ZAK-30 "Citadel": How a 30mm projectile learned to think for itself
In the May 2026 video recordings, everything happens in a matter of seconds. On the screen are capture markers, a short burst, and Ukrainian "Fierce" It scatters in the air. There is no direct hit: there is a cloud of fragments a few meters away. drone, at the point where it should have been. The anti-aircraft gun, it would seem, had long been displaced rockets, is returning. It didn't improve its accuracy. The shell simply now chooses its own moment to fire. This idea has a history spanning more than eighty years, and it began not in the Moscow region, but in the Pacific Ocean in 1943.
"Lyuty" in a cloud of shrapnel
The first public footage of combat use ZAK-30 "Citadel" appeared in military-related Telegram channels in May 2026, a few days before Rostec announced the system's premiere at the First International Security Forum in the Moscow region. The video shows the system's interface in semi-automated mode: trackers on targets, acquisition, and lead calculation. The target is AN-196 "Lyuty"A kamikaze drone of the aircraft type, according to open estimates, 4,4 meters long, with a wingspan of 6,7 meters, a weight of 250–300 kg, a cruising speed of approximately 150 km/h, and a range of over a thousand kilometers. Not fast or maneuverable, but low-altitude, stealthy, and capable of carrying out mass raids on rear-area targets.
Rostec talks about the complex itself:
The Citadel has both optical-electronic and radar detection and tracking systems. drones The optical channel operates in the visible and infrared ranges. Furthermore, the system can fire projectiles with controlled detonation, and its operation—from target detection to destruction—is highly automated.
Precise performance characteristics (firing range, rate of fire, ammunition capacity, reaction time) have not been officially published. Estimates in open sources vary, and most conclusions about the system are based on a few minutes of video and official press releases. But the main point in these videos is clear even without the numbers: the drone is not hit. There is a cloud of fragments at the calculated point of the trajectory, and this is enough to cause the Lyuty's plywood-composite structure to disintegrate in mid-air. The idea behind this is significantly older than the system itself: its basic principles were conceived even before the invention of the transistor.
A laboratory in Maryland and a V-1 over England
In January 1943, the American cruiser Helena shot down a Japanese aircraft near the Solomon Islands – one of the first documented cases of the use of a new type of projectile designated VT-fuse (Variable Time). Each contained a miniature Doppler radar: four vacuum tubes, an antenna, a power source, and a detonating mechanism. The projectile would automatically locate its target and decide when to detonate. Development was led by physicist Merle Tuve (since 1942 at the Johns Hopkins University Applied Physics Laboratory in Silver Spring, Maryland). The project was considered so secret that VT projectiles were initially only allowed to be used over sea, to prevent unexploded samples from falling into enemy hands.
Radar fuze (VT fuze), developed during World War II
Real glory came in the summer of 1944, when anti-aircraft batteries on the south coast of England met a wave V-1By September, the effectiveness of the projectiles had increased exponentially compared to projectiles with a conventional time fuse: cruise missiles were no longer shot down by hundreds, but by dozens. Before the VT, anti-aircraft gunners set the detonation time before firing, using firing tables and a stopwatch; now the projectile itself knew when the target was close.
ZSU-23-4 Shilka
The Soviet school chose the opposite path – the density of fire. C-60 (The 57-mm AZP-57 anti-aircraft gun, part of a battery system, was adopted in 1950) still maintained the classic design with radar guidance and a conventional projectile. But then the emphasis shifted radically. ZSU-23-4 «Shilka» (1962) – four 23-mm barrels, up to 3400 rounds per minute per installation, 1RL33 radar as part of the RPK-2 Tobol radio-instrument complex. Marine AK-630 (adopted into service in 1976) is a six-barreled artillery mount capable of firing up to 5000 rounds per minute. The projectile is a standard high-explosive fragmentation shell, without any electronics. The principle is the same: to create a wall of fire that a small target cannot penetrate.
AK-630M
The VT and AK-630 have the same mission—shooting down a small aerial target—but they accomplish it with diametrically different means: one smart projectile versus five thousand simple ones per minute. By the 2020s, both schools of thought had hit a wall. Missiles aren't available for every drone; they're more expensive than the drones themselves. And they push through a wall of fire with sheer mass; five thousand rounds per minute sound impressive until there are fifty targets in the sky at once.
AHEAD: Intelligence from a Projectile to a Ground-Based Computer
The third move was first found by the Swiss Oerlikon Contraves (joined by Rheinmetall in 1999). Ammunition AHEAD (Advanced Hit Efficiency and Destruction), perfected by the late 1990s: 35 mm, with 152 tungsten fragments inside, and an electronic timer in the tail. The key here is this: the detonation intelligence was moved from the shell to a ground-based computer, externally, near the gun barrel. Induction coils are located at the muzzle; as the shell passes through, the fire control system measures its initial velocity and records the precise time to detonation, calculated using radar data, in the timer. The shell is inexpensive: no radar, no target sensor, just a timer and propellant.
A programmable projectile using AHEAD (Advanced Hit Efficiency and Destruction) technology.
With AHEAD, the barrel-mounted anti-aircraft gun has become popular again. German MANTIS (in service since 2011, airbase protection) - six separate 35-mm artillery mounts (Revolver Gun) with programmable ammunition, rate of about 1000 rounds per minute per barrel. Marine Skyshield, recent wheeled Skyranger 30 already on the 30-mm AHEAD, with the addition of small SADM missiles from MBDA. By the mid-2020s, Germany's Nah- und Nächstbereichsschutz program effectively cemented the C-RAM gun layer as a standard element of the echeloned Defense.
Skyranger 30
The Citadel is based on the same concept, but with its own unique solutions. Caliber: 30 mm; according to some observers, it is based on the Spica module (in some publications with the index BM-30-D) with a 2A42 automatic cannon (Instrument Design Bureau, Tula, in production since the early 1980s, the base weapon of the BMP-2, Mi-28, and a good half of ground vehicles). Rostec has not officially confirmed this. The 2A42's rate of fire is 200-300 rounds per minute at low rates and 550-800 at high rates, compared to 1,000 rounds per barrel for the 35mm Oerlikon. For the classic C-RAM mission of repelling a salvo of rockets, this is insufficient. The programmable projectile partially compensates for the difference: fewer shots, but each one is smarter. Whether this makes up for the difference against truly massive air strikes is an open question.
Layout module BM-30-D arr. 2016 Photo by Central Research Institute Burevestnik
How exactly the Citadel programs the fuse is unclear from open sources. Some publications describe the system as laser-based: a pulse encoding the detonation time is transmitted to a receiver in the shell. Other observers suggest an induction system similar to AHEAD; both are technically possible with a 30mm caliber. The logic is the same in any case: the shell only needs to count the milliseconds and detonate the charge at the desired point.
Each of the three elements of the Citadel was conceived before it. That the projectile should automatically detect the moment of detonation was realized back in the 1940s with the VT. That it's more convenient to load time directly into it at the muzzle was figured out at Oerlikon in the late 1990s. So, all that remains of the Citadel is automatic target tracking, which media reports refer to as artificial intelligence. The video does indeed show trackers, classification, and automatic target acquisition. It's impossible to say from open sources what exactly the neural network does, or whether it does anything at all. It's a convenient term, but impossible to verify.
Cheap Shot and the Limits of the Concept
The main argument in favor of a cannon-based anti-drone layer is the cost per shot. A modern anti-aircraft missile, even a short-range one, costs tens of thousands of dollars; a programmable artillery A shot is an order of magnitude cheaper. With massive drone strikes, this ceases to be a pleasant bonus and becomes a condition for the system's existence: missile ammunition runs out before the drones do.
Rheinmetall's Skyranger 30, the American M-SHORAD program (Stryker with the XM914 30mm cannon), Korea's Hanwha with a programmable 30mm munition, the next-generation MANTIS in the German Nah- und Nächstbereichsschutz program with a 2028 target date, BAE Systems with its own variant—the list goes on. So, Citadel is more of a follower than a pioneer: the same challenge—making a programmable round less expensive than a conventional one—is being tackled in parallel in several countries, and it hasn't yet been fully resolved.
What the system can do is serve as the last line of defense for a specific facility. Its effective range, judging by similar systems with a 30mm cannon and an optical-electronic station, is approximately one to two kilometers. Long-range missiles are still held by the Pantsir-S, S-300, and S-400 systems. EW and fighter aviation"Citadel" is already working in the inner ring, due to what has broken through.
There are plenty of open questions. The main one is throughput: how many targets the system can track and engage simultaneously. This figure hasn't been published, but it determines what would happen during a truly massive attack: twenty to thirty Lyuty missiles at a single point from different directions. Then there's the performance of the optical channel in smoke, under laser illumination, and when operating under electronic warfare jamming; nothing is mentioned in the publicly available materials. The actual operational statistics are a few videos and a press release, which can't be used to judge either the average round rate per target or the tracking failure rate. This is the normal state of knowledge about the system just revealed, and it will change within the next year or two, either through official data or through what the other side begins to publish.
Information