 8 giving 375 rpm.

3000 rpm is the result of 50 cycles p/s (Hz) in 60 seconds. The standard European mains frequency is 50 Hz. In contrast, the US is relying upon 60 Hz and therefore their rotation speed being 3600 rpm.

Practically this speed can be made lower, but then the according generator poles have to be increased. But, the US like consumers being confronted with rotation speed of 3600 rpm (60 x 60 = 3600).


Jaap has promised that he would provide an extension adapter so that the current (5 mm) capstan diameter will be restored in what it standard had been 9.7 mm. This will of course increase tape speed. All depends upon whether my estimation mains rotation speed versus coil group numbers obeys to my assumptions

This adapter should be stable, but can be changed, at will.

Remaining a capstan diameter of 9.7 mm has the advantage, that the rubber-pressure-roller mechanism does not have to be changed (adapted).


I spend also some time in extending the various circuitries on our experimental amplifier board.


The board layout complexity is increasing

It has to be said, though, that all concerning circuits are of quite straight forward design technique. The accompanied potentiometers allowing gain adjustment for each OP-Amp stage separately; a very convenient provision in my perception.


The wiring is briefly showing where the contact pins being interconnected with

Please notice, that some minor errors might have been incorporated, but the first aim of this drawing is to allow conveniently interconnecting the various stages. For this occasion I have used German expressions;.but trust that technicians among you will understand what it all is about.

Last week Dick Zijlmans stressed, that at the entrance of the mixer stage, all signal phases should arrive 'in phase'. In my perception, this can be commenced (and checked) by supplying all channels from a mutual signal source (matched onto its application).

The consequence of Dicks point deals only with the current phase of signals, of course not with level amplitudes.

It is, nevertheless, my intension to adjust and maintain all internal signals at a 0 db level reference (maximum driving level with a + 5 ą 6 db margin upwards.



On 28 June 2016


My engagements mainly being still split  between the KLATT related affairs (touched briefly in the foregoing section), as well as the sluggy progress of our Magnetophon endeavour; lasting for more than 3 years. Neglecting - I also have to take care of our website, as well as the regular daily commitments in respect to our Foundation. The KLATT related subject presentation, Deo volente, being scheduled for autumn 2017; again a joined project together with my friend Phil Judkins. The overwhelming quantity of sources is really astonishing us. In many respect, we are coming so near to this intriguing subject - that we virtually have become part of it; albeit, far away from those gloomy bleak times, sitting on a comfortable couch or desk chair.



The current bottleneck being the adapter to the AW 2 capstan shaft, fitting it for 9.7 mm shaft diameter (38 cm/s).

All depends on Jaap Keijzer's commitments.

Nevertheless, I used the opportunity to proceed with the experimental audio interface board.


The board layout becomes a bit confusing, we therefore should record where the many interconnections being meant for

Two OPAMPs should still be wired (the two open sockets in the lower row), one a preamp for the microphone channel 2 (left) and its according buffer stage (second IC right of it); which's latter signal being fed onto the mixer stage. Mixing microphone channels 1 and 2 as well as the CD channel.

The quite heavy blue capacitors are constituting 10 µF types (63 V). Don't think, that the shop around the corner will have these on stock!

These are particularly meant for as to separate dc signal levels, because the rest of the many OPAMPS being dc linked mainly. Some channels really do need such provisions; in particular those providing a high channel gain. A small dc off-set to begin with will be multiplied by the amplification factors; hence, is loading the OPAMP output line with it. When its load would be a low impedance line-transformer one should implement such device. In most regular audio amplifiers they use electrolytic types; but these are very vulnerable to incorrect polarisation; therefore we adopt MKT types. 


That not all interconnections being placed in a simple manner is due to the fact that circuits being engendered (expanded) experimentally in the course of this project

The challenge always is, where and how to position your first IC socket on an experimental board!

However, helpful might be that the central ground-level of lines being marked as such, as well as their interconnection purposes being indicated. Please bear in mind, that we operate symmetrical in- and outputs against ground; because we operate the OPAMPs with + 12 V and - 12 V. 

Important is: that all channel phases should be identical; when inputs being fed from a common source (of course, each one adapted in level to their specs). That later signal phase can differ is then due to different signal sources. For example, microphone signal versus CD player signal content.


Because this page has become huge, we have decided to create a new webpage, dedicated onto our 'ongoing Magnetophon T 8 Saga'.



To be continued in due course


By Arthur O. Bauer