Home | Data Base | Technical | Radio | Search
DON-2N

Don-2N Don-2N
Don-2N Don-2N
Don-2N Don-2N


Radar 5N20 Don-2N
NATO reporting name: Pill Box
Project Orlan Data Base // public encyclopedia

WHAT IT IS

Don-2N (Дон-2Н; GRAU index 5N20, 5Н20; NATO reporting name Pill Box) is the battle-management radar of the Moscow anti-ballistic-missile system, A-135 «Амур» (Amur), in full "Московская система ПРО А-135 «Амур»". It stands in the woods near Sofrino-1, Pushkinsky District, Moscow Oblast, a few tens of kilometres north-east of the capital. It is a four-sided truncated concrete pyramid with a large fixed phased-array antenna on each face, so it has no rotating parts and looks at the whole upper hemisphere at once.

It does three jobs with one set of hardware. In the Moscow ABM role it finds incoming ballistic-missile warheads, tracks them with high precision, separates the real warheads from decoys, and sends command-guidance signals to the interceptor missiles. For missile-attack warning (СПРН) it is a precise confirming sensor inside the national picture. For space surveillance (СККП) it catalogues satellites and debris and, in peacetime, that is most of its work.

There is exactly one Don-2N. A prototype sector version, Don-2NP (Дон-2НП, NATO Horse Leg), stands at the Sary-Shagan range in Kazakhstan. Nothing else of this kind was built by anyone. As with every system in this database, almost every performance number below is a developer, ministry or reference-work figure and not a measured one; the places where sources disagree are marked.

Where this site already has a related article it is linked: see A-135 / A-235 Nudol for the whole missile system, Elbrus and Elbrus in command and control for the computers, PESA radar and AESA radar for the antenna types, SPRN at Solnechnogorsk, Voronezh and Dnepr for the early-warning side, and S-400, S-500 and 91N6E for the air-defence systems that Don-2N is often confused with.

For how the warning and defence picture relates to the last-resort command system, see Perimeter (Dead Hand) and Object 1335 / Grot. Don-2N is not described in open sources as part of Perimeter's trigger logic or as connected to Object 1335.

QUICK FACTS

PURPOSE AND PLACE IN THE MOSCOW ABM SYSTEM

The 1972 ABM Treaty let each side defend one region. The United States chose a missile-field site at Grand Forks, North Dakota, and later switched off its short-lived Safeguard system. The Soviet Union kept and modernised its defence of Moscow and the Central Industrial Region (Центральный промышленный район). Don-2N is the eye and the brain-stem of that defence.

A-135 is a layered system on paper. Early warning comes from the national network: satellites of the Oko type, then over-the-horizon and ground radars of the Dnepr, Daryal and Voronezh families (see Dnepr, Voronezh and SPRN). Those sensors give minutes of warning and a rough track. They are not precise enough to aim an interceptor. Don-2N takes over: it picks up the cloud of objects as it rises over the horizon, measures each object, decides what is a warhead and what is junk, and passes the result to the command-computing post (5К80), which assigns interceptors. Don-2N then tracks both the objects and the interceptors and sends the steering commands up to the missiles over its own data-link antennas.

The design goal, stated by its developers, was a single radar able to do both long-range (outside the atmosphere) and short-range (inside the atmosphere) interception. Sloka told the story in RTI veterans' memoirs: in the early 1970s it was still undecided whether one radar could guide both kinds of interceptor, and the bet that it could is what made Don-N win. The same account says the radar was built to work on its own if everything around it failed: "in case of missile attack it remains combat operational in stand-alone mode", and all its systems are duplicated so parts can be replaced without cutting power (official description given to the Valdai Club visitors in 2011).

BEFORE DON-2N: A-35, THE DUNAY RADARS AND THE FIRST MOSCOW SYSTEM

Moscow's first ABM system, A-35 (А-35, later A-35M) with the A-350 Galosh interceptor, was built in the 1960s and accepted in the 1970s. Its eyes were different radars. Long-range detection was done by the Dunay-3 (Дунай-3) stations at Kubinka (construction began in 1962) and the Dunay-3U (Дунай-3У) stations at Chernetskoye near Chekhov, which went on combat duty on 15 May 1978 (NATO Cat House; the test version Dunay-3UP at Sary-Shagan was Top Roost). They were sector radars with continuous linear-FM emission and mechanical-electronic scanning of a limited sector, and the Dunay-3U is reported to have looked towards China. Firing was done by the Yenisei and Tobol (Енисей, Тобол) radars. They are not ancestors of Don-2N: the veteran accounts trace Don-2N to a separate line, the centimetre-wave project 5N12N "Don", and not to the Dunay family.

The line of thought that led to Don-2N is documented in the RTI veteran history. In 1963 designer Vladimir Chelomei put his S-225 system into the "Taran" ABM concept and Mints set up a laboratory under Rimily Avramenko to build the detection radar. A decimetre set (5N12G) and then, from early 1964, a 7.5 cm centimetre-wave attachment (5N12N) were started. By 1965 Avramenko had been removed and the decimetre work went to Mikhail Veisbein. In February 1966 Viktor Kelzon was told to draw up the preliminary project "Don" for the second stage of A-35, and a 1968 sketch design of the centimetre station 5N12N followed. It was a 90-degree sector radar: four would be needed for all-round cover, and the 1960s estimate for one station was 600 million roubles, so about two billion for four. This is why the later four-faced single building was attractive.

HISTORY AND DEVELOPMENT

LOCATION AND CONSTRUCTION

Sofrino-1 is a closed military settlement in the woods of Pushkinsky District, about 56°10′ N, 37°46′ E. People also call the station "the Pushkino radar". It is well outside the Moscow ring road. The approach road runs between large concrete screens that shield it from the radar's radiation: a 2024 visitor report says that when the station goes "to high power" (на высокое, the crews' slang for the full-power working mode) a car caught in the beam can lose its electronics, and visitors are routed by a safe path. A 2024 report also describes a long reinforced-concrete tunnel under the field that was used during construction so that building materials and equipment could not be tracked from space (as claimed).

The command-computing post (КВП 5К80, unit 20007) is a buried bunker at the foot of the radar. It is described as a thick-walled underground complex with a communications node, a data-transmission combat post, and a machine hall of about 1,200 square metres. Elements of the full A-135 complex were spread over distances up to about 100 km and linked by a data-transmission system (5Я67, see below). A 53T6 silo group of twelve launchers is co-located with Don-2N at Sofrino.

THE BUILDING AND THE FOUR FACES

The structure is a regular four-sided frustum (a pyramid with the top cut off):

Each of the four faces carries two antennas, per the RTI description and English Wikipedia (citing the 2020 reference book by Borisov and Pervov):

Element counts: the 2021 RTI Systems and press descriptions say "hundreds of thousands" of radiators in total; TechInsider gives about 60,000 emitters per array, which is consistent (about 240,000 in all). An earlier version of this article said "more than 250,000"; treat that as a round figure for "a couple of hundred thousand". The arrays are made of replaceable modules: in 2011 Valdai visitors saw one array element being removed for repair from inside the building.

Why four faces: a flat electronically steered array loses gain as the beam moves away from straight ahead. About ±60° off broadside is the usual practical limit for a flat array (see PESA radar). Four faces, each covering a quarter of the horizon, give all-round cover without any moving part, and the pyramid's tilt helps point each face upward. The structure also has to be hardened and screened: there are concrete shields around it and a requirement that the radar keep working independently if outside links are cut.

HOW THE RADAR WORKS

Active or passive? Earlier versions of this site, and some western sources, class Don-2N as a passive electronically scanned array (PESA). The Russian developers do not. Sloka's colleagues describe it as an active phased array (АФАР) in which the transmit power is generated inside the antenna: the head engineer of the Dnepropetrovsk plant wrote that the older Moscow radars (Yenisei-Tobol, for A-35) used a chain of amplitrons feeding a semi-active antenna with up to about 120 MW in the pulse, while Don-2N "adopted the model of an active phased array with distributed power in the aperture". The transmit module (6ДГУ with its modulator 6ДГМ) contains a three-stage microwave amplifier chain codenamed "Servant" (Сервант) with megawatt-class pulse output, mounted on one base, about 300 kg. The memoir says tens of such modules in the array give 50-60 MW of pulse power per array. That is a very coarse, large-module active array, not the thousands of tiny transmit/receive modules of a modern AESA (see AESA radar). The memoir also mentions 300 transmit modules in a pricing example for the 5N20, without saying whether that is per array or total. My reading is "active, large-module"; the point is argued and the module count is not public.

Waveforms and processing. The station is described as a monopulse multi-function radar of the centimetre band with large-module active phased arrays, electronic control of both the position and shape of the transmit and receive beams, and digital processing of the radar signals (vpk.name, RTI). The decision to make all signal processing digital in the 1970s was Sloka's and was bold for the time. The station has a large library of probing signals that differ in energy, carrier frequency, repetition period, spectral width and duration, and it can change the beam width instantly. Different waveforms are used for searching a sector, for precision tracking and for discriminating objects inside a complex target; the stated design includes a wideband signal for fine range resolution (vpk.name says "широкополосным зондирующим сигналом"), which is what makes it possible to see the length structure of a re-entering object. The station has separate channels for five measurements of a target's coordinates (RTI).

Normal and emergency modes. A statement in the veterans' account describes the design assumption: in a "normal situation" the station would search for targets out to 3,500 km in a narrow sector; if the enemy exploded high-altitude nuclear warheads that blinded the metre-wave early-warning radars, Don-N would begin detecting at about 300 km in a wide sector. That is a rare official explanation of why one radar was given both a long-range and a short-range mode. It is a design intention from memoirs, not a test result.

Jam resistance. RTI lists high jam resistance from high frequency selectivity, a narrow beam, wide frequency range, automatic jammer cancellers, the ability to reduce sensitivity in the direction of jammers, and special signal structures. Its stated features also include modular construction, high automation, ability to find and track fast targets, and ability to detect low-observable targets.

SPECIFICATIONS (PUBLISHED FIGURES)

WHY THE RANGE NUMBERS DISAGREE

Range for radars of this kind depends on the size of the object, its radar cross-section, the viewing angle, the waveform and the time spent on it. Four different numbers circulate, and they are mostly answers to different questions:

Because a five-centimetre sphere has a far smaller radar cross-section than a re-entry vehicle, there is no contradiction between "800 km on a 5 cm sphere" and "3,700 km on a warhead". Don-2N is a high-power fire-control radar, not a long-range search fence. The long-range fence work belongs to the stations in the Voronezh and Dnepr articles.

DISCRIMINATING WARHEADS FROM DECOYS

This is the central reason the radar exists. A ballistic missile with a multiple warhead or penetration aids releases several real warheads and a cloud of lighter things: heavy and light decoys, chaff dipoles, jammers. In space, with no air, a feather and a hammer travel together. The Russian doctrine for A-135, set out in the memoirs of its designers, used two kinds of discrimination:

The system also relies on volume: the design is to track a hundred objects, discriminate, and shoot only what matters. In a real attack the number of objects could exceed that, and no source claims otherwise.

THE COMPUTERS

Don-2N needs fast computing for four separate things: real-time digital processing of the radar signal itself, the radar's control and tracking computer, the battle algorithm that discriminates and plans the engagement, and the guidance computation for the interceptors. The sources describe them as separate systems, and the old version of this page blurred that.

Signal processor (спецвычислитель, СВУ). The first plan was to use a special computer by Mikhail Kartsev (the M-13 line). In Sloka's account it "could not pull it" and was dropped. RTI and the design bureau of the Dnepropetrovsk Machine-Building Plant (КБ Днепровского машиностроительного завода, DMZ) built a special digital processor instead. It was led by Vladimir Yurko, took about a decade, and was made of tens of thousands of printed cells in over 500 cabinets, cables with gold-plated conductors, and was designed with a purpose-built CAD system called "Dnepr". The cells used multilayer boards of up to 18 layers. The test of the prototype on the Sary-Shagan Don-2NP "worked at first switch-on". This computer is not an Elbrus; it is the real-time signal front end.

Elbrus-2. The control and battle computers are the Elbrus-2 multiprocessor complex (МВК «Эльбрус-2») of the Institute of Precision Mechanics and Computer Engineering (ИТМиВТ), designed under chief designer Vsevolod Burtsev. It is also covered in Elbrus and Elbrus in command and control. Published figures:

How many? Academician Boris Babayan, the chief of the later Elbrus family, wrote that two ten-processor complexes were installed at the combat facilities at the command-computing post in Sofrino and two at the Don-2N radar, so four in all. Russian Wikipedia and RTI say the radar's computing complex "built on four ten-processor Elbrus-2 supercomputers" delivers up to a billion operations per second. Four times 125 million is 500 million, so "up to a billion" is a round-up of the headline numbers; one source on this site (A-135 / A-235 Nudol) gives about 250 MIPS each. The two accounts differ on where each complex sat, so I give the split as reported rather than as established.

Replacement in 2021. In January 2021 General Grabchuk told the Ministry of Defence newspaper Krasnaya Zvezda that the modernised command-computing post "works on the new computing system Эльбрус-90С with a performance of about 100 billion operations per second". He said the area occupied by the equipment would fall four times and power consumption forty times, and that heavy redundancy would cut the time to restore readiness. The article compared this with open-source data that the old Elbrus-2 "could do no more than 12.5 million operations". That comparison is apples to oranges: 12.5 million is a figure for one processor, and the old complex had many. The honest ratio of the old complex (125 million per ten-processor machine, up to a billion for the full set per RTI) to 100 billion is roughly 100 to 800 times, and the headline "10,000 times" is not a like-for-like figure. Whether "Эльбрус-90С" is built on a SPARC (Elbrus-90micro line) or on the later E2K processors is not stated in the sources found. The command post is also reported to use an Elbrus-90micro server network earlier.

2024. A May 2024 report from inside unit 20007 says the machine hall is 1,200 square metres of tall cabinets, the Elbrus-2 machines are obsolete but kept in working order in reserve, and "the modern intellectual core fits in just three small cabinets" with four computers each, one of which exceeds the entire Elbrus-2 hall. The unit was fully re-equipped about a year earlier. Whether the hall is truly kept in reserve or only partly so, no source says. This confirms the earlier version's summary and puts it into one picture: Elbrus-2 was the original brain, the 2021 command post moved onto newer Russian machines.

Reports in 2007 said a new computing system and its software were being brought to the radar, and the 2021 interview speaks of "high-performance specialised electronic computing means" coming into use. The more detailed claim, that the radar's own signal processor was replaced, is not stated in the sources found.

POWER, COOLING AND SHIELDING

COMMUNICATIONS AND CONNECTIONS

Public information on the links is thin. The following structure is what the open sources support.

THE A-135 SYSTEM AND ITS INTERCEPTORS

A-135 is built around one radar and two kinds of interceptor in underground silos. The full list below is from ru.wikipedia, English Wikipedia, vpk.name and the Military Balance 2023, and it matches the details on A-135 / A-235 Nudol.

Under the treaty-era limits of one site and 100 interceptors, the Moscow site has never been more than a small system. Pavel Podvig quoted a mid-1980s memo (by Vitaly Kataev) that the system "will be completed in 1987 to provide protection from a strike of 1-2 modern and prospective ICBMs and up to 35 Pershing 2-type intermediate-range missiles". Its capacity against a large strike is limited, as the memo implies.

HOW AN ENGAGEMENT WORKS, STEP BY STEP

This follows the published descriptions of the A-135 design. The real procedures and software are secret, and the system has never been used against a real attack.

NUCLEAR WARHEADS AND THE MOVE TO NON-NUCLEAR KILL

The 51T6 and the 53T6 carry nuclear warheads. In the 1970s hit-to-kill against a cloud of decoys was not trusted, and a nuclear blast can destroy warheads and decoys together. That means that if A-135 were used as designed, there would be nuclear explosions over the Moscow region. In 2011 Major General Vladimir Derkach told the Valdai visitors that Russia planned to move to conventional warheads only as detection, targeting and interception improved (reported by Richard Weitz). Modernisation materials since then have spoken of conventional or kinetic options for the new short interceptor and for A-235 as a whole. No source states that any silo round has actually been converted. The only safe statement is that conversion is the declared direction.

A-235 AND NUDOL: WHERE DON-2N FITS

A-235 is the planned modernisation of Moscow's defence. Public design history: a research project called "Samolet-M" (Самолёт-М) under the state contract of 31 January 1991 with the Radio Instrument Engineering Research Institute (НИИРП) of Vympel, due in 2015; after the 1990s stall it was taken up again by Almaz-Antey from about 2011. The planned elements on paper included a command-computing post, an information radar complex "Kiev", a multi-function radar "Narva", the Amur firing complex, the Kivach and Ilek firing complexes, an aircraft-based optical-electronic complex "Onega" for atmospheric discrimination, and three interceptor layers: an A-925 (51T6) modification at long range, a medium-range 58R6, and the PRS-1M (45T6) at short range. These are paper lists from ru.wikipedia and may not describe what was fielded.

Don-2N is the existing radar that all of these would use. Some Russian writers give a "Don-2M" as the radar of A-235 (ru.wikipedia, citing a 2012 article); others assume the old Don-2N, upgraded. The 2021 modernisation of the transmit and receive parts and the computers fits the second reading. The role of the station in the updated system has not been given officially.

"Nudol" (Нудоль) in Western writing usually means something narrower: the road-mobile 14Ts033 complex (14Ц033), with the 14A042 interceptor, 14P078 command post and 14Ts031 radar. The US calls the missile PL-19. It has been flown from Plesetsk, not Sary-Shagan, many times (see the testing section) and it destroyed the defunct satellite Kosmos 1408 at about 470 km on 15 November 2021. The Secure World Foundation assesses it as a separate direct-ascent anti-satellite project rather than the successor of A-135, and leaked-document reporting (UNITED24 Media) says the same question remains open. This article keeps them apart: A-135 / A-235 is the Moscow ABM belt that Don-2N guides; Nudol is a mobile ASAT-capable interceptor whose place in that belt is argued. See A-135 / A-235 Nudol.

OTHER JOBS: SPACE SURVEILLANCE AND THE ODERACS TEST

In peacetime the main business is space. The station tracks satellites for the Main Space Situation Centre and, as the 2011 and 2024 visitors were told, helps avoid collisions and follows launches, orbit changes and decays. In the 2024 report the crew tracked a satellite at 810 km falling to 720 km, probably before deorbiting. The same report claims the station follows every Starlink launch, which could not be checked.

The best public calibration of its sensitivity is the ODERACS experiment (Orbital Debris Radar Calibration Spheres). In February 1994 the US Space Shuttle mission STS-60 released six metal spheres, three pairs of 15, 10 and 5 cm. Russian and American radars saw the larger ones. Don-2N, according to a 2024 article, was the only participating radar to detect the smallest, 5 cm, spheres, and did so at an altitude of 352 km, with tracking at a range of about 800 km. ru.wikipedia says the detection and tracking range was 500-800 km, and says "1996" for the experiment, but the Shuttle release was in February 1994 (STS-60). The statement in the old version of this page, "1,500-2,000 km from the 1994 balloon test", is incorrect: they were solid spheres, not balloons, and the tracking range was up to about 800 km.

TESTING HISTORY

COVERAGE AREA

From the display shown to the Valdai visitors in May 2011: the station covers most of Europe except Spain and Portugal, the western half of the former USSR and a large part of the Middle East. The southern edge runs through the Mediterranean from the French-Spanish border, south of Sicily and along the Egyptian coast; Syria and Iraq are covered, northern Jordan too; in Central Asia it includes Turkmenistan, Uzbekistan and Kazakhstan but not Kyrgyzstan or Tajikistan. The hosts said it could see any missile launched from Iran towards Europe. This is a display seen by a visitor, not a published specification, and the shape is for the working mode shown that day.

TREATY CONTEXT

DON-2N COMPARED WITH OTHER RADARS AND SYSTEMS

LIMITS AND VULNERABILITIES

Public analysis of these points is thin and mostly comes from foreign writers. What follows combines what the sources say with my own assessment, marked as such.

COST (APPROXIMATE)

No official cost is published. Two data points exist. A 1960s estimate for one 90-degree sector station of the earlier 5N12N project was 600 million roubles; four would be about two billion. The head engineer of the Dnepropetrovsk plant wrote that the equipment of the prototype and the standard radar was "more than a billion dollars in prices of the time". Neither number is a verified total, and exchange rates in the Soviet period were administrative. The construction workforce (over 20,000 builders) and materials (tens of thousands of tonnes of metal and concrete) give a sense of the scale.

RUSSIAN STATUS AND MODERNISATION

STATUS AND OUTLOOK (OCTOBER 2026)

As of 16 October 2026 Don-2N is the operating battle-management radar of Moscow's A-135 system, modernised in place, with its Elbrus-2 computer hall obsolete and the work moved to new Russian machines. The 53T6 silos around Moscow are still the live shooters; the 51T6 layer is off duty. The new interceptors (53T6M / PRS-1M, and the A-235 long-range echelon in whatever form it takes) have been tested but no public statement says the full A-235 is on duty. The mobile Nudol has flown from Plesetsk and has hit a satellite once; whether it is part of the Moscow belt is not settled. The most defensible one-line status: Moscow still has a nuclear-era silo ABM defence with an upgraded radar and computers, and a long, partly secret programme to make it conventional and longer-ranged.

WHAT IS UNCERTAIN IN THIS ARTICLE

TIMELINE

RUSSIAN TERMS USED IN THIS ARTICLE

IN ONE SENTENCE

Don-2N is the single four-faced centimetre-wave phased-array radar at Sofrino that picks warheads out of decoy clouds for Moscow's A-135 missile defence and steers its interceptors by radio command, built over two decades from 1969 around a purpose-built digital processor and Elbrus-2 computers, moved onto new Russian computing in the early 2020s, and still the only radar of its kind in the world.

Project Orlan // Underground Information Database