On the early Russian manned spacecraft, cosmonauts were ejected at a height of about 11 kilometers and descended to the Earth by parachute. On spacecraft with several cosmonauts this landing option was unacceptable.
On board Voskhod soft landing was achieved by using a rocket engine which activated at a certain height and damped the descent rate, providing soft contact with the ground. A telescopic metal probe which pulled out from the body on the descent path closed a contact switch at the moment of touch down which activated the rocket engine. But the switch was found unreliable under testing depending on the type of soil or vegetation leading to no firing or premature operation. In addition, the probe often failed when the module was landing with high horizontal wind velocity component.
To reliably achieve soft landing it is necessary to measure the spacecraft altitude with high precision. Furthermore the system must have all-weather capabilities and function over water, ice, snow or firm ground with all possible variation of spacecraft attitude and horizontal speed. In 1964 a team of the Saint-Petersburg Polytechnic University headed by Dr. Eugene Yurevich was tasked to develop a gamma-ray altimeter (GRA) for the new Soyuz. A GRA measures the photon backscatter from ground resulting from the transmission of photons from a cobalt-60 gamma source. The gamma-ray altimeter records the distance to the ground and at just 1.5 meters it triggers the soft landing engines to slow down the capsule to 2.6-1.4 m/s impact which is also absorbed by the seat damping system. The early GRA soft landing system (called Kaktus) included a decision logic base on five computers. Only later the design baseline was changed to three computers working on a majority principle (two of three).
The Kaktus was successfully tested in the spring 1966. Mock-ups of the descent module were released from an aircraft on various ground and weather conditions to measure accelerations and dynamic loads. For man-rating certification it was necessary to perform three successful and successive tests onboard unmanned craft. The early Soyuz (unmanned) flight tests, failed for various causes unconnected to the soft landing system, which always worked nominally. During the second Soyuz test one of the first stage boosters failed at ignition. While the maintenance towers were being moved into position the spacecraft abort system spontaneously switched on and the rocket exploded destroying the launch pad and killing a technician. The descent module was automatically pulled away from the pad at some 300 metres.
Then the overall Soyuz certification was withdrawn after the ill fated Soyuz-1 flight which caused the death of Vladimir Komarov when the parachute failed to deploy. During one Kaktus recertification test there was a serious anomaly: an untimely activation followed by an early parachute release at height of several hundred meters, which caused the spacecraft to crash to the ground. This was a serious incident which was traced back to depressurization of the descent module. On the sixth day of the flight the pressure fell to 380 millimeters of mercury, and during descent it fell to just 25 millimeters. There was a leak due to a bad seal in the edging strip around the hatch. After power was supplied to the landing system, a corona discharge occurred in the gamma-ray altimeter circuit due to the low pressure. It issued a false command to start up the soft landing engines and, simultaneously, to shoot off the parachute. The pressure in the descent module was supposed to be normal; otherwise the cosmonauts (without pressure suit in the early times) would die. It was logical that it never occurred to anyone to test out the altimeter at a pressure of 25 millimeters of mercury. During flight operations the soft landing system failed on Soyuz-35 (3 June 1980) and Soyuz TM-25 (14 august 1997).
In the Soyuz-TMA version, an improved Kaktus-2V gamma-altimeter replaced the Kaktus-1V in the soft-landing system. In addition the landing system comprises three longer Kazbek-UM impact-absorbing crew seats with four-mode dampers that adjust the seat response according to the astronaut’s mass. Two (of six single-mode) soft landing engines (SLE) were replaced with two new three-mode engines (SLE-M) to improve soft landing performance. The touchdown speed was reduced to from 2.6 to 1.4 m/s versus 3.6 / 2.6 for the Soyuz TM. Landing with only the reserve parachute was reduced to 4.0 to 2.4 m/s versus 6.1 to 4.3 m/s for the Soyuz TM.