Just to be clear. Do you want me to connect the power supply to the charger port (after connecting the battery pack)? Or do you want me to power the BMS from the power supply? I suppose the former but just double checking.
Have a pack connected to the BMS so that it’s powered up, you can use your current 12S pack at 40V.
Set $5 high enough, that the voltage you’re going to connect to the charger side, doesn’t trigger an attempt to start charging, so say set it to 40000 or higher, if you have a max 24V coming in.
We’re just interested in seeing if the charger side voltage sense circuit is accurate or if it’s showing a clearly wrong voltage.
OK, done. I first connected the power supply while being turned off. Then I turned it on and gradually increased the voltage to ~24V. Then back to 0V and finally I disconnected it completely.
The voltage readings seem to be correct once the power supply is connected (before that it shows ~12V).
So here’s the schematic sheet for the charging Mosfets, the BQ76200 gate driver for them and the charger side sense circuit (battery side sense circuit blanked, as it’s not relevant in this case, but it’s nothing special). Pretty much the typical application circuit from the datasheet.
Based on Jan’s measurements and what I’m seeing in the state printout, I would say that the charger side Mosfet Q2 has been overvolted, I’m guessing/assuming by repeated short voltage spikes when connecting a powered on charger, slowly chipping away at it. There has been a punch-through event that has broken down the Drain-Source silicon, that has caused the Mosfet to become low impedance, whereas a Q1 still seems to be alright as it’s Drain-Source measures in the megaOhm/open-circuit region.
Based on the continuous 12V that the charger side is measuring, I’d say the Q2 has an effective resistance of ~330kOhm in the Drain-Source direction, so that the new, now existing resistive divider would produce the voltage for a measured 12V voltage.
This is explained that when you connect the power supply, it drains/sinks the voltage down to 0, Q2(R2) is still quite high resistance, so it can’t let a lot of current flow through it.
EDIT: Just, a sec I’ll measure my own units’ forward voltages and resistances for the Q1 and Q2
Unit1:
B+ to middle of fets: 2,7MOhm, 0,47V
C+ to middle of fets: 2,7MOhm, 0,47V
C+ to B+: open-circuit / infinite resistance in both directions
unit2:
B+ to middle of fets: 2,4MOhm, 0,46V
C+ to middle of fets: 2,8MOhm, 0,48V
C+ to B+: open-circuit / infinite resistance in both directions
I can’t say based on experience, if the Q2 is toasted enough that it can’t be opened for charging. Aka, is the now leaking state the most “closed” it can be, but it’ll still open and goes low-impedance (low-resistance) when driven by BQ76200.
My small (4A) charger trips an efuse if it gets any energy coming from the battery. Plugging the board in and then the charger in the wall is just not an option.
Again, moooore mosfets, pretty please
It would then seem that it’s “stuck” in that state. Probably still stays high impedance when the charging starts and the current flow is low enough, that it triggers the charge termination current limit.
That’s what I would recommend, as the battery side Fet, based on the multimeter measurement, seems to be still okay.
As the fault was what I had been suspecting, and I believe I’m correct what caused them, I have also thought about ways to protect against it. And I would place a TVS diode between the C+ and GND pad. There is a 54V TVS diode on the battery side B+ and GND, but the charger side doesn’t. This would be an easy addon-fix to protect from future voltage spikes, that is doable with a normal soldering iron, not requiring hot air.
I’ll likely add the TVS diode to the board design for the upcoming “Lite” → “Lite 12S” refresh and HW would then move on to 1.0 state. There some small other changes I would be doing as well.
@Zach I’d say you’re in the same situation as @janpom , one or both of your charging Mosfets are toast and most likely have been zapped by your charger.
With the breakdown voltage at about the Vds-max rating for the FETs and the clamping voltage up over 80V you feel that TVS would still protect the FET?
Has anyone scoped out the charger voltage transient to estimate its power? That might be important for the max power spec for the TVS.
Im really trying to think here. Assuming it’s not the charger (just for the sake of the argument), is there something else that may have affected these mosfets? Perhaps something I soldered?
Considering the charger side doesn’t really have any capacitance apart from the stray one, there isn’t really a lot of energy in the current rush that might happen, so even if we are not in the voltage region of peak clamping capability, it’ll be able shunt that short and sharp pulse transient, but at the moment there isn’t anything absorbing the voltage spikes.
Only realistic thing I could come up with is a static electricity zap from the human body (now even better with synthetic material clothing!) to the C+ pad, that could be in the thousands of volts and absolutely be capable of causing the punch-through event for the Mosfet.
I’m gonna improve the state printout logging, so that for the battery and charger voltages, it’ll print out lowest-latest-highest values, because there could be something drastic going on, but due to the printout currently only printing the latest value once every second, there is quite a lot things that might be missed based on it. In your case the worrying thing was that the charger side sense circuit had measured over 60V at one point.
)