The Type 346 radar: China’s answer to SPY-1

Type 346 active phased array radar is the mainstay of PLAN’s destroyer fleet long range missile defense. It first appeared in 2004 onboard the Type 052C destroyer as the principal sensor for the indigenous HHQ-9 long range SAM system, a Chinese derivative of the Soviet S-300. The phased array radar came as somewhat of a surprise and was immediately engulfed in mystery and speculation. In this article I analyse in detail its potential performance based on publicly available information and compare it to the USN’s SPY-1 radar.


Initially, the theory was that this radar is a derivative of a Ukrainian Kvant-Radiolokatsiya’s (State Research Institute on Radar Systems) APAR design. Norman Friedman speculated that the radar most likely operates in C-band (4-6 GHz), on the grounds that other (Western) radar manufactures struggled to package sufficient radiative power in S-band (2-4 GHz) Tx/Rx modules small enough to fit this kind of radar1. Furthermore, there was the fact that the HT-233 target tracking and fire control radar for the land based HQ-9 SAM operated in C-band and it was reasonable to assume the same for the naval variant of the system.

Only known image of the Type-346 array. The rectangular arrays on the top and bottom are speculated as either C-band missile tracking/control radar/datalink arrays or IFF arrays.
Source: Chinese Internet

According to unverified information, the main array within the Type 346 radar is a 4m octagon. This is close, but somewhat above my own estimate of 395cm x 436cm for the entire assembly, assuming that the sailor above the bridge stands 185cm tall to the tip of his hat (3.7cm per px). Based on the photo above, the width of octagonal array is 95% of the assembly width, which puts the array width at 371cm. The height of the octagonal array appears to be 82% of the assembly height, which puts the array height at 358cm. This is just slightly below the SPY-1 internal array dimensions of 384cm in height and 367cm in width.

Measurement of Type 346 radar panel dimensions.
Source: Wikipedia


According to unverified reports, the Tr/Tx modules are quad-packed (one controller for 4 Tx/Rx) similar to UK’s Sampson radar. Because of the technological state-of-the-art at the time, this was the only way to manufacture modules small enough to fit within the confines of the antenna aperture. The same source claims that the peak power per quad pack module is 100W4. The total number of transceivers per array is claimed as “greater than 5000”.

However, based on my estimate the aperture of the S-band array is around 10.5m2. Assuming an operating frequency of 3400 MHz (like SPY-1) and a dense rectangular layout of T/R elements with half lambda spacing, there is room for just 4877 elements. The distance between T/R elements is driven by three primary factors: power output, scanning field-of-view (FOV) and cost. For a wide FOV of ± 60º and a given operating wavelength λ, the separation between T/R elements in a square distribution of elements must be at most 0.536λ to avoid grating lobes (large radar beams adjacent to the main beam). This implies an antenna area of 0.278λ2 per element3.

We can use the above formula to estimate the number of T/R elements for the given aperture of 10.5m2. At 3500 MHz (8.5cm wavelength), there is room for 5227 elements and at 3600 MHz for 5447 elements. Based on the shape of the radar array cooling hood, the spacing between elements is densest in the centre of the array and progressively sparser towards the edges. This is a common feature in phased array radars. Therefore, an operating frequency centered around 3,600 MHz is a plausible guess.

Comparison in radar power to SPY-1

Let’s assume that the transceiver count is 5000 per array. A quad pack with peak power of 100W implies that each array has a peak power of 125kW. Peak power per T/R element is 25W. In the late 90s, the British MESAR-2 S-band AESA radar had 10W peak power T/R elements, so this constitutes a major improvement in a span of about 6 years. If the true, that would imply that China was not significantly behind the West in the development of competitive naval AESA radars as was once believed.

If we assume a reasonable duty cycle of 20%, the average power per array (before antenna) comes out to 25kW. It is unclear if these figures represent radiated power or power delivered from the TRM to the antenna. If the former, then we would have to account for antenna losses which can range from 20%-40%.

Comparing it to the USN SPY-1 radar is not easy, because there is quite a bit of seemingly contradictory information on the radiated power of that family of radars. According to a IEEE paper from 1988, the original SPY-1A radar generated pulses with peaks of 5MW, but the average pulse power is 32 kW. It is not clear from the paper whether that was radiated power or transmitter power. Furthermore, unlike an AESA radar, the SPY-1A radar did not radiate simultaneously from all 4 radar faces. It has two transmitters sets: one in the forward deckhouse and one in the aft deckhouse (Ticonderoga cruiser). Each transmitter is timeshared between two radar faces. It is unclear whether the 32kW figure is per deckhouse or the combined average power of all 4 radar faces.

The updated SPY-1B radar on the Ticonderoga class cruisers doubled the duty cycle of the transmitter and apparently upped the peak power to 6MW. The average power, according to Friedman, is 58 kW. This is significantly less than 76.8 kW that would be derived by scaling the SPY-1A accordingly. We do know that the phase shifter was upgraded from 4 bits to 7 bits and that this resulted in a higher insertion loss, but that would explain only a part of the difference.

SPY-1D is the destroyer version of the SPY-1B radar. In this system there is only one transmitter set, and it is timeshared by 4 radar faces so that at any single point in time just one radar face is emitting. As an additional piece of information a 2004 report stated “the average radiated power aperture for the Aegis System is 485 kwm2.” Based on the publication date, it can be assumed that the reference was to the SPY-1D radar. Assuming a SPY-1 antenna area of 12 m2, the average emitted power comes out to about 40 kW, or 10kW per radar face.

40kW radiated power is actually a surprisingly high output for a transmitter power of 58kW, given that the 7 bit phase shifter alone has a maximum allowed insertion loss of 1.35dB or ~36%. This is not taking into account the waveguide loss, the corporate feed losses and the antenna loss.

The SPY-1D (V) is a further improvement, wherein the duty cycle of the radar was increased by at least 33%, so that the average power is at least 58kW*1.33 = 77 kW.11. According to the source, this is apparently the combined transmitter power of all 4 radar faces. This radar also received the capability to dual-beam: that is to emit from two opposite radar faces simultaneously, ostensibly with pulses half the power compared to single beaming.

SPY-1 radars use a cross-field type amplifier characterised by very high efficiency (up to 70%) and compact size but low gain. If the SPY-1D(V) radars still have the same peak power of 6MW, then the average power of 77kW implies a duty cycle of 1.28%. This is a plausible number, but is less than half of what CFA tubes of comparable power and bandwidth were capable in the late 80s14.

Assuming that the CFAs have 60% efficiency, the input power into the HPAs would be 128kW. This appears inconsistent with the claim that the SPY-6 radar uses roughly double the electrical power of SPY-1 which resulted in an increase in total ship power consumption by almost 1.5 MW18. A part of the power increase also went into a greater need for cooling, but it is safe to assume that the SPY-1 radar requires at least 1MW of input power.

Based on the publicly available figures it is possible to estimate the antenna efficiency of the SPY-1 radar from the formula: G=\eta * D, where G is gain and D is directivity. According to Friedman, the antenna gain is 42dB and the beam width is 1.7×1.7 degrees. Therefore, the antenna efficiency \eta=0.8. In the table below, we will assume the same efficiency for the Type 346 radar:

RadarAperture [m2]Radiated power per antenna [kW]PA product [kWm2]Round trip sensitivity ratio
SPY-1D(V) 1213.3 = (40kW*1.33)/4159.61
SPY-1D(V) hypothesis12759005.64
SPY-1D(V) hypothesis with antenna loss12607204.51
Type 346 (3db advantage on receive)10.4625261.53.28
Type 346 with antenna loss10.4620209.22.62
Comparison of key radar parameters assuming all radars radiate at the same frequency and the AESA radar has a 3dB loss advantage on receive

Large PESA radars generate a very powerful signal from a vacuum tube amplifier, transport it through a waveguide, divide it in a cascade of corporate feed networks and channel the subdivided signals through ferrite phase-shifters: a process that can halve the input power by the time it reaches the antenna. In comparison, AESA radars place the phase-shifters before the high power amplifiers on the transmit path, and after the low noise amplifier on the receive side, minimising losses:

Taking into account the assumed 3dB sensitivity advantage of AESA on receive, the Type 346 radar may be over 3 times as sensitive as the SPY-1D(V) radar on the Flight IIA Arleigh Burke destroyers. The caveat being that the claimed average radiated power for SPY-1 appears suspiciously low given its power consumption of at least 1MW. If we assume a HPA efficiency of 60% and a transmission loss of 3dB (50%) after the transmitter, the total radiated power could be 300kW or 75 kW per radar face. In the early 2000s, vacuum tube amplifiers held a significant advantage in efficiency over solid state amplifiers. Today, that is no longer the case as GaN HPAs routinely reach 70% PAE efficiency, although vacuum tubes still hold an advantage in power and efficiency at very high frequencies.

Even though PESA radars typically have good sidelobe management, AESA radars go a step further by being able to adjust both the amplitude and the phase of each radiating element. Small sidelobes are critical in reducing clutter and decreasing a radar’s susceptibility to jamming. Therefore, such radars will have longer detection ranges against targets when subjected to jamming. Furthermore, the narrower the beam the greater the likelihood a radar can distinguish a target from a jamming platform. Another advantage of AESA radars is that they operate with peak powers that are an order of magnitude smaller compared to PESA radars. This significantly reduces the range/probability at which enemy ESM will detect them.

Power consumption

Type 346 uses GaAs solid-state HPAs (high power amplifier). The added power efficiency of these devices is typically in the range of 20%-60%, depending on radiated frequency, bandwidth and power. The higher the three parameters, the lower the efficiency. T/R modules typically account for 70-80% of thermal losses in an active array.

Assuming a 35% PAE efficiency, the quadpack T/R module efficiency would likely be around 25%17, and each TRM would need 25W/0.25 = 100W input power per module. Because there are 1250 such modules per array, they need to be supplied with 125kW of input power. Furthermore, assuming the T/R module losses account for 70% of the total array losses, the total input power into the array comes out to 100kW/0.7 (losses) + 25 kW (useful work) = 167.86 kW. For the combined 4 arrays, that is 671.4 kW of electrical power. But that’s not all. Because active arrays generate large amounts of heat, they need to be actively cooled to remain within operating temperature range. To dissipate 571kW of heat =671.4kW-100kW (transmitted power) of heat, requires about 162 tons equivalent of chilling (1 ton = 3.517 kW). That much refrigeration can require close to 200kW of input power. Therefore, the entire radar system could conceivably consume just under 900 kW of power.

Mysterious rectangular arrays

Another unanswered question is the function of the narrow rectangular arrays at the top and bottom of the Type 346 radar arrays. The mainstream theory is that these are C-band arrays for target illumination and/or missile tracking radars with uplink/downlink command function. Alternatively, the main array could perform both missile tracking and uplink/downlink (like SPY-1) and these could be IFF transponders instead. On the Type 052D, the Type 346 array face is more square in shape, and what appear to be the same rectangular arrays have now been repositioned outside the radar and on top of the bridge:

IFF interrogators or missile tracking/uplink radars on the Type 052D destroyer. Source: CGTN video

According to information floating around the Internet, these arrays are designed to pickup the beacon signal emitted from the rear of the HHQ-9 missile and keep track of the missile as it flies towards the target12. They are also present on the PLAN’s carrier Shandong. On the Type 055 destroyer these arrays have grown significantly in size with the potential implication that the Type 346 radar variant on that ship is substantially more powerful compared to the radar on the Type 052C/D destroyers and therefore its range exceeded the capability of the smaller IFFs.

The much enlarged IFF arrays/FCR
Source: Chinese Internet

Type 346A

The radar panels on the Type 052D destroyer have been estimated at about 4.3mx4.3m: roughly the same in height and almost 10% wider than those on Type 052C. Because the rectangular arrays have been moved out, there is now significantly more area left for the S-band octagonal array. The radar faces are flat and do not significantly protrude from the hull, indicating that this design uses liquid cooling. This new radar has received the designation Type 346A.

Assuming the same 5% edge clearance, by my estimate, the array area is about 13.1m2. While we know that the updated variant of the radar uses liquid cooling, we have no information whether this led to a significant increase in the radiated power or was primarily used to mitigate the weight impact of the larger radar arrays: liquid cooling does not require as large a heatsink to be attached directly to the arrays as in an air cooling solution because it is more flexible in the placement of heat dissipating surfaces.

Type 052DL destroyer. Visible under the bridge are the large square shaped Type 346A radar faces.
Source: Chinese Internet via

If we scale the original Type 346 array to the size of the Type 346A array, we will arrive at a T/R element count of 6262 and a radiated power of 31.3 kW. This is probably a lower limit estimate, as it is safe to assume that in the ensuing years the efficiency and/or the radiated power of each element was improved.

Another uncertainty in making a power-aperture product comparison are the operating frequencies. According to Norman Friedman, the SPY-1 radar operates in the range of 3.1 GHz to 3.5 GHz. Based on the alleged element count of at least 5000 elements for the original Type 346 radar, its operating frequency band must be higher (~3,600 MhZ), in the upper reaches of S-band. A higher frequency makes narrower beam widths possible, but can result in inferior atmospheric attenuation characteristics impacting the effective range of a radar. Furthermore, many targets have higher radar cross sections at higher frequencies negating part of the gain advantage.

RadarAperture [m2]Radiated power per antenna [kW]PA product [kWm2]Relative roundtrip sensitivity to SPY-1D(V)
Type 346A (low estimate)13.131.31410.25.14
Power aperture product comparison of Type 346A and SPY-1 operating at the same frequency

If we assume a 3 dB (~2x) relative loss on the receive side for SPY-1 compared to Type 346A, the Type 346A has the potential to be about 5 times as sensitive as the SPY-1D(V), absent differences in signal processing. Furthermore, on account of Type 346A being an AESA, it is reasonable to assume that this radar will have superior performance when subjected to adversarial jamming due to better sidelobe suppression.

Given that the Type 052D is a relatively small platform that has significantly grown since its origins in the Type 051B destroyer hull (6200 tons) the latest radars might not be able to demonstrate their full potential due to a lack of electrical and cooling power. Ships of Type 052D class are significantly smaller than the Arleigh Burke class destroyers: 7,700 metric tons and 9300 metric tons6 of displacement, respectively. The latter have 9MW of onboard electrical power generation capacity, a third of which is reserved for redundancy purpose with close to 2MW of extra growth headroom: the total operational electric load is slightly above 4MW.

Therefore, it is more realistic to compare the Type 052D to ships closer to its displacement. We should also keep in mind that the Type 052D is essentially a burdened Type 052C hull (7,000 tons displacement) and does not appear to have any additional internal volume.

ClassDisplacement (full)Generator numberTotal Power
Akizuki6,600 mt37.2 MW
Horizon7050 mt46.4 MW
Hobart7000 mt48 MW
Kolkata7400 mt44 MW
Comparison of power generation capacity on destroyers comparable in displacement to the Type 052C destroyer. Source: 7,8,9,10

Based on empirical data, it is reasonable to assume that the Type 052C/D destroyer class has anywhere between 4-8 MW of electrical power generation onboard. This comparison did not take into account fully electric or hybrid electric ships like the FREMM.

Weapons guidance

The principle air defence weapon associated with the Type 346 family of radars is the HHQ-9 missile. The original HHQ-9 (naval HQ-9) missiles were speculated to have the kinematic performance of either the S-300F (90km range) or the S-300FM (150km). The latter were bought from Russia and installed on the two Type 051C destroyers, possibly as a back up measure or even as a yardstick for performance validation of PLAN’s new SAM and radars.

S-300 missiles use a variant of command guidance called TVM (track-via-missile). In this system the fire control radar tracks both the target and the missile and commands the missile to follow an optimal impact trajectory. Unlike standard command guidance, the missile itself has a radar receiver which detects the pings from the fire control radar and relays them back to the ground control station. In effect, this is a bi-static radar system with one transmitter and two receivers producing a more accurate position and velocity of the target. Because the fire control radar does not need to illuminate the target as in a semi-active guidance configuration the target finds it much more difficult to discern whether it is merely tracked or fired upon with missiles incoming.

HHQ-9 missile launched from a Type 052C destroyer. Source: Chinese Internet via

It is understood that the original HQ-9 systems adopted the same form of guidance. However its HT-233 FCR operates in C-band, while the Type 346 operates in S-band. As a result the latter has both worse angular and radial resolution and would not be able to generate a fine enough solution on the target to guide the missile to impact at long range. This is the principal reason why fire control radars operate in C, X or higher bands. In absence of an obvious illuminator on the Type 052C two hypothesis present themselves:

  • the rectangular arrays serve as target illuminators for the HHQ-9 missiles
  • the HHQ-9 missiles employ active radar homing (ARH)

In the late 2000s, during the Zhuhai expo China offered a derivative of its HQ-9 system called FD-2000 for export. A notable feature of this system is that the missiles employ active radar homing, as can be discerned from the placard below:

FD-2000 missile with active radar homing
Source: Chinese Internet via

Type 052D class are equipped with an updated HHQ-9B missile. These missiles have dual-mode ARH and IR seekers. Their range is estimated at 230km+5, which puts them close in kinematic performance to USN’s SM-6 missiles and is well within the detection range capability of its radar.


The Type 346 radar remains shrouded in mystery. On the face of it, it appears to be a very capable system. In some aspects, like its AESA architecture it is technologically ahead of the USN’s AEGIS SPY-1 radar.

The new variant on the Type 052D destroyer appears to not only have larger outward dimensions, but is also internally larger on account of moving out the IFF/missile trackers and placing them on top of the bridge. The new radar is physically more powerful and likely more advanced in terms of signal processing. Based on a back-of-the-envelope estimate, its performance should be at least equal to that of the SPY-1D(V) radar on USN Arleigh Burke destroyers, but most probably is superior in most roles, except possibly BMD where the SPY-1 radars can channel all their power through a single radar face.

The USN will soon get its first SPY-6 radar with the new Arleigh Burke Flight III destroyer. The Navy has reported that the new radar is almost 100 times more sensitive than SPY-116 it intends to replace. That means that SPY-6 has the potential to have a range 3.16 (\sqrt[4]{100}) times superior to SPY-1 against a target of equal radar cross section.


  1. The Naval Institute Guide to World Naval Weapon Systems, 5th Edition; Norman Friedman; Naval Institute Press, 2006
  2. John A. Adam, “Pinning Defense Hopes on Aegis,” IEEE Spectrum, June 1988
  3. The LRDR: (Not) The Best Discrimination Money Can Buy? (January 30, 2019)
  15. Defense Science Board Task Force, “Contributions of Space Based Radar to Missile Defense,” June 2004, p. 2.
  17. On the use of AESA (Active Electronically Scanned Array) Radar and IRST (InfraRed Search&Track) System to Detect and Track Low Observable Threats; George-Konstantinos Gaitanakis, George Limnaios, and Konstantinos C. Zikidis; 2019

Leave a Reply

Your email address will not be published.