Introducing Pylontech Battery Health Node-RED integration

Really timely topic this — been down a similar rabbit hole recently with my van build transitioning to a more permanent off-grid setup at home.

Running a pair of Pylontech US2000C units managed through Victron Cerbo GX, and whilst the VRM portal gives you some visibility into pack health, it's honestly quite shallow once you start digging into individual cell-level data. The built-in dashboard doesn't flag early cell divergence particularly well in my experience — by the time something looks obviously wrong, you've potentially already got a warranty situation developing.

The Node-RED angle is genuinely interesting here. I've got a basic flow pulling data from the Cerbo via MQTT already, feeding into a Grafana dashboard, but I haven't pushed it as far as building dedicated battery health logic. Would be curious whether anyone has built flows that specifically watch for:

• Cell voltage spread across individual modules
• SOC drift between parallel packs
• Temperature delta warnings before the BMS actually trips

The Pylontech RS485 interface exposes quite a lot if you query it directly rather than relying purely on what gets passed upstream to the Victron stack. There's a question of whether the Cerbo intermediary strips out any of the more granular data.

Has anyone integrated this with Home Assistant as well, or are people keeping it purely Node-RED → InfluxDB → Grafana? I'm trying to avoid sprawl in my setup but equally don't want to miss something a proper HA integration would catch.

Would also be interested to know if anyone's caught a failing module early through custom monitoring before it became a full warranty claim — that's the real value proposition here as far as I can see.

Great thread @PaulaFisher91! I've got a similar setup with US2000C units and Node-RED, and one thing I'd add that's worth knowing — the Pylontech RS485 console port gives you far more granular data than the CAN bus alone. I'm pulling individual cell voltages and temperatures per module which has been genuinely useful for spotting early imbalance before it becomes a headache.

If you haven't already, have a look at the pylontech-battery Node-RED contrib package on npm — saved me writing custom serial parsing from scratch. Just be mindful the baud rate needs setting to 115200 for the US2000C specifically, caught me out initially!

What are you using to visualise the data — Grafana with InfluxDB, or keeping it all within Node-RED dashboard? I've gone the Grafana route and the long-term trending has been brilliant for understanding seasonal patterns.

Picking up where @IvyDai left off — the piece nobody talks about is data persistence across Node-RED restarts. If your flow loses context, your historical state-of-health trending becomes meaningless noise rather than a genuine picture of battery degradation over time.

I spent an embarrassing number of evenings debugging this exact problem with my US2000C pair before I realised the node-red-contrib-influxdb node was silently dropping writes when InfluxDB wasn't quite ready on boot.

The fix that actually stuck: a proper startup delay node gating the Pylontech RS485 polling until the database confirms it's accepting connections. Obvious in retrospect, but the Pylontech BMS will happily spit data into the void without complaining.

Worth checking your Grafana dashboards for suspicious gaps around system restarts — that's usually the tell.

@DriftGal raises a fair point — I lost three weeks of SOH trend data when my Pi decided to have an existential crisis mid-winter, which is exactly as fun as it sounds.

One thing nobody's flagged yet: if you're pulling RS485 data from the Pylontechs, watch your polling interval. Too aggressive and the BMS gets grumpy and drops the connection — ask me how I know. I settled on 10-second intervals and it's been rock solid since.

Also worth wiring in a Victron Cerbo GX if budget allows — it speaks Pylontech natively and feeds straight into VRM, so Node-RED becomes more of a bonus layer than your only lifeline. Redundancy on a narrowboat or home setup isn't optional, it's just good sense.

@CornishNomad three weeks of data gone sounds like my Pylontech's way of saying "fresh start, no regrets" — meanwhile I've got mine logging to a little SQLite DB on a USB stick because apparently I trust a £3 stick more than an SD card after losing my whole Victron dashboard config to card corruption last winter. Node-RED flows backed up to GitHub too, learned that one the hard way. The cabin setup basically runs itself now, which is either impressive engineering or a cry for help depending on who you ask.