Felicity Solar

Felicity Solar FLA48100 51.2 V 5.12 kWh LiFePO4 Battery

Great for the price, good cell quality, limited or nonexistent BMS balancing

TWO FELICITY SOLAR LPBA48100-EU (LiFePO4 16S, 51.2 V, 100 Ah) IN PARALLEL In short: first-rate build, full readable comms, but a balancer that doesn't actually balance, and no BMS settings the user can change. USAGE - Pack 1 closely monitored ~20 days, pack 2 paralleled ~9 days. Luxpower hybrid inverter over CAN. - ESP32 reads the BMS via RS485 every 0.5 s and Bluetooth hourly; 300k+ samples. - ~7 equivalent cycles on pack 1 (36 kWh out), 2.4 on pack 2 (12.5 kWh). Early review: BMS behavior and comms, not durability. BUILD (STRENGTH) Solid finish and terminals. Cells ran 22-30 °C, up to 93 A charge / 78 A discharge, no alarms. ~1.2 mOhm per cell, ~3 mOhm cell-to-terminal (BMS incl.), ~3 mOhm parallel cable. Two RJ45 ports wired in parallel inside; DIP-switch address. Verified pinout: 1 isolated GND, 2 +5 V (remote power, untested), 3/4 CANL/CANH to inverter, 5/6 RS485 B/A, 7/8 CANL/CANH battery link. THE PROBLEM: BALANCING - Passive, only with cell average above 3.38 V. The minimum-difference parameter ships as 65535 (invalid, i.e. not configured). - Above 3.38 V, cell 1 shows a ~10 mV square wave every 25-40 s on both packs, whichever cell is highest. - After dozens of hours above 3.38 V, imbalance only grew. At median cell 3.43 V, ~10 A: pack 1 high cell went from +5/+15 to +68/+80 mV in two weeks; pack 2 from +13/+24 to +103/+117 mV in three days. Leading cells are 0.3-0.65 Ah ahead, not shrinking. Top-balance test: no effect. - Fleet (cloud data): 28 of 40 packs above 3.38 V show the same wave, always cell 1. Well-matched packs sit at 6-8 mV at the top (median); mine hit 60-90 mV, among the worst of ~125. Unbalanced from the factory stays unbalanced. Effect: the high cell hits 3.60 V first and the BMS stops charging early. In parallel one cell reached 3.639 V, 11 mV from OVP (3.650 V). SUPPORT AND FIRMWARE - Support was fast and helpful. They pushed Wi-Fi/BT module firmware 2.10 to 2.15 remotely. That's only the comms module; BMS firmware (where balancing lives) is unchanged. The square wave vanished, the Ah gap didn't. - Balance start voltage can be changed remotely (3.38 to 3.34 V on one pack). Minimum difference stays 65535, not changeable. - No BMS parameter is user-editable (LAN or BT): protections, alarms, currents, temps, balancing. Sending the set command directly: module acknowledges, value doesn't change. - Of ~490 cloud batteries reporting BMS version, all but one run 524. One BMS version in the field, no update in sight. COMMUNICATION RS485 (9600 8N1, Voltronic BMS 485): 16 cells, 4 temps, V, I, SOC, capacity, alarm/protection bits, status bits, dynamic limits sent to inverter, FW/HW versions. 0.5 s, most stable. No SOH, serial or config. Only the 117 documented registers exist, none writable. Responses aren't standard Modbus (register count instead of byte count), so common libraries reject them. BT / local Wi-Fi: all of the above but slower (8-16 s), plus SOH, 2-decimal SOC, serial, parallel state and the full config table, read-only. Since 2.15 BT needs pairing, one device at a time. CAN: identifies as "PYLON" (Pylontech LV). Inverter receives dynamic current limits (master sends the sum), combined SOC, SOH (always 100%), V, I, max/min cell V and temp (despite forum claims they're RS485-only), module count, and force-charge requests, which worked even with grid charging disabled. REAL-WORLD BEHAVIOR - Capacity: 98.56 Ah / 5.10 kWh full to empty at BMS request (NOT UVP/OVP) at 10-13 A, 25-29 °C; both together 195.3 Ah. Meets rating. - No float. Charging pauses when the high cell hits 3.60 V and shows 100%; drops to 99%, charges again 10-20 min at 10 A: "mini-tops" all day. - Requested charge current: 100 A up to 80% SOC, then 5 A less per point, 10 A at 98-99%. SOC runs ahead, so end of charge is throttled (saw 42 to 99% with only 27.7 Ah in). - SOC drifts: a spontaneous +19 point jump; 26% with only 2.4 Ah left, then straight to 0%. Fixes itself only at full/empty; do periodic full charges. - Discharge pauses at 2.75 V lowest cell; MOSFETs open at 2.50 V. - In parallel a pause is only a request to the inverter: a paused pack keeps flowing current until the other pauses too. Saw one go below 2.75 V and another reach 3.639 V. VERDICT Pros: excellent build, full capacity, low resistance, smart parallel ports, complete and stable RS485, CAN that works with a third-party inverter, responsive support. Cons: balancer that doesn't fix imbalance, invalid factory parameter nobody can change; no user BMS settings; single BMS version with no update; inaccurate SOC between recalibrations; cloud account required for the app. Recommended if you can live with the imbalance or add an active balancer. BMS firmware with working balancing (or a configurable minimum difference) would raise the score a lot. I will install a JK Smart Equalizer on each pack to work in parallel with the original BMS for the best balancing.