GK IIIb system



                                                                                                                                                                            Photo Cryptomuseum.com

Iv. - Gerät , likely stood for: Invertier - Apparatus; which actually is exactly expressing what its purpose is.

Our second device obtained end of April 2019, covered on its back information that it concerns a stationary apparatus of the occupying forces in Denmark. Which is in accordance with where it came from, and that was once from late Ebbe Pedersen.


However, now usually known as: scrambler.


Today we would like to show you how it experimentally operates.



Page initiated on: 8 June 2019

Status: 16 September 2019




Test_3    (16 September 2019)




First was to check whether both GK IIIb oscillators being tuned in line

For this some time ago, I have left our firstly obtained oscillator frequency unchanged; as it was oscillating at 2094 Hz. Therefore the newly obtained second GK IIIb (Set II) has been tuned at this latter frequency.

It isn't yet known what once the exact switching frequency was.  It doesn't bother much as long as its operating frequency is above, say, 2000 Hz. The internal band-pass filters do cut-off already above 1800 Hz. I doubt that it was 2000 Hz exactly, as the second set was oscillating >2200 Hz and I therefore doubt that both oscillator tunings allow about 10% frequency-deviation downwards. It all is arbitrary, but as long as both switching frequencies are equal, as is shown above, the system should operate correctly.

Paul Reuvers told me last Friday, when I visited them in Eindhoven:- that Churchill's wartime (secret) telephone scrambler operated at 2500 Hz, a quite common frequency, in the early 1970s also used by the Dutch police.


However, the screen-shot proves that without re-aligning the painted lissajous of the first order is showing that it all had been adjusted correctly. 

Fore it the two scope channel-selector-switches being pressed both. Creating that the X and Y axis being fed from the two input channels. A circle is pained when both amplitudes and signals phases do differ 90°. When there exists an amplitude difference between both generators the circle is becoming an ellipsoid. However, essential is, that a painted ellipsoid does not rotate, because this is indicating a difference in signal-phase.



Please notice on the left-hand side the second wire (quasi in the top middle) where some forgoing soldering is visible was the (invisible) trouble causing faulty connection

Please bear in mind, on the left-hand switch connections we notice 6 wire connections which might correlate with the wiring of the OB and ZB switching contacts concerned. Though we recognise at least 3 extra wires on the right-hand side, which aren't shown on the following block schematic.


The wire-would circular potentiometer, is the one shown in the lower receiver channel just controlling the input level fed onto Rö 3 G1. The vertically mounted wire-wound resistor on the right-hand side, is the fixed potentiometer in the upper (transmitter) channel (controlling G1 level of Rö 1).

The switch visible up in the centre is the actual 'On - Off' (Ein - Aus) mode switch.



The faulty wire-connection concerned one out-of the three wires around the selector-switch marked OB (Orts Batterie)(local battery mode)

When the GK IIIb's are operating, operating on each side the OB/ZB selector this is resulting in some change of the signal-amplitude.




Test condition one: both GK IIIb sets (apparatus I and II) being set in 'off' mode

Hence, the in- and output signals  fed onto the local telephone terminals and line connections and line terminals are internally by-passed and thus do equal, even galvanic.



Demonstration condition 2

Set I being switched on in scramble (Invertierer) mode and Set II is left in 'off' mode.

In this case: Set I is scrambling in two directions (both in the receiving and sending channel). Hence input and outputs are being scrambled.



Set I is operating in the off mode (by-passing signals in either direction), whereas Set II is scrambling

What will be noticed on both side is equally noticed in the foregoing drawing. Scrambling  is noticed.


What next is happening, is that both - Set I and II are scrambling, but on their receiving side the scrambling has happened twice; reversing the side-bands two times is reconstructing the original message or signal.     In other words: one time scrambling means higher frequencies transferred into lower frequencies. When this process of inverting the spectrum is repeated again what was transferred to a higher spectrum will be transferred into a lower spectrum v.v.; and the output spectrum should equal what once have been entering the system. This is, of course, valid for both transmission directions. 

A single tone will be reconstructed quite well, but without considering the quite limited (typical) 'Feldtelefon 33' (FF 33) voice-spectrum and the necessary band-pass filtering as to cut-off the upper side-band components, the voice reconstruction will still differ a bit concerning its bandwidth spectrum; albeit that when operating FF 33s on both sides of the system, there might be little difference encountered.         However, the signals passing the by-pass modes ('Off' modes) being not limited (hampered) in its bandwidth spectrum.



Now system I and II are both operated in the scrambling- as well as descrambling modes

Thus signals fed on either side onto the telephone terminals, albeit scrambled twice (in set I and II) are now being restored into their original pitch.

In the next two YouTube films (00034 and 00035) you will see and hear the series of demonstrated modes, as described in the foregoing  4 drawings.

The first YouTube film 0033, is showing that what was concluded months ago was truly valid that the Gl 101/6 Z was really a faulty device as was the second built-in device as well.



The defect ring-modulator module


Experimentally the ring-modulator being replaced by 4 BA 324 diodes, which, actually, here not directly being wired in series, but in two groups of two diodes in series 


What might prove to be necessary later, is replacing also the two old ring-modulators (diodes) in Set II by adapting also BA 324 diodes. But leaving both Gl 106/6 Z modules still being mounted; only disconnecting the according wirings.


YouTube films:


Film 00033:    Now all is operating correctly, I placed back the once removed Gl 101/6 module. It proved, however, that my former conclusion that it malfunctions was correct.

Film 00034:    Explaining the block diagram and how all is working; also showing what bothered us for several months. A single wire being faulty at the OB/ZB mode switch. I have certainly tested whether all wires at least weren't disconnected. What might have happened is that we have encountered a kind of: cold soldering or that like.

Film 00035: Demonstrating the various modes selected by set I versus II, and the according consequences. When both sets are operating the input signal has been restored. Also shown is the situation when both sets being set at the 'Off' mode and consequently the input signal is passing unchanged through the two GK IIIb units.   The tone-generator feeding, say, Set I. Parallel onto the generator output, a cable is linked onto the frequency counter; on top of the oscilloscope. The scope is monitoring the output at the telephone connection-terminals of Set II. Conditions: signal is passing through two sets. First both mode-switches set at 'Off' mode. Thus what is being fed onto is passing unchanged and is shown on the CRT screen. Now one set is being switched in the 'On' mode the second set stays at the 'Off' mode. Since the tone pitch is being reversed. Thus a high pitch becomes low and a low input pitch becomes a higher tone. Then both sets being switched in 'On' modes. Now the input tone pitch is becoming identical with that on the output side.




On 21/27 June 2019

Today we would like to show you our first experiments with two genuine German wartime field telephones in conjunction with two GK IIIb voice-scramblers. The results being, however, yet quite meagre. The reproduced bandwidth is still too limited. What also counts, is, that the loudspeaker unit is directly connected onto the telephone terminal.


YouTube film

Film 00041: Fred, our Diemen neighbour friend, is supporting us with speaking into the microphone of a genuine wartime military telephone set; for this occasion placed in the depot, as to separate him from what is heard from the loudspeaker. Albeit that some of his voice can still be noticed in the background. First listening through both GK IIIb "Invertier-Geräte" (scrambler) units. And occasionally switching one apparatus off. This is causing that what is coming out of the loudspeaker is now noticed being "scrambled". However, when both sets being set in off-mode - both sets being by-passed and the limited field telephone quality is now directly audible.  The twice scrambled signal is becoming de-scrambled; albeit clearly of quite limited reproduction quality. At the end we notice a Lissajous painted at the CRT screen; the X and Y channels being fed from the sine wave signals supplied onto the ring-modulator modules. We notice clearly first order operation. Phase deviations are visible, but are roughly, say, +/- 45° maybe a bit more. 

Deo volente, I would like to repeat the test - but now operating a 1970s analogue telephone, albeit that I have to supply additionally, say, 48 Volts onto such system. I suppose also, that the mode-selector 'ZB' (Zentralbatterrie) is allowing such type of operation.

We also have increased the switching frequency; albeit from 2094 Hz to 2100 Hz. I had hoped for a wider tuning, but the build-in tuning facility does not allow tuning at >2115 Hz. When we would have had a higher frequency the pass-band would have started at a higher frequency (pitch). Let us consider we operate between, say, 400 Hz - 1750 Hz we  transfer now: 406 Hz - 1756 Hz; after all a senseless move.  

Passing through music does not make sense with such a narrow pass-band.

Therefore, please consider this contribution as a step in a chain of experiments due to commence.





Status 16 September 2019



As to test whether a more modern telephone apparatus would do a better job, I tested a Dutch T65 telephone, once used in the 1960s and 1970s even beyond

It actually performs better.

The unit in from is a regular 48 V line current power supply. A bit too big, but this is what is currently at hand.

It performs far better than the old Feldfernsprecher 33 (FF 33)



In front we notice an FF 33 apparatus and in the background our T65 under test

On the right hand the 48 V line current power supply



Our third photo

Down in the background of the far left-hand side, we notice our second GK IIIb scrambler apparatus; which telephone connection being loaded with the LSg B loudspeaker, which reporduces the T65 telephone sound.


Our aim is to put it later on display during our Secret Communications 3 Exhibition in the following manner:

On one side we would like to operate the T65 and on the other end, we load the second GK IIIb with a louspeaker.

When both GK IIIbs are being operated the content of the speech being conveyed audible.

Though, when one of the two GK IIIbs being switched off the loudspeaker conveys a scrambled signal. When, however both GK IIIb scramblers being switched off the the dierct voice tone is being heard.

We always should be aware, that the GK IIIb only conveys a passband betwwen, say, 400 Hz to, say 1820 Hz.

Consequently a rather narrow badwidth and therefore limited speech quality.

But this clearly is the consequence of providing such limited passband.

The regular telephone passs 300 - 3400 Hz, which is considered sufficient.


YouTube film

Film 00042:    Demonstrating the German GK IIIb wartime scrambler system, with a quite modern Dutch PTT  T65  type telephone, successfully.



Deo volente, to be continued in due course



By Arthur O. Bauer