The TSEC/KL-7 is an American off-line crypto machine, developed by the Army Security Agency (ASA) and Armed Forces Security Agency (AFSA) under the name AFSAM-7, and introduced in 1953 by its successor, the National Security Agency (NSA). It was the first tactical lightweight rotor crypto machine with electronics, developed as standard crypto device. In 1955, the AFSAM-7 was renamed TSEC/KL-7. The machine was widely used by the U.S. military, its NATO partners, the CIA and FBI until the mid-1960s, and later mostly served as backup, often superenciphered on other systems, until its withdrawal from service in 1983.
The KL-7 machine is a 1950s marvel of technology that combined the latest cryptologic knowledge and electronics. Its cleverly designed keyboard changes the signal direction through the rotors. Notch rings on its rotors control the rotor movement in a most complex fashion and vacuum tubes control the timing signals of the newly designed miniature printer. The U.S. Navy KL-47 was cryptographically compatible to the KL-7 but had a different upper-case system.
The machine had excellent cryptographic properties and was designed to resist any cryptanalytic attack by the Soviets, even when its technical details were compromised. Unfortunately, some serious security breaches occurred in the 1960s. Both U.S. Army Signals Warrant Officer Joseph Helmich and Navy Chief Warrant Officer John Walker not only compromised the KL-7 design by selling the complete repair and maintenance manual to the Soviets, but also provided them with the secret key lists, adding even more to the Cold War myth surrounding this fascinating machine.
The history of the KL-7 and its technical specifications were kept secret for decades and information about the machine only gradually surfaced in recent years. This page is a tribute to the ASA, AFSA and NSA engineers and cryptologists who developed the KL-7 and the men who worked with this machine while serving their country in all corners of the world. On this website you can also download an accurate and realistic TSEC/KL-7 simulator.
The TSEC/KL-7 is an off-line non-reciprocal rotor cipher machine with electro-mechanical and electronic components (vacuum tubes). The machine measures 12 x 12 x 3.37 inches (30,5 x 30,5 x 16,2 cm) and weighs a mere 20.5 lbs (9,3 Kg). The KL-7 operates on 24 Volts DC for the motor-generator (DC motor that drives an AC generator).
The 24 Volts to power the KL-7 could either come from batteries, as available in vehicles, or from the mains. For use on mains there is a separate AC power converter that can be set to 100-125 V or 200-250 V at 50-60 Hz and its transformer with rectifier delivers 21 to 31V DC to the KL-7. The batteries or AC power converter output are attached to the connector of the 24 Volts power cord, located at the front-right of the machine (see also electronics section).
In 1955 the AFSAM-7 was renamed TSEC/KL-7, following NSA's nomenclature. TSEC stands for Telecommunications Security (COMSEC), K for Crypto equipment and L for Literal because the encrypted output is in letters (A through Z). For components of a machine, the TSEC comes after the item designation. Below the NSA naming of the main KL-7 parts, followed by the AFSA naming from its predecessor agency. See also NSA nomenclature table .
The other important parts are the mode selector and sliding contact board underneath keyboard to switch the signals through the rotors and select the different cipher modes, and the set of rotors to perform the encipherment and notch rings to control their movement. The pulse generator, the timing unit and vacuum tubes, three 2D21 thyratron and one 12AX7 double triode, provide the timing signals for the printer, which prints the text or code groups on a gummed paper tape.
Depressing a key will ground one of the pulse coils. The rotors encipher that signal on its way to the pulse generator. The encipherment is non-reciprocal and the sliding contact board must therefore switch the direction of the signal through the rotors to switch between enciphering and deciphering. When the armature of the pulse generator passes a grounded coil, it induces a pulse that is passed on to both the Gate vacuum tube and to the step-up transformer. The transformer sends the pulse to Sharpener tube, which cleans up the signal and sends it to the Print tube to activate the print hammer.
If the Shift tube is in FIG mode, it tells the Sharpener tube to only react on high pulses from figure coils and ignores any low pulse from letters. The single-shot Gate tube prevents double or stray pulses by disabling the Print tube after the desired single pulse is processed until the printing and rotor stepping is completed and ready for a new pulse.
The Print tube also activates the timing unit clutch, causing the timing unit to perform one single cycle in 0.1 seconds (600 rpm). This cycle controls four cam switches with timing signals for the letters and figures function, the print control circuit, and to prevent multiple cycling.
The notch rings on
the rotors are sensed by actuator switches of the
logic that activates
some of the stepping magnets. These magnets
enable drive pawls, and the timing unit provides
mechanical power to the drive pawls to move the
rotors. More details and the complete circuitry
are found in the electronics section.
The Contact Panel Assembly comprises spring-loaded contacts across the bottom plate that mate with contacts at the bottom of the keyboard, stepping unit and printer. This modular design allows easy disassembly of all main parts of the KL-7 for maintenance and repair.
To switch between enciphering and deciphering, the KL-7 has a simple and compact solution to change the direction of the signals through the rotors. The keyboard contains a large Sliding Contact Board which is operated by the Selector handle with the positions O-P-E-D (Off - Plain - Encipher - Decipher).
The sliding contact board with its T-shaped contacts moves from right to left between the keyboard keys and the spring-loaded contacts on the contact panel assembly. The keys are all grounded and depressing them will ground the T-shaped contact plate at the top of the sliding contact board.
The selector handle has a pawl that grasps into a vertical slot on the left of the contact board. Turning the selector clockwise moves the sliding contact board from right to left. Two rails on the sliding contact board press it down onto the spring-loaded pins, meanwhile ensuring smooth movement of the contact board.
The sliding contact board also contains the main power switch. The KL-7 power cable with connector (on the right) passes the radio interference filter (vertical block right-side machine), enters at the right-side top of the keyboard (see flipped keyboard) and connects to two slide contacts that touch two copper stripes (double-T contacts) at the far right of the contact board. The board also switches various parts of the electronics.
In Encipher mode, the depressed G key is connected via the T-shape and its GE contact to the EG (Encipher letter G) contact on the left side of the cipher unit that holds the rotors, and exits at the DH (Decipher H) contact on the right side, passes the small connecting bar of the H key and its goes from its Print contact to the H pulse coil.
In Decipher mode, the depressed H key is connected via the T-shape and its HD contact to the DH (Decipher letter H) contact on the right side of the cipher unit and exits at the EG (Decipher G) contact on the left side, passes the connecting bar of the G key and its goes from its Print contact to the G pulse coil.
The use of two neighboring T-shapes for each key enables the O-P-E-D sequence from right to left. The above is a simplified example with 3-pin rotors. In reality, the KL-7 uses 36-pin rotors. Note that, to perform the piggyback functions, some E, P and D connections from J, V, X, Y, Z, SPACE, FIG and LET are swapped, and additional contacts on the sliding contact board switch some piggyback wires and other control functions.
Note on direction of current! The arrows in the above image are only used to easily visualize how to keys are connected to the pulse coils. In reality, the current travels in the opposite direction! Each coil generates its own current when the armature of the pulse generator passes that coil. The positive side of the coil is connected through the key contacts and rotors to the common. However, this is not actually "grounding" the signal in the usual sense.
A pulse coil only uses the common to transport its signal to the step-up transformer, which inverts the signal before it is processed by the electronics. Within the circuit, each pulse coil can be considered as a separate power source with its own circuit.
When a key is depressed, and depending the Selector position, the positive side of the coil that represents its enciphered or deciphered version is connected to the general common. When energized, the coil sends its positive signal through the "common" and the step-up transformer (in opposite polarity) to end its route at the negative side of the same coil (see red arrows). See also the complete electronic circuit and timing signals.
How can depressing key G, connected to EG on the left of the rotor pack, ground the correct coil to print the encrypted letter H? For a given position and wiring of the rotors, when EG is connected to DH, then DH is also connected to EG. In fact, to encrypt the letter G, the KL-7 does not "encrypt" the letter G but grounds the letter G, and by doing so, the rotor pack grounds its counterpart DH, which is connected to the H coil that is energized by the pulse generator and sends the signal to print the letter H.
Since EG and DH are reciprocal, this works also for decryption. This however is only true for this particular position of the rotors, which changes after each letter of the message. This is what makes the KL-7's sliding contact board so clever. This is also why in the first example the black arrows are opposite to the actual direction of the current, because following the path, starting from the H coil, seems illogical when you must first depress key G.
The sliding contact board also switches some additional functions in the electronic circuit. Also, depending the selected mode, the signal from some keys, involved in the piggyback system, is rerouted. The board also has a notched part in front of the printer mechanism. With the mode selector in Encipher position, this cam pushes a pin into the printer mechanism, causing the KL-7 to print a space after each fifth character, the standard format for cryptograms.
* Note on contact problems. To pass the signal from a depressed key to a pulse coil requires many contact points to work flawlessly (a contact point is where two switching or sliding contacts make contact). In Plain mode, there are only two contact points: key to sliding contact board and from there to contact panel assembly p contact, connected directly to the pulse coil. In Encipher or Decipher mode, there are 13 contact points: one from key to sliding contact board, one to contact panel assembly, 9 through the rotors, and two contacts on the sliding contact board's small bridge, connected to the pulse coil.
However, for each time that the signal exits the rotor cage at one of the re-entry contacts, an additional pass through the rotors adds 9 contact points to the signal path. Theoretically, up to 10 additional passes are possible, thus adding 90 additional contact points. Also, using the LET or FIG key in Encipher or Decipher mode will add two additional contact points on the Letters Cam switch or Figures Cam Switch. In any mode, one single failing contact will prevent the machine from processing the character and stepping the rotors to process the next letter. This is the notorious so-called dead-rove, usually caused by poor maintenance. Regular and thorough cleaning of the rotors and keyboard contacts was therefore essential.
There are 324 contact points between the eight rotors mutually, and the rotor contact plates from the cipher unit. Between the keyboard and the top of the sliding contact board are 29 contact points (letter keys, FIG, LET and SPACE). Between the bottom of the sliding contact board and the top of the contact panel assembly are 119 contact points. In total, 472 contact points move (rotors, sliding contact board) or switch (keyboard keys). This does not take into account the numerous contacts between the bottom of the modular parts (motor-generator, pulse generator, stepping unit) and the top of the contact panel assembly. These are less prone to dirt because the modular parts are fixed onto the contact panel assembly and are only removed when a module is defective or requires maintenance. The only exception is the cipher unit, removed daily to load the new rotor settings.
An optional TSEC/HL-1 Tape Reader was available for the KL-7. In NSA nomenclature, the H means ancillary device that helps processing, and L means letters output. The HL-1 reads five-bit punched tapes with plain or ciphertext and sends it to the KL-7. This enables fast processing of message tapes, received from or created on a standard teleprinter, avoiding the laborious manual typing of messages on the KL-7.
The use of the HL-1 tape reader requires the KLX-7/TSEC Keyboard Adapter, mounted between the keyboard and the Sliding Contact Board. The installation of the KLX-7 therefore requires the removal of the KL-7 keyboard cover.
To install the KLX-7, the Mode Selector must be in Decipher position. The five screws of the keyboard are removed, the keyboard lifted without moving the Sliding Contact Board and the standoffs removed.
The KLX-7 is placed on the contact panel while ensuring its Selector handle extension engages in the notch of the Sliding Contact Board. The KLX-7 is then fixed with screws in its five hex nuts. Finally, five standoffs are placed in the hex nuts and the keyboard is fixed back onto the chassis by tightening its screws into the standoffs.
Once the KLX-7 is installed, its connector is located at the center front, just underneath the KL-7 keyboard. The KLX-7 board has 30 pressure contacts with their caps through the upper part of the board. Each of these contacts is connected to the KLX-7 connector.
Also, two contacts connect the two sliding 24 Volts power contacts from the KL-7 to the KLX-7 connector to pass the 24 Volts to the HL-1. Moreover, depending the KL-7 Mode Selector position, a 24 Volts signal is sent to the HL-1 to either process letters, figures and spaces in Plain and Encipher mode, or only letters in Decipher mode, and ignore all other 5-bit characters on the tape.
The HL-1 Tape Reader senses five-bit punched tape and converts the bits, through a series of relays, into connecting the common to one of the 26 letters, space, LET or FIG wire, and passes that common through the adapter cable to the KLX-7 adapter underneath the keyboard. This has the same effect on the KL-7 as depressing the keyboard key and grounding its corresponding pulse coil(s). Note that in reality the connector on the HL-1 side is twice the size of the connector on the KLX-7 side. The connector also has an alligator clip for grounding. Before 3-wire sockets became standard, this clip was fixed onto some metal part near the socket. More on the HL-1 tape reader at Jerry Proc's Crypto Machines and Nick England's US Navy Crypto Equipment.
The enciphering in the KL-7 is performed by the rotors. These are set according to the key list. Each individual rotor performs a substitution cipher. The rotor core wiring is still classified, and most surviving rotors are either stripped of their wirings or inaccessible.
Its wiring was also changed on a regular basis, in contrast to, for instance, the German Enigma where the wiring never changed during its whole service time (the U.S. cryptologists learned the lessons). The KL-7 rotors were easily rewired manually by simply plugging the contacts into other positions without the need of soldering. These were designated by NSA as Red Rotors. Due to persisting problems, these were replace by the Orange Rotors in 1956.
The KL-7 had a set of 12 rotors to choose from, later expanded to 13, labelled "A" through "M". (see CORE on part of the key list at the right). Each rotor core has 36 flat contacts on the left side (as seen in cipher unit windows) that are wired in a scrambled fashion with 36 spring-loaded contacts on the right side. The wiring performs a substitution encipherment. An adjustable black alphabet ring is attached to the rotor core. These letters are visible through the little windows of the cipher unit. The drive-pawls from the rotor stepping unit use the notches on the alphabet ring to advance the rotors.
The side of the alphabet ring is also marked with the numbers 1 through 36. By depressing the alphabet ring, it can rotate relative to the rotor core wirings. This is done by aligning an alphabet ring number with the white arrow on the side on the rotor (ALPHA RING SET). The 36 positions on all alphabet rings are labelled as show in the table below. Note that 10 of the positions on the alphabet ring are left blank.
The KL-7 also has set of 11 white plastic notch rings. The notches and cams on these rings control seven stepping switches in the rotor stepping unit. They are responsible for the highly irregular movement of the rotors. These notch rings were also part of the key settings.
The notch rings are labelled 1 through 11 on their side, next to a little black arrow (NOTCH RING & SET number). The notch ring is attached on any of the rotor cores by aligning the black arrow of the notch ring with the little hole near the edge of the rotor, then depress and align the notch ring with the required letter on the alphabet ring (NOTCH RING & SET letter). Since there are some blanks on the alphabet ring, these positions are marked on the key list as the letter at its left with plus sign (e.g. position 18 is M+).
As part of the key settings, a selection of seven notch rings are attached to the seven moving rotors. The stationary rotor at the fourth (non-moving) position in the cipher unit must always carry the special the wide ring and is aligned according to the key list (WIDE RING SET).
The rotors had persistent contact problems, caused by particles from the beryllium copper contacts that wore off and turned into non-conductive abrasive copper oxide. The beryllium copper dust particles are toxic and inhaling these should be strictly avoided. More on the rotors in the development section.
The detachable KLK-7 cipher unit or rotor cage holds the eight rotors that are selected from the key list. The KL-7 uses a complex re-entry system that can cause multiple enciphering of a single character. When the signal leaves the exit rotor there are two possible situations: the signal either is passed immediately to the pulse generator through one of the 26 wires, or it leaves the exit rotor on one of the 10 re-entry contacts. In the latter case, the signal is sent back to one of the 10 re-entry contacts at the entry rotor, to perform a new pass through the rotors. When the signal leaves the exit rotor again, the process is repeated.
Depending on the internal wiring and current position of the rotors, the signal performs one or more passes (theoretically up to 11 passes) through all rotors before leaving the exit rotor towards the pulse generator. This results in a most complex signal path that constantly changes in both number of passes and its way through the rotors. Note that the above drawing is a simplified example with three 6-pin rotors and 2 re-entry connections.
In reality, the cipher unit has eight 36-pins rotors and 10 re-entry wires. Also, the letters and re-entry wires are not connected straightforward to the end plates of the Cipher Unit but ordered in a scrambled fashion (see table base and end plate connections below). The combination of rotors and their positions creates a most complex signal path for each position of the rotors.
The cipher unit has a fixed end plate on the left with 36 spring-loaded contacts in a circular fashion, and a fixed spindle attached to it. The end plate on the right of the cage has 36 flat contacts in a circular fashion and is removable to load the rotors.
Both end plates have at the bottom (see image) a connector with 26 flat contacts that make contact with spring loaded pins on the Base, which connect with the 26 letters from the sliding contact board.
The end plates also have a connector with 10 flat contacts (re-entry) to connect with spring loaded contacts on the rotor stepping unit, which connects the wires of both sides straightforward between left and right (0-0, 1-1, 2-2 ...) to act as re-entry loops.
Once all rotors are inserted, the cipher unit is placed in the rotor stepping unit and fixed with two locking levers at the front left and right of the rotor stepping unit.
The table below shows the
wiring order between base and cipher unit end
plate pins. The pins are numbered clockwise (seen
from the left) and pin 01 is aligned with the
white index stripe on the cipher unit windows.
Both cipher unit end plates are wired
identically. The letter Q from the
sliding contact board wired to end plate pin 01,
letter P to pin 02 and so on. The
digits 0 through 9 represent the re-entry wires.
|End Plate Pins||01||02||03||04||05||06||07||08||09||10||11||12||13||14||15||16||17||18||19||20||21||22||23||24||25||26||27||28||29||30||31||32||33||34||35||36|
When a key is depressed, the pulse generator converts the signal that passed through the rotors and the keyboard into accurate timing signals for the electronics to control the printer and to activate the clutch of the timing unit which also provides mechanical power to the rotor stepping unit.
The print hammer must activate at the exact moment a given letter on the rotating printer drum passes the hammer. To achieve this, the pulse generator provides exact timing. He consists of a magnetic armature, rotating clockwise at 2200 rpm, and two stators with pulse coils. The print wheel with all letters and figures rotates on the same axle. When the electronics receive a pulse from a coil it will activate the print hammer at the appropriate moment. There are 37 coils for 26 letters, 10 figures and the space.
The front stator has 19 coils. The rear stator has 18 coils and is positioned exactly behind the front stator. The rotor armature (blue at above image) has two separate magnetic poles, one for each stator. The rear pole lags 9.47 degrees behind the front pole. This ensures that only one coil at a time is induced, front coil first and rear coil behind it next.
The KL-7 combines the 10 figures with 10 letters in the top row alpha-numeric keys. The electronics must therefore detect the difference between printing a letter or its corresponding figure. To do so, the alpha-numeric keys have two coils in series, a negative 5 Volts low pulse coil (green) for letters and a negative 10 Volts high pulse coil for figures (yellow). One special coil combination is the letter V low pulse coil in series with the SPACE high pulse coil and a junction between the coils. These are part of the piggyback system.
All other letters have a single negative 10 Volts high pulse coil (red). The difference between the low and high coils is the size of their core and winding. Note that the red and yellow high pulse coils are identical and only differ in color to distinguish letter and figure coils in the drawing.
Depressing a key grounds its corresponding coil, generating a transient pulse that activates the Gate tube. The Gate tube determines the time frame in which the Print tube is allowed to process a character pulse. This ensures that only the current pulse activates the Print tube, and any following stray pulses are ignored. For more details on the generation of pulses, see signal timing.
The common - negative - side of the pulse coils is connected to the step-up transformer. When the pulse generator armature passes the grounded coil, it will induce a pulse. The 1:3 step-up transformer inverts the negative pulse phase and raises the -5 to +15 Volts or -10 to +30 Volts to prepare the pulse for the Sharpener tube control grid (see images). The Sharpener tube passes the cleaned-up pulse to the Print tube.
To conduct, the Sharpener tube control grid potential must be higher than the tube's cathode potential. If the control grid potential is lower than the cathode, this represents a negative bias, and the Sharpener cannot conduct. To make the Sharpener conduct, the incoming positive pulse on its grid must overcome or "pull up" the negative bias enough. This negative bias acts as a threshold (i.e. hurdle) enabling the Sharpener tube to distinguish low and high pulses. More on how vacuum tubes work is found in the electronics section.
The Shift tube holds the current shift mode. In LET mode, the Shift tube puts 8 volts on the Sharpener cathode. Without pulse, the transformer grounds the Sharpener grid to 0 Volts, causing a minus 8 bias (0 - 8 = -8). In FIG mode, the Shift tube increases the Sharpener cathode potential to 22 Volts, causing a minus 22 bias (0 - 22 = -22).
In LET mode, the default minus 8 Volts bias is a low threshold and the Sharpener tube will accept both high pulse letters ( -8 + 30 = +22) and low pulse letters from the alpha-numeric double-pulse (-8 + 15 = + 7) but its associated high pulse figure, although high enough ( -8 + 30 = +22), does not trigger the Sharpener tube because the tube's 11 ms recovery time is longer than the 2.88 ms between the two pulses (only one coil distance, e.g. Q and 1).
In FIG mode, the Shift tube increases the bias to approximately minus 22 Volts, resulting in a higher threshold. The Sharpener tube then ignores the alpha-numeric low pulse letters as they don't pass the threshold (-22 + 15 = -7) and will only react on alpha-numeric high pulse figures (-22 + 30 = +8).
The following situations will occur when depressing any key:
The Gate tube prevents repeating or stray pulses, but not using this function also enables the repeat key. The RPT repeat key forces the Gate tube to keep the Print tube ready. When pressing the RPT simultaneous with another key, that character is printed continuously.
Depressing a key will always activate the printer and the timing unit clutch. Note that the space key only activates the clutch and advances the paper one step without printing, and the space and piggyback letters are swapped and print other characters, depending on the state of the sliding contact board (see also letters and figures).
The fixed pulse coils for 26 letters, 10 figures, a space and blank are placed clockwise on the pulse generator. with the double stator alternating between front coil first and rear coil behind it, in the order as shown below, starting with the Q front coil at the top (space and blank represented by a dash).
The KL-7 has a continuously clockwise rotating print drum at 2200 rpm, fixed on the same axle as the pulse generator. The print drum has the same character sequence on its circumference, but counterclockwise, which is the exact reverse order to the pulse coils.
This seems counterintuitive. However, seen from the front, the clockwise rotating print drum passes the print hammer at the bottom from right to left. The characters Q, P, W, etc. must therefore pass one by one at the bottom, from right to left. Thus, the characters Q, P, W, etc. are placed at the bottom of the print drum from left to right, i.e. counterclockwise.
The moment the magnetic armature of the pulse generator passes a grounded coil, the transformer, Sharpener and Print tube pass this signal to the print hammer and the timing unit clutch. The print hammer pushes the paper and inked ribbon upwards against the print drum, at the exact moment that the required character passes the print hammer. The two ink ribbon spools are behind the removable black cap.
A pin, controlled by the sliding contact board, mechanically switches between continuously printing (plaintext) and five-letter groups with a space between each group (ciphertext). The paper roll is stored in the black circular casing between the motor block and the cipher unit.
The timing unit controls four cam switches that generate timing signals for the electronics, provides mechanical power to advance the paper strip, the ink ribbon, and the pawl system of the rotor stepping unit to advance the rotors. The timing unit clutch is engaged by the trip magnet.
When the Print tube energizes the trip magnet, its rod* (yellow) is briefly removed from under the clutch pawl stop bulge* (light green). The spring-loaded clutch pawl can now move inward and its single tooth (light green) grabs into the clutch drive gear (dark blue), connecting the running DC motor through the gears to the camshaft. Note: in reality, the trip magnet is located directly below the Clutch.
Once a revolution is completed, the clutch pawl arrives back at its original position above the meanwhile returned rod, which now pushes the clutch pawl back upwards, causing the clutch pawl single tooth to disengage from the gear.
As the gears have a 11: 1 motor/timing ratio, the timing unit performs one single revolution at 600 rpm in 0.1 second. During this cycle, the four cam switches send their timing signals to the electronics, and the paper strip and ink ribbon move one step further. The rotation of the timing unit and cam switches is in sync with all other mechanical processes of the KL-7.
Also during this cycle, only those rotors that have been released by their associated stepping magnet are advanced one step. Thanks to the timing signals, the electronics know when the mechanical cycle has been completed.
The cam switches on the timing shaft control the following functions:
More about the cam switches in the electronics section.
The motor-generator, providing both mechanical power and AC voltage, is a combination of a 24 Volts DC motor and a 400 cycle AC power generator delivering 5 Watts and 150 Volts rms at 6600 rpm, together in one single housing.
The motor exists in two versions. The CE 87420 was replaced from serial number 13149 by the improved CE 88000. The difference is how the current through the field winding is controlled.
The components to control the motor speed are not included in the general electronics because the controller assembly is housed inside a cover at the back of the motor-generator (see image, resistors in ohm, capacitor in microfarad).
The CE 87420 is a shunt wound DC motor where the armature and shunt field are connected parallel, which typically produces a modest starting torque but has good speed stability under varying load because its speed is self-regulated.
When the speed of the rotating armature drops, the counter-electromotive force (CEMF) also drops and more current can flow through the armature, increasing torque and motor speed. When speed rises, the CEMF also rises and less current can flow through the armature, decreasing motor speed. This feedback creates speed stability, regardless the load.
The motor speed is adjustable by changing the current through the field winding, which also changes the amount of CEMF and thus motor speed. This is done by the adjustable resistor R202 to compensate manufacturing variations of the field winding and the NTC resistor R201 (Negative Temperature Coefficient).
The newer CE 88000 is also a shunt wound DC motor with self-regulating speed, but has an additional centrifugal governor switch to maintain a constant speed while operating at voltages ranging from 21 to 31 Volts DC. The Governor speed adjustment is set at 6600 +/- 100 rpm.
The governor switch is open when motor speed is low. Resistors R135 and R136 are then in series, causing a lower current through the field winding. The resulting lower CEMF speeds up the motor. When the motor reaches its optimal speed, the governor closes and short-circuits R135. The resulting higher current through the field winding causes a higher CEMF that slows down the motor.
The motor speed is adjustable with the upper screw on the governor. The lower screw adjusts the air gap between the contacts. Capacitor C121 suppresses sparks on the governor and R136 limits current surge when the governor closes.
The AC generator is mounted in front of
the motor on the same shaft. The generator rotor is a
permanent magnet that is also fixed to the shaft. The
rotating magnet induces the high-voltage in the
stationary field coils.
The KLA-7 rotor stepping unit supports the cipher unit (rotor cage) and controls the stepping of the rotors. At the rear of the cradle, there are eight large stop pawls that prevent any non-moving rotor from moving along with neighboring moving rotors. The fourth stop pawl normally isn't used, but keeps that fourth rotor in place when testing the rotors on a special axle without the rotor cage shell.
In the middle of the cradle are seven drive pawls to advance the rotors by grasping the notches in the rotor alphabet ring. These pawls are mechanically powered by the timing unit, under control of seven stepping magnets from the stepping logic.
In front of the drive pawls are seven tiny cams from actuator switches that read the notch rings on the rotors. These pile-up switches control the stepping logic for the stepping magnets. Just before the cradle are seven Set Key buttons to manually advance each individual rotor.
When a key is depressed, the pulse generator passes the signal to the Print tube which in turn will activate the timing unit that executes a single cycle. The stepping of the rotors is mechanically driven by the timing unit side shaft that rotates the crankshaft assembly for the rotor stepping unit.
The crankshaft assembly has an eccentric crank pin that is connect to the drive link assembly and converts the rotation into a back-and-forth movement that is passed to the drive shaft by the rotor stepping crank. There are 7 ball-bearing roller cams fixed to the drive shaft, that's one for each moving rotor (forth rotor doesnt move).
The roller cams hold the drive linkage forward. When the timing unit cycles, the drive shaft rotates 75 degrees in clockwise direction and the 7 rollers move downwards, away from the 7 drive linkages. Each drive linkage has a spring that pulls the linkage backwards. However, when the stepping magnet of an associated drive linkage is not energized, the magnet prevents that drive linkage from moving backwards.
During a cycle, some of the stepping magnets are energized by the stepping logic, under control of the 7 actuator switches. When a stepping magnet is energized and the drive shaft turns the roller cams away from drive linkages, the associated drive linkage moves backwards, causing its drive pawl to move backwards.
At the same time, the drive linkage also releases the associated stop pawl, which is then pushed downwards by the stop pawl spring to enable its rotor to move. Simultaneously, the drive pawl sloped end slides backwards out of the alphabet ring notch. When the drive pawl arrives at the next notch, the small spring on the drive linkage will force the drive pawl upward into the next alphabet ring notch.
When the drive shaft has reached it maximum swing angle backwards, the roller cams return forward and push any released drive linkages back to the forward position. The flat end of the drive pawl pushes its rotor one step further. The drive linkage also forces the stop pawl upward into a free alphabet ring notch. During any stepping cycle, these stop pawls prevent non-moving rotors from moving along by the friction of moving neighboring rotors.
On the right of the drawing are the 7 actuator switches that sense the white notch rings on the rotors. In the Plain mode, the operator can use the 7 Set Key buttons to manually step the individual rotors and set them in the proper position (see procedures enciphering and deciphering). During enciphering and deciphering, each key stroke causes some of the rotors to step in a highly irregular fashion under control of the 7 actuator switches.
In the circuit diagram below, all actuator switches are shown inactive. Each switch is a pile-up of the two sections, Sa and Sb. Note that the order of the upper Sa section is as actually positioned in the rotor stepping unit. The order of the lower Sb section and the stepping electromagnets is mixed to make the circuit diagram more readable, and these are actually placed from left to right according to their number. There are only 7 switches and stepping magnets because the stationary fourth rotor is skipped. The pins on all switches are numbered the same as on switch S301a and b.
The stepping circuit is powered by the same 24 Volts DC that powers the DC motor and the tube filaments. The switch is part of the sliding contact board and determines which part of the circuit is powered, depending on the cipher mode.
In Encipher (E as shown in drawing) and Decipher (D) mode, only the upper Sa section of the switch pile-up, used for sensing the notch rings, is powered (route K1 - K12 or K10 - K11 - A3 - B3). Magnets 2 through 6 are each controlled by two switches in OR logic. These magnets are activated when at least one of the two switches has the appropriate state. Magnets 1 and 7 are controlled by three switches and are activated when one switch is inactive AND at least one of two other switches is inactive. At least two magnets are always active at any given moment.
In Plain mode, only the lower Sb section is powered (route K1 - K12 - K13 - A4 - B4). Depressing a Set Key button first energizes its stepping magnet through contacts 4 and 2 to release the corresponding drive linkage. A fraction later, contact 4 is pushed against contact 1 that energizes the trip magnet, causing the timing unit to perform a cycle and provide mechanical power to step the corresponding rotor. If the Set Key button is still held down after completing the cycle, the rotor will continue to step until the button is released.
The stepping logic is wired in such way that a situation where none of the rotors move is avoided as this would cause the rotors to halt permanently. The logic table on the right shows the required notch combinations to move a rotor. The stepping of a single rotor is controlled by two or three separate notch rings. Two notch rings can produce a maximum period (unique movement sequence) of 1,296 and three rings a maximum period of 46,656. This is for one single rotor. The combination of seven notch rings therefore provides a most complex stepping sequence.
To enable enciphering 37 different characters into letter-only code groups, the KL-7 uses a system, similar to the teletype code. Two signals, LET and FIG, switch the machine between letters and figures. Both character sets use the same signals and they are distinguished only by the FIG or LET mode on that particular moment. To show the current state, a neon bulb lights up when the KL-7 is in FIG mode. The characters QWERTYUIOP are processed as 1234567890 in FIG mode.
This still gives 26 alpha (-numeric) keys, the additional SPACE, LET and FIG. The KL-7 must encipher these three additional characters into a letters. Therefore, the KL-7 design permits the special functions to piggyback on some of the existing alphabet letters. The letters J, V, X, Y and Z were selected because they are some of the less frequently used letters.
To accomplish the piggyback of the SPACE key on letter Z, the sliding contact board reroutes the X, Z and SPACE signals. The following drawings explain the rerouting of those letter in the different modes.
1. Below, any unspecified letter, key
or pulse coil (A through Z) is denoted by Δ (delta). Underneath each key
are three contacts, named e-p-d (encipher - plain -
decipher. Thus, the contacts underneath the A key are
named Ae, Ap and Ad. The left side contacts of the cipher unit that holds the rotors are named EA through EZ
(Encipher this letter) and the right-side contacts are
named DA through DZ (Decipher this letter). See sliding contact board for routing through rotors.
2. The actual direction of the current is opposite to the direction of arrows in the diagrams, because the pulse coils produce a negative pulse. The arrows are used solely to demonstrate the path from key to transformer. See pulse timing signals for more details.
In plain mode, the p-contact of each key is connected to its corresponding pulse coil (black route). Zp to the Z coil and Xp to the X coil. Both are high pulse coils (red). Depressing the letter Z or X will close the circuit of its corresponding coil. When the magnetic armature of the pulse generator passes that coil, the induced signal is stepped up by transformer T101, passed to the sharpener and print tube, and printed.
The SPACE key is connected to a junction on the V-SPACE coils pair. This special feature is also used in the separate FIG circuit. By using the junction, depressing the SPACE key will only close the SPACE coil circuit.
All other keys are also connected via their p-contact to the corresponding pulse coil. These coils can be high pulse letter coils (red) or low pulse alpha-numeric coils (green). The low pulse V coil from the V-SPACE pair is connected to the V key's Vp contact.
In plain mode, the signals do not pass
through the rotors in the cipher unit.
In encipher mode, the e-contact of each key is normally connected to the corresponding E-contact on the cipher unit left side. This however is not the case with the piggyback Z key.
The Z and X key e-contacts (red route) are now both switched to the EX-contact (Encipher X) on the cipher unit left side. This doesn't affect the X, but the letter Z also becomes X, because the SPACE needs the Z to piggyback.
Therefore, the SPACE key's e-contact (blue route) is connected to the cipher unit left side EZ-contact and actually enciphers the letter Z that will represent the SPACE.
The enciphered version of all letters exits the cipher unit right side at one of the D-contacts. This can be any of the 26 letters DA through DZ, even the depressed key itself, depending on the rotor positions and wiring.
The signal then travels to the key that
corresponds with the cipher unit's D-contact, enters that
key's d-contact, passes its small connecting bar and its
p-contact to the pulse coil and prints the enciphered
In decipher mode, the d-contact of each key is connected to the corresponding D-contact on the cipher unit right side (black route). The only exception is the SPACE key, which is not used, as only letters are to be deciphered.
The deciphered version of all letters exits the cipher unit left side at one of the E-contacts. This can be the any of the 26 contacts EA through AZ, even the depressed key itself, depending on the rotor positions and wiring.
The signal then travels from that E-contact to the key that corresponds with the deciphered letter, enters that key's e-contact, passes its connecting bar and p-contact to the pulse coil, and prints the deciphered letter.
When a depressed key represents the enciphered X, the signal exits the cipher unit as deciphered letter X on the left side EX-contact. The signal (red route) then travels to the Xe contact and via a connecting bar to the X coil, to print the X. Thus, both originally plain X, and the Z (changed into X) are deciphered and printed as X.
When the depressed key represents the enciphered Z, the signal exits the cipher unit left side as deciphered Z at the EZ-contact. Since the deciphered Z actually represents a SPACE, the signal (blue route) does not travel to the Ze contact, but via the SPACE Key's e-contact and connecting bar to the SPACE coil to print a space.
The KL-7 can be powered in two ways. The Power Cable, connected to a 24 volts DC source (e.g. vehicle battery) or the AC Power Converter that converts 110V or 220V AC into 21 to 31 volts DC (24 volts is ideal). Both power cable and converter have an Amphenol AN 3101A Mil Spec connector (2 pins female) with AN 3057-4 cable clamp that mates with the KL-7's power cable with male connector.
The AC Power Converter has a switch to select the AC input voltage. For 100-125 volts, the two primary windings are switched parallel (routes 1w2-11-14-15 and 1-13-12-3w4). For 200-250 volts, the primary windings are switched in series (route 1w2-11-12-3w4). Since the primary windings switched in series have twice as many turns as in parallel, the output voltage is inversely halved and thus remains 24 Volts.
The secondary winding is connected to two selenium full-wave bridge rectifier stacks connected parallel. The first version of the power converter produced 2.4 Amps output current. The newer version with 4.5 Amps (as of sn15413) is more efficient for the KL-7 with second generation motor-generator with governor switch. The higher current is also better when the selenium rectifiers are aging. The AC Power Converter measures 10.6 x 4.7 x 4 inches (27 x 12 x 10 cm).
The KL-7 itself has a power cable with AN 3106A connector (2 pins male), connected to the Radio Interference Filter which is mounted in the right-side rack (see photo mode selector). From the filter, a short cable enters the KL-7 at the right-side top of the keyboard. Via two fuses and two slide contacts, the 24 volts arrive at the double T contacts (see also circuit diagram below). The rack also carries a dummy female connector to secure the power cable during transport.
The filter suppresses voltage variations on the power cable as these can cause interference on the mains grid or even piggy-back on the modulated signals of nearby radio equipment, leaking signals that could be exploited by the adversary, even at great distance. The Pi-type filter attenuates these signals over a wide band of frequencies. The filter assembly was sealed, and maintenance or repair was not allowed. If the filter malfunctioned, the complete filter assembly had to be replaced. More about the security risks of stray signals is found on our TEMPEST page.
To follow the KL-7 circuit diagram below, we need to know the type of vacuum tubes and how they function. These are crucial to process the timing signals for the printer. The KL-7 has one 12AX7 and three 2D21 tubes. These were produced by many manufacturers like RCA, General Electric, Brimar or Mullard. For military supply contracts, the tubes were designated JAN 12AX7 and JAN 2D21 (Joint Army-Navy electronics standard nomenclature).
The 12AX7 Twin Triode consists of two separate triodes (three electrodes each) making this tube very suitable for use as multivibrator. For each triode, the current between its positive anode, called plate (pin 1 and 6), and the cathode (3 and 8) is controlled by a very small voltage on the control grid (2 and 7). The more positive the control grid bias is, the more current flows from plate to cathode. A negative control grid bias stops the flow of current. Note that electrons flow in opposite direction from the conventional flow of current. A positive control grid bias (2 or 7) will therefore attract the electrons from the cathode (3 or 8) and lead them towards the plate (1 or 6). The 12AX7 also has two separate heater filaments in series (4 to 5) with a center-tap (9), making it possible to operate the filaments parallel at 6.3 volts, as used in the KL-7, or in series at 12.6 volts (data sheet).
Although the physics are quite different, you could compare the triode with the modern NPN transistor, where the current from collector to emitter is controlled by the base.
The 2D21 Screen-grid Thyratron works quite differently from a regular triode or tetrode vacuum tube. This thyratron is a xenon gas filled tetrode (four electrodes) that operates as electronic switch and can handle higher currents than vacuum tubes. A negative bias on the screen grid (pin 5 or 7) will repel the electrons from the cathode (2) and prevent them from reaching the plate (6) and firing the tube, regardless the bias on the control grid (1). When both control grid and screen grid bias are positive, the xenon gas ionizes and the electrons multiply (i.e. Townsend discharge). The resulting plasma creates a high conductivity (virtually a short-circuit) between the plate (6) and the cathode (2). Once the current flows, the only way to stop that flow is to either reduce the plate voltage below the critical threshold or to cut off the plate current completely, as for instance done by the KL-7's timing unit Charge Cam switch. The heater filament operates at 6.3 volts (data sheet).
The 2D21 therefore acts as a fast switch, either on or off, and you could compare the thyratron with a relay switch, or with the modern thyristor or FET transistor where the current is switched by its gate voltage. Note that a regular (non-gas filled) tetrode does not act as a switch but works similar to a triode to which a screen grid is added. See also how vacuum tubes work at Mr Carlson's Youtube video.
Below the full circuit diagram. All electronics are located underneath the Contact Panel Assembly with spring loaded contacts, except the vacuum tubes, located behind the Rotor Stepping Unit. The components in the circuit diagram are labeled in the same way as the original maintenance manual and circuit diagram. Since the original diagram is pretty hard to read, the different sections of the electronics are presented here well separated. The orange dots represent parts of the sliding contact board.
At the top left of the circuit diagram is the power supply for all electronics. The 24 Volts DC arrives at two double T contacts on the sliding contact board and powers the Motor-Generator that provides mechanical power and high-voltage. The 24 Volts that power the DC motor is also used for the filaments of the four vacuum tubes and the stepping magnets in the rotor stepping unit.
The full wave bridge rectifier, followed by two diodes, provides +220 Volts at terminal W4. The bridge rectifier + is also directly connected through terminal W8 to resistors R109 and R110 to provide +200 Volts at terminal W2 for some parts of the circuit. A half wave rectifier provides the negative 72 Volts.
RCA 12AX7 Twin Triode and 2D21 Thyratron
Original General Electric JAN 12AX7 WA
Joint Army-Navy electronics standard
© Photo TubeDepot
The pulse generator in the center left converts a key stroke into one or two negative pulses when that pulse coil is grounded and induced by a rotating magnetic armature. 15 letter keys have a single high pulse coil. 10 alpha-numeric keys have a low pulse and high pulse coil in series. For the piggyback system, the V key has a low pulse coil in series with the SPACE key high pulse coil and includes a junction.
When a key is depressed, a negative pulse triggers the V104 Gate tube, which acts as single-shot multivibrator that controls the Print tube screen grid bias, to limit when the Print tube is allowed to activate the printer. Next, when the armature passes the coil of the depressed key, a negative pulse is sent to the T101 step-up transformer in the center, which inverts the pulse (see also signal timing).
In the center of the diagram is the V102 Sharpener tube that receives the pulses from the step-up transformer. The Sharpener serves as threshold to distinguish low and high pulses. When V102 is not ionized, its cathode has the same potential as terminal W5, determined by the V103 Shift tube via the R119 - R122 - R123 resistor network.
In LET mode, a low pulse is sufficient to overcome the minus 8 volt bias on the Sharpener tube control grid. When in FIG mode, the increased voltage drop between terminal W5 and common makes the Sharpener control grid bias more negative, requiring a high pulse to fire the sharpener.
The V101 Print tube only fires when the triggered Gate tube makes the Print tube screen grid bias positive, and the Sharpener tube pulse simultaneously makes the Print tube control grid bias positive.
When the Print tube has activated the timing unit, the Repeat Cam switch disconnects the positive voltage at the Gate tube A-plate anode, turning the Print tube screen grid bias negative. This prevents the Print tube from firing successive pulses or stray pulses from keyboard or from moving rotor pins.
Once the cycle and rotor stepping is completed, the Repeat Cam switch closes again and both Gate and Print tube are ready for the next cycle.
When the RPT key (repeat) is depressed, the negative bias on terminal X1 and on the Print tube screen grid is removed and the tube is ready to fire at any pulse it receives from the Sharpener tube. Depressing the RPT key together with a letter key will therefore continuously print the depressed plain or enciphered letter.
When the Print tube fires, the print and trip magnet are activated. Both magnets are energized in a quite special way. The Print capacitors C101 & C102 and Trip capacitors C106 & C107 are constantly charged though R109 and R110 to +200 Volts. Since charged capacitors don't draw current, no current flows through the Trip Magnet circuit. When the Print tube fires, this shorts the circuit and discharges the capacitors though the tube-print-trip circuit, in series with the capacitors. Their discharge current energized the Trip Magnet.
This activates the timing unit to perform one cycle, and switches on the camshaft provide four timing signals. During this cycle, the Charge Cam switch shorts the 10K resistor R110, leaving only the 1K resistor R109 between the 220 Volts at W8 and the capacitors at W2, ensuring a fast recharge of the capacitors to be ready for the next machine cycle. The Charge Cam switch also disables the Print tube until the cycle is finished, preventing the Print tube from firing multiple times during a cycle.
At the bottom of the circuit diagram is the Shift tube section that memorizes the current shift mode, shows the current mode with a neon lamp and controls the threshold of the Sharpener to know what signals to accept in LET or FIG mode.
The switches with an orange dot in the above circuit diagram represent parts of the sliding contact board and shows how they influence the stepping of the rotors, the shift mode circuit and some of the keys. How those switches work is shown below.
The letters cam switch and figures cam switch from the timing unit control how the piggyback letters for LET and FIG keys are processed. Since we have the Plain, Encipher and Decipher mode and two shift modes, there are six different ways these keys are processed. To accomplish this, the sliding contact board reroutes the FIG and LET keys, and the figures and letters cam switches.
Note: Below, the unspecified key, letter or pulse coil (A through Z) is denoted by Δ (delta). Also, the actual circuit uses a few more switches on the sliding contact board to avoid conflict with other key circuits in the different modes, but the route of the signals is as shown below.
The FIG key operation:
The LET key operation:
The reason for printing a space when deciphering the LET or FIG key instead of only changing the Shift mode is that activation of the Timing Unit is required for the timing signals and stepping of the rotors to correctly encipher or decipher the following letter.
The timing signals for the electronics are generated in two different ways. When depressing a key, the corresponding pulse generator coil is grounded, causing a first pulse. This pulse in not generated by the pulse generator itself, but by the key at the moment he closes the circuit! While the key is still depressed, the pulse generator armature passes the grounded coil and induces a second pulse (which could be single or double) that initiates the print and rotor stepping cycle. On the right the relevant parts of the circuit (see also the full circuit diagram).
The Pulse Generator coils' common (test point V14) is connected through R131 and R114 to the +220 Volts. Therefore, depressing any key connects its pulse coil, which has low DC resistance, to the general commons of all electronics. This causes a large voltage drop across that pulse coil and the resulting counter-electromotive force (CEMF) makes the pulse coils' common negative to the general common of the electronics. This negative pulse is the so-called "transient pulse".
The voltage drop on key closure is passed through R131, terminal X3 and C105 to the Gate tube B section's cathode, causing A section to cut off and B section to conduct. The Gate tube single-shot multivibrator is now triggered, but the voltage drop also discharged C116, causing terminal X1 to stay 10 milliseconds longer negative. Terminal X1 is connected to the Print tube's screen grid, thus preventing a little longer the Print Tube from firing.
The negative 10 Volts transient pulse is also inverted by the transformer into a positive 30 Volts and sent to the Sharpener tube. However, the processed pulse will not initiate printing because the Print tube's screen grid bias is still negative due to the 10 milliseconds delay at terminal X1. The transient pulse thus only initiated the multivibrator.
When the 10 ms delays has passed, and the key is still pressed down, the rotating pulse generator's armature will eventually pass the grounded pulse coil. This creates again a negative 10 Volts pulse that enters the transformer (see direction i arrow) which in turn inverts the pulse into a +30 Volts pulse and fires the Sharpener tube. This is the actual pulse of a letter or the initial pulse of an alpha-numeric double pulse.
Since the Print tube's screen grid bias is no longer negative, the Sharpener can now fire the Print tube. Note that after the pulse armature has passed the grounded coil, the decaying magnetic flux also induces a current in the opposite direction, but that pulse is blocked by the three diodes.
Firing the Print tube now energizes both the Printer Magnet to print the corresponding character and the Trip Magnet to rotate the Timing Unit that controls the cam switches. The Repeat Cam switch then cuts off the Gate tube A section's plate to prevent the multivibrator from completing a full cycle of pulse sensing, thus blocking stray pulses from the moving rotor's contacts in the Stepping Unit. More about the route of the signals through the rotors is found in the Sliding Contact Board section.
We can calculate the theoretical strength of the KL-7 by taking all cryptographic variables for a complete machine set-up. We assume the machines general principle of operation is known, but don't have any information about the internal wiring of the rotors and shape of the notch rings.
The 8 rotor cores can be wired in 3.66322 different ways. This comprises all positions, relative to the machine, making the alphabet ring superfluous, because it only serves as visual representation of the rotor alignment. This also comprises all positions of the stationary 4th core, set by its wide ring.
The notch rings can be shaped in 7.2375 different ways. This comprises all combinations of notches, relative to the seven actuator switches. This gives a total of 2.65408 purely cryptographic combinations or 1357 bit key. However, the machine and its wiring can be known to the adversary.
Therefore, we look at the practical settings for the operator. He must select 8 rotors from a set of 13, giving 51,891,840 combinations. He has 78,364,164,096 ways to set 7 alphabet rings. There are 1,663,200 ways to select 7 notch rings from a set of 11. The 7 notch rings and the wide ring on the 4th position can be set in 2,821,109,907,456 different ways.
Finally, there are 78,364,164,096 possible rotor alignments at the start of a message. In total, this gives 1.4948 ways to determine a key setting, both internal and external. This represents a 161 bit key.
When the machines specifications are known to the adversary (espionage, capture) he has to find 8 cores from a set of 13, giving 51,891,840 combinations. There are 1,663,200 ways to combine 7 notch rings from a set of 11.
There are 2,821,109,907,456 ways to set 7 notch rings and 1 wide ring. There are 78,364,164,096 ways to set all core and notch combinations, relative to the machine. The alphabet ring, only a visual representation of the rotor positions, is disregarded. This gives the adversary a total of 1.9037 combinations, representing a 124 bit key.
Trying out all possible keys on a 124 bit key, a so-called brute-force attack, is considered infeasible with all present and future computer power. However, cryptanalysis is more than key size, theoretical security and brute-force attacks. Rotor cipher machines have proven vulnerable to certain types of cryptanalytic attacks, performed on fast computers. Therefore, the KL-7 is no longer considered secure. Nevertheless, even skilled cryptanalysts with current resources would today still face a huge task to mount a successful attack against the KL-7, especially when they only have a limited number of messages at their disposal.
Encrypting a large number of messages with the same key always poses a risk, as there is more chance of messages being in depth, giving the cryptanalyst more statistical clues about that specific key setting. During the KL-7s service time, the numbers of message encrypted with the same key settings was limited by compartmentalization. For a given day, instead of using one key list for all users, different key lists were distributed to different echelons and services, thus reducing the use of one specific key setting. An additional method to limit "in depth" messages is to encrypt each message with a unique random starting position of the rotors, called Message Rotor Alignment. More in the message procedures below.
There were two types of operation for the KL-7, called POLLUX (1950) for low-level Army and Air Force traffic, and ADONIS (1951) for high-level traffic. The difference between them was the set of rotors and the method to determine the random starting position of the rotors for each new message, called Message Rotor Alignment.
Each ADONIS rotor set consists of 12 (later 13) rotor cores, 11 notch rings and one stationary wide ring. Before using the KL-7, the internal settings must be set according to a key list that contains multiple keys, each valid for a period of 24 hours. Each daily key list determines the choice of rotor cores and its position in the cipher unit, the setting of its alphabet ring, and the type of notch ring and its setting on that rotor core. The rotor core that is selected for the fourth position must always be fitted with the stationary wide ring, which can also be set in any position on that rotor core. This wide ring prevents the fourth rotor from moving.
Below, the ADONIS key list example, in the format as documented in the declassified KAO-41C/TSEC. Usually, a second cipher unit, with the rotors arranged according to the key list of the previous day, was readily available. When they received a message from the previous day, the operator could quickly remove the current cipher unit from the rotor stepping unit and replace it with the old cipher unit.
Note: you can set the example key for day 31 on the KL-7 simulator and perform the encipher example shown below, to experience the encipher procedure as actually used by the operators. Be advised that the rotors of the simulator have their own core wirings because the secret wiring of the real rotors was never declassified, and the wiring of surviving rotors has been removed.
After placing all prepared rotors in the cipher unit, the new settings are checked with the 36-45 letters check. With the mode selector in Plain mode, all rotors are set in the A position. Next, the machine is set to Encipher mode (some rotors will move one step). The counter is reset, and the letter "L" is typed 45 times. The last two code groups should match the 36-45 letter check on the key list for that day.
The system indicator, shown at the bottom-right of the key for that day, consisted of five digits (ADONIS) or five letters (POLLUX) and was used before and after the enciphered message to identify the system and key.
For each new message, the operator must use a new unique and completely random start positions for the rotors. The method to convey these rotor positions to the receiver is called the Indicator System. The random start positions, called Message Rotor Alignment, is visible through the little windows of the cipher unit and crucial for the security of the message because using the same Message Rotor Alignment for multiple messages leads to patterns that can be exploited by codebreakers.
To produce a Message Rotor Alignment, the operator takes a previously prepared random five-letter group called Message Indicator (see note further down). For our example message, we use the random message indicator "ELXNO".
1. Set the KL-7 mode selector to Plain
2. Set AAA-AAAA as rotor alignment with the Set Key buttons on the front of the cipher unit.
3. Switch to Encipher mode (some rotors will move one step)
4. Encipher the Message Indicator "ELXNO". The resulting enciphered indicator will be "BHLDO".
5. Switch back to Plain mode.
6. Set "BHL-DOBH" as Message Rotor Alignment by re-using the first two letters at the end. Again use the seven Set Key buttons to adjust the rotors.
7. Switch to Encipher mode (some rotors will move one step), tear of the printed tape and reset the counter.
8. Encipher the example plaintext message "TOP SECRET MESSAGE 123 TEST".
Any incomplete final group should be completed by one space and, if required, followed by enough random letters to complete that last code group.
Below, the resulting ciphertext. Note that if you enciphered the message with the KL-7 sim, the result might partly differ, depending on whether you switched to FIG and LET mode before or after the space.
The complete message includes the system indicator, the Message Indicator spelled-out (NATO alphabet), the enciphered text and repeated system indicator:
The standard format for enciphered messages was the so-called CODRESS, documented in the publication ACP 127 (unclassified). In such messages, the full originator, all addressees and security classification were included in the enciphered text. These messages were always unclassified, although the coded groups contained secret information.
Below our example message with the priority pro-sign (here routine), routing indicator(s) of the addressee(s) and serial number, priority again with date and time group, groups count, and finally the message between two BT breaks.
The receiver has a KL-7 with the same rotor arrangement, as provided for that particular day in his key list. The system indicator 28604 in the received message identifies the ADONIS crypto system and key for this particular message. He will proceed as follows:
1. He switches his KL-7 mode selector
to Plain mode.
2. He sets "AAA-AAAA" as rotor alignment with the Set Key buttons on the front of the cipher unit.
3. He switches to Encipher mode (not Decipher mode!).
4. He enciphers the spelled-out message indicator "ELXNO" to again produce the sender's original Message Rotor Alignment "BHLDO".
5. He switches back to Plain mode.
6. He then sets "BHL-DOBH" as Message Rotor Alignment, also repeating the first two letters at the end.
7. He finally switches to Decipher mode and deciphers the ciphertext "GOATJ ZPFJZ RGDET FKCSB TCMTD XTQLP" to retrieve the original message.
In the KL-7 simulator help file Appendix A you will find two fascinating enciphered training messages, related to the Cuban missile crisis, to practice the deciphering procedure. They use the KAO-41C/TSEC procedure and example key as described here.
The random five-letter Message Indicators are created beforehand. An unused KL-7 is loaded with a randomly chosen rotor arrangement that is never used in any existing key list. Next, all rotors are set in a random start position and the mode selector is set to Encipher mode. A large number of random words or letter is typed. The KL-7 prints a large number of random five-letter groups on the paper strip. The operator later uses these Message Indicators to produce a different Message Rotor Alignments for each new message. To avoid re-use, each Message Indicator on the paper strip is used only once, then torn off the strip, and pasted on the corresponding message form.
The POLLUX rotor set consists of only 8 rotors, 7 notch rings and one stationary wide ring. The POLLUX procedure only uses letters as system indicator on the key list. Its 36-45 letters check is performed with the letter A instead of letter L for ADONIS. POLLUX also uses a random five-letter Message Indicator but prescribes setting the random Message Indicator directly as Message Rotor Alignment, instead of enciphering the Message Indicator. The document adds the notice "do not encrypt using this system only. This system gives true indicator in clear and then refers to unspecified supplementary instructions. Various other methods to securely convey the Message Rotor Alignment are possible.
1945 - 1976
The development of the KL-7 involved several agencies. The Signal Intelligence Service (SIS), established in 1929 as part of the Army Signal Corps, was responsible for cryptanalysis under direction of the renowned cryptologist William F. Friedman. In March 1943, the SIS was renamed Signal Security Service (SSS) and in July 1943 again renamed Signal Security Agency (SSA). Its successor, the Army Security Agency (ASA), was establishment in September 1945 and existed until late 1976.
In 1949, the Armed Forces Security Agency (AFSA) was established to merge all Communications Security (COMSEC) and Communications Intelligence (COMINT) efforts. William Friedman led its cryptologic division. This agency gave the machine its initial name AFSAM-7. However, the means and responsibilities of AFSA were scattered over many different military and civil services. This eventually resulted in the 1952 establishment of its successor, the National Security Agency (NSA), with Friedman as chief cryptologist. The development of the KL-7 therefore involved the ASA, AFSA and NSA.
The roots of the KL-7 are found in the Second World War when the U.S. Army SIGABA rotor cipher machine, called ECM (Electric Cipher Machine) by the Navy, and the SIGABA CCM (Combined Cipher Machines) had set a new standard for secure high-level communications between the Allies. At tactical level, the lightweight mechanical M-209 was widely used. By the end of the war, the M-209 was no longer considered secure, and the U.S. Army expressed the need for a lightweight secure crypto machine that could replace the M-209, but would have a cryptographic strength, similar to the SIGABA.
The U.S. Navy was also seeking a small cipher machine with the qualities of the ECM, with a focus on saving weight. In March 1945, the Army headquarters requested the Signal Security Agency (SSA), soon after renamed Army Security Agency (ASA), to develop a machine that would fit their needs. Meanwhile, the CCM, based on the AJAX crypto principle and used by both the U.S. and United Kingdom, was outdated and needed replacement.
The project was designated MX-507 and ASA saw it as a long-range research project. The ASA researchers quickly decided to opt for a rotor-based machine. They also had to design a completely new lightweight printing system, as the new machine was required to operate off-line and print out the messages on paper. Eventually, they were able to reduce a printer system to one quarter of its original size and weight.
ASA applied a new cryptographic principle called re-entry or re-flexing with 36-pins rotors. The idea was to take parts of the cipher output, re-enter the output back into the enciphering process, and re-encipher it again. Cryptanalyst Albert W. Small conceived the system in 1940 and filed it for patent in 1944. However, his patent had been placed under Patent Office Secrecy Order. This would cause a patent conflict in 1957.
The rotors were a further development of the early WW2 rotors. The so-called Blue Rotor, used until the late 1950s, was a fairly large Hebern type 26-pins rotor, simple and rugged. The regular rewiring of those rotors, required for security reasons, was quite complicated. A modified version of the Blue Rotor, called White Rotor, carried an alphabet ring and notch ring.
The U.S. Navy also developed a smaller Hebern type 26-pins rotor, called Yellow Rotor, for their successor of the CCM. There was also a study to use printed rotors, where the circuits were etched onto the rotor body, but that project ended in 1953 and was ultimately discontinued.
The Armed Forces Security Agency (AFSA) was created in 1949. It was the first American central cryptologic organization and one of its goals was to provide standardization of secure communications devices and determine a general policy for crypto equipment. The research of the ASA was transferred to the AFSA in December 1949. The MX-507 was renamed AFSAM-7, which stands for Armed Forces Security Agency Machine No 7.
After a series of cryptologic studies, already initiated in 1946, AFSA decided to use the new 36-pins Red Rotor with rotatable alphabet ring and notch ring as standard rotor for the AFSAM-7 and AFSAM-9. However, the Red Rotor had two major problems. Tolerance issues with the plastic molding process and contact problems. The rotor used beryllium copper contacts, of which particles wore off and turned into abrasive non-conductive copper oxide, which both exacerbated the wear even more, and caused contact problems.
From 1946 on, external contractors studied the Red Rotor problem the next ten years, costing $1,250,000 to arrive at the Orange Rotor. Tests with 200 contact materials did not find better materials than beryllium copper and the plastic compound was still the best suitable. After more modifications and improvements, the Red Rotor was accepted but contact problems persisted. The Orange Rotor would not be introduced until 1956. It's important to note that the beryllium copper dust particles from the rotor contacts are toxic and could cause the lung disease berylliosis. Inhaling these particles should be strictly avoided.
In April 1949, the United States and its allies had formed the North Atlantic Treaty Organization or NATO, and deteriorating relations with the Soviet Union resulted into a grim Cold War. Secure communications between the NATO members was an important part of making a front against the USSR. An additional challenge faced, by AFSA, was to design for themselves a machine that could also be distributed among their NATO allies, without disclosing vital secret crypto technology that could come into Soviet hands, either directly or through infiltration in NATO.
With such a large organization as NATO, it was more than likely that this machine or its specifications would sooner or later reach Russian soil. The design had to resist by far any possible cryptanalytic attack by Soviet codebreakers, even when the technical details of the machine were disclosed. The security of the machine had to depend solely on the secrecy of the key settings, thus obeying Kerckhoffs well-known law on cryptography.
In September 1950, AFSA demonstrated an engineering model. The final design used eight 36-pins rotors, a re-entry of ten rotor signals, and a most complex irregular stepping of the rotors, electrically controlled by notch rings on the rotors. The problems with the printer timing and the shift system were solved by a clever design with vacuum tubes, making the AFSAM-7 the first tactical cipher machine ever to use electronics.
The AFSAM-7 was approved, and the Army could build prototype models. By December 1950, the Army declared the AFSAM-7 ready for production. The machine would become the first standard crypto machine in the U.S. Armed Forces. The crypto system was designated POLLUX. Contractors were selected and operational and maintenance manuals were composed. In February 1951, contracts were signed to produce 25,000 AFSAM-7s at a rate of 5,000 per year. The first repair and maintenance course for Army and Air Force personnel was scheduled in September 1951. However, due to tooling problems and material shortages, delivery of the AFSAM-7 was delayed to June 1952 and then delayed again to January 1953.
In 1951, the BRUTUS crypto principle was proposed as replacement for the CCM's outdated AJAX principle. The BRUTUS rotor stepping maze controlled the irregular movement of the rotors, with rotors 2 and 6 rotating in opposite direction, and differed from the POLLUX stepping logic. BRUTUS used seven 26-pins rotors from a set of ten, with removable cams and alphabet rings. The number of notches on the notch pattern had to be 7, 9, 11, 15, 17 or 19 (co-primes). Meanwhile, the Navy had been developing its own machine, initially named Portable Cipher Machine (PCM) and later renamed AFSAM-47. They had already adopted the BRUTUS crypto principle earlier on for their AFSAM-47, but production, planned for late 1950, was already delayed.
The upper-case system on the British TYPEX cipher machine was non-standard and a combined U.S./U.K. system was therefore impossible until TYPEX was replaced. The CCM Replacement's Working Party suggested a system for combined use to achieve compatibility between AFSAM-7, AFSAM-47, the British SINGLET and other U.K and U.S. machines. This comprised the Space key to piggyback on letter Z, switch to figures on J and switch to letters on V.
However, the design of the AFSAM-47 used the eight upper-case character - ( ) / : ? , . and was only compatible with the British SINGLET. Neither the limited nor extended upper-case system could be introduced until the British stopped using the TYPEX with BRUTUS adaptor. The limited upper-case system with numerals and space was eventually adopted for all combined cipher machines. Until a combined policy was agreed, all cipher machines designed for U.S./U.K. use should at least include the limited upper-case system.
In October 1951, AFSA announced two types of operation. The AFSAM-7 traffic for high-level communications was designated ADONIS and the traffic for the Army and Air Force designated POLLUX. The differences between the two crypto systems were the rotor sets and the Message Rotor Alignment procedure at the start of each individual message.
In 1952, the British services wanted to use the BRUTUS crypto principle to replace the CCM, as agreed in 1951. However, analysis showed that initial and long-term costs for NATO requirements, parts and rotors were less expensive for ADONIS with 36-pins rotors, compared to BRUTUS. Plans were made for a phased introduction of the ADONIS principle in combined machines by January 1955. ADONIS equipment would be made available to the U.K. until they could produce their own version with ADONIS principle. The final production contract for the AFSAM-7 was signed on February 9, 1952.
The U.S. urged to standardize ADONIS with 36-pins Red Rotors as they could apply the re-entry principle, impossible with 26 pins. ADONIS also avoided the use of rotor cage adaptors. AFSA's successor, the newly formed National Security Agency (NSA), preferred ADONIS, as the AFSAM-9 teletype encryption with nine 36-pins rotors, later renamed TSEC/KW-9, was also in development. As turned out later, the TSEC/KW-9 pushed the speed of electromechanical encryption to its limit and suffered regular synchronization loss.
In December 1952, the U.S. Office of Communications Security Conference discussed the replacement of the CCM. Participants were, among others, the U.S. Army, Navy, Air Force and cryptologists William Friedman and Albert Small. By then, the British used the early POLLUX principle. Although technically identical, ADONIS conveyed the Message Indicator (i.e. rotor start positions) in encrypted form to the receiver and the earlier POLLUX used a less secure method in clear. The question was raised whether they should say nothing to the British about POLLUX being inadequate and ADONIS more secure.
Meanwhile, the production of the Navy AFSAM-47 kept delaying and the security of BRUTUS was questioned. One proposal was to improve security of the BRUTUS with its 26-pins rotors by adding a plugboard. Although this could make it more secure than ADONIS, the AFSAM-7 was already developed and in production by the Burroughs corporation. The BRUTUS based Navy AFSAM-47, manufactured by Teletype Corporation and subcontractors, was two years behind. The U.S. Army and Air Force preferred the AFSAM-7 and it could be made available for combined and NATO use by early 1955. The use of a plugboard for the AFSAM-7 was also briefly discussed, but William Friedman argued that operators highly disapprove the idea because setting a plugboard was prone to errors, and also creates problems when a message from the previous day would arrive.
In the long term, the AFSAM-7 with 36-pins rotors was more secure than the Navy AFSAM-47 with 26-pins rotors, because the AFSAM-7 would resist cryptanalysis longer than the AFSAM-47. According to William Friedman, cryptologists from both the U.S. and U.K agreed against BRUTUS for the AFSAM-47. Albert Small also preferred the ADONIS principle, but the U.S. Navy insisted to continue production of the AFSAM-47 with BRUTUS principle. Although the British preferred BRUTUS, it was practicality, production costs and quick replacement that prevailed.
It was officially agreed that the SIGABA CCM machine, using the less secure AJAX principle, urgently required replacement, as all cryptanalytic attacks that worked on AJAX also worked on the CSP 2200, a modified SIGABA ECM Mark II. Friedman made it clear that ADONIS and AFSAM-7 were the answer to the CCM problem. Meanwhile, tests on a modified version AFSAM-47B with 36-pins rotors, compatible with ADONIS, were underway and already performed 100 hours on the 47 printer without error. However, any production of the AFSAM-47B was at least two years behind on the AFSAM-47 production.
The Joint Chiefs of Staff therefore believed that replacement of the CCM by a machine with the BRUTUS principle should be suspended until service tests of the AFSAM-7 were completed. The Navy, however, insisted on keeping the AFSAM-47 with BRUTUS and wait to see if the AFSAM-7 ADONIS tests and production would succeed or fail, before redesigning the AFSAM-47 into AFSAM-47B to accept ADONIS with 36-pin rotors.
By November 1953, the British part of the COMSEC Conference, which assessed the security of cryptographic equipment, was not in favor of POLLUX, because using Message Indicators in clear posed the risk of in-depth messages and recovery of the key settings, certainly with high traffic volumes. In contrast, ADONIS was considered secure for all classifications for at least ten years, if good operating standards were maintained. U.S. cryptologists even considered the machine secure for the next twenty years. However, the British considered the AFSAM-47 with BRUTUS principle only secure for the next five years. They also recommended to replace the British CCM as soon as possible, as it was regarded insecure within three years. The TYPEX II and Typex Mk 22 remained secure for the next five years.
The AFSAM-7, favored by the NSA, eventually proved successful and the ADONIS principle was also chosen for the Navy AFSAM-47B. The NSA introduced the AFSAM-7 in the U.S. Armed Forces, and a smaller number of AFSAM-7s was also purchased by the Federal Bureau of Investigation (FBI) and Central Intelligence Agency (CIA). The AFSAM-7 could resist any cryptanalytic attack at the moment of its release. In 1955, the AFSAM-7 was renamed TSEC/KL-7, according to the new nomenclature for crypto equipment.
An ancillary Baudot paper tape reader called TSEC/HL-1 was developed for the KL-7 to enable directly reading and processing five-bit level punched tape, as produced by standard teleprinters. This required the removal of the KL-7 keyboard and the installation of the KLX-7/TSEC keyboard adaptor between keyboard and chassis.
The AFSAM-47B was later renamed TSEC/KL-47. This machine was compatible with the KL-7. Individual components of the KL-7 and KL-47 were manufactured by several different U.S. government-contracted companies. After final assembly at different locations, the machines became the property of the NSA and were distributed within the U.S.
Although cryptographically compatible with the KL-7, the KL-47 still had the extended upper-case system with punctuations, while the KL-7 only had the limited upper-case without punctuations. Therefore, to avoid decryption errors, any message sent from a KL-47 to a KL-7 required spelling out the punctuation marks or omitting them.
In 1957, Boris Hagelin, engineer and founder of the Swiss firm Crypto AG, told NSA cryptologist William Friedman that he had filed a patent application for the re-entry principle in 1953. Hagelin's U.S. patent No. 2.802.047 was issued in 1957 and conflicted with the patent from Albert Small, who had already filed the application for patent in 1944, at the insistence of no less than William Friedman. Moreover, in 1956, Small had requested declassification of his pending patents, still under Secrecy Order.
The issue of conflicting patents had to be resolved. The NSA feared that sensitive information would be disclosed and preferred that the pending application from Albert Small was not made public if he would not acquire legal claims for compensation. Another option was to release only unclassified portions of the patents. Eventually, the conflict was solved, and Albert Small's Patent 2.984.700 was issued in 1961. Crypto AG also used the re-entry principle in its HX-63, twelve years after the introduction of the AFSAM-7.
TEMPEST, the procedures and techniques to shield devices against eavesdropping on unintentionally emitted signals, was in its early stages of research when the KL-7 was being developed. Although Bell engineers recognized the risk of unwanted stray signals as early as 1943, the initial attempts to reduce these signals were limited to filters on the power supply, and shielding as much as possible. The first breakthrough came in 1956 with the introduction of low voltage circuits with transistors, but this was four years after the introduction of the KL-7, and the first extensive TEMPEST regulations were only drafted in 1958.
De KL-7 has a radio interference filter between the external power supply and its electronics, but some electrical contacts and coils could still be a source of unwanted signals. The NSA therefore conducted a study in 1955 to determine whether the coil of the printer magnet, which activates the print hammer, would emit signals that could be exploited. These print coil signals were detectable 25 feet from the machine.
Analysis of recorded signals during decipherment of a message on the KL-7 showed that measuring the interval between the intercepted pulses of the print magnet, and knowing the order of the letters on the print drum, enabled the recovery of the plain text. Variations in motor speed between pulses could complicate measurement of the intervals, but other signals, radiated at the same distance, could determine the change of motor speed, making recovery of the plain text much easier. They also found a correlation between the number of rotors that stepped and print drum speed deviations. See also NSA radiation study KL-7 .
The KL-7 remained in use without additional technical changes to reduce unwanted signals, but the 1958 TEMPEST regulations undoubtedly advised operating the machine at fixed or tactical locations were eavesdropping at close range was unlikely. Nevertheless, even at secure locations, unwanted signals could unexpectedly piggyback on other equipment and enable eavesdropping from far greater distances, as you can read at our TEMPEST page.
The KL-7 was initially only intended for the U.S. Army, Air Force, Navy, CIA and FBI, but during the 1953 Communications Security Conference in London, the NSA proposed to share the ADONIS crypto-principle with their NATO allies. The goal was to improve communications security and interoperability, and to replace the less secure Combined Cipher Machine (CCM) by the AFSAM-7.
U.S. AIR FORCE
From 1951 to 1954, the Army Security Agency (ASA) procured 6547 units. Once delivered, they would replace the less secure M-209. The FBI placed an order for 120 AFSAM-7s, 120 Office Cases at $100 each, an additional 120 AFSAM-207 Cipher Units (i.e. KLK-7/TSEC) at $50 each, 250 sets of Rotors (two per machine to enable swift daily key change), 5 spare AFSAM-107 Stepping Units (i.e. KLA-7/TSEC), 5 spare AC Power Converters at $25 each, paper tape and ink ribbons. The order totaled $258,900. However, by 1953, the cost of the FBI order had risen to $299,232 ($3,365,084 in present 2022) due to rising production costs.
ASA initially received 650 AFSAM-7s, of which 120 for the FBI. These were gradually issued, two per FBI office. Meanwhile, the twenty FBI offices with the highest message volume received AFSAM-7s on loan from the NSA, and the NSA also trained FBI personnel in operating the AFSAM-7. The CIA received its first four AFSAM-7 for local testing in 1954 but use in the field only started the following year.
By 1954 all FBI offices, Quantico, the White House Signal Detachment (WHSD), the Seat of Government and President Dwight Eisenhower's Air Force One were equipped with the AFSAM-7. However, many AFSAM-7s from early production runs had several technical issues. In April 1954, the NSA director received a list of deficiencies, noted by the Chief Army Field Forces during testing.
In order to meet the performance standards, ASA requested modifications for a total of 2339 AFSAM-7s from the 1st, 2nd and 3rd production runs, at that moment in storage facilities. These were returned the next month to the Burroughs Corporation. When the second shipment arrived at ASA, it was discovered that 615 from the 1400 already delivered AFSAM-7s required modifications. As a result, machines from the White House, ASA Europe and ASA Pacific had to be replaced by modified versions.
The U.S. Joint Chiefs of Staff approved the use of the AFSAM-7 by NATO in 1954. The plan was to introduce the machine to medium and high levels by mid-1956. The NSA did recognize that the AFSAM-7, or reproductions with the same cryptographic principle, would eventually also find their way to non-military use in those countries, or might even end up in Soviet hands. The NSA was confident the AFSAM-7 was secure against any attempt by the Soviets to decipher the messages, even when its cryptographic principles and specifications were compromised. The machine was therefore certified for Top Secret messages at the start of its career.
Meanwhile, a new British crypto machine was developed with identical crypto principle as the AFSAM-7 to replace the outdated CCM (LUCIFER), which was the British CCM-Typex, interoperable with the American CCM/SIGABA. The new BID/60 SINGLET greatly resembled the AFSAM-7 and used rotors that were identical to those of the AFSAM-7, but was not expected to be in production before 1960. The U.S. decided in 1954 to make available 3,500 units of the AFSAM-7 to the United Kingdom and 3,000 units to other NATO countries. These machines were on loan and remained property of the NSA.
In early 1955, the Standing Group of the North Atlantic Military Committee (NAMC), which provides policy guidance, decided to supply the AFSAM-7 to Supreme Allied Commander Europe (SACEUR) for further distribution to all NATO members. Target date to replace the CCM by the AFSAM-7 was 1 July 1956. Meanwhile, the AFSAM 7 was renamed TSEC/KL 7. The U.S. Army Security Agency Europe also assigned two military instructors to NATO in 1955. They assisted in the training of NATO personnel, designated to repair and maintain the KL-7. Those who attended the maintenance school had to be qualified as teletypewriter mechanic and have the proper security clearances. Basic knowledge of electronics was also desirable.
In 1956, NATO decided to order the HL-1 tape reader and KLX-7/TSEC keyboard adaptor to cope with the increasing volume of encrypted traffic. The NATO members had to determine their required stock of KL-7 spare parts and could also order a kit with basic spare parts at the price of $150 per machine. The spare parts and kits were gradually delivered between 1957 and 1960.
Also in 1956, CIA's Operations and Training Division planned the AFSAM-7 for mobile message centers. Noise issues were solved with a soundproof container and a keyboard adaptor became available in 1957. That year, CIA O&T personnel also visited the NSA to observe the HL-1 tape reader that could process perforated tapes. The HL-1 was later installed on loan at the CIA Signal Center.
In 1957, NATO agreed to adopt the KL-7 for second level NATO use and also for first level NATO use with ADONIS key lists, to replace the Typex (SIMPLEX) traffic that used the Typex II with SIMPLEX pads. This agreement also comprised each NATO member's Ministry of Defense and Foreign Affairs, their embassies in Paris and Washington, and their National Military Representative in Washington. In 1958 they extended the use of the KL-7 to minesweepers, fast patrol boats and long-range maritime aircraft.
|NATO MEMBER STATES IN 1957||OUTSIDE NATO|
Besides the United States, the KL-7 was used by its NATO allies Australia, Belgium, Canada, Denmark, England, France, Federal Republic of Germany (former West Germany), Greece, Italy, Luxemburg, Netherlands, Norway, New Zealand, Portugal and Turkey. Outside NATO the KL-7 was also on loan to South Korea, South Vietnam and Nationalist China.
Some political background: South Vietnam, officially called Republic of Vietnam (RVN), existed from 1955 until the 1975 North Vietnamese victory and formation of the current Socialist Republic of Vietnam. South Korea, officially the Republic of Korea (ROK), was formed in 1948 with the division of the Korean peninsula in two political entities along the 38th parallel, with the U.S.-backed South Korea and Soviet-backed North Korea, officially the Democratic People's Republic of Korea (DPRK). The Republic of China, often called Nationalist China, was formed in 1912 by the Kuomintang. After the 1949 communist takeover the Kuomintang fled to Taiwan, since then officially called Republic of China (ROC), which is disputed by mainland People's Republic of China (PRC) to this day.
In 1958 the price per KL-7 totaled $1458. This comprised the KLB-7 base at $814, KLA-7 stepping unit $328, KLK-7 cipher unit $80, CE87054 Carrying Case $161, CE87066 AC Power Converter $75 and a set of rotors $100. A complete KL-7 would cost $14,313 when converted into present 2022.
The extended use of the KL-7 was discussed, and NATO member Canada pointed out that the KL-7 had not reached 100% reliability yet due to problems with the pulse generator. A book cipher as back-up would be essential. The KL-7 was also used on NATO submarines and in 1959 they approved the Basic Submarine Code as back-up system for the KL-7. The code, in itself not secure enough, was used in conjunction with letter one-time pads. In December 1959 NATO authorized the use of KL-7 ADONIS for first level military and diplomatic traffic. By 1966 some 25,000 KL-7 were produced for the U.S. and their allies.
When France left NATO's military structure in 1966, NATO needed a separate crypto system to exclude France from their most sensitive communications. Initially, two new KL-7 ADONIS key lists and a new set of rotors with other internal wiring was introduced. NATO did continue to distribute COSMIC TOP SECRET key lists to France but introduced separate key lists from which the French were excluded. The KL-7 key lists for General Small Ships, the Maritime Patrol Aircraft, Atlantic Channel North Sea and Baltic Area remained available to France.
For transport, accessories and operating, the KL-7 had three types of casings.
The KLB-7 Base with all electronics was not classified. The KLA-7 Rotor Stepping Unit with actuator switches and wiring was classified CONFIDENTIAL. The KLK-7 Cipher Unit was also CONFIDENTIAL. The maintenance rotors were CONFIDENTIAL, the operational POLLUX rotors CONFIDENTIAL and ADONIS rotors SECRET.
Usually, two Cipher Units were available for each KL-7, one with the current day key settings and one with the previous day settings. The KL-7 also came with an operator maintenance kit in a small square metal box, containing a few tools, a cleaning block to clean keyboard and rotor contacts, contact cleaner and lubricant, a brush and bristle, spare vacuum tubes and neon bulb for the Shift indication.
* Despite its extensive use, the KL-7 was not the most popular machine. On start-up, the vacuum tubes need 16 seconds to heat up before you can type on its keyboard, as the printer timing is controlled by the electronics. The KL-7 also has a high acoustical signature, and the motor gears produces a high-pitched noise. The KL-7 was notorious for its regular contact problems, caused by the keyboard and rotors. The operator often had to push firmly on the keys to get the machine cycling, not allowing him to get any speed. Dirty contacts could cause the machine to halt, the so-called dead-rove.
To prevent or solve contact problems, the rotors and keyboard contacts had to be cleaned regularly and meticulously. In Plain mode, there are only two contact points from any key to a pulse coil. In Encipher or Decipher mode, the signal passes 13 contact points, and two additional contact points when using the FIG or LET key. Moreover, each pass through the re-entry wires adds another 9 additional contact points in the rotors. To use all keys, a total of 472 contacts must work flawlessly to prevent the notorious dead-rove. More about the signal path in the Selector section.
When the KL-7 is turned on, the DC motor slowly takes speed, and the reduction gears produce the characteristic high pitched noise. The advancing rotors also produce their typical sound. On the audio you hear at 0:00 the startup, 0:05 manually advance one rotor with Set Key button step by step and at 0:07 continuously, 0:11 typing plain text, 0:17 switch to cipher, 0:20 encipher, 0:25 switch off the KL-7.
During its service time, the rotors of the KL-7 and KL-47 were rewired on a regular basis. Some rotors were rewired on a yearly basis on national or NATO level and some rotors, often referred to as the NSA rotors, were to be sent directly to the NSA and were rewired by NSA personnel only. It was strictly forbidden to operators, even to the maintenance technicians with crypto clearance for KL-7, to check out the internal wiring of the rotors.
The technicians were not allowed to test the rotors pin-to-pin. They were instructed to place a defective rotor on a large conductive plate that made contact with all rotor pins at once, and then check out the connection on each pin at the other side with an Ohm meter. This way, the technician would see if a wire was broken, but didnt know to which pin it corresponded on the other side.
There were some incidents where a KL-7 or KL-47 was compromised. One well-known incident is the seizure of USS Pueblo by North Korea in 1968. Officially the ship was an AGER-2 (Auxiliary General Environmental Research). In reality, the ship was stuffed with SIGINT (Signals Intelligence) and ELINT (Electronic Intelligence) equipment to eavesdrop on North Korean and Soviet communications. When the North Koreans attacked and boarded the ship, the U.S. Navy immediately stopped all communications with the KL-47 until the NSA had distributed new key lists. The machine itself was designed to resist cryptanalysis, even when the technical specifications were known to the adversary. What they didnt know was that the KL-47 keys were already compromised.
During the Vietnam War, KL-7s were also loaned to the Army of the Republic of Vietnam (ARVN). In 1965, the U.S. 101st US ASA Security Detachment, operating under cover designator 7th Radio Research Unit, was based in Saigon (today Ho Chi Minh City).
The 7th RRU conducted Signal Security (SIGSEC) and also performed the cryptographic security analysis of ARVN's use of the KL-7 to ensure the machines were used as prescribed. 7th RRU concluded that the ARVN became quite proficient in using the KL-7.
The U.S. forces in Vietnam used the KL-7 from division down to company level. However, a 101st ASA Detachment COMSEC analysis revealed that Communications Security (COMSEC) was often neglected in the heat of battle. Operation SILVER BAYONET with the famous 1965 Battle of Ia Drang showed that a combination of underestimated enemy strength and poor COMSEC can cause heavy losses.
They didn't use the KL-7 for intra-battalion and lower echelon communications and used manual systems instead, often cryptographically less secure, or even plain unencrypted messages over radio.
In the course of war, some KL-7s were captured by the North Vietnamese Army (NVA). One of those, belonging to a U.S. Marine unit, was handed over to the Russians who sent it to the Soviet Socialist Republic Poland for analysis. After the dissolution of the Soviet Union, Polish officials handed over that KL-7 to the NSA and the machine is now in NSA's National Cryptologic Museum collection.
After the 1973 withdrawal of most U.S. troops and subsequent defeat of the ARVN in 1975, all kinds of crypto equipment fell into the hands of the NVA, including some KL-7s.
KL-7 that was captured by NVA and returned to NSA in 2000
(left of the Carrying Case you see a KLX-7 keyboard adaptor)
NSA - National Cryptologic Museum (Notice)
Although the KL-7 was only meant to be used by the U.S. military, its NATO allies and some state departments, there are some rare cases where civilians operated the KL-7. One such case was during the 1982 Falklands War. Within a few days, the British Navy had to sail a huge naval task force across the South Atlantic. They quickly chartered merchant ships to support the operations. One of them was the Eburna tanker that carried fuel oil, diesel and aviation fuel to transfer fuel at sea. The civil radio officer had no experience with naval communications or crypto systems and had to learn the basics of cryptography and operating the KL-7 within very short time. Read the full story in KL-7 on Merchant Ships during the Falklands War
Advances in technology and the introduction of miniature electronic components increased the computational power to support cryptanalysis tremendously in the next decades. As a result, the KL-7 had become operationally insecure by the mid-1960s and vital message traffic, enciphered with the KL-7, was often superenciphered on other systems. From the 1970s on, the KW-26 and KW-37 online cipher equipment largely replaced the outdated KL-7. Some KL-7s stayed in service, mostly as back-up, and retired in the mid 1980s.
The last known recorded message, enciphered with a KL-7, was sent by the Canadian armed forces in June 1983. The fully electronic KL-51 RACE off-line cipher machine could be regarded as its successor. The KL-7 machine itself was unclassified but the cipher unit wiring, the entry plates and the stepping circuitry were confidential. After its service time, most KL-7s and KL-47s and their rotors were recalled. The surviving KL-7s were carefully stripped from their stepping mechanism wiring and rotor entry wiring, commonly denoted as sanitized.
* The above timeline shows the development from MX-507 project to AFSAM-7 and TSEC/KL-7. The ASA developed, procured, and issued the machines. From 1949 on, the AFSA and its successor NSA were responsible for communications security. Although the AFSAM-7 existed on paper in 1949, it took two years from the 1950 engineering model to the actual production start in 1952, and delivery to the Armed Forces only started in early 1953. The POLLUX procedure for low-level traffic was adopted in 1950 and the more secure ADONIS procedure for high-level message traffic was added in 1951. Due to persistent problems, the Red Rotors were replaced by the Orange Rotors in 1956.
Although the NSA had taken over from AFSA in 1952, the machine's name only changed to TSEC/KL-7 in 1955. The newer machines, produced from 1955 on, carried the name TSEC/KL-7 and older production runs were gradually retrofitted with the new name labels. Exact dates of final decommissioning of the KL-7's are unknown, as they were gradually recalled from the many different services and countries, with Canada as the last known user until 1983.
The KL-7 is a unique machine in many ways. It was the first machine, developed under one centralized cryptologic organization and introduced as a standard crypto device in all parts of the armed forces. At that time, the KL-7 used the latest cryptologic techniques and was the first ever cipher machine with electronics, yet its rotor-based design would soon lose the battle against miniaturization of electronics and computational power. It proved to be the last of a breed of true cipher machines.
Many operators cursed the machine for its quirky keyboard and regular contact problems. They welcomed its electronic successors, but today they speak with sentiment about that wonderful machine and even vividly remember the typical sound of its stepping rotors. Maybe because of the era in which the KL-7 and the operators gave their best. Or maybe because the KL-7 served all over the world, collecting secrets and memories about the Cold War, companionship, and even exciting stories about treason and espionage. Because this was not the end of the KL-7 story...
In 1974, a highly sensitive well-placed source told the FBI that the Soviet foreign military intelligence directorate GRU had an agent with codename "Greenwood", who was an American from the U.S. military that had been posted in France and Vietnam. The FBI started counterintelligence operation "Hookshot" to identify the person. The Army Intelligence and Security Command INSCOM narrowed down the search to former U.S. Army Signals Warrant Officer Joseph Helmich (1937-2002), who served in 1963 as crypto custodian in France, in 1964-65 in Vietnam with crypto clearance in a supply unit, and later at Fort Bragg, North Carolina. The FBI started an extensive investigation with surveillance.
Being faced with financial problems, Helmich had contacted in 1963 the Soviet Embassy in Paris, France. He received $131,000 in return for technical information on the KL-7, at that moment the most widely used cipher machines in U.S. military. After returning to the United States, Helmich continued to provide KL-7 key lists to the Soviets until 1966, enabling them to decrypt KL-7 messages from U.S. troops and Intelligence in Vietnam. Although already under suspicion in 1964, because his wealth did not match his pay grade, it was only during FBI surveillance in early 1980 that they observed him visiting the Soviet embassy in Canada to contact the KGB. After extensive interrogations, Helmich eventually confessed in 1981 and was sentenced to life imprisonment.
In 1985, the FBI received a tip from the ex-wife of John Anthony Walker (1937-2014), a retired U.S. Navy communications specialist. Later on, he was observed by the FBI while dropping a grocery bag alongside a road north of Washington D.C. The bag contained 129 copies of stolen secret U.S. Navy documents. At the same moment, and a few miles further, a Soviet KGB agent left a grocery bag with $200,000. It was clearly a dead drop exchange to covertly exchange documents and money, without meeting face-to-face. The following night, John Walker was arrested by the FBI in a motel.
The investigation shook up the military intelligence community. As later turned out, already in 1967, Chief Warrant Officer John Walker simply walked into the Soviet Embassy in Washington DC with a KL-47 key list, and offered the Soviets to sell secret Navy documents for cash. It was the beginning of a spying career of no less than 18 years. During a search of his house after his arrest, the FBI discovered a special device, provide by the KGB, to read the internal wiring of the KL-47 rotors. During interrogations, Walker admitted providing the Soviets with complete technical maintenance manuals, which enabled the reconstruction of a fully operational KL-47, cryptographically identical to the KL-7. He was also sentenced to life imprisonment.
The importance Soviet Intelligence gave to the key lists, despite possessing all technical details of the KL-47, shows that they were probably unable to break the KL-47 or KL-7 message traffic purely by cryptanalysis, or that they had no sufficient computer power to decipher them within reasonable time for practical use, at least in the early 1960s.
You can download a realistic simulation of the TSEC/KL-7 on this website. This simulator is based on the most recent information and developed in collaboration with Crypto Museum. The simulator operates in exactly the same way as the real machine. With most surviving KL-7s sanitized, this simulator is the only remaining way to actually work with this beautiful machine and serves to keep this machine and its history alive. The simulator has an extensive 20-page manual that includes the use of the simulator, the technical and historical details on the KL-7 and some example messages. Please visit the KL-7 Simulator page to download the software,
Technical documents, released by NSA
Related technical information
Patents and declassification
CIA and FBI documents AFSAM-7 and TSEC/KL-7
Reports on sharing the KL-7 with NATO and used crypto principles.
Miscellaneous historical documents
Intelligence and espionage
© Dirk Rijmenants 2004. Last changes: 03 October 2023
|Home KL-7 Simulator|