Pondering a High Voltage DIY Battery solution for a Sandi 3-phase inverter

[Spelunker]

After much deliberation, I'd really like to implement a 30kw Sandi inverter for my off-grid workshop. I haven't found many adventurous souls attempting to tackle a high voltage ESS. Mike G's YouTube channel got me thinking... and high voltage does indeed have its advantages. His homebrew battery solution seems to be working fine, but I'd like to implement something a little less intimidating.

Not really keen on re-purposing an EV battery. Prefer to have a more maintainable, aesthetic solution that will be deployed in a Mechanical Room.

Some have suggested using 48V battery systems in series to achieve the voltage needed but are advised against doing so because the BMSs are only compatible with parallel banks.

I'm attempting to determine what the real limitations of using multiple 48V battery systems connected in series other than the obvious fact that there is no specific total aggregate data communicated to make sure the units stay somewhat synced and are in proper working order.

Could you not monitor each 48V BMS separately with custom software, and then determine the status of the whole series? If one of the units were to become out of balance, you could make settings adjustments to it or the other units.

If one were to use common DIY battery box kits with identical BMSs (JK Inverter BMS?), it should be doable to monitor things effectively and make sure orderly shutdowns happen safely if necessary.

Trying to evaluate starting with 5 "assemble yourself" boxes like the Seplos or Yixiang boxes using 280Ah or 302Ah cells. That starts things out at a nominal 256V system which is within the Sandi inverter specs.

If anyone has some thoughts on this, it would be much appreciated.

Thanks!
- TM

 

[Solar Guppy]

I'm attempting to determine what the real limitations of using multiple 48V battery systems connected in series other than the obvious fact that there is no specific total aggregate data communicated to make sure the units stay somewhat synced and are in proper working order.

When one of the batteries goes offline ( open circuit ) in a series string, the voltage presented to other batteries will exceed the rated voltage of the BMS and literally will blow up the output mosfets, possibly fires on the PCB as well.

For example: When the string of 3 or more is all connected, there should be 48V present across each battery

3 in series all have charge and discharge enabled:

0 48 96 144 Voltage across battery terminals
- 48V+ - 48v+ -48V+ Battery each with its own BMS


When one battery goes offline in the series string ( open circuit ), you now have a larger potential due to the inverter ( load ) positive going to towards 0v.

Lets say the first battery BMS turns off discharge you now have a potential -96V across the first battery

0 -96 -48 0 Voltage across battery terminals
- 48V+ - 48v+ -48V+ Battery each with its own BMS

As the series strings have more elements its get even higher potentials

HV batteries have no FETS in the string, all cells are always connected. Additionally, there is the need for isolated communications and HV batteries typically use contactors not FETs

In theory, if you have multiple parallel series strings this might be avoided, but can't be guaranteed


Seplos make HV batteries, I doubt you could build one for less, check out my system for links

Home made HV batteries are not something most should attempt, even with my experience, it is something I would not build, to many issues and hazards to contend with.

 

[Spelunker]

0 -96 -48 0 Voltage across battery terminals

Ahh yes, I see the issue. Since each BMS can independently switch the charging and/or discharging function of that bank (using FETs) then we have the problem with unexpected voltage potentials being experienced by the other BMSs.

So... We really never want a 48V bank to become disconnected, the series string must be operated as a whole and not bank switchable. Maybe we need to lobotomize the BMSs to only watch cell voltages and balance the 16 cells in its purview. Bypass (not connect) the main FET switch. I wonder if it can power on and work in that mode?

Then each bank can communicate via CAN/RS485/RS232 to a small custom controller that is responsible for switching contactors for enabling charge and discharge modes on the whole string. The communications would need to be verified as isolated or made to be so.

Or... Ditch the BMSs and simply use a smart balancer in each box. Setup and monitor cells again with a custom controller. This might be the better approach.

With LiFePo4 prices currently, my rough calculations would seem to indicate a 256V, ~78kwh system should be feasible to build at significantly less than $10k. Maybe I'm dreaming.

Seplos make HV batteries, I doubt you could build one for less, check out my system for links

I checked out your system... Very nice setup!

 

[Solar Guppy]

Orion BMS has what you would need for a HV battery, then lots of parts for lots of cells, custom cases and all the little things that add up is cost.

Sure it can be done, but for the hours involved its isn't a true savings over just buying one on Alibaba ready to go.

Also your looking at much higher than 250V as the current is limited to 100A for that inverter your looking at. Add in surge needs and your looking at 409V for sure ( 128 cells ), which is what I have. Seplos has a rack mount option that could meet your storage needs, when you price it out to building your own there isn't any savings to DYI