The Lorenz Schlüsselzusatz 40/42 or SZ for short went through a few changes during it's lifetime which were designed to add extra security to the encryption. This page gives a little detail on what changed on each model.
The Lorenz SZ-40, originally built in 1940 had two stages. The initial pre-production SZ-40 had 10 wheels but it was decided to update this to include the extra two motor wheels to try to stop a repetitive pattern.
The first transmissions were intercepted in early 1940 by a group of policemen on the South Coast who were listening out for possible German spy transmissions from inside the UK and an experimental link using SZ-40 machines was recorded in June 1941.
The regular SZ 40 model consists of five "spaltencaesar" (Chi) wheels which all stepped in unison for each character enciphered. These drive two "vorgelege" (control / Mu) wheels which then step the five "springcaesar" (Psi) wheels irregularly.
The SZ 42 was brought into use from 1942 onwards to improve upon the security of the SZ 40. The 12 wheels are exactly the same as the original model, but an additional limitation was introduced to alter the pattern of the Psi wheels.
Instead of just using the 37-pin motor wheel setting to tell when the Psi wheels should turn, the setting of the second Chi wheel(No.9) from one position back was also checked. If the 37-pin wheel active setting was off and the limitation value from Chi2(1 back) wheel was on, the Psi wheels would not move, any other combination would move the all of the Psi wheels together forward one position.
A further optional limitation, which was based on the fifth impulse from the plaintext character 2 back, was also available which could be added.
The Plaintext Limitation (Klartextfunktion) represents an additional security measure. This can be turned on or off by a switch located under the hood in a box at the upper right of the Lorenz. It has a screwed on lid marked "KT-Schalter" (KT-Switch).
This means an addition of the plaintext fifth impulse from two characters back to the limitation affecting the Psi wheels (1-5). It was used on the SZ42a and SZ42b later in the war, but caused issues when used due to radio interference which would cause the message to be garbled on reception. Bletchley Park named this the P5 limitation.
Also at this time in 1942, the Germans switched from using the Ablesetafel 40 to a QEP book of codes to stop the limits imposed by the fixed number of options by the original method and also to stop giving away extra information with the full set of twelve indicator letters being sent each message. Fortunately (for the Allies and Bletchley Park), Bill Tutte had already worked out methods of calculating the pin settings without having to rely on the indicator values.
The SZ 42B was almost idenitcal to the SZ-42A with the further addition of another wheel into the limitation calculation. The limitation value was now set using both the Chi2 wheel, 1 back (wheel 9) and also the Psi1, 1 back (Wheel 1). These affected the movement of the Psi wheels as on the SZ-42A. The optional P5 (KT-Switch) limitation was also available as before.
This is a simulation of the Lorenz SZ 42c (also known as SK-44). This was never in production but was being built to replace the SZ42a/b. it was not completed by the time the war ended. The information comes from sparse TICOM reports - I am not aware of any photos or models of this version running so this could be the first time this action has been shown since 1945. There were to be 10 wheels which moved constantly unless interrupted. Interuption of wheels 1-5, which move in unison, was by the settings of wheel 9+10, wheel 6 by 8+2, 7 by 9+3, 8 by 10+4, 9 by 6+5 and 10 by 7+1. Should all wheels become locked together, the machine would count to 3 then step wheels 1 to 5.
Another feature was that the sending Lorenz would continue to operate, sending a pseudo-random string of character between messages and was to be kept in sync with the receiving machine by using a crystal-controlled teleprinter. This was to try to hide the exact position of the real encoded message thereby making searching for cribs within the text much more difficult. There were plans for a further additional machine beyond this one (SK-45) which had an additional eleventh wheel which was to step the machine past dead spots so the three count device would not be required.