diff --git a/YamBMS_RP_BMS_EG4_RS485.yaml b/YamBMS_RP_BMS_EG4_RS485.yaml
index be1e8a2..c64b567 100644
--- a/YamBMS_RP_BMS_EG4_RS485.yaml
+++ b/YamBMS_RP_BMS_EG4_RS485.yaml
@@ -1,4 +1,4 @@
-# Updated : 2026.08.11
+# Updated : 2026.08.29
# Version : 1.8.0
# GitHub : https://github.com/Sleeper85/esphome-yambms
@@ -286,6 +286,7 @@ packages:
bms_name: 'EG4-BMS 1'
## BMS address must start at 2 with EG4 LLV2, ID 1 is a special case and has different output.
bms_address: 2 # BMS 1 modbus address
+ bms_balance_trigger_voltage: '0.030' # V. Cell spread at which the BMS balances, set yours
bms_status_led_id: 'esp_light' # LED used to show status; default = esp_light, comment out to disable
# BMS 2
- path: 'packages/bms/bms_combine_EG4_LLV2_RS485_bms_full.yaml'
@@ -293,6 +294,7 @@ packages:
bms_id: '2' # must be a number
bms_name: 'EG4-BMS 2'
bms_address: 3 # BMS 2 modbus address
+ bms_balance_trigger_voltage: '0.030' # V. Cell spread at which the BMS balances, set yours
bms_status_led_id: 'esp_light' # LED used to show status; default = esp_light, comment out to disable
# BMS 3
- path: 'packages/bms/bms_combine_EG4_LLV2_RS485_bms_full.yaml'
@@ -300,6 +302,7 @@ packages:
bms_id: '3' # must be a number
bms_name: 'EG4-BMS 3'
bms_address: 4 # BMS 3 modbus address
+ bms_balance_trigger_voltage: '0.030' # V. Cell spread at which the BMS balances, set yours
bms_status_led_id: 'esp_light' # LED used to show status; default = esp_light, comment out to disable
diff --git a/documents/README/Charging_logic.md b/documents/README/Charging_logic.md
index a9e18f4..0ba6027 100644
--- a/documents/README/Charging_logic.md
+++ b/documents/README/Charging_logic.md
@@ -71,12 +71,22 @@ The Bulk or Cut-Off **conditions must hold continuously for 10s** before the cor
Once the **Cut-Off phase** is entered:
-- if the BMS reports that cells are **equalizing**, the cut-off timer is stopped: the charge is held so the balancer can work;
+- if the cells are reported — or inferred — as **equalizing**, the cut-off timer is stopped: the charge is held so the balancer can work;
- otherwise the **cut-off timer** starts (`180s` default): if the Cut-Off conditions still hold when it expires, the **EOC** (End Of Charge) is declared;
- in parallel, if the `EOC timer` switch is enabled, the **EOC timer** (`30min` default) guarantees that the Cut-Off phase ends at the latest when it expires, even if cells are still equalizing.
Returning to the Bulk phase stops both timers.
+Some BMS balance their cells but never expose a balancing flag. YamBMS infers it for
+them, from the max cell reaching the bulk setpoint together with a cell spread that has
+reached `bms_balance_trigger_voltage`, the threshold at which that pack's own balancer
+engages, so the cut-off timer is held exactly as it would be with a
+BMS that reports the flag. A real external balancer, when one is online, always takes
+priority over the inference. This is enabled by default for Seplos V3, Seplos V1/V2,
+EG4 LL v2 and JK RS485 DISPLAY — see
+[Changelog_Fake_Equalizing.md](changelog/Changelog_Fake_Equalizing.md) for the references
+it uses and how to tune or disable it.
+
## LFP Cut-Off Values
- Nominal : 3.20 V
diff --git a/documents/README/changelog/Changelog_Fake_Equalizing.md b/documents/README/changelog/Changelog_Fake_Equalizing.md
new file mode 100644
index 0000000..1b44c76
--- /dev/null
+++ b/documents/README/changelog/Changelog_Fake_Equalizing.md
@@ -0,0 +1,249 @@
+# YamBMS - Changelog about the new Fake Equalizing
+
+[](https://www.gnu.org/licenses/gpl-3.0)
+[](https://github.com/Sleeper85/esphome-yambms/releases/latest)
+
+
+
+
+`packages/bms/bms_options_fake_equalizing.yaml`
+
+Original idea by [@haysdb](https://github.com/haysdb), who reported the problem and
+proposed a first implementation. The package shipped here is a much smaller rewrite of
+that idea — see [Why not the original pull request](#why-not-the-original-pull-request)
+at the end.
+
+---
+
+## 1. The problem
+
+Several BMS balance their cells but never expose a balancing flag over their protocol.
+The YamBMS packages for those BMS had no choice but to declare the flag as a constant:
+
+```yaml
+ # equalizing
+ - platform: template
+ id: bms${bms_id}_bms_equalizing
+ lambda: return false;
+```
+
+That constant is not a neutral placeholder, it is a wrong value. Those packs *do*
+balance, and `equalizing` is not a cosmetic entity: in `yambms_core.yaml`, while it is
+true the **Cut-Off timer is stopped** and the charge is allowed to continue up to the EOC
+timer. Reporting `false` during a real balancing phase means the charge is terminated
+while the pack is still equalizing, cycle after cycle, and the user has no indication
+that anything is wrong.
+
+## 2. What the package does
+
+It infers the flag from the pack itself, and — this is the part that matters — **it
+hardcodes neither of the two references it needs**. Both are already in your
+configuration.
+
+**The voltage reference** is the per-cell bulk setpoint:
+
+```cpp
+cell_bulk_v = bulk_voltage / cell_count
+```
+
+the exact derivation `yambms_core.yaml` already uses for its own end-of-charge
+comparison. It follows `yambms_battery_chemistry`, the cell count and the bulk voltage you
+configured, with no per-chemistry constant to maintain. LFP, Li-ion and LTO work out of
+the box.
+
+**The spread reference** is `bms_balance_trigger_voltage` — the very spread at which your
+pack's own balancer engages. That number is not a matter of taste, and it is not the same
+from one BMS to the next: **10 mV on a JK or a Seplos, 30 mV on an EG4 LL v2**. A fixed
+constant would have fitted none of them — it would have reported balancing on a pack whose
+balancer is idle, or stayed silent on one that is working. `bms_base_balancer.yaml`
+already resolves this value to the external balancer's setting whenever one is online, and
+to the BMS's own otherwise.
+
+A pack that reports no trigger voltage at all never starts. There is no basis to claim it
+is balancing, and no threshold worth inventing for it: the inference simply stays out of
+the way and the behaviour is the one you have today.
+
+```
+ START when BOTH are true
+ max cell >= cell_bulk_v the pack is at the setpoint
+ AND (max - min) >= trigger voltage its own balancing threshold
+
+ STOP when EITHER is true
+ max cell < cell_bulk_v - hysteresis the pack dropped off the setpoint
+ OR (max - min) < trigger - delta hyst. the spread closed back below it
+```
+
+The **spread condition is what keeps this honest**. A pack that reaches the setpoint
+already balanced has nothing to equalize: it reports `false` and the Cut-Off timer does
+its job normally. A plain voltage comparison, with no spread condition, would report
+balancing forever and push every single charge to the EOC timer ceiling.
+
+The **two hysteresis margins** exist because both references are sat *on* at top of
+charge: the max cell rests on the setpoint for a long time, and the spread hovers around
+the trigger voltage. Without them the state would toggle on every tick.
+
+The spread and its threshold are compared as whole millivolts, the resolution the cells
+are reported at. This is not cosmetic: `3.450f - 3.445f` evaluates to `0.005000114`, so a
+raw float comparison against `0.005` would hold the spread above the threshold forever and
+never release the Cut-Off timer.
+
+### Where it plugs in
+
+The package extends `bms${bms_id}_bms_equalizing`, the **raw** BMS flag, and not the
+combined `bms${bms_id}_equalizing` sensor. Consequence: `bms_base_balancer.yaml` is
+neither modified nor duplicated, and it keeps giving an **external balancer priority**
+whenever one is online. If you run a NEEY or a Modbus balancer on that `bms_id`, its
+reading wins; the inference only applies when no balancer reports for this pack.
+
+## 3. Which BMS are concerned
+
+The four BMS below shipped a hardcoded `return false;`. The package is now **included by
+default** for them — there is nothing to add to your YAML:
+
+| BMS | Packages |
+|---|---|
+| EG4 LL V2 | `bms_combine_EG4_LLV2_RS485_bms_full.yaml`
`bms_modbus_EG4_LLV2_RS485_bms_full.yaml` |
+| Seplos V1 V2 | `bms_combine_SEPLOS_V1_V2_RS485_bms_full.yaml`
`bms_modbus_SEPLOS_V1_V2_RS485_bms_full.yaml` |
+| Seplos V3 | `bms_combine_SEPLOS_V3_RS485_bms_full.yaml`
`bms_modbus_SEPLOS_V3_RS485_bms_full.yaml` |
+| JK RS485 DISPLAY | `bms_combine_JK_RS485_DISPLAY_full.yaml` |
+
+The `bms_modbus_*` variants are the multi-node server packages, so a secondary node
+relays the inferred flag to the client exactly like a real one.
+
+The **Fake BMS** (`bms_combine_FAKE.yaml`, `_FAKE_extended.yaml`, `_DEMO.yaml`) also uses
+it now. It previously carried an inline voltage comparison with no spread condition,
+which the package replaces.
+
+### BMS that are *not* concerned
+
+- **JK** (BLE, RS485 Modbus, UART 4G-GPS), **JBD**, **Ecoworthy**, **PACE**, **Deye CAN**
+ and the YamBMS Modbus client node all report a real balancing flag. Nothing changes.
+- **BASEN** already infers the flag inline, from its own `balance_starting_voltage` and
+ `balance_trigger_voltage` reported by the BMS. Those are a better source than a
+ derived anchor, so it is left as is.
+- **KS BLE** derives the flag from the balanced-cell bitmask, which is real information.
+
+The package can be added manually to another BMS, once for each `bms_id`:
+
+```yaml
+ - path: packages/bms/bms_options_fake_equalizing.yaml
+ vars:
+ bms_id: '1' # must be a number
+```
+
+but only where `bms_bms_equalizing` is declared as a plain `platform: template`
+binary_sensor, which is what it replaces the lambda of. A BMS that exposes the flag
+through its own component — `jk_bms_ble`, `jbd_bms`, `modbus_controller` and the like —
+cannot be extended this way: those platforms take no `lambda:`, and the configuration
+will not validate. That is not a real limitation, since those are precisely the BMS that
+already report a genuine flag.
+
+## 4. Tuning
+
+### The one setting that matters: `bms_balance_trigger_voltage`
+
+This is your BMS's own balancing threshold, and it is the spread the inference compares
+against. Some BMS report it over their protocol; those that cannot expose it as a var to
+declare at import — **Seplos V1/V2, Seplos V3, EG4 LL v2 and JK RS485 DISPLAY all work
+that way**. It also feeds `Auto CVL`, so it was already worth getting right.
+
+EG4 LL v2 and JK RS485 DISPLAY used a hard-coded value.
+They now expose it like the others with the following default values:
+
+| BMS | Default |
+|---|---|
+| EG4 LL V2 | 0.03 |
+| SEPLOS V1 V2 | 0.01 |
+| SEPLOS V3 | 0.01 |
+| JK RS485 DISPLAY | 0.01 |
+
+```yaml
+ # BMS 1
+ - path: 'packages/bms/bms_combine_your_BMS.yaml'
+ vars:
+ bms_id: '1'
+ # ... other vars of your BMS ...
+ bms_balance_trigger_voltage: '0.010' # V. Your BMS's balancing threshold
+ bms_fake_equalizing_enabled: 'true' # Optional : true/fasle, true by default
+ bms_fake_equalizing_hysteresis: '0.030' # Optional : 0.030 by default
+ bms_fake_equalizing_delta_hysteresis: '0.002' # Optional : 0.002 by default
+```
+
+Set it to what your BMS is actually configured for. On a JK you can read and change it in
+the BMS itself; on an EG4 it may not be adjustable, in which case leave the default.
+
+### The rest
+
+An on/off switch and two margins around the references. No threshold among them, and the
+defaults are fine for any pack.
+
+| Substitution | Default | Meaning |
+|---|---|---|
+| `bms_fake_equalizing_enabled` | `'true'` | set to `'false'` to turn the inference off for this BMS |
+| `bms_fake_equalizing_hysteresis` | `0.030` | how far below the bulk setpoint the max cell may fall before balancing stops |
+| `bms_fake_equalizing_delta_hysteresis` | `0.002` | how far below the trigger voltage the spread may fall before balancing stops |
+
+All of them can be passed per BMS as above, or set once for every BMS in the root
+`substitutions:` block. The spread values are resolved to whole millivolts, the resolution
+the cells are reported at, so a value finer than `0.001` has no effect.
+
+**To disable the inference** on a BMS without editing any package:
+
+```yaml
+ bms_fake_equalizing_enabled: 'false'
+```
+
+The flag then reports the constant `false` its package used to hardcode, which is the
+behaviour you had before this release. Do *not* try to disable it by raising
+`bms_balance_trigger_voltage` instead — that value also feeds `Auto CVL`, where it means
+something else entirely: how far the highest cell may overshoot the per-cell setpoint
+before the PI clamps.
+
+## 5. What changes for you on upgrade
+
+If you own one of the four BMS above, the `balancing` entity will now turn on at the top
+of charge as soon as your cells are spread by as much as your BMS's own
+`bms_balance_trigger_voltage` — 10 mV on a Seplos or a JK, 30 mV on an EG4 unless you
+declared otherwise. The charge then runs until the spread closes back below that, or
+until the EOC timer expires, instead of being cut short by the Cut-Off timer. That is the
+intended fix, but it *is* a change in charge termination, so watch a couple of cycles
+after updating.
+
+The risk is bounded on the safe side. A false positive can only extend the charge up to
+the EOC timer, which exists for exactly that purpose. A false negative simply restores
+the previous behaviour.
+
+## Why not the original pull request
+
+[@haysdb](https://github.com/haysdb) opened
+[PR #304](https://github.com/Sleeper85/esphome-yambms/pull/304) with a *Fake Balancer*
+package. The diagnosis was right and the idea is his. The implementation was shaped
+around his own setup — a NEEY balancer that does not run every day — and carried more
+than YamBMS needs:
+
+- **Absolute thresholds as entities.** Four `number` entities held the start / sleep /
+ trigger / stop-diff voltages, hardcoded to LFP values (3.45 / 3.42 V). They lived
+ alongside the bulk voltage the user had already configured, so the same physical
+ setpoint existed twice and could drift apart. Taking the voltage reference from
+ `bulk_voltage / cell_count` and the spread reference from
+ `bms_balance_trigger_voltage` removes the duplication entirely, and makes the
+ package track each BMS instead of a constant that fits none of them.
+- **A companion stub package.** The lambda referenced `balancer${bms_id}_online_status`
+ and `balancer${bms_id}_equalizing`, which only exist when a balancer package is
+ included, so a second package had to declare two always-false sensors just to make it
+ compile without one. Reading the raw BMS flag instead makes the stub unnecessary.
+- **A duplicated core lambda.** It extended the *combined* `bms${bms_id}_equalizing` and
+ recopied the branch from `bms_base_balancer.yaml` to add its OR. Extending the raw flag
+ keeps a single copy of that logic.
+- **A side effect on another entity.** The interval wrote
+ `bms${bms_id}_var_ext_balancer_online = real_online || fake_eq`, which made
+ `bms_base_balancer.yaml` report the external balancer's trigger voltage — `0.000 V`
+ when no balancer had ever been online.
+- **Eight entities and a sub-device per BMS**, for a function that needs none.
+
+The `Force` switch and the `Max Run Time` number were dropped as well: they answer the
+need to launch a weekly top-balance by hand when the hardware balancer is idle, which is
+a balancer feature rather than a BMS one.
+
+Thanks to @haysdb for finding the problem and for the threshold model — start and stop on
+a spread with hysteresis is his, and it is the part that makes the inference trustworthy.
diff --git a/examples/single-node/LP_BMS_example_EG4_LLV2_RS485_Modbus.yaml b/examples/single-node/LP_BMS_example_EG4_LLV2_RS485_Modbus.yaml
index b23d152..a67c596 100644
--- a/examples/single-node/LP_BMS_example_EG4_LLV2_RS485_Modbus.yaml
+++ b/examples/single-node/LP_BMS_example_EG4_LLV2_RS485_Modbus.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.02.16
-# Version : 1.1.4
+# Updated : 2026.08.29
+# Version : 1.1.5
# GitHub : https://github.com/Sleeper85/esphome-yambms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -39,6 +39,7 @@ packages:
bms_name: 'EG4-BMS 1'
## BMS address must start at 2 with EG4 LLV2, ID 1 is a special case and has different output.
bms_address: 2 # BMS 1 modbus address
+ bms_balance_trigger_voltage: '0.030' # V. Cell spread at which the BMS balances, set yours
bms_status_led_id: 'esp_light' # LED used to show status; default = esp_light, comment out to disable
bms2: !include
@@ -47,6 +48,7 @@ packages:
bms_id: '2' # must be a number
bms_name: 'EG4-BMS 2'
bms_address: 3 # BMS 2 modbus address
+ bms_balance_trigger_voltage: '0.030' # V. Cell spread at which the BMS balances, set yours
bms_status_led_id: 'esp_light' # LED used to show status; default = esp_light, comment out to disable
bms3: !include
@@ -55,4 +57,5 @@ packages:
bms_id: '3' # must be a number
bms_name: 'EG4-BMS 3'
bms_address: 4 # BMS 3 modbus address
+ bms_balance_trigger_voltage: '0.030' # V. Cell spread at which the BMS balances, set yours
bms_status_led_id: 'esp_light' # LED used to show status; default = esp_light, comment out to disable
diff --git a/examples/single-node/RP_BMS_example_EG4_LLV2_RS485_Modbus.yaml b/examples/single-node/RP_BMS_example_EG4_LLV2_RS485_Modbus.yaml
index 53e0dfa..5849961 100644
--- a/examples/single-node/RP_BMS_example_EG4_LLV2_RS485_Modbus.yaml
+++ b/examples/single-node/RP_BMS_example_EG4_LLV2_RS485_Modbus.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.02.16
-# Version : 1.1.4
+# Updated : 2026.08.29
+# Version : 1.1.5
# GitHub : https://github.com/Sleeper85/esphome-yambms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -43,6 +43,7 @@
bms_name: 'EG4-BMS 1'
## BMS address must start at 2 with EG4 LLV2, ID 1 is a special case and has different output.
bms_address: 2 # BMS 1 modbus address
+ bms_balance_trigger_voltage: '0.030' # V. Cell spread at which the BMS balances, set yours
bms_status_led_id: 'esp_light' # LED used to show status; default = esp_light, comment out to disable
# BMS 2
- path: 'packages/bms/bms_combine_EG4_LLV2_RS485_bms_full.yaml'
@@ -50,6 +51,7 @@
bms_id: '2' # must be a number
bms_name: 'EG4-BMS 2'
bms_address: 3 # BMS 2 modbus address
+ bms_balance_trigger_voltage: '0.030' # V. Cell spread at which the BMS balances, set yours
bms_status_led_id: 'esp_light' # LED used to show status; default = esp_light, comment out to disable
# BMS 3
- path: 'packages/bms/bms_combine_EG4_LLV2_RS485_bms_full.yaml'
@@ -57,4 +59,5 @@
bms_id: '2' # must be a number
bms_name: 'EG4-BMS 3'
bms_address: 4 # BMS 3 modbus address
+ bms_balance_trigger_voltage: '0.030' # V. Cell spread at which the BMS balances, set yours
bms_status_led_id: 'esp_light' # LED used to show status; default = esp_light, comment out to disable
diff --git a/packages/bms/bms_combine_DEMO.yaml b/packages/bms/bms_combine_DEMO.yaml
index 1cb4eae..f03140f 100644
--- a/packages/bms/bms_combine_DEMO.yaml
+++ b/packages/bms/bms_combine_DEMO.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.1.2
+# Updated : 2026.08.28
+# Version : 1.1.3
# GitHub : https://github.com/Sleeper85/esphome-yambms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -20,6 +20,7 @@
packages:
bms_sensors: !include bms_sensors_FAKE.yaml
bms_numbers: !include bms_sensors_FAKE_extended.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_demo_cs: !include bms_demo_charge_simulation.yaml
bms_combine: !include
file: bms_combine.yaml
diff --git a/packages/bms/bms_combine_EG4_LLV2_RS485_bms_full.yaml b/packages/bms/bms_combine_EG4_LLV2_RS485_bms_full.yaml
index fb46268..148aa20 100644
--- a/packages/bms/bms_combine_EG4_LLV2_RS485_bms_full.yaml
+++ b/packages/bms/bms_combine_EG4_LLV2_RS485_bms_full.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.1.3
+# Updated : 2026.08.28
+# Version : 1.1.4
# GitHub : https://github.com/Sleeper85/esphome-yambms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -21,6 +21,7 @@
packages:
bms_sensors: !include bms_sensors_EG4_LLV2_RS485_bms_full.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_combine: !include
file: bms_combine.yaml
vars:
diff --git a/packages/bms/bms_combine_FAKE.yaml b/packages/bms/bms_combine_FAKE.yaml
index b0f1eee..ddd7dfc 100644
--- a/packages/bms/bms_combine_FAKE.yaml
+++ b/packages/bms/bms_combine_FAKE.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.1.2
+# Updated : 2026.08.28
+# Version : 1.1.3
# GitHub : https://github.com/Sleeper85/esphome-yambms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -19,6 +19,7 @@
packages:
bms_sensors: !include bms_sensors_FAKE.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_combine: !include
file: bms_combine.yaml
vars:
diff --git a/packages/bms/bms_combine_FAKE_extended.yaml b/packages/bms/bms_combine_FAKE_extended.yaml
index ae9db17..987a65f 100644
--- a/packages/bms/bms_combine_FAKE_extended.yaml
+++ b/packages/bms/bms_combine_FAKE_extended.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.1.2
+# Updated : 2026.08.28
+# Version : 1.1.3
# GitHub : https://github.com/Sleeper85/esphome-yambms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -20,6 +20,7 @@
packages:
bms_sensors: !include bms_sensors_FAKE.yaml
bms_numbers: !include bms_sensors_FAKE_extended.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_combine: !include
file: bms_combine.yaml
vars:
diff --git a/packages/bms/bms_combine_JK_RS485_DISPLAY_full.yaml b/packages/bms/bms_combine_JK_RS485_DISPLAY_full.yaml
index 90fe836..1bbc555 100644
--- a/packages/bms/bms_combine_JK_RS485_DISPLAY_full.yaml
+++ b/packages/bms/bms_combine_JK_RS485_DISPLAY_full.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.2.0
+# Updated : 2026.08.29
+# Version : 1.2.2
# GitHub : https://github.com/syssi/esphome-jk-bms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -19,6 +19,7 @@
packages:
bms_base: !include bms_base.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_combine: !include
file: bms_combine.yaml
vars:
@@ -27,6 +28,13 @@ packages:
bms_temperature_sensor: !include bms_temperature_sensor_2.yaml
bms_battery_soh: !include bms_base_battery_soh.yaml
+substitutions:
+ # (V) cell spread at which this BMS starts balancing. Not readable over the display
+ # port, so declare your unit's actual setting at import. Used by 'Auto CVL' and, when
+ # bms_options_fake_equalizing.yaml is included, as the spread reference for the
+ # inferred balancing flag.
+ bms_balance_trigger_voltage: '0.01'
+
# +--------------------------------------+
# | DISPLAY port schematic |
# +--------------------------------------+
@@ -193,7 +201,7 @@ sensor:
device_id: bms_${bms_id}
# name: "state of charge"
delta_cell_voltage:
- device_id: bms_${bms_id}
+ device_id: bms_${bms_id}
name: "delta cell voltage"
mosfet_temperature:
id: bms${bms_id}_temperature_sensor_1
@@ -384,7 +392,7 @@ sensor:
- or:
- throttle: 10s
- delta: 0.001
- lambda: return 0.005;
+ lambda: return ${bms_balance_trigger_voltage};
# FAKE charging_cycles
- platform: template
diff --git a/packages/bms/bms_combine_SEPLOS_V1_V2_RS485_bms_full.yaml b/packages/bms/bms_combine_SEPLOS_V1_V2_RS485_bms_full.yaml
index 7451793..367e2db 100644
--- a/packages/bms/bms_combine_SEPLOS_V1_V2_RS485_bms_full.yaml
+++ b/packages/bms/bms_combine_SEPLOS_V1_V2_RS485_bms_full.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.1.2
+# Updated : 2026.08.28
+# Version : 1.1.3
# GitHub : https://github.com/syssi/esphome-seplos-bms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -19,6 +19,7 @@
packages:
bms_sensors: !include bms_sensors_SEPLOS_V1_V2_RS485_bms_full.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_combine: !include
file: bms_combine.yaml
vars:
diff --git a/packages/bms/bms_combine_SEPLOS_V3_RS485_bms_full.yaml b/packages/bms/bms_combine_SEPLOS_V3_RS485_bms_full.yaml
index 37e374f..140394d 100644
--- a/packages/bms/bms_combine_SEPLOS_V3_RS485_bms_full.yaml
+++ b/packages/bms/bms_combine_SEPLOS_V3_RS485_bms_full.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.1.2
+# Updated : 2026.08.28
+# Version : 1.1.3
# GitHub : https://github.com/syssi/esphome-seplos-bms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -19,6 +19,7 @@
packages:
bms_sensors: !include bms_sensors_SEPLOS_V3_RS485_bms_full.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_combine: !include
file: bms_combine.yaml
vars:
diff --git a/packages/bms/bms_demo_charge_simulation.yaml b/packages/bms/bms_demo_charge_simulation.yaml
index d49f9e8..2793395 100644
--- a/packages/bms/bms_demo_charge_simulation.yaml
+++ b/packages/bms/bms_demo_charge_simulation.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.06.29
-# Version : 1.1.3
+# Updated : 2026.08.29
+# Version : 1.1.4
# GitHub : https://github.com/Sleeper85/esphome-yambms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -38,6 +38,9 @@ substitutions:
min_temperature: '4.8'
max_temperature: '5.0'
delta_cell_voltage: '0.005'
+ # Spread held during the two balancing steps. Must stay above this BMS's
+ # `bms_balance_trigger_voltage` for the inferred equalizing flag to assert.
+ delta_cell_voltage_balancing: '0.020'
switch:
# DEMO switch
@@ -118,14 +121,14 @@ interval:
// Step 4 : 98% (10sec)
id(bms${bms_id}_state_of_charge).publish_state(98);
id(bms${bms_id}_current).publish_state(5);
- id(bms${bms_id}_min_cell_voltage).publish_state(requested_cvl / cell_count + 0.01 - ${delta_cell_voltage});
+ id(bms${bms_id}_min_cell_voltage).publish_state(requested_cvl / cell_count + 0.01 - ${delta_cell_voltage_balancing});
id(bms${bms_id}_max_cell_voltage).publish_state(requested_cvl / cell_count + 0.01); // equalizing
}
else if (seconds <= 5 * step_sec) { // Entering Cut-off phase...
// Step 5 : 99% (10sec)
id(bms${bms_id}_state_of_charge).publish_state(99);
id(bms${bms_id}_current).publish_state(3.5);
- id(bms${bms_id}_min_cell_voltage).publish_state(requested_cvl / cell_count + 0.005 - ${delta_cell_voltage});
+ id(bms${bms_id}_min_cell_voltage).publish_state(requested_cvl / cell_count + 0.005 - ${delta_cell_voltage_balancing});
id(bms${bms_id}_max_cell_voltage).publish_state(requested_cvl / cell_count + 0.005); // equalizing
}
else if (seconds <= 15 * step_sec) { // End of Cut-off phase...
diff --git a/packages/bms/bms_modbus_EG4_LLV2_RS485_bms_full.yaml b/packages/bms/bms_modbus_EG4_LLV2_RS485_bms_full.yaml
index 9e12bd1..6c5cb98 100644
--- a/packages/bms/bms_modbus_EG4_LLV2_RS485_bms_full.yaml
+++ b/packages/bms/bms_modbus_EG4_LLV2_RS485_bms_full.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.1.4
+# Updated : 2026.08.28
+# Version : 1.1.5
# GitHub : https://github.com/syssi/esphome-jk-bms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -19,4 +19,5 @@
packages:
bms_sensors: !include bms_sensors_EG4_LLV2_RS485_bms_full.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_combine: !include bms_modbus_server.yaml
diff --git a/packages/bms/bms_modbus_SEPLOS_V1_V2_RS485_bms_full.yaml b/packages/bms/bms_modbus_SEPLOS_V1_V2_RS485_bms_full.yaml
index 5a82e6f..4ad85cd 100644
--- a/packages/bms/bms_modbus_SEPLOS_V1_V2_RS485_bms_full.yaml
+++ b/packages/bms/bms_modbus_SEPLOS_V1_V2_RS485_bms_full.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.1.2
+# Updated : 2026.08.28
+# Version : 1.1.3
# GitHub : https://github.com/syssi/esphome-pace-bms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -19,4 +19,5 @@
packages:
bms_sensors: !include bms_sensors_SEPLOS_V1_V2_RS485_bms_full.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_combine: !include bms_modbus_server.yaml
diff --git a/packages/bms/bms_modbus_SEPLOS_V3_RS485_bms_full.yaml b/packages/bms/bms_modbus_SEPLOS_V3_RS485_bms_full.yaml
index d4eed66..93be0d9 100644
--- a/packages/bms/bms_modbus_SEPLOS_V3_RS485_bms_full.yaml
+++ b/packages/bms/bms_modbus_SEPLOS_V3_RS485_bms_full.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.14
-# Version : 1.1.2
+# Updated : 2026.08.28
+# Version : 1.1.3
# GitHub : https://github.com/syssi/esphome-pace-bms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -19,4 +19,5 @@
packages:
bms_sensors: !include bms_sensors_SEPLOS_V3_RS485_bms_full.yaml
+ bms_fake_equalizing: !include bms_options_fake_equalizing.yaml
bms_combine: !include bms_modbus_server.yaml
diff --git a/packages/bms/bms_options_fake_equalizing.yaml b/packages/bms/bms_options_fake_equalizing.yaml
new file mode 100644
index 0000000..e9f8b47
--- /dev/null
+++ b/packages/bms/bms_options_fake_equalizing.yaml
@@ -0,0 +1,214 @@
+# Updated : 2026.08.29
+# Version : 1.1.0
+# GitHub : https://github.com/Sleeper85/esphome-yambms
+
+# YamBMS ( Yet another multi-BMS Merging Solution )
+
+# This YAML is free software: you can redistribute it and/or
+# modify it under the terms of the GNU General Public License
+# as published by the Free Software Foundation, either version 3
+# of the License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+# See the GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see .
+
+
+
+# +--------------------------------------------------------------------------------+
+# | Fake Equalizing - infer the balancing flag a BMS does not report |
+# +--------------------------------------------------------------------------------+
+# | Some BMS balance their cells but never expose a balancing flag, so
+# | `bms_equalizing` stays false and the core cannot tell a pack that is still
+# | balancing from one that has finished. That matters : while equalizing is true
+# | the core stops the Cut-Off timer (yambms_core.yaml) and lets the charge run to
+# | the EOC timer instead of terminating early.
+# |
+# | This package infers that flag from the pack itself, and hardcodes neither of the
+# | two references it needs :
+# |
+# | the voltage anchor is the per-cell bulk setpoint, `bulk voltage / cell count`,
+# | the same derivation as cell_bulk_v in yambms_core.yaml, so it follows
+# | `yambms_battery_chemistry`, the cell count and the configured bulk voltage ;
+# |
+# | the spread anchor is `bms${bms_id}_balance_trigger_voltage`, the very spread at
+# | which this pack's own balancer engages. bms_base_balancer.yaml resolves it to
+# | the external balancer's value whenever one is online, to the BMS's otherwise.
+# |
+# | START when BOTH are true :
+# | max cell >= cell bulk setpoint the pack is at the setpoint
+# | AND (max - min) >= trigger voltage the pack's own balancing threshold
+# |
+# | STOP when EITHER is true :
+# | max cell < setpoint - hysteresis the pack dropped off the setpoint
+# | OR (max - min) < trigger - delta hyst. the spread closed back below it
+# |
+# | Nothing is configurable but an on/off switch and the two hysteresis margins, so
+# | the inference tracks the hardware instead of a constant that would fit no BMS in
+# | particular : the
+# | balance trigger voltage is 10 mV on a JK or a Seplos, 30 mV on an EG4. A pack that
+# | reports none at all never starts - there is no threshold worth inventing for it.
+# |
+# | The spread condition is what keeps this honest : a pack that reaches the setpoint
+# | already balanced has nothing to equalize, so it reports false and the Cut-Off
+# | timer does its job. A plain voltage comparison would report balancing forever and
+# | terminate the charge on the EOC timer every cycle.
+# |
+# | It extends `bms_equalizing`, the RAW BMS flag, not the combined
+# | `bms${bms_id}_equalizing` sensor. So bms_base_balancer.yaml is untouched and
+# | keeps giving an external balancer (NEEY, modbus client) priority whenever one
+# | is online : the inference only applies when no balancer reports for this pack.
+# |
+# | REQUIRES the BMS package it is included alongside to expose :
+# |
+# | bms${bms_id}_bms_equalizing the raw flag, declared as a template
+# | binary_sensor (extended here)
+# | bms${bms_id}_max_cell_voltage
+# | bms${bms_id}_min_cell_voltage
+# |
+# | plus, both always present :
+# |
+# | bms${bms_id}_balance_trigger_voltage from bms_base_balancer.yaml
+# | yambms${yambms_id}_bulk_voltage from the core
+# | yambms${yambms_id}_cell_count
+# |
+# | A BMS that cannot read its balancing threshold over its protocol exposes it as the
+# | `bms_balance_trigger_voltage` var instead, to be set at import - Seplos V1/V2 and
+# | V3, EG4 LL v2 and JK RS485 DISPLAY all work that way. Declare your unit's actual
+# | setting there : it is what this package compares the spread against.
+# +--------------------------------------------------------------------------------+
+# | Included by default - these BMS need nothing added to your YAML
+# +--------------------------------------------------------------------------------+
+# |
+# | EG4 LL V2 bms_combine_EG4_LLV2_RS485_bms_full.yaml
+# | bms_modbus_EG4_LLV2_RS485_bms_full.yaml
+# | Seplos V1 V2 bms_combine_SEPLOS_V1_V2_RS485_bms_full.yaml
+# | bms_modbus_SEPLOS_V1_V2_RS485_bms_full.yaml
+# | Seplos V3 bms_combine_SEPLOS_V3_RS485_bms_full.yaml
+# | bms_modbus_SEPLOS_V3_RS485_bms_full.yaml
+# | JK RS485 DISP. bms_combine_JK_RS485_DISPLAY_full.yaml
+# | Fake BMS bms_combine_FAKE.yaml, _FAKE_extended.yaml, _DEMO.yaml
+# |
+# | The four real BMS above declared `bms_equalizing` as a hardcoded
+# | `return false;` : they do balance, they just never say so. The Fake BMS used
+# | to carry an inline voltage comparison, which this package replaces.
+# +--------------------------------------------------------------------------------+
+# | Adding it to another BMS - once for each bms_id
+# +--------------------------------------------------------------------------------+
+#
+# - path: packages/bms/bms_options_fake_equalizing.yaml
+# vars:
+# bms_id: '1' # must be a number
+#
+# | Only where `bms${bms_id}_bms_equalizing` is declared as a plain
+# | `platform: template` binary_sensor, since this replaces its lambda. A BMS that
+# | exposes the flag through its own component - jk_bms_ble, jbd_bms,
+# | modbus_controller and the like - cannot be extended this way : those platforms
+# | take no `lambda:` and the configuration will not validate. No real loss, as those
+# | are exactly the BMS that already report a genuine flag.
+#
+# +--------------------------------------------------------------------------------+
+# | Tuning - add the vars to your BMS include, they override the defaults below
+# +--------------------------------------------------------------------------------+
+#
+# # BMS 1
+# - path: 'packages/bms/bms_combine_your_BMS.yaml'
+# vars:
+# bms_id: '1'
+# # ... other vars of your BMS ...
+# bms_balance_trigger_voltage: '0.010' # V. Your BMS's balancing threshold
+# bms_fake_equalizing_enabled: 'true' # Optional : true/fasle, true by default
+# bms_fake_equalizing_hysteresis: '0.030' # Optional : 0.030 by default
+# bms_fake_equalizing_delta_hysteresis: '0.002' # Optional : 0.002 by default
+#
+# | The one worth setting is `bms_balance_trigger_voltage` : it is your BMS's own
+# | balancing threshold, it also feeds 'Auto CVL', and the rest is fine as it is.
+# |
+# | Or set them once for every BMS in the root `substitutions:` block.
+# | To turn the inference off on one BMS : bms_fake_equalizing_enabled: 'false'
+# +--------------------------------------------------------------------------------+
+
+
+
+substitutions:
+ # +--------------------------------------------------------------------------------+
+ # | Nothing here is a threshold : the two references come from elsewhere.
+ # | The voltage anchor is the per-cell bulk setpoint, from the core.
+ # | The spread anchor is `bms_balance_trigger_voltage`, from the BMS itself.
+ # | What is left is an on/off switch and the two hysteresis margins.
+ # +--------------------------------------------------------------------------------+
+ # 'true' / 'false'. Set to 'false' to turn the inference off for this BMS : the flag
+ # goes back to the constant false its package used to report. Needed because the
+ # package is included by default for the BMS that cannot report the flag, and a
+ # Remote Packages user cannot edit a package to drop it.
+ bms_fake_equalizing_enabled: 'true'
+ # (V) how far below the bulk setpoint the max cell may fall before balancing stops.
+ # Without it the state toggles on every tick while the max cell sits on the setpoint.
+ bms_fake_equalizing_hysteresis: '0.030'
+ # (V) how far below the balance trigger voltage the spread may fall before balancing
+ # stops. Resolved to whole millivolts, the resolution the cells are reported at :
+ # a value finer than 0.001 has no effect.
+ bms_fake_equalizing_delta_hysteresis: '0.002'
+
+binary_sensor:
+ # equalizing (inferred)
+ - id: !extend bms${bms_id}_bms_equalizing
+ lambda: |-
+ static bool balancing = false;
+
+ // Turned off : report the constant false this BMS package used to hardcode.
+ if (!${bms_fake_equalizing_enabled})
+ return false;
+
+ float cell_count = id(yambms${yambms_id}_cell_count).state;
+ float bulk_v = id(yambms${yambms_id}_bulk_voltage).state;
+ float max_cell_v = id(bms${bms_id}_max_cell_voltage).state;
+ float min_cell_v = id(bms${bms_id}_min_cell_voltage).state;
+ float trigger_v = id(bms${bms_id}_balance_trigger_voltage).state;
+
+ // The spread reference is the balance trigger voltage : the very spread at which
+ // this pack's own balancer engages, so the inference tracks the hardware instead
+ // of a constant that would fit no BMS in particular. bms_base_balancer.yaml
+ // resolves it to the external balancer's value whenever one is online.
+ //
+ // Hold the last state while any reference is missing, rather than dropping out of
+ // balancing on a single missing reading and releasing the Cut-Off timer. A pack
+ // that never reports a trigger voltage never starts : there is no basis to claim
+ // it is balancing, and no threshold worth inventing for it. Some packages do
+ // publish 0 before their first read (PACE, the Modbus client node), which this
+ // covers as well.
+ if (isnan(cell_count) || cell_count <= 0 || isnan(bulk_v) ||
+ isnan(max_cell_v) || isnan(min_cell_v) ||
+ isnan(trigger_v) || trigger_v <= 0.0f)
+ return balancing;
+
+ // Same derivation as cell_bulk_v in yambms_core.yaml
+ float cell_bulk_v = bulk_v / cell_count;
+
+ // The spread and its threshold are compared as whole millivolts, which is the
+ // resolution the cells are reported at. A float subtraction of two mV-aligned
+ // readings lands a fraction of a uV off - 3.450 - 3.445 gives 0.005000114 - which
+ // would hold the spread just above the threshold forever and never release the
+ // Cut-Off timer. The voltage gates keep their float comparison : their margin is
+ // the hysteresis, orders of magnitude above that error.
+ int delta_mv = (int) roundf((max_cell_v - min_cell_v) * 1000.0f);
+ int trigger_mv = (int) roundf(trigger_v * 1000.0f);
+ int hyst_mv = (int) roundf(${bms_fake_equalizing_delta_hysteresis} * 1000.0f);
+
+ if (!balancing)
+ {
+ // Start : at the setpoint AND spread up to what makes this BMS balance
+ balancing = (max_cell_v >= cell_bulk_v) && (delta_mv >= trigger_mv);
+ }
+ else
+ {
+ // Stop : fell off the setpoint OR the spread closed back below the trigger
+ balancing = (max_cell_v >= cell_bulk_v - ${bms_fake_equalizing_hysteresis}) &&
+ (delta_mv >= trigger_mv - hyst_mv);
+ }
+
+ return balancing;
diff --git a/packages/bms/bms_sensors_EG4_LLV2_RS485_bms_full.yaml b/packages/bms/bms_sensors_EG4_LLV2_RS485_bms_full.yaml
index 404e634..b3c1204 100644
--- a/packages/bms/bms_sensors_EG4_LLV2_RS485_bms_full.yaml
+++ b/packages/bms/bms_sensors_EG4_LLV2_RS485_bms_full.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.08.15
-# Version : 2.0.4
+# Updated : 2026.08.29
+# Version : 2.0.5
# GitHub : https://github.com/RAR/esphome-eg4-bms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -22,6 +22,13 @@ packages:
bms_temperature_sensor: !include bms_temperature_sensor_4.yaml
bms_errors_bitmask: !include bms_errors_bitmask_EG4_LLV2_RS485_Modbus.yaml
+substitutions:
+ # (V) cell spread at which this BMS starts balancing. Not readable over the protocol,
+ # so declare your unit's actual setting at import. Used by 'Auto CVL' and, when
+ # bms_options_fake_equalizing.yaml is included, as the spread reference for the
+ # inferred balancing flag.
+ bms_balance_trigger_voltage: '0.030'
+
# +--------------------------------------+
# | Component settings |
# +--------------------------------------+
@@ -281,7 +288,7 @@ sensor:
- or:
- throttle: 10s
- delta: 0.001
- lambda: return 0.030;
+ lambda: return ${bms_balance_trigger_voltage};
- platform: template
id: bms${bms_id}_cell_ovp
diff --git a/packages/bms/bms_sensors_FAKE.yaml b/packages/bms/bms_sensors_FAKE.yaml
index fcd7804..57fa521 100644
--- a/packages/bms/bms_sensors_FAKE.yaml
+++ b/packages/bms/bms_sensors_FAKE.yaml
@@ -1,5 +1,5 @@
-# Updated : 2026.06.29
-# Version : 1.3.0
+# Updated : 2026.08.28
+# Version : 1.3.1
# GitHub : https://github.com/Sleeper85/esphome-yambms
# YamBMS ( Yet another multi-BMS Merging Solution )
@@ -66,7 +66,6 @@ interval:
float current = id(bms${bms_id}_current).state;
float battery_capacity = id(bms${bms_id}_battery_capacity).state;
float battery_soc = id(bms${bms_id}_state_of_charge).state;
- float bulk_v = id(yambms${yambms_id}_bulk_voltage).state;
// Voltage
float voltage = cell_count * (min_cell_v + max_cell_v) / 2;
@@ -76,12 +75,6 @@ interval:
float power = (voltage * current);
id(bms${bms_id}_power).publish_state(power);
- // Equalizing
- bool equalizing = false;
- if (max_cell_v > bulk_v / cell_count)
- equalizing = true;
- id(bms${bms_id}_bms_equalizing).publish_state(equalizing);
-
// Capacity Remaining
id(bms${bms_id}_capacity_remaining_ah).publish_state(battery_capacity * battery_soc / 100);