Current status: 27 May 2019
1 + 2 + 2a + 3 + 4
GK III b
Field telephone-line scrambler
Looking for such an apparatus for, say, 40 years.
This survey should, Deo volente, being a mutual project together with Hans Goulooze.
Siemens & Halske
Typ: G K III b
Zeichnung 11Rel.trgb.11b (tragbar)
Fabrikations Nr.: 424028
From outside it might not look fine, but inside it may be regarded untouched
Looking what is inside, it does not give the impression that someone has touched it
I hope that you do agree that it is in a perfect shape, inside
I suppose that the scale it adjusting the carrier frequency of 2000 Hz, we actually measured 2094 Hz.
Viewing the right-hand side, also untouched and looking quite nice and genuine
Up the three RV 2P800 valves
The centre valve constitutes the (suppressed) carrier oscillator.
We have pulled-out the middle RV 2P800 and found that the Al. screening cylinder looks like new.
In the centre the "control panel"
I tend to believe, that ZB and OB is deliberate designated for deception, as it most likely is the selector switch which determines whether this station is receiving or transmitting a particular side-band. The station on the other side of the strip has to select just opposite mode.
In German telephone exchange techniques name: Post A or B; where A sends with lower-sideband towards B and B sends with upper-sideband towards A.
Thus when one side has switched at, say, ZB the other station should select mode OB.
"Fernsprech-Apparat" = the Telephone connection; "Leitung" is the connection onto the exchange system, thus to outside.
The wire-wound resistor is the input potentiometer
This one causes problems as it axis stocks. However, I have measured the signal of in- and output, and at least it functions, albeit that it is impossible to turn it in either direction.
Viewing the case inside
Hans Goulooze was convinced to so set-up some trials
For it we need a range of supply sources, such as + 90 V, -3 V and 2 V filament voltage for the RV 2P800 valves.
After we noticed, that in some way or another current consumption is quite normal, we continued our experiments
Our first attempt was to measure what signal is to be find at the valve in the middle; I expect it constituting the 2 kHz (suppressed) carrier oscillator
The signal at the anode is rather sound and we need a counter for exact frequency measurement as well.
The sine wave looks sound
Not yet 3 x 50 Vtt is what being measure at the anode of the oscillator valve RV 2P800 valve, with an anode voltage of about 90 V dc.
Therefore additionally a counter has to be implemented
We see that the carrier oscillator oscillates at 2094 Hz; instead of what is known 2000 Hz.
The flat-topped signal is found around the (expected) ring-modulator.
Some of you might notice the application of two 10 : 1 probes in series. This has been dome because the RACAL-DANA counter allows only 5 V rms. We now have a voltage reduction of: 1 : (10 x 10) = 1 : 100.
Dieter Beikirch very kindly supplied this photo of the GK III b block schematic, provided inside the cover-panel (Deckel) of his own GK III b apparatus
Umg. may stand for: "Umweg", or by-pass. However, I suppose this is a selective means, as it should only conveying bell- and/or dial signals (pulses).
Up to now, this is the only information at hand. Meagre? Maybe, but it is increasing our challenge.
A search on Google was rather disappointing, as non was found that had anything in common to where we are looking for.
I must admit, this is in the very most cases true in our endeavours.
Are we so exceptional in our demanding?
Maybe; but on the other hand, it is also a drive to bridge the apparent gaps.
On 29 March - 1 April 2019
Later this afternoon, I returned to the Klooster again to take some YouTube films, which are adding.
I took up the challenge again to get first an idea about the functions of the various modules as well as their interconnections.
AOB: Vermoeden LSB and USB proved the day later to be pure nonsense, because it does invert the audio spectrum in one direction only, which was proved on 29th March 2019.
However, I would like to guide you through the progress of our Survey:
Using a tone-signal generator, on both the telephone and and line inputs. Confusing is, that you see my connection onto two different pins. But the signal is also to be found on some pins of the lower module C 120 on the right-hand side
This screen pattern was not yet first understood
The set-up is becoming a bit complex, as we are using several power supplies; but I have introduced my note book again
Various measurements and considerations, pointed into the direction of faults in the ring-(diodes)-modulators, for both signal channels
Because when no signal being supplied onto the two terminal blocks (telephone and line) we still encounter a quite strong distorted signal of about 2094 Hz
Therefore, I had first to take precautions for removing the ring-modulator module:
Since our Tfb 2 Survey I have got some feeling for as to how Siemens was building-up their military telephone systems.
Rel b Kb 3x
Rel Bv GL 1-1/6
Undisturbed looking at the connections of this magic module
Drawing its brief schematic took far more efforts than expected
Because all looks the same.
I don't know the rectifier type, as I did expect Siruter like diodes, but these?
Viewing it from the opposite side
The small bobbins are special wound resistors; as on two white ceramic bodies we saw hand-written 24,4 and 23,8; I guess their particular dynamic resistance in the conducting mode; which had to be equalled to one another.
Today I have measured the serial resistor onto the 4 ring-modulator-diodes; and expected to find various values, which astonishingly is not. All prove to be 10.9 ohm!
I expected a real ring-modulator circuit, but found two series of diodes wired in series; shown down in by brief notes
X-are representing the connections marked with a red dot.
Why using especially BA 324 diodes, simply, because this type was available.
You can see the four diodes being connected experimentally
Afterwards: I instantly switched-on the HT again and discovered that the remaining "switching carrier" signal was hardly noticeable.
I measured by some of the GL 101/6 diodes a real short-circuit! (really measuring across every diode separately)
However, I cannot imagine what might have caused it, because the switching voltage was approx. 2 Vtt. The effective signal amplitude could, however, being a bit higher within the actual circuitry.
My aim is, to keep all a "genuine" as possible, and being restrictive in disconnecting - and opening modules. But totally preventing, proves to be impossible; as long as our aim is to turn it into an operational device.
On 15 April have been added a scientific Siemens paper:
Der Kupferoxydul-Gleichrichter in der Messtechnik
Von R. Tamm und F. Bath
Nach dem Vortrag in der Gesellschaft für technische Physik zu Dresden am 6.12.1935
Der Siemens Kupferoxydul-Gleichrichter
Please notice: that on the latter PDF page 14 shown Ring-modulator like divice - both diode-ends (ceramic disks) are being coloured black. In my perception, this may point at the very fact that just the way the construction of the diode contacts and what is in between is one of the secrets of their construction; maybe even belonging to one of their patent claims.
This latter paper is to be found in:
Siemens Nachrichtentechnik 6.Jahrgang 1936 1.Folge
Instantly disconnected the wires between nodule Ü 1480 and the ring-modulator module GL 101/6 and wired in similar was the diodes in the circuit
Time for experiments:
As I was alone in the room, I hardly could operate two "Field-telephones" at the same time; each one at either terminal connections (telephone - line).
Therefore I started with measuring what was expected, as at least into one direction clearly it was heard strange signals.
It was, to my astonishment, found that it is converting the speech spectrum such, that high frequencies turning-up at a low audible frequency spectrum, and the lower the generator being tuned the higher the output frequency went. The essence of a speech-inverter is a device in which the speech spectrum being reversed.
The Germans might have consiederd it like:
Voice - Inverter
For instance: input 400 Hz at the output terminals I measured about 1680 Hz (the counter didn't display a steady reading) . Ca. 1 kHz gave about 1 kHz (which I consider a weakness)
1500 Hz provides about 570 Hz. (deviations cause by the fact that truly our oscillator operates at 2094 Hz) maybe its signal phased being inversed.
Please remember: that our ring-modulator switching frequency, measured recently being 2094 Hz; but the counter expects (pure) sine-waves and therefore is not providing a steady read-out.
Than we can predict what every supplied frequency will cause at the output of the system.: 1500 Hz will produce 2094 Hz and 1500 Hz = 594 Hz
400 Hz will produce 2094 Hz - 400 Hz = 1694 Hz
That I could not measure exactly the correctly converted frequency, might be due to the remaining (switching) carrier (2094 Hz) versus the gating within our Dana-Racal frequency counter.
Therefore, please notice, that these figures are brief, because the counter being influenced by a mixture of other signals (maybe distortion products as well).
Some figures may have been influenced by the current oscillator frequency of 2094 Hz, whereas it should be 2000 Hz. What I would like to check in due course - whether the field-telephone inductor signal is passing unhampered through the GK IIIb system.
Today, after I had put this chapter on the web this morning, I have decided to return to the Klooster and among other matters I have also made some YouTube films.
Which I would like to add today.
Film 00028: The aim today is let you listen onto a German wartime voice coder. Speaking in a field telephone microphone. It is clear that the output is "scrambled" in the way as we have measured before. Switching frequency currently at 2094 Hz.
Film 00029: Viewing what also had been photographed, I suppose self-explaining.
Film 00030: Looking at the removed defect ring-modulator module, the rest has been explained by photos as well.
On Saturday 25 and Tuesday 28 May,
After so many annoying hours, I wanted desperately to approach our newly obtained GK IIIb apparatus again; that once came via-via from late Ebbe Pedersen's collection in Denmark.
Unexpectedly, the less sound GK IIIb, swopped before, gave actually less trouble than the new (looking) mint set.
Please consider this schematic carefully again. The fault found was: that into one direction it seemingly operates fair, but into the opposite direction, it does not
Last week, I discovered that likely someone encountered before a problem, as the soldering was clearly more bright than in our other set (please notice that the clamp is set at the switch-centre though it should be connected onto one of the outside contacts; though in this position the selector being set as to function with the newly found line disruption
Please notice the plain colour and its length of the (too long) mounting-screw a bit down in the background, which clearly indicates that this (these) aren't genuine. On the front-side, it becomes clear that these screws had once been slightly painted as to get it match with the rest. A quick view at our other set, shows that indeed these hardly come through the metal front-plate and were clearly colour-locked with grey paint.
All is a bit strange, but I compared both switch numbers and these do match, thus this switch must also be considered genuine.
What was discovered, is, that the kind of wiring of the similar switch in the first set looks quite different. This might point at a different period of production. Our first set carries all valves of the year 1943 and valve number 1 and 3 even the acceptance stamps do not differ much. Whereas the newly obtained set (this one), is in a better general shape, carries valves accepted in 1941. In both cases the valves '2' the valve in the centre for the oscillator, are respectively also pointing at 1943 and 1941, albeit that both weeks of acceptance are a bit different.
The valves found inside contain for us the only hold of an indication of its period of production. Nowhere inside we find a product date indication.
What is the function of the switch at the front panel?
Simply: to switch the speech inverter on-off; in the latter mode, the signal is being by-passed.
Please reconsider the foregoing block schematic. The switch is shown on the left-hand side; where the position Aus indicates the by-pass mode, whereas the switch here is set in the operational mode.
The other switch contacts are directly connecting the power sources, such as HT and filament supply.
As into one direction it operates sound and not into the opposite direction I had to go back onto the signal-input which is incorrectly conveyed.
After a range of tests, I came at the module designated BÜ1110
In German language: BÜ may stand for: Brücke, which actually is the regular function of this circuitry, curious in English language being designated: Hybrid circuit.
In Dutch and German language it is known as Vork (fork) and the German name for such a device is Gabel. And this is the designation 'G' used in the above block diagram.
The two open wires constitute the filter enter the low-pass filter part of the signal-processing.
Let us now look what I discovered after due considerations:
the cross on the left-hand side line indicates what I so far have deduced.
I am still doubting a bit, whether my conclusions are truly valid
The cross is drawn between the front-panel selector on-off and in the latter mode by-passing the system.
But still I am not 100% convinced, because how could have then operated it in the good-direction?
A solution for it might be: that the wire colours at BÜ1110 and at the out-side terminal differ and in between another connection does exist.
The circuit should split two-wire system into: one receiving channel and a transmitting channel; also known as: two-wire - four-wire technique.
That it works now sound is good visible
To be continued in due course
By Arthur O. Bauer