ECM Mark II Curatorial Report

The ECM that was displayed aboard U.S.S. Pampanito is a CSP-889-2900 formerly on loan from the Naval Security Group. It has "12-29-43 BTS" stamped into the bottom, and "CONT. AX? 1728", "ACCEPTED JUN 1943" with an indecipherable mark printed in orange ink. The print unit ENG-108 is serial number 999 which is consistent with 1943 manufacture. There is no name plate on the unit or the rotor cage. Although the holes for the screws that once held the plate exist. The top housing has holes and an outline in the appropriate location and size to hold a the cleaning instruction plate added in Dec 1943, these holes have been painted over. Inside the machine were two pieces of paper. The first was a 3x5" card on which was printed in ink "Bacchus / Gorgon". The second is a "memorandum of call" form revised in 1967 on the back of which was printed in pencil "Baccus CSP 2900", "Basket CSP 2899". The code names Bacchus and Gorgon were used during the 1950s for CSP 2900 based systems.

The machine arrived during summer of 1996 in pretty good condition, after mechanical and electrical safety checks, a new ribbon and about 30 hours of cleaning and lubrication we have tested its operation successfully. The cipher and control rotors provided were test rotors (wired straight through) so the cipher wheel stepping was not very erratic. The index rotors are wired, changing their setup does change the cipher wheel stepping. After exhausting our resources to find additional rotors we have made a temporary and reversible change to two of the cipher/control rotors. The modified rotors have been used to provide an adequate simulation of random stepping of cipher wheels and unintelligible cipher text. The modified cipher/control rotors wiring is included in the Java ECM emulation as rotor 1 and 2:


Since we have only two wired rotors, we have put black marker on the second rotor to uniquely distinguish between the two.

Table below represents the physical rotor wirings that were in the index rotor set. We do not know for what systems they were used. They are all embossed with "CTT 68AAC" on the edge. Note that index rotors labeling on their circumference increases in the counter clockwise direction.
0,1,2,3,4,5,6,7,8,9 Left side of rotor, below are the right hand sides.
7,5,9,1,4,8,2,6,3,0 Rotor # 10
3,8,1,0,5,9,2,7,6,4 Rotor # 20
4,0,8,6,1,5,3,2,9,7 Rotor # 30
3,9,8,0,5,2,6,1,7,4 Rotor # 40
6,4,9,7,1,3,5,2,8,0 Rotor # 50

The machine was cleaned and lubricated according to SIGKKK-2, 1945. To minimize realignment the main rotor shaft was lubricated in place and the printer unit was removed, but not separated. The printer unit should be removed, separated, cleaned and the center shaft lubricated. We are trying to find drawings or descriptions of "pawl release rod" 100707, "assembly studs" 100708, and "assembly ring" 100706. We will attempt separating the printer unit when we have determined if we can find information on these tools that will facilitate assembly. When received, during encrypt the tape was not spacing in 5 character groups, it advanced to the space and stopped advancing. This corrected itself after cleaning and was probably caused by a weakened spring on the cam follower. Lubricants were used as described: 100983 oil (SAE 20) - We used SAE 20 synthetic bicycle fork lubricating oil. 100984 grease (light grease). 108607 "Lubriplate #105" Lubriplate #105 has been in continuous production without change in formula by Fiske Bros. Refinery, Newark, NJ (201-589-9150) since 1933. It was in stock at Coast Marine with "space age" printed on the tube. A thin coat of DeOxit from CAIG was used on exposed leaf contacts in the build up switches. Cleaning was done with tech wipes and a non-residual electrical cleaning spray (tested on the plastic first). No abrasives were used.

The CSP-2900 model was developed by the Navy ca. 1950 (SRH-360). The unit has two switches in the position that a CSP-889 has only the Zeroize/Operate switch. The new switch is marked 889/2900. There is also a knob marked 889 F/2900 R extending out on the left of the keyboard. Near the 889 F/2900 R knob the housing looks like it underwent some hand work to fit the shaft, this is all painted. The unit was built as a CSP-889 and later modified to CSP-889-2900. There is an added (not CSP-889) mechanism that after small tweaking was found to increment a cam each time a character is enciphered or deciphered. The counter is cleared when either the number 1 or number 5 cipher rotor turns or the machine is in the O, P or R master switch position. If the counter reaches 21 the machine stops working. I speculate that this is a check to detect when the index rotors are not properly set up and no cipher rotor motion is generated.

Floating loose in the machine was a single metal stud that appears to be unrelated to the machine, possibly a piece of construction debris. I have replaced the ribbon with an Okidata printer ribbon (this ribbon is on a plastic spool.) The original ribbon, stud, the notes are in a plastic bag that is attached to the wooden bottom of the carrying case. These are being stored with the electrical spares below the radio room. The machine was received bolted to the bottom of the carrying case, we did not get the top of the carrying case. There were no wooden shims in the case to protect the rubber shock mounts. There was no cover for the print head.

The details below elaborate on the general description provided in History Of Converter M-134-C (1949 Army History). I have only included the details that I noticed where different from the Army History's description. Cipher and control rotor contacts on the rotor cage are labeled as when the rotor has A on top position, not reversed, as printed (i.e. clockwise). I label index rotor contacts with the units digit zero on top, i.e. 10, 20, 30, 40, 50 showing as the starting position, unit digits increase in a counter clockwise manner. I do not know if these are the labeling conventions used elsewhere.

Additional details can be found in the comments of the Java language source code to ECM Applet.

With first switch in the 889 position, the second switch in the OPERATE position, the control switch on E and the knob selecting 889 F:

The right of the control rotor bank has the FGHI contacts at 60 VAC, the DE contacts are energized to 16.2 VAC,. I believe the voltage on the DE contacts is unintended leakage from the 889/2900 switch, before cleaning these were at 25 VAC.

The connections between the left plate of the index bank (number) and the left plate of the control bank (letters) are below.
1-B, 2-C, 3-DE, 4-FGH, 5-IJK, 6-LMNO, 7-PQRST, 8-UVWXYZ, 0-no connection
Below is another representation of the same information:
91233444555666677777888888 index

All cipher rotors turn in a clockwise rotation.

With first switch in the 2900 position, the second switch in the OPERATE position, the control switch on E and the knob selecting 2900 R:

The right of the control rotor bank has DEFGHI contacts are energized to 60.3 VAC.

The connections between the left plate of the index bank (number) and the left plate of the control bank (letters) are below.
1 - B, 2-C, 3-DE, 4-FGH, 5-IJK, 6-LMNO, 7-ST, 8-WXYZ, 9-A, 0- UV
Below is another representation of the same information:
912334445556666---77008888 index
P, Q and R are not connected.

Cipher rotors 2 and 4 turn counter-clockwise, 1, 3, 5 turn clockwise. The 2900 R knob is mechanically linked to the cams that turn rotors 2 and 4. The knob and first switch both must select either 889 or 2900 to operate.

In either position of the first switch (889 or 2900) or knob (889 F or 2900 R), second switch in OPERATE position, control switch on E:

The cipher rotor solenoids (first number) are connected to right of the index bank (second number).
1-09, 2-87, 3-56, 4-34, 5-12

The keyboard is wired to the left plate of the cipher bank in a sequential manner with A on top, proceeding clockwise. Z position of the cipher bank is occupied by the spacebar. (During encipher typing Z generates an encrypted X.) When the control switch is in D (decipher) position the right plate instead of left plate of the cipher bank is connected to the keyboard as expected.

References for this page may be found on the ECM Mark II page, they are not repeated here to save space.

Return to the ECM Mark II page.


Copyright © 2006, Maritime Park Association
All Rights Reserved
Legal Notices and Privacy Policy
Version 2.00, 11 Sep 2006