(c) 1999,2022 Peter McCollum

The CK-8 Coder/Keyer (military AN/GRA-71)

The CK-8 components comprise a device that allows the user to record a message composed of dits and dahs onto a small tape cartridge, then the message is 'played back' at a rate of about 300 WPM, to electrically key the transmitter. The purpose is to 'burst transmit' a message, so that the operator is on the air for a minimum amount of time. For CIA users, the primary reason is so that the enemy doesn't have enough time to RDF your location. Also, you gain security from a burst message in areas that are less technically sophisticated, since the message is not copy-able unless you are able to record it, then play it back at a reduced speed.


A military GRA-71 burst-coder set, shown with all of the pieces stowed.

Development History

The design of this keyer/coder system was occurring prior to October 1956, as part of the development of “Semi-Automatic Two-Way Radio Station AS-3”, known as “Project 2108”. At that time, and into 1957, the components were referred to as Keyer AK-3, Coder AC-3, and the CA-3 tape cartridge; and their physical size was larger than the final units. A “Final Report” on the development of the AS-3 was available in July 1959, so it is presumed that the keyer components were finalized by then, with the final nomenclature of KE-8, CO-3, and CA-3. The AS-3’s AT-3 transmitter manual includes only the CO-3 and CA-3, and their size and shape appears to match the GRA-71 units seen today. The KE-8-type keyer components were an integral part of the AT-3 transmitter, rather than a separate device.

Meanwhile, by late 1958 a related project was underway to develop “Magnetic Coder/Keyer CK-8”, under contract RD-151 to Stenographic Machines Co. A memo from December 1958 states that “[the CK-8’s] tape cartridge must be electrically and mechanically compatible with the AS-3 equipment developed by [redacted] under Project 2108”. This means that the CK-8 system evolved from the AS-3’s coder/keyer components which were already under development.

In July 1962, the Army Special Forces Material Division started procuring KE-8, CA-3, and CO/B-8 units. It appears that the military adopted the CK-8 components, added the MX-4498 (KA-3), and assigned the “AN/GRA-71” nomenclature. This likely happened in 1963 (a common GRA-71 contract date). No evidence has been found that the CIA used the KA-3 adaptor/case unit.

The last stocks of CIA CK-8 units were consumed in 1974.








Coder/Keyer System

The overall system




Writes Morse letters to CA-3 tape. Alphabetic wheel; magneto-mechanical only (no active electronics)




Writes dot/dash/space to CA-3 tape. ‘dot’, ‘dash’, and ‘space’ keys only; magneto-mechanical only (no active electronics)




Keys xmtr using data on CA-3 tape; wind-up motor; gets DC power from xmtr



Tape cartridge

Also CA-3, CA/A-3, CA/A-3B



Power adaptor & case

Only seen as MX-4498 with GRA-71. Uses 6.3VAC filament supply from T-784 to create 12VDC to operate keyer




Early unit developed for AT-3 system




Early unit developed for AT-3 system


Other CIA Coder/Keyer Devices

CK-7 – Coder/Keyer. An improved version of the military “Perforator Set, Tape, Telegraph AN/UGT-1”. First deliveries in March 1959.

CK-12 – Coder/Keyer. Prototypes seen in November 1960. Intended to key the RT-6 xmtr.

CK-100 – Coder/Keyer. November 1964. A special configuration of CK-8 components for use with the RS-100. It is a re-packaging, such that the CO-8 and CA-3 units are inside the RS-100 case, making them waterproof while in operation.

CK-30 – Coder/Keyer. Under development December 1964, engineering models expected mid-1965. Stores 300 groups on magnetic tape; keys at 300 WPM. Much smaller than the previous CK-8 system. Design utilizes 18 integrated circuits. Early proto used stainless steel tape; later used mylar tape. Compatible with RT-49.

CK-34 – Coder/Keyer, to be used with the AS-12. Protos in the lab September 1965.

CK-36 – July 1966. Used with ASR-100 VHF transceiver.

CK-33 - Morse coder/keyer, employing a photographic key film technique. Introduced in early 1970. Unit cost is $6000. These units have “not shown a high level of reliability”. The last stocks of CIA CK-33 units were consumed in 1974.

CK/A-33 – A Baudot version of the CK-33. None used operationally as of March 1973, but expected to be used in 1974 with telephone communications system TC-33, automatic print out base system RS-536, and RS802/RS-803; all of which require Baudot code inputs.

CK-42Baudot coder/keyer, considerably smaller than the CK/A-33. Outputs synchronous Baudot at any speed up to 10,000 wpm. Uses a photographic key film technique. Prototypes originally expected in June 1973. The replacement for the CK/A-33, at a cost of $3000-$3500. In the future, the CY-2 electronic key generator may eliminate the key film.

NOTE: The CK-42 first appears in memos as described above. However, a later memo from March 1974 says that the “first CK-42 prototype was delivered 27-Mar-1974. It will be evaluated prior to starting phase two of the contract.” A similar comment appears again in December 1975: “Two prototype CK-42 Coder/Keyers were delivered by [redacted], on 27 November.” So, it appears that CK-42 development was likely re-started at least once, with some new design requirements. See the details documented on the CryptoMuseum site here: http://www.cryptomuseum.com/crypto/usa/ck42/index.htm , which likely represents the final version from late 1975.

CK-43 – A 2-way Baudot coder/keyer with automatic encipherment and decipherment, using a one-time magnetic tape as the cryptographic key. Message entry is by means of a stylus operated keyboard. Plain text message readout is by lights displayed on a keyboard. Speeds up to 2400 wpm. Under development (breadboard) in March 1973.

CK-45 - Morse coder keyer. Speed is factory set to either 300 wpm or 600 wpm. Performs automatic encipherment using one-time magnetic tape cryptographic key. Message entry is by means of a stylus operated keyboard. Compatible with the RS-59 system, and directly interchangeable with the CK-33 in [redacted]. Considered as a low-cost replacement for the CK-8 and CK-33, as of March 1973.

KE-35 – ‘Deception Keyer’. Allows the complete automation of deception broadcasts. A 24-hour program is prepared on Teletype tape. Includes settings for call sign and other params.


Military Application of CK-8 (AN/GRA-71)

For the S.F. "A teams" in Vietnam, burst transmissions were a practical necessity, since they were required to send morning weather reports and evening operations reports daily, and the size of these reports could be significant. With 20 or more teams competing for the same bandwidth, it was necessary for each team to get their message through quickly and efficiently.

Several military radios, including the T-784, PRC-64, PRC-74A, and PRC-104 support the GRA-71 system. Newer, solid-state transmitters such as the PRC-64 are cabled directly to the KE-8, since the electronics in the KA-3 (MX-4498) is only needed when keying tube cathodes directly, where high voltages/currents may exist (such as with the T-784).

Apparently some GRA-71's were made with non-Roman letters on the coder wheels, for use in S.E. Asia. The Army's cost for the GRA-71 was $759.14.


Two of the components of the GRA-71 set: MX-4496 (CO-8) tape coder (with tape access door open, and note the scrambled alphabet in smaller red letters on the wheel, which implements a Vigenere's square); MX-4495 (CO-3) tape coder, with dot, dash, and space buttons.


Notes on the GRA-71 (from the manual):

·         NSN: 5820-00-056-6856, listed in FM 24-24 in 1988.

·         TM 11-5835-224-12: "Coder - Burst Transmission Group AN/GRA-71", dated 27 May 1964. Bill Howard has a copy of the manual that Stenographic Machines sent to the government - it is spiral bound, and dated 1-Jan-1964.

·         Manual mentions that it is copyrighted by Stenographic Machines, Inc. They are the original maker, but Arvin apparently made the low bid on the second lot.

·         The manual specifically mentions the T-784/GRC-109 transmitter - it does not say that the GRA-71 can be used with any other radio.

·         The "IDY" button sends out a continuous stream of "dots" at 300 WPM. It is the same on all units, not a unique identifier signal.

·         The 'handle extension' on the MX-4496 (CO-8) is supposed to be opened while in use, to allow you to more easily and smoothly operate it.

·         The tape in the MA-9 is designed to be replaceable. It contains 12.5 feet of 3M Type 428 "Instrument Grade" recording tape.

·         The tape moves at 4.5" per second, dots are 3.3 ms, and dashes are 10 ms.

·         Transistors in the KY-468 (KE-8) are 2N207, 2N492, 2N884, 2N799, 2N887 (SCR), and 2N1377; a total of 13 transistors and 1 SCR. A later version KE-8F described in the UK/PRC-316 manual uses 2N207 (4), 2N884 (SCR, 4), 2N3980 (UJT), 2N799 (2), 2N720A (2).

·         The MX-4498 (KA-3) contains a converter for changing 6.3 VAC into 12 VDC.

GRA-71 notes from Jeffrey Leopard:

At least one of the pieces of equipment that was used to record and play back the burst message was the AN/GSH-17 Recorder-Reproducer Set, Sound (NSN 5835-00-901-4924 [and TM 11-5835-227-12]). Except for the name and NSN I am working from memory, but I believe the following description is fairly accurate. The "gish 17" is basically a [3] track tape recorder/player with two tape decks and two inputs to record from two receivers at once [diversity reception]. This was not a piece of equipment that would have been used in the man-pack mode as it is quite large and heavy (approx. 30"x22"x18", over 100 pounds) and operates from 115 volts 60 Hz. It only has the capability to record and play back burst messages, it does not send them.

As far as I know there never was any device suitable for man-pack operation that would record the burst messages transmitted from the base station to the teams in the field until the appearance of the OA/8990 Digital Message Device Group made by RACAL in the 1980's which could both send and receive burst messages. It was part of the Special Forces Burst Communication System. An interesting note on this device (the OA/8890) which was designed to be used with the PRC-70 and PSC-3 generation of equipment. Even though this piece was designed some 30 years or so after the GRC-109, I have seen some references that suggest that it may have been used to send burst transmissions with the GRC-109 transmitter through the use of a locally manufactured connecting cable, but would not receive them.

The GSH-17 receiving system mentioned above also includes the RD-265/GR Recorder-Reproducer (NSN 5835-00-901-1086), 2 each of the CV-1716/GR Frequency Converter (converts 455 KC or 1.75 MC receiver IF to an audio tone), and the RP-138/GR Sound Reproducer. The equipment manual is TM 11-5835-228-34 or 11-5835-227-12. Typical radio receivers used with the GSH-17 were the R-390A/URR and the RT-662/GRC. Two receivers were supported to allow for diversity reception. The tape decks used 3 tracks on a 1/4 inch tape cartridge; 2 tracks for the redundant receiver signals, and a 3rd track for recording queing and indexing marks.

GSH-17 notes from a user:

The AN/GSH-17 was used in the AN/GRC-26D and the AN/GRC-122(*) V1 and V2 RATT rigs. These configurations were used in Forward Operating Bases (FOB) or Area Control Bases (ACB). Typically a forward deployed C Team/Battalion Headquarters used this configuration.

The Communications Central AN/TSC-26 had the major components of the AN/GSH-17. It consisted of three S-280 size shelters (VAN's); one RCVR VAN, one XMTR VAN, and one Control VAN. The AN/TSC-26 was used mainly at the Special Forces Operating Base (SFOB) and could be deployed to support a Battalion FOB.

In the original AN/TSC-26 configuration the RCVR VAN had three positions. Each had two R-390(*) RCVR's, two CV-1716/GR's, two Kahn ISB converters, one RD-265/GR, a IDY intercept alarm, and a control panel to control the three XMTR's in the XMTR VAN. The play back unit used in the AN/TSC-26 was a table top version of the RP-138/GR, the RP-149/GR.

The GSH-17 system.


A CV-1716 Frequency Converter. Image courtesy of Robert Kinney.


GRA-71 notes from the late Bill Howard:

As to the AN/GRA-71 Code Burst Transmitter. I first came in contact with these sets when the 100th MTC in Louisville, KY put on a training exercise for the Ranger Infantry Company in Pontiac Michigan in the late 1970's. Each patrol was equipped with an AN/GRA-71 and was at a separate table. The controller put out various models such as tanks, railroad tracks with a train, etc. The patrol had to observe, click out a message and at specified times, transmit the message (probably used PRC-74A's). This was received at the company on a special receiver, decoded, and the information transmitted to the Corps G2 (that was me). We then posted the information to the situation map and at pre-determined times, we conducted a "briefing for the Corps Commander" and allowed the patrols to observe the briefing. They all learned how critical their information was to the intelligence effort. I was interested in the AN/GRA 71 and wrote to the manufacturer, Stenographic Machines. They wrote back that they were the original maker and when the contract was put out for bid for a second batch, Stenographic Machines was not the low bidder [apparently Arvin was]. They were kind enough to send me a copy of their instruction manual with the caveat that it was protected property and could not be copied or reproduced. They also said that from time to time, people found parts of the set in flea markets and wrote to them asking for information or repair parts, etc. of which they had none.

GRA-71 notes from Bob McCord:

[Regarding the coding schemes used:] There were basically two cryptographic schemes I remember we used, and neither of them were straight replacement codes. Both depended on outside text which was not part of the encrypting/decrypting scheme and was also separate from the message text.

[Regarding the contents of a coded message found on a GRA-71 tape:] This could of course be just a test message that was put on the tape at the maintenance shop; but, the message being present on both tapes would be consistent with operating procedures (you always put the message on both tapes so that, when you got to the transmit site, if one didn't work there was a backup). The first part does indeed look like "IR" followed by a group count. I don't know what the "IR" means though. It might be the last two letters of the sending units code name/designation. Example: "DK5A DE WA9IR COUNT ONE SIX BT..." or "TOPHAT DE FLAIR COUNT ONE SIX BT...". Sometimes this info was lost at the head of the tape, so I personally used about 20 spaces at the head of the tape to avoid this problem. I would not assume that because the same groups were present on two different tapes, that a straight substitution code was used. Also, while looking at the letter-frequency characteristics indicates that it MIGHT have been a substitution code, code base text was generated randomly so that it might give the same indication. In my day, it would have been highly unusual to send a message that had been encrypted with a simple substitution code, so I have to think that the person who made this tape would not have used one of those either.

[Regarding the use of code wheel, etc.:] I never used the code wheel/alpha wheel unit myself (although I knew those who swore by it) but opted to carry the dit, space, dah unit because it was smaller and more lightweight than the other. Either one works, but every one has their own preferences. I always assumed (although I don't remember ever being directly told) that the IDY was just an attention-getter for the person waiting for the message to start the tape rolling. As I remember, we sent so many seconds of IDYs, then a burst, then so many more seconds of IDYs, another burst, and then ran like hell... If it was a training mission, we would cut the run like hell part and resend the entire message twice using the leg key... The only way to keep your speed up on code is to practice.


The following table shows the connector pinouts for the Keyer and Transmitter (T-784/GRC-109) cables on the MX-4498 (keyer adaptor KA-3), and the KY-468 (keyer KE-8):

Connector pin


Wire color


Osc. cathode



Final cathode



(Final screen)






(B+ from xmtr)



(Final screen)



6.3 V from xmtr






(screen B+ in)



Keyclick filter





Keyer Adaptor-A



Keyer Adaptor-B



Keyer Adaptor-C

-12V supply for keyer


Keyer Adaptor-D

Keyed signal from keyer (high)


Keyer Adaptor-E



Keyer Adaptor-F

Keyed signal from keyer (low)


Keyer Adaptor-H

+12V supply for keyer






Chassis/RF GND






-12V supply in



Keyed signal out (NPN collector, floating)






Keyed signal out (NPN emitter, floating)



+12V supply in (positive electronic ground)



Message transcript

Following is a transcript of messages found on the tapes of the GRA-71 unit shown above. The first tape has a complete message, the second tape's message is incomplete.

I bought the unit from Fair Radio years ago, and it came sealed in one of those foil-lined paper bags - it was a refurb unit from Tobyhana depot. Some of the pieces looked new, others (such as the tapes) were clearly used.

I transcribed the tapes by 'playing' the tapes on the GRA, keying a T-784 xmtr, receiving the signal on an R-390A, and recording the audio to a cassette tape. Then I played the audio tape into the computer's sound card, and used a WAV file-editor to 'see' the dits and dahs in the audio waveform.

Tape one transcript:




Tape two transcript:



1) The 2nd tape's message seems to be missing the beginning portion.

2) Note that the beginning of the 1st message reads "IR count one six" in plain text. The message has a length of 16 five-letter groups, plus one letter.

3) I inserted the spaces in the transcript. The original does not have any spaces.

4) The following phrases appear in both messages: QZVD, QZGMXXUVXU, UTOCY.

5) The "=" character is < -...- > .

6) The <error> is < ........ > .

7) The <end> is < .-.-. > .

8) The letter-frequency is not random. Here are percentages for various letters:

  Z - 8.6%

  OUX - 7.4% each

  KMRV - 6.2% each

  CIQ - 4.9% each

  DTW - 3.7% each

  GHJPY - 2.5% each

  ABES - 1.2% each

9) There were three letters across the two messages where the letter was garbled, so I may have mis-read those (O vs J, W vs P, and Z vs O).

My guess is that these are 'training' messages of some sort. A true message would not have long duplications across 2 different messages, and would likely have a random distribution of characters. So, the message may be a 'code', but not a 'cipher' -- for example, the word UTOCY may stand for something specific.