|
|
|
|
|
|
|
Crypto Rotor Germany Siemens
The machine is built around a
Siemens T-36 teleprinter,
of which
the keyboard and the paper strip printer are visible in the image
on the right. Towards the rear is a large removable cipher
unit with 10 irregularly stepping cipher rotors, each of which has
a different number of segments.
The T-52 is mounted on a large and heavy die-cast chassis,
which is fitted onto a wooden base plate. For transport,
the complete assembly could be placed in a wooden transit case
with large metal grips at the sides. The total weight, inluding
the transit case, is well over 100 kg.
|
|
|
During WWII, the T-52 was one of the main cipher machines of the
Third Reich. It was used for battalion-level traffic and was mainly
operated over land lines. Only the traffic that was send by radio
could be intercepted. It was occasionally broken by
Swedish codebreakers, and later also by the
codebreakers of Bletchley Park (BP),
but only when messages had been received in depth. 3
The intelligence derived from the broken messages
was not considered vital, as BP was already able to read the
tactical messages encrypted on the Enigma, and
also the high-grade traffic of the German High Command
which was encrypted with the Lorenz SZ-40/42.
The rest was surplus.
As initial version (T-52a) was not very secure,
several improved models were released over time.
The final versions (T-52d and T-52e) were so good that they
were virtually unbreakable.
After the war, these models were recovered from surplus supplies
and were reused by several countries.
|
 |
-
Schlüsselfernschreibmaschine = Key Teletypewriter.
-
Geheimschreiber = Secret Writer.
-
Two or more messages that have been encrypted with the same key.
|
Although there are many different versions of the Geheimschreiber,
we'll use the T-52d to show the position of the various parts and controls,
as it is the most sophisticated and secure version of the machine.
In the image below, the T-52d is shown from the front right.
At the heart of the machine is a Siemens T-36 teleprinter (Fernschreiber)
with its keyboard sticking out at the front.
The 10 cipher wheels are clearly visible in the image above.
The cipher mechanism is a complete unit that is mounted on the chassis
to the rear of the teleprinter.
Rather than printing onto sheets of paper, the T-52 produces its output
on a paper-strip, which is fed from a holder at the right into the printer
that is located just above the keyboard, leaving the machine on the left.
|
|
During WWII, the German war machine heavily relied on cipher machines for
secure communication, both over land lines and radio. Apart from a number
of hand cipher methods, they mainly used three (later four) cipher machines
for the bulk of their secure messages:
|
- Enigma
The enigma machine
was used at the lowest level of the command chain.
Between 20,000 and 30,000 machines were built and used by several parts
of the German Army and related organisations. Due to the sheer size of the
German war theatre, the majority of enigma messages was sent over radio.
- T-52 Geheimschreiber
The Geheimschreiber made it possible to encrypt a teleprinter line (telex).
Although it was possible to use such links over radio, the majority of T-52
messages was sent over land lines. As a result, such messages were difficult
to intercept. It was cryptographically more secure than the Lorenz SZ-40.
- Lorenz SZ-40/42
Like the Geheimschreiber, the Lorenz Schlüssel Zusatz (SZ) was used for the
protection of telex signals. It was a stand-alone wheel-based unit that was
connected between the teleprinter and the line. The Germans used it at
the highest level. The Lorenz machine was used over land lines as well as
over radio.
It was broken by Bletchley Park
by means of the Colossus
computer.
- Siemens T-43
This was a one-time pad machine (OTP) that was introduced relatively late
in the war. It was used only on a few networks.
Bletchley Park probably called this machine
Trasher.
Machines like the T-43
are often called 'Mixers'.
|
- T-52a
This was the first version of the T-52. It was based on the T-36
teletype and was only built in small quantities from 1932-1934.
It appeared to cause radio interference. It was later modified
with a filter and was then called T-52a/b.
- T-52b
The T-52b was a slighly improved version of the T-52a in wich a filter
against radio interference was added. As the machine is otherwise
identical to the T-52a, the two versions are generally identified as
T-52a/b. The T-52b was built from 1934-1942.
- T-52c (Cäsar)
The T-52c was developed in 1941. It had a simpler setting of the
message key, but had the nasty side-effect that the number of
possible alphabets was reduced dramatically.
The T-52c had a switch to make it backwards compatible with
the T-52a/b. The T-52c had the same cipher period as the T-52a/b
and is sometimes called: the Cäsarmaschine.
- T-52ca
This was an improved version of the T-52c in which the number
of alphabets was increased again by fixing a flaw.
- T-52d (Dora)
This is a serious improvement of the earlier T-52a/b.
It features irregular wheel stepping of the cipher wheels
and a so-called Klartextfunction (KTF).
The T-52d was developed in 1942/43.
- T-52e (Emil)
The same improvements that converted the T-52a/b into the T-52d,
were also applied to the T-52c. This resulted in the T-52e.
- T-52f
This version of the T-52 was developed but never taken into
production. In May 1945 the design was ready but by then the
war had ended [4].
|
|
Due to the nature of the various models and the later modifications,
the T-52 can be divided into four distinct functional groups:
|
- T-52a, T52b (T-52a/b)
- T-52c, T52ca
- T-52d
- T-52e
|
|
The following elements of the machine affect the encryption algorithm:
|
- Key generator
Each of the five bits of a plaintext character can be inverted under control
of the contacts of certain notched discs of the 10 cipher rotors. This can be
seen as a pseudo-random character stream – produced by the Key Generator (KG) –
with is XOR-ed with the selected plaintext character. In the original
German description, this operation is known as Vertauschen (swapping).
- Bit-swapping
The individual bits of the encrypted character can be swapped under control
of the contacts of certain notched discs of the 10 cipher rotors.
In the original German description, this function is called
Würfel-Verfahren (dice operation, or scrambling).
- Plugboard
The function of the various notched discs of the 10 cipher rotors can be
altered with the 20 plugs of the plugboard, that are arranged in a 2 × 10
pattern.
The configuration of the plugboard is specified in the current key list.
- Rotor start position
At the start of an encipherment, each of the 10 cipher rotors have to be set
to a given start position, as specified in the current key list.
- Message key
Each message must be encrypted with a unique message key, identified with
a trigram.
- KTF selector (mit/ohne)
This selector, which is not present on the T-52a/b/c, affects the stepping
pattern of the rotors. KTF stands for Klartekstfunktion
(clear text function).
It has two settings: 'mit' (with) and 'ohne' (without). When set to 'mit'
(with), the selected cleartext character affects the rotor stepping.
For this reason, it was generally recommended to set it to 'mit'.
|
Below is the block diagram of the T-52d. Note that for each character, 5
bits are required which are shown here as a single line.
The upper half shows the transmission path, which starts at the top right with
the (+) and (-) lines, representing the logical (1) and (0) levels.
These lines are fed to a key generator, described in the German
documentation as Sender Tauschen (ST), which means transmitter
swap. Although this is technically correct — it swaps the 1s and 0s
an arbitrary number of times — it is better described as a 5-bit pseudo-random
generator (PRNG). The ST key generator is formed by the rightmost five
rotors (N6-N10) of which the bit order can be swapped with 10 cables of the
plugboard.
The resulting bit pattern is then fed to the keyboard, which consists
of five single-pole double-throw (SPDT) switches (K1-K5). This effectively
adds the pseudo-random bit stream from the ST to the selected character by
means of an exclusive-OR operation (XOR). This principle is also known as
modulo-2 addition and as Vernam Cipher [7].
From the keyboard, the five bits are passed through a bit-swap
function, described in the German documentation as Sender Würfel (SW),
which means transmitter dice. It is controlled by the leftmost
five rotors (N1-N5) and offers 25 (32) possibilities 1 to swap the bits.
In addition, the bits can be swapped by means of 10 cables on the plugboard.
As there five plugs at either side of the SW block, this offers an extra
5! · 5! = 14400 static 2 settings. Next, the resulting five bits are
serialised by means of a rotating commutator, and then transmitted as a
serial data stream (TX).
|
| |
Block diagram of T52d shown in cipher mode
|
The lower half of the diagram shows the reception path. The serial data (RX) is
first converted to five individual bits, by means of a rotating
commutator and a 5-bit memory consisting of five capacitors. From there,
the bits are passed through a bit-swap function (EW), which is the inverse
of the bit-swap function (SW) in the transmission path. To maintain
reciprocity, the plugboards at either side of the EW are wired
identically to the SW plugboards. This is why each patch cable contains
four wires. 3 Next, the data is fed through the ET function, which
– including the plugboards at either side – is identical to the ST function
in the transmission path. It swaps the logic (1) and (0) levels an arbitrary
number of times, which is basically another XOR operation.
The output of the ET now holds the decrypted plaintext character,
but as this consists of just the charge of the memory capacitors, it does not
have enough energy to drive the printer. For this reason, the bits
are first copied into a latch, consisting of five polar relays 4 (R1-R5).
The data is then fed to the translator (German: Übersetzer) — a rotating
cam shaft with switches — which releases the print hammer at the right
moment, in synchronism with the rotating print head.
The machine (transmission and reception) can be switched
between Klar (plain) and Geheim (cipher). This is achieved with six switches
(KGa-f), each of which consists of a 5-pole double-throw switch (one
pole for each bit). This means that a total of 30 contacts must be swapped when
switching from plain to cipher.
The KG-switch is shown here in Geheim (cipher) position.
|
|
-
Theoretically there are 5! (120) possible bit orders, but due to the circuit
— there are just 5 switches with 2 position each — only 25 (32) orders are
used.
-
In this context, 'static' means that these settings are not changed during
encipherment.
-
Entering a plug into a socket of the plugboard, configures the transmission
path and the reception path simultaneously.
-
A polar relay is also known as a bistable relay. It can be driven by short
pulses (in this case the charge from a capacitor) and retains the last
setting, also when power is removed. In surviving T-52 machines, these
polar relays are the most common cause of reliability issues.
|
|
The operation of the T-52 is not easily explained. It is a very complex
machine and there are significant differences between the various
models. The basic principles are best explained by first looking at
the initial design and then introducing the improvements.
|
|
The T-52a was the first of the Siemens T-52 machines to see the
light of day. It had limited cipher security, mainly because it
exhibited regular stepping of the cipher wheels.
|
|
The T-52c was an attempt to improve security of the rather insecure
T-52a/b. A large box with 5 levers, used for setting the message key,
was added to the left of the keyboard.
However, the wheel combining logic, that was meant to improve
security, did exactly the opposite: it weakened the cryptographic
strength of the machine, as it reduced the total number of alphabets.
|
The image on the right shows the extremely rare T-52c variant.
The 5-lever unit for setting the message key is
clearly visible at the front left.
Also note the compatibility switch
at the front right. It allows the machine to be used in combination
with the older T-52a/b models.
According to the serial number plate,
it is a 50 baud version that
was built in 1944. This is remarkable, as by this time most
machines had been replaced by the improved
d and e models.
|
|
|
|
The T-52d was a well-designed machine. It was in fact a T-52a/b
with much improved cipher security. Its cryptographic strength was
considerably better than the Lorenz SZ-40.
Consequently, it was never broken by Swedish cryptanalists.
It was however broken by Bletchley Park,
but only if they had messages in depth.
If the T-52d had been used from the beginning, and its operators had
been better instructed, it seems unlikely that the machine would ever
have been broken.
|
|
The image on the right shows the interior of a T-52d machine, after
the black protective cover has been removed.
At the right hand side of the machine, just behind the paper tape
holder, the plugboard is clearly visible. This plugboard was normally
protected by a lockable hood.
|
|
|
|
Although the T-52 provided extremely good security for its days, it was
nevertheless broken during WWII. The first to break Siemens T-52a/b
traffic, was a group of Swedish cryptanalists led by Professor Arne
Beurling in June 1940.
The Swedish were lucky in that, after the occupation of Norway and Denmark,
the Germans started using land lines for their communication.
|
The lines ran through Sweden and as early as May 1940,
the Swedes tapped these lines and
intercepted all T-52 traffic to and from Oslo.
The machine was broken with hand methods, based on a set of messages
in depth, that were intercepted on 25 and 27 May 1940 [2].
Once the messages were broken and the key was known, the bulk of information
was deciphered on a T-52 emulator built by Vigo Waldemar Lindstein of the
Ericsson
company. After the war, Lindstein worked for
Hagelin for several years as head of the
Engineering division, until he started
Transvertex. He developed the
HC-9 cipher machine and was CEO of
Transvertex until the company was taken over by
Ericsson in 1969.
|
|
|
In 1943, the Germans discovered the weaknesses of the T-52a/b and c,
and were informed that their ciphers were being read by the Swedes.
They then introduced the T-52d, which featured irregular stepping of
the cipher wheels. It appeared to be too much for Beurling and his team.
|
|
The British cryptanalists at Bletchley Park (BP)
were less fortunate than their Swedish collegues, mainly because the
T-52 was primarily used over land lines, to which the British, unlike the
Swedes, had no access.
Over time however, the T-52 also occasionally appeared on radio links.
|
The first break by BP came in mid-1942.
In the summer of 1942, the Germans started using the T-52 on the radio link
between Sicily (Italy) and Libya, and later between Aegean and Sicily.
As the operators were sending multiple messages in depth (i.e.
with the same initial settings) BP cryptanalyst Michael Crum 1 managed
to achieve a break and reconstruct the machine [4].
BP called all German teleprinter traffic 'FISH'.
Whilst the Lorenz SZ-40
traffic was codenamed 'TUNNY',
the intercepted messages from the T-52 Geheimschreiber
were called 'STURGEON'.
|
|
|
Over time, all T-52 models encountered by BP, were eventually broken,
including the much improved T-52d,
but only if they had received messages in depth.
In practice, BP found that most messages that were sent over T-52 links,
were also sent via Enigma
or Lorenz SZ-40.
As they were better equipped to break these two,
the value of broken T-52 messages was limited.
|
The story of the T-52 does not finish at the end of WWII. Instead a large
number of machines found their way into the armies and security services
of a number of countries. The machine is known to have been used by the
French Foreign Office and by the Dutch Navy. The East-Germans intended to
use the machine, but it is uncertain whether they actually did.
Even the British Navy considered using the machine, but turned it
down for unknown reasons after a series of tests [4].
|
|
The French were by far the largest post-war user of the T-52.
Approximately 380 machines survived the war, 280 of which were left behind
in Germany. Although these machines had to be destroyed, they were 'just'
dismantled and the various parts ended up on the surplus market.
Around 1948, electromechanical firm
Willy Reichert
in Trier (Germany) bought large
amounts of the surplus stock and re-assembled a substantial quantity
of T-52d and T-52e machines. More than 235 re-assembled machines were
subsequently sold to the French Foreign Office between 1949 and 1953.
Some of these machines appeared on the surplus market again years later.
|
The Dutch Navy started using the T-52d during the 1950s. It is currently
unknown how long and for what purpose they were used, but it is most
likely that they were used on teleprinter lines between The Netherlands
and the Dutch East-Indies (Indonesia). It is also unclear at present who
supplied the machines to the Dutch. It's entirely possible that they
were also supplied by Reichert,
but they may also belong to the 100
machines that were not left behind in Germany after WWII.
|
|
A good simulation of a T-52d has been produced by the Crypto Simulation
Group (CSG). The simulator runs under Windows and is available from
Frode Weierud's website.
|
On 23 March 2000, the CSG even managed to interface the simulator to a real
T-52d machine, after which the compatibility of the simulator was confirmed.
For this they used the RS-232 port of the PC and a simple relay-based
telex interface.
The simulator is suitable for Windows.
➤ Download T-52d Simulator (off-site)
|
|
|
|
AME
|
|
Allgemeine Magnet Empfänger — General Magnet Receiver
|
|
AMS
|
|
Allgemeine Magnet Sender — General Magnet Transmitter
|
|
AMÜ
|
|
Allgemeine Magnet Übersetzer — General Magnet Translator
|
|
GEK
|
|
Generator Empfanger Kontakt Key — Generator Receiver Contact
|
|
KG
|
|
Klar/Geheim — Plain/Cipher
This lever selects between 'Klar' (plain) and 'Geheim' (cipher).
It consists of 30 double-throw switches that are mounted on a common axle.
|
|
KLI
|
|
Klinken — Socket
This are the 20 sockets of the plugboard, into which the plugs (Stöpsel)
must be inserted.
|
|
KTF
|
|
Klartekstfunktion — Clear text function
This function determines how the rotor stepping mechanism advances. It has
two settings: 'mit' (with) and 'ohne' (without). When set to 'mit',
rotor stepping is influenced by the selected plaintext character.
|
|
STÖ
|
|
Stöpsel — Plug
These are the 20 plugs, or jacks, at the end of the 20 patch cables,
that should be installed into the 20 sockets of the plugboard. Each pair of
cables are related to one of the ten rotors. Each plug has 4 wire contacts:
two for the transmitter and two for the receiver.
|
|
The following names were used to identify the T-52 and/or its traffic:
|
- Ttyp52
- T-52
- G-Schreiber
- Geheimschreiber
- Geheimfernschreiber
- SFM
- Schlüsselfernschreibmaschine
- Ln 25330
- Sägefisch 1
- Sturgeon 2
|
-
Sägefisch (sawfish) was the German name for the T52 traffic.
-
Sturgeon was the British name for the T52 traffic.
|
- T-52a
- T-52b
- T-52c
- T-52ca
- T-52d
- T-52e
T-52f 1
|
-
Never taken into production, as the war ended before the design was finised.
|
- DE615016 - 18 July 1930
First patent related to the T-52, filed by Siemens und Halske
in Berlin-Siemensstadt.
It lists August Jipp and Ehrhard Roßberg as the inventors.
It was filed on 18 July 1930 and published on 29 May 1935.
- US1912983 - 16 June 1931
The same patent was also filed in the US, where the device was
called 'Secret Telegraph System'. The patent was filed on
16 June 1931 and was published on 6 June 1933.
It lists August Jipp, Ehrhard Roßberg and Eberhard Hettler
as the inventors.
- DE591974 - 11 October 1930
This patent is similar to DE615015. It was filed later but has
a lower file number. The patent was approved 2 years before
DE615015.
- DE666436 - 13 September 1930
The is the patent for the additional security added to the
T-52d, the so-called KTF (Klartextfunction,
Clear Text Function). It was filed on 13 September 1930 and
was published on 29 September 1938. Eberhard Roßberg is listed
as the inventor.
|
|
-
This document is stamped by the German Kriegsmarine (Kommando der Marine
an der Nordsee),
and has an inlay of 12 January 1944, with stamps of the Dutch Navy.
-
These documents contain the signature of
Willi Reichert and were kindly supplied
by Wolfgang Mache. Included is a copy of the circuit diagram of the
T-36 teletypewriter,
dated 1 September 1932.
-
Document kindly provided by Günter Hütter [6].
|
-
Document kindly provided by Günter Hütter [6].
|
- Foundation for German Communication and Related Technologies
T-52d featured on this page courtesy Arthur Bauer.
- Bengt Beckman, Codebreakers,
Arne Beurling and the Swedish Crypto Program During WWII.
2002, ISBN 0-8218-2889-4.
Original Swedish Title: Svenska Kryptobedrifter.
1996, ISBN 91-0-056229-7.
- CG McKay and Bengt Beckman, Swedish signal intelligence 1900-1945
2003. ISBN 0-7146-5211-5 (hard cover).
- Frode Weierud, BP's Sturegeon, The FISH That Laid No Eggs
The Rutherford Journal, Volume 1, 2005-2006. PDF version, p. 29.
- Wolfgang Mache, Der Siemens-Geheimschreiber,
Ein Beitrag zur Geschichte der Telekommunikation,
1992: 60 Jahre Schlüsselfernschreibmaschine (German).
Archiv für deutsche Postgeschichte Heft 2, 1992, pp. 85-94.
➤ English translation 1
- Günter Hütter, T-52 documentation (listed above)
Austria, 2008-2021.
- Paul Reuvers and Marc Simons, The Vernam Cipher
Siemens & Halske, Wernerwerk. 11 August 2012.
|
-
Published in NCVA Cryptolog, Corvallis, Oregon, Winter 1990.
|
|
|
|
Any links shown in red are currently unavailable.
If you like the information on this website, why not make a donation?
© Crypto Museum. Created: Wednesday 05 August 2009. Last changed: Friday, 16 August 2024 - 13:58 CET.
|
|
|
|
|
![T-52d Geheimschreiber, courtesy Arthur Bauer [1]](img/300355/008/full.jpg)
![Layout of the T-52d Geheimschreiber. Courtesy Günter Hütter [6].](img/t52d_large.jpg)
