SOK Troubles (user error)

[mrzed001]

Just for perspective, an entire bluetooth SOK BMS is $100 retail which is one reason why nice contactors are normally not used due to the additional cost it would add to the battery (along with power draw).

If I was doing a high voltage solar installation or definitely a large DIY battery then I would be looking at a good contactor with a external BMS but I would also not be skimping on the solar controller.

Even the "best" (most expensive) Victron MPPTs do not have galvanic isolation. If FETs fall short, same result.
Only one, the hyper expensive 450Voc Victron MPPT has it.

https://www.victronenergy.com/upload/documents/Datasheet-SmartSolar-MPPT-RS-EN.pdf

And at this moment I could not think of any other standalone MPPT that has galvanic isolation.

Better all-in-one off-grid and real hybrid inverters have galvanic isolation from battery (to DC BUS, to MPPT, to AC sides)

 

[HighTechLab]

Excellent work here tracking down root cause and showing us and even better getting SOK to improve the design, thank you.

Does seem like they could design the IC such that it is driven off the battery side so that once the FET's are shutdown due to over voltage the IC would never be in the circuit on the high side and the FETs would have to breakdown before it would be exposed.

The IC needs to sense the voltage (in this case that killed it) in order to release the shutdown.

know the MCU that 90% of these BMS units which is either an ATMEGA328P or STM MCU,

It's an SIT8995, feel free to refer to spec sheet below.

Even the "best" (most expensive) Victron MPPTs do not have galvanic isolation. If FETs fall short, same result.
Only one, the hyper expensive 450Voc Victron MPPT has it.

https://www.victronenergy.com/upload/documents/Datasheet-SmartSolar-MPPT-RS-EN.pdf

And at this moment I could not think of any other standalone MPPT that has galvanic isolation.

Better all-in-one off-grid and real hybrid inverters have galvanic isolation from battery (to DC BUS, to MPPT, to AC sides)

It doesn't require full galvanic isolation, so searching for galvanically isolated MPPTs is a waste.

Most quality MPPTs use magnetic coupling via a transformer to step down voltage, and the drive for this transformer is just like a switch mode power supply. If the drive circuit fails shorted, you have DC to the transformer, therefore no current flows across (need changing magnetic fields in order to have flow). The reason galvanic isolation is not achieved in most controllers is because PV- and B- are coupled...It's perfectly safe if one side of the PV is coupled, but as long as BOTH aren't, you are fine. By having the negative side coupled, you technically loose this galvanic isolation rating.

 

[jharrell]

The IC needs to sense the voltage (in this case that killed it) in order to release the shutdown.

Isn't the ADC going to have a voltage divider anyway? Doubling the voltage division would allow it handle 80v instead of 40v.

 

[HighTechLab]

Isn't the ADC going to have a voltage divider anyway? Doubling the voltage division would allow it handle 80v instead of 40v.

No, just through a 1k resistor, not a divider network.

 

[jharrell]

No, just through a 1k resistor, not a divider network.

Ok the IC is specific for BMS's (I need to learn to read Chinese) and has a charge sense pin rated at 40v most and probably internal divider to ADC, technically you could put a another divider in front of it and change the software to read the new lower value properly, its not like the charge sense has to have a lot of resolution. However 40v for a 4s BMS is a reasonable amount of protection IMO although it would be nice if it where closer to the FET's rating.

 

[mrzed001]

Most quality MPPTs use magnetic coupling via a transformer to step down voltage, and the drive for this transformer is just like a switch mode power supply. If the drive circuit fails shorted, you have DC to the transformer, therefore no current flows across (need changing magnetic fields in order to have flow). The reason galvanic isolation is not achieved in most controllers is because PV- and B- are coupled...It's perfectly safe if one side of the PV is coupled, but as long as BOTH aren't, you are fine. By having the negative side coupled, you technically loose this galvanic isolation rating.

Most MPPTs do not use a transformer. Especially low V MPPTs like 70-150Voc. Only a Buck/Boost converter with FET, capacitor and inductor.

I can not tell what is inside a Victron MPPT
It is not serviceable. Full with a strong blue foam. You have to cut the metal case open and scratch the foam.
Never saw a full teardown of a Victron MPPT (yet ).

 

[Alkaline]

Ok I have changed my stance on this topic, SOK and current connected have made a good case, this appears to be an issue with furion mppt charger.

So now, what can we do to stop this form happening, I think these faulty MPPT can burn many batteries.

 

[Bud Martin]

Most MPPTs do not use a transformer. Especially low V MPPTs like 70-150Voc. Only a Buck/Boost converter with FET, capacitor and inductor.

I can not tell what is inside a Victron MPPT
It is not serviceable. Full with a strong blue foam. You have to cut the metal case open and scratch the foam.
Never saw a full teardown of a Victron MPPT (yet ).

I do not believe the small Victron SCC uses isolated buck converter.

Do NOT discard a Dead Victron MPPT 100/30 or its kind

All, For those who have a failed Victron MPPT, please do not discard it. I have few experiences to resurrect the dead Victron MPPT 100/30.

diysolarforum.com

 

[Mahendra Gomanie]

I feel bad for this guy.

View attachment 87699

View attachment 87700
View attachment 87701

Looks like poor settings on charge controller .cheap charge controller with really high voltage for long period/s

 

[RCinFLA]

It will charge and charge ... and boil the battery.

Worse case scenario: FLA battery explodes.

Worst case scenario: if lot of battery or small not vented room then Hydrogen (HHO) explosion :

Just a side FYI,

Carbon Monoxide alarm monitors are sensitive to hydrogen gas. I have had two occasions over the years where it sounded the alarm on stacked 48v lead acid battery array when one cell went shorted.

 

[Hedges]

Ok I have changed my stance on this topic, SOK and current connected have made a good case, this appears to be an issue with furion mppt charger.

So now, what can we do to stop this form happening, I think these faulty MPPT can burn many batteries.

We've done a lot of thought experiments and napkin designs of possible protection circuits:

Why not to use Daly BMS with MPPT controllers

So there is a major problem with these type of BMS. Any thoughts?

diysolarforum.com

Simplest approach is to buy a top-tier SCC, which is less likely to fail. And be sure to stay within Voc & Isc limits when designing PV array.

Some SCC may be available which are transformer-coupled, eliminating the issue that failed transistor causes PV to battery connection.

 

[BatteryNut]

This is not good... BMS should never continue to send power to a battery once it fails, these things need a hard switch or fuse, the MCU on the controller should stop all current going in and out.

These type of events are going to give lithium a bad name and I have seen enough posts like to know that MPPT controllers are having voltage spikes.

Also not singling out SOK, I mean any BMS that allow current to flow when it has failed. Not good at all.

This has nothing to do with the MCU. You understand what FIELD EFFECT in FET means, right? You can output whatever voltage you want from the MCU's IO to the gate and it won't do a damn thing. What you're trying to say is people should substitute FETs with more robust mechanisms such as quick switching relays/mechanical contactors with secondary aux switches backed up with a resistor ladder to detect partial weld conditions and throw in something like a class T fuse for failsafe. Maybe you want to go big and add things like HVIL and a full ISO26262 compliant software stack. But you forget something very fundamental: most people are PRICE SENSITIVE. You think the majority of people here build their systems because they're professionals in EE and know what they're doing? No, they do it because they get more JUICE/DOLLAR rolling their own.

 

[Hedges]

System architecture is going to include battery, BMS, SCC, inverter.
Various protection can be included in each.

BMS has to handle very high current to supply inverter. Unless it has two ports (charge/discharge) high current and high withstand voltage in one circuit makes it costly and inefficient to implement there.

SCC is the place to put this protection. It handles lower current, and has a switching power supply already. That can be made isolated (at some cost and efficiency loss) completely solving this particular issue. Transformers are easily rated for 600V isolation and tested to multiple kV, so PV voltage can't break through. Only severe coupling or strike from lightning would be a problem. Remaining issue of SCC losing its head and "regulating" to excessive voltage is much less likely (and could have supervisor to crowbar gate of FET and shut it off.)

Other products like grid-tie inverters have redundant measurements and control. One error message from my Sunny Boys is difference between the two circuits, so it won't drive grid unless both agree everything is good.