p_i

It's been unstable since 04:40 this morning. 04:40 was my last reading in sequence and then there's several readings missing after that.

I've been looking at those too since the outage this past weekend. Those registers are 16bit (2 bytes) values where the high 8bit (byte) is the first value and the low 8bit (byte) is the second value. In simplified code this means that: month = register22%256; year = (register22-month)/256+2000; hour = register23%256; day = (register23-hour)/256; second = register24%256; minute = (register24-second)/256; % = modulus function. I'm not sure how you're going to do this in HA as I'm using node-red to do these calculations. I still need to test my time updating code but I need to do this close to midnight so I don't mess up my daily totals if I get it wrong.

Just turn off the "Time sync" option on the inverter. That way the totals won't be mucked up like that.

Even my own monitoring were affected by the randomly changing time on the inverters. It appears that when the time is adjusted in the wrong direction the daily totals are zero'ed on the inverters. I managed to get past this by unticking the "Time sync" option and manually correcting the date & time. Are you telling me that Solar Assistant still had the correct daily totals? Please note: I'm not attacking the product and merely wondering.

I've observed that the time is randomly changing on my inverters during this debacle and since schedules are time driven I've turned off the time sync option and manually corrected the current time and my schedules are once again working. Does anybody know what server(s) the inverters use to sync time?

Photo? Never mind, found it here:

Oh thank goodness. I was beginning to think it was due to local changes I've made (pi-hole installation). I should have known it was the Sunsynk data handling again... 🤬

I'm using mbusd on a RPI3b and I can connect to it from node-red running on another device using the same settings you listed (apart from the IP). Maybe the docker container is preventing you from accessing it? Maybe try to telnet to that IP on port 502 from another device on the same subnet? If telnet fails to connect, you might have a docker issue.

I've used 2x 120mm fans on each of my 5kw inverters. I simply cable tied them together and they're resting on top of the inverters. Yes, at some stage I'll have to come up with a proper mounting kit, but there's no reason to not use larger fans.

Try this thread here:

I'd hazard a guess that this site is a scam. They list their address as in KZN (2 different addresses in header and footer), yet they claim to be located in Westlake in Cape Town and offer free delivery in the Cape Town area. Other than that, the price is way too low so this is most likely it's a scam.

Is it just the photo or is the protective plastic film still covering the batteries?

The modbus queue in node-red filled up too quickly and by the time the next run is triggered, the previous run is still busy. So the queue just grows and grows and you start getting errors. It's entirely possible to do this if you only have a few registers you want to read, but if you have as many as I have (25+ per inverter), it will exceed the limits of rs485. However, if you still exceed the limits, you can simply increase the delay between each run, which is what I had to do back then.

If you have a single register you're polling, that might work. However, when you have about 25+ registers you poll per inverter, it gets busy quite quickly. Also remember that rs485 is a serial interface so you have limited bandwidth and signal issues when you try to exceed that.

I've managed to use a single rs485-usb device on my dual inverter setup which by the nature of parallel installations have different modbus ID's. It worked fine as long as I didn't poll the inverters at the same time, ie. I had to offset the polling times by a few seconds. I used that for a day or two until I could get a second rs485-usb device so I could get quicker updates (I pull data every 5 seconds at present).

You do get devices like these, if you only have AC available there in a DB board or something: https://za.rs-online.com/web/p/din-rail-power-supplies/1457871

No 12V from inside the Sunsynk that I know about.

That value IS in slipx's yaml: - platform: modbus_controller # 186 PV1 Power modbus_controller_id: inverter name: "${friendly_name} PV1 Power" id: inverter_esphome_pv1_power register_type: holding address: 186 unit_of_measurement: "W" accuracy_decimals: 0 device_class: power state_class: measurement value_type: U_WORD They're not in there in numerical order, so you have to search.

Power Solar PV1 = 186 and PV2 = 187.

Which variable/value are you looking for? On which inverter model?

Totally agreed.

I suppose this is what they meant with "fighting against each other", eg. 8kw start feeding power to a device in the house and then the 5kw also sense that something needs power and then there's too much power in total and it leaks out via the CT and the 8kw and/or the 5kw senses that and backs off. Maybe when they're properly parallel'ed (with comms cable between them) there's a bit more communication between the inverters so the one doesn't send more than what's needed.

A CT is simply a device to measure current passing through. It doesn't stop anything. It's merely there so the device using it can measure and take action, eg. throttle back output. But I guess you know this already?

I went a different route. I have one RPI3b in my garage with 2 waveshare rs485 usb adapters that's wired to the 2 inverters in my garage. I'm running mbusd software on that RPI3b which basically serves as an rs485 gateway via wifi. On another NUC inside the house I'm running node-red which grabs the stats from mbusd on the RPI3b and after some mild massaging it stores that in an influx database. The data is then visualised from influxdb with grafana. I experimented with HA for a while but all the restarting due to semi frequent updates put me off. I prefer something that runs 24x7 without constant interruptions due to software updates.

My own monitoring solution is giving me updated data every 5 seconds, without having to get up to walk to an inverter, so the modbus route is the best way in my opinion. Even with quicker stats you will still observe stuff that seems slightly off since no CT/meter will be 100% accurate, but you'll get better stats than the Sunsynk options. But unfortunately we're still at the mercy of the people coding the Sunsynk firmware...