camelot-forum.co.uk

I received my Pylon US3000C batteries a couple of weeks ago and have been looking at the interconnection of the inverters with the Pylon batteries.

I'm running SMA Sunny Island 5048 inverters and wasn't sure if they would 'talk' OK with the batteries. I remember seeing a little bit of info in the SMA manuals about Li-ion. The SMA sunny Backups do have a setting for them but I couldn't find it in the Sunny Island manuals. The firmware for the Sunny backups and Island look to be virtually the same but I then noticed the US Sunny Islands have a newer firmware that includes Li-Ion settings. Comparing the CANBUS protocol for the SMA and Pylon looked very promising. Loading this firmware on a spare inverter went well and a test with a single battery pack worked perfectly.

These Sunny Islands don't have approval for connection to the Grid so the next thing to look at was the charging of the batteries in the winter using the Agile Go tariff. The SI do have a connection for a grid input which could be used with suitable protection but as I have a few ELTEK SmartpackHE chargers I wanted to look at using these. One 2Kw or 3Kw charger pre phase would be brilliant. There's a bit about the CANBUS protocol on the internet so I looked into that.

Unfortunately it wasn't going to be as simple as the SI! The Pylon and SMA run at 500kb/s and the Eltecks run at 125kb/s. So you can't just connect them together. The CAN ID's are also different for different data and the DATA bits are also in different orders. Some of the data, eg voltage on the Eltek has 52.4v expressed as 5240 but the Pylon would be 524. The main data the Eltecks need are the Charging Voltage, Charging Current and a way to be shut down. Time to get the Arduino out. Bought a couple of CAN shields and did a lot of research and then programming.

The result;



The chargers will be on a timer for the 00.30 to 4.30 time slot and the SOC that they will fill the batteries to will be selectable via a pot. That may well be expanded to be settable by a weather forecast in the near future.

I have had some fun with my SI, originally it would only run with a 6kW Australian standard, then there was a 6kW German standard, but for the past year UK G98 and G99 has become available.

I just run my Pylontechs stand alone and set them as lead acid on the SI. I set the max charging voltage to the recommended charging voltage for the Pylontechs and set the temp compensation to 0mv/degree.

I also connect a Sofar ME3000SP without any sensor clamps and run it in Timed Mode in order to run it as a 3kW supplementary battery charger, I guess similar to your "ELTEK SmartpackHE chargers". So I can get 6kW of charging from the SI and another 3kW of charging from the Sofar, a total of 9kW or approx 165A of charging to my 3 battery banks.

Whilst I have been on Go Faster I have not used the supplementary battery charger as there is enough time in winter on Go Faster (5hrs) to charge up the batteries on the SI alone, but if I had to return to a 4 hr regime, I need the extra charging power to get the battery bank up. I can set the SI charging regime via a webpage which includes a weather report for tomorrow.

You run the newer SI now, don't you?

On the darkest winter day, with no PV, I think my 3 x 2Kw chargers on for 4 hrs should easily get us through the next day, if not I'll just swap them for the 3Kw ones, or I could even run both together. They are hot swappable anyway, and I've put all their serial numbers in the program. The program will also limit their charge current as required.

I had the Lead acid method as last resort but, as I understand it, Li-Ion also seem best with float and bulk settings. I also note that Vicron limit the max charge voltage of Pylon batteries to 52.4v whereas Pylon have the max voltage as 53.2v. Vicron say this is too high. Now I've done the one Arduino I may do another to sit between the Pylon and the SI to limit the max voltage. I'm not sure if the UC3000C are different but their seems to be a lot of warnings about over voltage with them and how it will damage them.

My next thing is how to still use the Lead acids that I have. I know if you have a SI cluster the second cluster uses a separate battery pack but can that SI have a different charge setting e.g. lead acid? I suspect not as I know, as a cluster, they do battery balancing between the packs.

I have noted one thing with the SI is that it shows a charge voltage of 51.8v when the Pylon is requesting 52.8v. This is with nothing connected other than the batteries and does go up after being connected to a mains supply after a while.

I have a SMA Home manager to connect in so may be able to use that weather feature as well. I'm using a Webbox at the moment and do have the more modern equivalent so could pull data from that for the Elteks.

I have the newer v12 but that has now been superseded with a v13, although they share the same firmware.

Because I use Lead Acid settings, I set bulk and float to be the same, so the charge just tapers off when it gets to the correct voltage level. The only issue I have is every 8 battery cycles the SI insists on doing a full charge (again set to same voltage) but it insists on using grid power if not enough solar PV is available. So the work around is to occasionally force a manual EQ charge (again set to same voltage) for a few seconds and then cancel it. This then resets the cycle counter, so puts off the full charge via the grid.

From my own table of SOC % vs voltage on the Plylontechs, 53v is about 100% and 52v is about 99% so I would agree with Victron on their conservatism.

I run 3 banks of batteries in parallel, one bank of 20kWh of Plylontech batteries with a DC disconnect relay to take them out of circuit at around 50.5v (about 90%). The other two banks are re-used Growatt batteries of 25kWh each and I charge them up to 55.2v (about 90%). I used to only run them to 54v (80%) when I ran them without cell balancers, but now with cell balancers all cells are within about 5mv so I am happy to go a bit higher.

One thing I found with the SI (at least with my model v12) is that you can set the resistance of the actual cables you use so it boosts the voltage at high charge rates to take into account the voltage drop in the cables so the actual batteries get the correct charge voltage. I wonder if that is why you are getting a different charge voltage showing on the SI vs the Pylon request.

The difference in voltage is with nothing at all connected to the SI other than the battery. I'm wondering if it's a SI ramp up sort of thing.

I've only tried charging the battery pack with the SI the once and that was under observation all the time and from the grid. It behaved faultlessly with the battery pack upping the voltage as the SOC got higher.

I have found with the soft start on the US2000C with one battery isn't enough to startup first time from cold. The first try part charges the capacitors in the SI enough for the second start to work. When I have it all installed it won't be a problem.




Next- to add an LCD screen so the laptop doesn't have to be connected.

Tinbum wrote: ↑Wed Aug 18, 2021 11:55 pm I have found with the soft start on the US2000C with one battery isn't enough to startup first time from cold. The first try part charges the capacitors in the SI enough for the second start to work. When I have it all installed it won't be a problem.

Thats probably the overcurrent protection in the Pylontechs as the SI has quite large capacitors to fill up.

The US3000C have a new soft start system for the inverters that limits the inrush current. I think that has a time limit on it so if the current is still increasing it turns off. As I have 5 packs per inverter it won't be a problem.

Final version ready to be put in a nice box. I've now incorporated MQQT and a setting to enable the maximum charge voltage to the SI to be limited to that advised by Vicron (or any other setting below that set by the battery BMS).

Now incorporated Solcheck to manage the nightly SOC setting for charging the batteries from the grid. Once I start importing I'm sure I will have to fettle the calcs a bit to get it right.



Manual control also available over Node Red;



I found I had to make a hardware addition to the Arduino MEGA by adding a DUE between it and the batteries. The MEGA wasn't able to keep up with the CAN data burst from the batteries and was missing some of it. The DUE acts as a bridge and has a larger buffer and better speed. It puts the data in its buffer and then retransmits it to the mega at a slightly slower rate. Not ideal but I've yet to figure out how to get 3 CANbus shields to run on a DUE and the MQQT kept on disconnecting.

Next stage is to change the Aduino Mega and Due over to a Teensy 4.1 Should work well and be a lot simpler.

https://www.skpang.co.uk/collections/te ... d-ethernet