Xiaomi BE3600 1.25A revision: less shielding, the same core platform

ACWiFi opened a white Xiaomi BE3600 sample using a 12V/1.25A adapter, produced in April, and compared it with the earlier 1.5A-powered version. This article focuses on observations from that sample, not a new measurement table.¹

What the opened sample shows

The sample has one 2.5G port and three gigabit ports. ACWiFi recorded about 6 W with no network cable, 6.4 W with a gigabit WAN connection, and 7.9 W with a 2.5G WAN connection. Similar figures to the 1.5A version suggest that the core platform did not change, but they cannot prove that every production batch uses the same PCB.¹

The obvious changes are mechanical: both sides of the board have three fewer shield cans, two electrolytic capacitors are gone, and the aluminum spreader is shorter and narrower. That explains why two BE3600 units with the same product name can show different teardown photographs or thermal behavior.¹

Chip and radio platform

The CPU remains a Qualcomm IPQ5312 with four A53 cores up to 1.1 GHz. The integrated 2.4 GHz radio supports 2x2 MIMO and up to 688 Mbps at 40 MHz, with a KCT8245SD FEM. The RS128M16V8DB-107AT memory is DDR3-1866 with 256 MB capacity.¹

The 5 GHz side uses a QCN6402, 2x2 MIMO, 160 MHz, and a stated maximum of 2882 Mbps at 4096-QAM, paired with a KCT8576HE FEM. The gigabit switch is YT9215S, the 2.5G PHY is QCA8081, and the GD5F1GM7REYIG flash has 128 MB.¹

ACWiFi also lists gain, noise figure, and output-power conditions for both FEMs. Values such as +20.5dBm, +23.0dBm, and +24.0dBm are tied to HE/VHT/HT modes and 5 V conditions; they are not a universal transmit-power result for every region or firmware.¹

Practical reading

For a low-cost BE3600 with 2.5G, the 1.25A sample appears to retain the CPU and radio design while reducing shielding, capacitors, and heatsink size. I would verify the label revision, included adapter, firmware, and return policy before buying; I would not use this board photo to infer flashability of another revision.¹

The source tested five phones at its K point in a normal home environment, but it was not a professional benchmark. I therefore do not infer aggregate throughput, long-term temperature, shield durability, or OpenWrt support from the teardown and a few speed-test screenshots.¹

Source and limits

The figures below describe the sample and conditions documented by the source; another revision or batch may differ.

Sources