diy solar

diy solar

Amperetime BMS voltage sag -- is this normal?

dogfud

Doing the Needful
Joined
Dec 8, 2022
Messages
13
Location
Emerald Triangle
Sorry if this is a noob question, but I'm trying to understand if this is typical BMS behavior or not. When charging a 4S 48V bank of Amperetime 200AH batteries, if I remove the charge current suddenly there is a major voltage sag that lasts about 1.75 seconds and causes my inverter to fault. It seems like the BMSs take a while to switch from charging back to forward current ..

Here's a picture of the drop measured at the battery when I disconnect my bench supply (120VDC/3A) from the PV input on the MPPT charger (Victron 150/70). Voltage drops from around 54V to under 10V then recovers after 1700mS or so. Inverter load on the DC side is maybe 1A, and the batteries are connected to the bus with 1/0, and 4AWG to the inverter, so it really seems like a BMS thing...

removing_pv_input.png


And here's what it looks like when I just flip the breaker disconnecting the MPPT charger from the bus bar, again DC load is maybe 1A from the inverter. The inverter (Victron Phoenix 48/1200) faults and shuts down of course.

open_mppt_breaker.png

Maybe its no big deal since PV panels don't just cut out abruptly in the wild, but I just wasn't expecting the BMS to react so sluggishly like that?

Any thoughts appreciated, thanks.
 
Thanks for the youtube link, that sounds like exactly the problem I'm running into. Aye carumba, now I get to fight with Amperetime to get them to pay for return shipping ;)

FWIW, heater is necessary in this application since this will be powering a Dankoff slow pump and stuff like that in a remote off-grid location. I wonder if the Trophy server rack battery would be a more solid option.
 
Last edited:
The problem in the video is the battery has a cell below low voltage cutout which is turning off the discharge MOSFET's. Charging current is still allowed through the discharge MOSFET body diode.

This is normal for a BMS, however if more than a few amps of charging current is applied, the body diode voltage drop of the turned off MOSFET's will generate too much MOSFET heating so they turn the discharge MOSFET back on to reduce the voltage drop.

If charging current drops, or reverses to a discharge current, the discharge MOSFET is re-disabled.

The quirk in this BMS is apparently its functional mode state change operation. It still has a locked in low cell trip that is apparently not reset as long as charging current is happening.

Add to this, on low-cost hardware controlled BMS, the MOSFET gate drive is often only a high value pull up resistor. All the paralleled power MOSFET's have a lot of gate drive capacitance. With only a high value resistor for MOSFET gate turn on there is a long R-C time constant to turn on disabled MOSFET's. This is often done on purpose to prevent a sudden load surge current by turning on the MOSFET's slowly.

At least in the video demonstration, I don't think it will have this quirk if battery is not fully discharged where the BMS has a low voltage cell triggered when charging is first applied.
General BMS charge-discharge Disable function.png
 
Last edited:
The problem in the video is the battery has a cell below low voltage cutout which is turning off the discharge MOSFET's. Charging current is still allowed through the discharge MOSFET body diode.

This is normal for a BMS, however if more than a few amps of charging current is applied, the body diode voltage drop of the turned off MOSFET's will generate too much MOSFET heating so they turn the discharge MOSFET back on to reduce the voltage drop.

If charging current drops, or reverses to a discharge current, the discharge MOSFET is re-disabled.

The quirk in this BMS is apparently its functional mode state change operation. It still has a locked in low cell trip that is apparently not reset as long as charging current is happening.

Add to this, on low-cost hardware controlled BMS, the MOSFET gate drive is often only a high value pull up resistor. All the paralleled power MOSFET's have a lot of gate drive capacitance. With only a high value resistor for MOSFET gate turn on there is a long R-C time constant to turn on disabled MOSFET's. This is often done on purpose to prevent a sudden load surge current by turning on the MOSFET's slowly.

At least in the video demonstration, I don't think it will have this quirk if battery is not fully discharged where the BMS has a low voltage cell triggered when charging is first applied.
View attachment 127314

If that example was the only one, I would agree. The OP of this thread, and another have experienced the same at higher states of charge. The OP's average voltage is 3.375V - possible they have one cell at 2.XV, but it's not likely.

This individual charged theirs all the way to 14.6V, and it still did it:

 
If that example was the only one, I would agree. The OP of this thread, and another have experienced the same at higher states of charge. The OP's average voltage is 3.375V - possible they have one cell at 2.XV, but it's not likely.

This individual charged theirs all the way to 14.6V, and it still did it:



Happy New Year guys, thanks for the help.

Just to drive it home I pulled out one of the batteries and shot a quick video this morning. Results are pretty much same behavior as the one you shared from @ddracing07.

SOC is 90%+ and battery is 19C, so no issues with heating or low voltage cutoff. Applying a charge of 14.4V/1.0A and then removing it, battery voltage dips to 5.75V for about a second and a half. Playing with the charge current it does the same thing down to somewhere between 500mA and 750mA.


So this battery is definitely not going to work for my application. Glad to get to the bottom of this inside of the 30 day return period, now I hope Amperetime doesn't give me a hard time about taking these back. I'll follow up with how that turns out.
 
@RCinFLA 's concern is that while you have 14.4V set, the battery is at 13.4V under charge, and there's a remote possibility a single cell is triggering LVP.

Your test would be more definitive if you confirmed X Ah input into the battery based on current and time, and 14.4V at battery terminals indicated during charge.
 
@RCinFLA 's concern is that while you have 14.4V set, the battery is at 13.4V under charge, and there's a remote possibility a single cell is triggering LVP.

Your test would be more definitive if you confirmed X Ah input into the battery based on current and time, and 14.4V at battery terminals indicated during charge.

Gotcha, yeah so I know they were under charge for quite a long time before this, but to capture it I'll hook up my smartshunt and top it off. Also this is one random battery out the 4S bank, so I could try a second one also. Or maybe recreate it with 2P ?

That said, even if there was one low cell triggering LVP then I'm still in a defect situation, just not a design defect.
 
Last edited:
They need to fix that bug, before shipping any more out. Or they will be refunding a lot.
 
Gotcha, yeah so I know they were under charge for quite a long time before this, but to capture it I'll hook up my smartshunt and top it off. Also this is one random battery out the 4S bank, so I could try a second one also. Or maybe recreate it with 2P ?

Charge each of the other batteries inputting 10Ah - guaranteeing no possibility of LVP. Test 'em.

That said, even if there was one low cell triggering LVP then I'm still in a defect situation, just not a design defect.

I'm 100% convinced these batteries have a major defect, design or other (maybe OCP due to an issue with the heater design?). Maybe it's just a batch of defective BMSs. Seeing this from three independent sources now indicates it's not isolated.
 
Happy New Year guys, thanks for the help.

Just to drive it home I pulled out one of the batteries and shot a quick video this morning. Results are pretty much same behavior as the one you shared from @ddracing07.

SOC is 90%+ and battery is 19C, so no issues with heating or low voltage cutoff. Applying a charge of 14.4V/1.0A and then removing it, battery voltage dips to 5.75V for about a second and a half. Playing with the charge current it does the same thing down to somewhere between 500mA and 750mA.


So this battery is definitely not going to work for my application. Glad to get to the bottom of this inside of the 30 day return period, now I hope Amperetime doesn't give me a hard time about taking these back. I'll follow up with how that turns out.

That looks like a relay, I think what is happening is that when ou are charging and its low, the relay kicks on and diverts the power to the heating pads, once temperature raises to the correct temp the relay is closed and you can send power back to the BMS to charge the power.

This is the same as JBD and the SOC / BMS they all function this way.
 
Okay here's another test, this time the battery is topped off and the Victron shunt shows it took at least 20Ah. BMS still drops out when you remove the charge, same as before.


And here's a picture from the scope where you can see it drops out for about 1500mS. Something's going on with the ripple at the bottom at around 4.2V, not sure what that is but it takes a while to figure out what to do.

I'll test the other three batteries just to confirm they all do the same thing, but at this point I'm pretty certain they will.


SDS00011.png
 
With the 10 amp charge, the terminal voltage rise on battery is only 50-80 mV so both back-to-back MOSFET's must still be on which is what the condition should be.

They are definitely shutting down, at least the discharge side MOSFET's, when charge is removed.

It is going into the disable discharge MOSFET's mode for 1.5 seconds after removing charge. Should only disable discharge if there is still a cell undervoltage when charge is disconnected.

Looks like, under charge, logic is assuming a low cell then after removing charge and detecting there is not a low cell, it re-enables discharge 1.5 seconds later.

I would think this would cause an immediate major uproar in customer complaints soon after beginning use of battery, unless it does not do this when there has never been a low cell voltage condition on BMS since it was first powered up when installed with battery pack, making it a problem waiting to happen after customer runs battery down to a low cell condition the first time.

Maybe doing a BMS hardware reset, by disconnecting BMS from battery pack for a minute, clears the problem, at least until another low cell voltage trigger happens again.

Everybody's inverter/charger will shut down when charging battery completes. That is really screwed up.
 
Last edited:
Just from seeing what people are saying. It could be that the BMS is switching to heating mode briefly, before switching to discharge. But heating should be tied to charging, not discharging.
 
Back
Top