diy solar

diy solar

Epever MPPT killed MPP Solar inverter and LFP pack

mrzed001

Voice of reason
Joined
Apr 10, 2020
Messages
1,003
Location
Hungary - EU
This happened to one of the members in the Hungarian off-grid community.

OP has two 24V LiFePO4 packs connected paralel :
- 280Ah EVE cells with JK BMS
- 160Ah with cell level BMS

Batteries where charged by:
- MPP Solar 3024GK inverter with 6 solar panels
- Epever 6420AN MPPT with 4 solar panels
So really a simple DC coupled system

The system is running for more than a year now.
With smart home so if batteries are full then extra load (heater) is starting automatically to consume the extra sun power.

But yesterday suddenly battery voltage rises to 34V and later 36V.

fesz_felugrik.jpg



MPP Solar 3024GK is in FAULT 08: DC BUS voltage too high.
Inverter seems to start but without MPPT. Some capacitors are blown inside.

poped_capacitators.jpg



The smaller LFP pack is heavily damaged. Cell level BMS could not protect it.

160Ah-s_LiFePO4.jpg

There was no fire.


The larger LFP pack has now a JK BMS (old pic with the old Daly), and it survives

LFP_pack_2.jpg


The Daly BMS was replaced to the JK BMS after the post here, where the Daly did not protect the pack after MPPT failure.
JK BMS did disconnect the charge as it should (and survived):

JK_BMS_disconnect.jpg

( For some reason Cycle capacity is counted wrong in the JK BMS for some time. No clue why :D )


We still do not know the cause of the error.
Maybe MPP Solar inverter's MPPT gone to short?
We tested, no continuity between battery and MPPT inputs.

MPP Solar was set to 16-Charger source priority = Solar+Utility.
There was a firmware problem when this caused the DC bus voltage to slowly creep up to 500V (in a rainy day) and the inverter stops with the ERROR 08.
But the inverters usually survive this without damage

So still no clue what caused this.


TLDR; 2 batteries, 2 MPPTs charging it, one is damaged and one battery overcharges
 
There has been some discussion of this problem and how to protect against this kind of MPPT charger failure coinciding with BMS failure. I think the most promising solution is an NO relay on either PV in or SCC out with a voltage monitor on the DC bus. If the DC bus goes above some safe cutoff point (probably just higher than BMS cutoff), then a monitoring circuit triggers the relays on all SCCs and sounds a major warning.
 
There has been some discussion of this problem and how to protect against this kind of MPPT charger failure coinciding with BMS failure. I think the most promising solution is an NO relay on either PV in or SCC out with a voltage monitor on the DC bus. If the DC bus goes above some safe cutoff point (probably just higher than BMS cutoff), then a monitoring circuit triggers the relays on all SCCs and sounds a major warning.

I know. I posted the double solar string Voltage meter with relay solution for the problem :)
Here it is not the case.

First it is an all-in-one inverter, not an MPPT.
In the all-in-one inverters the DC bus is about 373-450Vdc (at least in the 5kW inverters, the 3kW inverters I do not know as good).
So even if the MPPT of the unit is shorted that should not make any other damage inside the inverter.
The DC bus can handle a little bit more than 500Vdc

Second there is no continuity in the inverter between MPPT and battery ports. Not like in the MPPT case.
It is strange how could this overcharge the batteries.

Also because the charger is a separate DC/DC converter (AFAIK isolated in 5kW) in the inverter. So no power should be flowing to the battery.
All MOSFETs and IGBTs look good and working.
 
Last edited:
I've always been under the impression that the only real difference between discrete SCC and inverter vs AIO is the latter is all in one box. This would fit with the "DC Bus high" error - all the Vbatt generation/draw are tied together with only overcurrent protection, some of those connections being internal to the AIO.

That doesn't explain where the excess voltage is coming from. Only thing I can think of is to try each component separately. Start with battery at 100%SOC, then separately connect each of the following and test:
  • AIO with AIO AC input charge (panel disconnected)
  • AIO with AIO MPPT charge (no grid/gen AC)
  • Epever with AIO completely disconnected
Watch like a hawk for overvoltage to determine which component. OTOH, if your AIO is under warranty just test again the Epever, and then see about RMA for the AIO black box.
 
I've always been under the impression that the only real difference between discrete SCC and inverter vs AIO is the latter is all in one box. This would fit with the "DC Bus high" error - all the Vbatt generation/draw are tied together with only overcurrent protection, some of those connections being internal to the AIO.

That doesn't explain where the excess voltage is coming from. Only thing I can think of is to try each component separately. Start with battery at 100%SOC, then separately connect each of the following and test:
  • AIO with AIO AC input charge (panel disconnected)
  • AIO with AIO MPPT charge (no grid/gen AC)
  • Epever with AIO completely disconnected
Watch like a hawk for overvoltage to determine which component. OTOH, if your AIO is under warranty just test again the Epever, and then see about RMA for the AIO black box.

We have a new theory :)

The Epever is a 12/24/48V auto Voltage setting MPPT

Since really nothing seems to be wrong, our new theory is that Epever changed itself from 24V to 48V mode ? (for an unknown reason).
Epever tried to charge the battery to 48V and that burned out the capacitors in the MPP Solar inverter.
Also the small 24V BMS could not handle the 48V so this is why the small pack overcharged.
The JK BMS can way more than 48V so it simply shut down the big pack (and this is why that survived).

In the MPP Solar inverter the capacitors are replaced. It is working again.

We want to find the cause of the error to protect the system against it :)
 
That could do it. The voltage ranges should keep it in its track, but a deliberate software settings change could also derail it.
 
You have discovered a good reason not to use cell level BMS if the charge controller failed and applied too much voltage when the fet opens up on one of the cells in the pack it may not be able to handle the higher voltage and Zenner and continue to overcharge the pack.
 
You have discovered a good reason not to use cell level BMS if the charge controller failed and applied too much voltage when the fet opens up on one of the cells in the pack it may not be able to handle the higher voltage and Zenner and continue to overcharge the pack.

It is a good thing that JK BMS has 100V MOSFETs.

 
We did some tests. And blew up the already repaired MPP Solar inverter again ?


The Epever MPPT has an RBVS (remote battery voltage sensor) connection. To get the real battery V.
OP connected it directly to the battery terminals.
And why not?
It says in the user manual to connect it to battery terminals.
And this was the problem somehow.

If connected AFTER the BMS everything works OK
But if connected BEFORE the BMS (directly to battery) the Epever does not stop the charge at full battery.
Also I think it jumps internally from 24V to 36V charge (auto V selection).
Maybe that small V drop in the BMS makes it work crazy ?
This is our current theory.

So If you use this type of Epever MPPT:
DO NOT CONNECT VOLTAGE SENSOR BEFORE THE BMS !
 
Would have thought that would go without saying. If you connect something around the bms, the bms cannot be effective.
If you bought Battleborn, or any other pre made battery, the bms is inside. They won't let you bypass.
 
Wow. Yeah, if there is a way to force the system, you should always use that. I've never understood why anyone thought auto-sensing battery voltage was a good idea.
 
Had a similar thing happen to our old lead acid offgrid system. Auto-sensing SCC 12/24v (generic chinese brand) charged 12 batteries to full, and then switched into 24v mode and proceeded to boil off the lead acid batteries. Was a good prompt to upgrade to lithium cells and a renogy SCC.
 
Back
Top