Adding DIY battery to Bluetti AC300

[Zedart]

@sideeffect yes, you need a P090D To P150D and then I cut the P150D end. Locking only on AC300 side. For me looks like AC300 calculating capacity using voltage. So far my setup AC300+B300+Fogstar working well. Fogstar is connected to second AC300 input, so is charged direct from there.

 

[plympton]

I have my AC300 connected to 2 B300 + 2 Vatrer Rack batteries. It all works well, but the B300 SOC goes crazy when it fluctuates in cloudy conditions for a few days. If I get the batteries to fully charge, it all works great. If not, 1 B300 might say 3%, the other 30%, and the Vatrer's might say 60%. I wager the voltages are all OK, but it's a bummer if I truly have 60% battery left, because the AC300 will switch to grid power prematurely - just $, not safety, fortunately.

I've also had one B300 think it's charged, and the other at like 50%, and will jump up 30% in a few minutes. I think it's a Lifepo4 thing, with the voltage curve being so flat.

Either way, I hope we get more sun before EOD, as I'm stuck at 27% and won't make it through the night yet!

 

[Zedart]

I have my AC300 connected to 2 B300 + 2 Vatrer Rack batteries. It all works well, but the B300 SOC goes crazy when it fluctuates in cloudy conditions for a few days. If I get the batteries to fully charge, it all works great. If not, 1 B300 might say 3%, the other 30%, and the Vatrer's might say 60%. I wager the voltages are all OK, but it's a bummer if I truly have 60% battery left, because the AC300 will switch to grid power prematurely - just $, not safety, fortunately.

I've also had one B300 think it's charged, and the other at like 50%, and will jump up 30% in a few minutes. I think it's a Lifepo4 thing, with the voltage curve being so flat.

Either way, I hope we get more sun before EOD, as I'm stuck at 27% and won't make it through the night yet!

Do you tested with different batteries configuration? 1. B300+B300 2. Varter+Varter or 1. B300+Varter 2. B300+Varter?

 

[plympton]

Do you tested with different batteries configuration? 1. B300+B300 2. Varter+Varter or 1. B300+Varter 2. B300+Varter?

I have tested with 2 B300 + 1 Vatrer, and 2 B300 + 2 Vatrer. The Vatrer's seem to work really nicely together - the BMS's stay in sync pretty well. The B300's can be wildly out of sync with what they're reporting. I've seen a B300 go from 70 -> 100% in a matter of minutes. I've also seen 1 battery at 28% and the other at 100%! It's not for long, but the rack batteries seem to have a better tracking algorithm.

That's not terrible, per-se (I have it on good authority that the voltages are balanced), but the logic of the AC300 says to switch to line when below 10% (PV Priority), so it likely starts too early. I also worry if that 3% SOC I saw the other day was real or not.

I've pondered selling or reusing one of the B300's and getting another rack battery. This way the AC300 doesn't have 2 BMS's SOC's to average out making decisions. Dunno. For now, it is beautifully boring.

 

[Andrewr05]

That's odd that Bluettis own batteries are so wildly out of sync.

I wonder if there's a way to spoof the B300 connection and not run with one of their batteries at all?
That would be the ultimate hack right there.
Setting up a little dongle in-line to just trick the AC300 (and presumably this would work with the AC500 as well?) to believing that there's ones of its own batteries connected.

 

[plympton]

That's odd that Bluettis own batteries are so wildly out of sync.

I wonder if there's a way to spoof the B300 connection and not run with one of their batteries at all?
That would be the ultimate hack right there.
Setting up a little dongle in-line to just trick the AC300 (and presumably this would work with the AC500 as well?) to believing that there's ones of its own batteries connected.

I'm sure it's... possible. I'm sure it's not worth it unless it's a serious passion project. By the time you get there, the time and $ involved would buy a new inverter.

The nice thing about this hack is that I could swap from the AC300 to a different off-grid AIO pretty easily. I've been pretty pleased with it overall, outside of the SOC reporting.

 

[Kremer]

That's odd that Bluettis own batteries are so wildly out of sync.

I wonder if there's a way to spoof the B300 connection and not run with one of their batteries at all?
That would be the ultimate hack right there.
Setting up a little dongle in-line to just trick the AC300 (and presumably this would work with the AC500 as well?) to believing that there's ones of its own batteries connected.

I think its the B300/B300S BMS that is in charge of all the SOC and battery "brains" of the combination. It is what controls the charge rate, termination, etc... Spoofing it enough to satisfy the head unit wouldn't just be a manner of jumpering something.

 

[adirondack_wanderer]

Post removed - answered my own question

 

[sideeffect]

@sideeffect yes, you need a P090D To P150D and then I cut the P150D end. Locking only on AC300 side. For me looks like AC300 calculating capacity using voltage. So far my setup AC300+B300+Fogstar working well. Fogstar is connected to second AC300 input, so is charged direct from there.

Thanks for that info it was really helpful. Did you change the DIP switches at all on the Fogstar for example making it battery 2? Or did you just leave it at default.


Edit- looks like Inverter protocol is only needed for communication so I guess I can ignore that.

 

[Zedart]

Thanks for that info it was really helpful. Did you change the DIP switches at all on the Fogstar for example making it battery 2? Or did you just leave it at default.

I left dip switches at default.

 

[chromedshark]

I don't know if this is helpful for anyone, but I did some digging into the B300 communication with the AC300, and you can read it with just a normal CAN transceiver. CAN HIGH is pin 3 and CAN LOW is pin 4, where pin 3 is top right pin on the B300 battery port and pin 4 is the top left pin (numbering from the Bluetti B300 manual - just zoom in on the PDF). It communicates at 100kbps using the extended CAN format (29-bit ids), with CAN ids that look like 0x000000000CF50116 and 0x000000000CF50205.

It's using this 0x000000000CF50XXX form for all ids, with the first battery at 0x1XX and the second battery at 0x2XX. From a monitoring perspective, it's pretty easy to find the 16 balance voltages (4 two byte fields in 0x03, 0x04, 0x05), the full pack voltage (first two bytes of 0x02), and the state of charge percentages in the data (first two bytes of 0x12). However, making some non-Bluetti battery show up properly would likely require way more research. I'm guessing this is something totally custom, but I don't have any other experience with BMS CAN protocols so maybe it's actually similar to something else.

I did an analysis session where I recorded all CAN messages and monitored the device over Bluetooth while fully discharging two B300 batteries and then charging them back up to full, and I'd be happy to share the files for that if anyone is interested (the CAN log is 100MB). My primary interest was in reusing my batteries with a bigger inverter like an EG4 6000XP, but with server rack battery prices as low as they are, I don't know if I'll actually take any of this research further. Still, I figured I might as well post _something_ to save others some time.