ESS Protocol (BMS block)¶

Source document: Growatt xxSxxP ESS Protocol, rev 2.3, 2017-11-28 (Protocols/1xSxxP_ESS_Protocol_rev2.3_20171128.pdf)

Why this page exists¶

Protocol V1.39 documents the BMS block at input registers 1082-1124 by name only. Against the first row, register 1082 BMS_StatusOld, it carries this note and nothing else:

Detail information, refer to document: GrowattxxSxxP ESS Protocol

So V1.39 gives you the register names for the whole block and defers the units, scales and encodings to a separate document. That document is the one above, and it is the reason several of these registers were mapped by guesswork for a long time.

This is not a register map you can poll

The ESS Protocol describes the link between the inverter, the battery and an application program — a separate Modbus conversation in its own address space (0x0010-0x0052). You cannot read these addresses from the inverter.

What it gives us is the meaning of the values the inverter republishes in its own input registers at 1082+. The address correspondence below is the useful part.

Address correspondence¶

Evident from the names, and confirmed on hardware where marked.

ESS address	ESS name	V1.39 input register	Unit / encoding
`0x0010`	Gauge IC current	1100 `BMS_GaugeICCurr`	10 mA
`0x0013`	Status	1083 `BMS_Status`	bitfield
`0x0014`	Error	1085 `BMS_Error`	bitfield
`0x0015`	SOC	1086 `BMS_SOC`	% ✅
`0x0016`	Voltage	1087 `BMS_BatteryVolt`	10 mV ✅
`0x0017`	Current	1088 `BMS_BatteryCurr`	10 mA, two's complement ✅
`0x0018`	Temperature	1089 `BMS_BatteryTemp`	°C, -127 to 127 ✅
`0x0019`	Max charge/discharge current	1090 `BMS_MaxCurr`	—
`0x001A`	Gauge RM	1091 `BMS_GaugeRM`	10 mAh
`0x001B`	Gauge FCC	1092 `BMS_GaugeFCC`	10 mAh
`0x001C`	YW / FW	1093 `BMS_FW`	byte 1 hardware, byte 2 software
`0x001D`	Delta	1094 `BMS_DeltaVolt`	V, cell voltage
`0x001E`	Cycle Count	1095 `BMS_CycleCnt`	—
`0x0020`	SOH	1096 `BMS_SOH`	bits 0-6 counter, bit 7 flag
`0x0021`	CV Voltage	1097 `BMS_ConstantVolt`	10 mV
`0x0022`	Warning	1099 `BMS_WarnInfo`	bitfield

✅ = confirmed against instruments on an SPH3620 (#397): SOC against the inverter's own sensor, voltage against a meter reading 54.2 V, current against a clamp DC ammeter reading 16.4 A, temperature against a thermal camera.

The correspondence is not a fixed offset — 1082 BMS_StatusOld and 1084 BMS_ErrorOld are inverter-side copies with no ESS equivalent, which shifts the run.

The three that matter most¶

Current is 10 mA, not 100 mA¶

0x0017 is documented in units of 10 mA, so the scale is 0.01 A. Raw 1640 is 16.4 A.

This settled a real disagreement. Four SPH maps use 0.01, confirmed against a clamp meter; SPH_8000_10000_HU used 0.1, which was neither documented nor measured and was most likely taken by analogy from register 3170 Ibat, which genuinely is 0.1 A on the MIN/MOD range. Corrected to 0.01.

Current is signed, and the encoding is explicit¶

The document's "See Current explain" table:

Range	Meaning
`0x0000`-`0x7FFF`	current is positive
`0x8000`-`0xFFFF`	current is negative

Plain two's complement. That is why these registers carry 'signed': True.

What it does not define

The spec fixes the numeric encoding and never states which physical direction is positive. Whether charging reads positive is a property of the firmware, not of this document.

On an SPH3620 it is the standard one — positive charging, negative discharging — confirmed in both directions against the reported power (#397):

State	Current	Voltage	Implied	Reported
Charging	+44.30 A	54 V	2392 W	~2.4 kW
Discharging	-54.00 A	53 V	2862 W	2.8 kW
Discharging	-4.20 A	53 V	223 W	220 W

Those also confirm the 0.01 scale across a 13x range of magnitudes, from 4.2 A to 54 A.

Still one model, though. See Invert Battery Power if your battery power reads backwards — noting that option affects battery power and not battery current.

Temperature is whole degrees¶

0x0018 is documented as °C with a range of -127 to 127 — whole degrees, signed. Not tenths.

This matters because V1.39 documents the inverter's own battery temperature at register 1040 as 0.1C, and at least one SPH3620 populates 1040 from the BMS value without rescaling it — so 25 °C arrives as raw 25 and the documented scale renders it as 2.5 °C. The integration detects and corrects that at read time rather than changing the scale, because spec-compliant inverters do report tenths there.

Bitfields¶

The document also defines the status, error and warning bitfields behind registers 1083, 1085 and 1099. The integration does not currently decode them; the tables are in the PDF if anyone wants to.

Register	ESS	Contents
1083 `BMS_Status`	`0x0013`	bits 0-1 state (soft start / standby / charging / discharging), bit 3 cell balance, bit 4 sleep, bits 5-6 discharge and charge output, bit 7 terminal status, bits 8-9 master box mode, bits 10-11 SP status
1085 `BMS_Error`	`0x0014`	15 protection flags — over/under voltage, over-current charge and discharge, short circuit, over/under temperature for charge and discharge separately, MOS and environment temperature
1099 `BMS_WarnInfo`	`0x0022`	warning-level equivalents of the above, plus bits 14-15 giving battery chemistry (00 LiFePO4, 01 ternary lithium, 10 lithium titanate)

Applicability¶

The document is titled for the 1xSxxP family (SPF 5 kW, MHP 5 kW), and V1.39 references it generically as "GrowattxxSxxP" for the BMS block. Treat the units as reliable for the BMS block on SPH; treat anything model-specific with the usual caution, and prefer a measurement where one exists.