|
4 | 4 |
|
5 | 5 | --- |
6 | 6 |
|
| 7 | +## v1.5.0 |
| 8 | + |
| 9 | +Issues: #360, #362, #367 |
| 10 | + |
| 11 | +> ### Some control entities are renamed |
| 12 | +> |
| 13 | +> Twelve number, select and time entities gain the sub-device they belong to in their |
| 14 | +> displayed name — **"Growatt Battery Work Mode"** rather than "Growatt Work Mode". |
| 15 | +> |
| 16 | +> **Nothing breaks.** Entity IDs live in Home Assistant's registry and do not change, so |
| 17 | +> automations, scripts, template sensors and dashboard cards keep working exactly as they |
| 18 | +> are. Only the label shown in the UI moves. If you search for a control by name and don't |
| 19 | +> recognise it, look for the sub-device word. |
| 20 | +> |
| 21 | +> This also fixes WIT inverters showing **"Growatt Growatt TOU Period 1 Start"** — the |
| 22 | +> device name was being applied twice. |
| 23 | +
|
| 24 | +### Problems now surfaced in the UI instead of the log |
| 25 | + |
| 26 | +Two conditions used to leave users running a degraded setup with no way to know. Both now |
| 27 | +appear under **Settings → Repairs**. |
| 28 | + |
| 29 | +- **Settings being reverted.** Local changes that the inverter silently discards — usually |
| 30 | + a ShineWiFi dongle restoring cloud settings, sometimes a prerequisite that isn't enabled. |
| 31 | + Previously a notification that scrolled away; a repair persists until the cause is fixed. |
| 32 | + |
| 33 | +- **An RS485 gateway returning mismatched responses.** Some adapters answer a request with |
| 34 | + a complete, valid response to an *earlier* one. Since v1.3.7 these are discarded, so your |
| 35 | + data stays correct — but the reads are lost and the only evidence was a log line. One |
| 36 | + reporter's gateway was doing this on roughly one poll in three and only found out by |
| 37 | + reading logs. Raised once per session at 5% or more across at least 200 reads, with the |
| 38 | + gateway address and rate, linking [the gateway guide](docs/troubleshooting/rs485-gateways.md). |
| 39 | + |
| 40 | +### Sensor names can now be translated |
| 41 | + |
| 42 | +All 169 sensor names moved out of Python and into the translation files. The integration |
| 43 | +already shipped 22 languages, but they only covered the setup and options screens — sensor |
| 44 | +names were hardcoded English regardless of your Home Assistant language. Other languages |
| 45 | +can now be contributed without touching code. |
| 46 | + |
| 47 | +English text is unchanged, and a test asserts that by comparing every string against the |
| 48 | +original definitions. |
| 49 | + |
| 50 | +### Model identification |
| 51 | + |
| 52 | +- **DTC 5001 and 5002 corrected.** `MID 33-36KTL3-X(Pro.E)` and `MID 3-33KTL3-X3` were |
| 53 | + documented under 5002; they belong to 5001. In Growatt's table those two rows fall at the |
| 54 | + top of the next page under a merged cell, so they read as 5002 unless you check where the |
| 55 | + merge begins. The corrected split is also the sensible one — 5001 is every MID model, |
| 56 | + 5002 every MOD. |
| 57 | + |
| 58 | +- **The MAX / MAX-X family was missing** from the published protocol page along with DTC |
| 59 | + 3501 and 5401, and `3735` was named "SPA 3000-6000TL BL" where the specification says |
| 60 | + "SPA 3000TL BL-UP". That page held its own copy of the table, a fourth alongside two code |
| 61 | + modules and the troubleshooting page. All four now derive from one registry, three of them |
| 62 | + enforced by tests. |
| 63 | + |
| 64 | +### Internal |
| 65 | + |
| 66 | +- **Every entity now shares one base class.** Twenty classes each carried their own unique |
| 67 | + ID, device assignment and naming flag, and the copies had drifted into two conventions at |
| 68 | + once. Consolidating them is what exposed the naming inconsistency above. |
| 69 | + |
| 70 | +- **Contributor documentation** gained twelve verification rules drawn from real defects — |
| 71 | + check the protocol documents before inferring a register, state which register space you |
| 72 | + mean, and treat "Read OK" as evidence the address responded rather than that the value |
| 73 | + means anything. |
| 74 | + |
| 75 | +--- |
| 76 | + |
7 | 77 | ## v1.4.1 |
8 | 78 |
|
9 | 79 | Issues: #362 |
|
0 commit comments