https://endless-sphere.com/forums/viewtopic.php?f=2&t=84856&hilit=bms+switch
Autopsy of a dead BestTech Power HCX-D131 14S 80A BMS
- Thread starter Matador
- Start date
https://endless-sphere.com/forums/viewtopic.php?f=2&t=84856&hilit=bms+switch
izeman
1 GW
thanks. funny to see that i posted in that thread as well (w/o having read the first pages of course).amberwolf said:
https://endless-sphere.com/forums/viewtopic.php?f=2&t=84856&hilit=bms+switch
imho i already tried to lift the gate signal (2y ago) and put a switch in line. can't remember why it didn't work. but i'll try again
izeman said:thanks. funny to see that i posted in that thread as well (w/o having read the first pages of course).amberwolf said:
https://endless-sphere.com/forums/viewtopic.php?f=2&t=84856&hilit=bms+switch
imho i already tried to lift the gate signal (2y ago) and put a switch in line. can't remember why it didn't work. but i'll try again
depending on the type of mosfet, it turns off/on differently. The gate can't ever be left floating, it typically needs to be pulled up or down with a resistor to keep the mosfet off unless turned on, also through a resistor but smaller value, typically few Ohm - 100 ohm. Floating gate often turns mosfet on/off by it's own. Be sure of what you are doing, you don't want oscillations when dealing with these high currents.
izeman
1 GW
thanks for clarification. i would have directly pulled down GATE to GND to shut down the bms. so a toggle switch in the gate line, toggling the GATE between whatever it was connected to before, and GND. correct? or do i need the resistor?
izeman said:
Ok so let’s presume the switch has 2 positions and the mosfet is n-type (check this!)
This is probably simplification of what you have
http://tahmidmc.blogspot.se/2013/02/n-channel-mosfet-high-side-drive-when.html
edit: see figure 1: low side switch. R1 is the pull down resistor which will keep mosfet gate low at all times, even with power off and when there is no signal coming through R2. This is the effect you wish to recreate with your switch, but you also wish to disconnect the potential signal coming from R1 when doing this otherwise you might fry gate driver in case it was trying to turn mosfets on while you are wishing to turn them off.
these are the 2 positions of the switch and how they should be connected:
Off position:
Gate disconnected from “old” gate drive
Gate connected to GND through 1kOhm resistor. (a similar pull down resistor already exists between gate driver and mosfets. If you can find this resistor than you can keep it and insert switch between gate driver output and pull down resistor. Ideally if the gate driver was a discrete element you would influence it’s logic, but it is probably not the case. If by any case the gates are decoupled mutually with resistors there would be some changes to this but this is probably not the case.
On position: to return the circuit in the state as if before you did any modifications. On position is actually “BMS Governing On state”
izeman said:
I've seen the temp sensor work if you add a resistor from B- to P- to bring the output up to pack voltage (this resets the BMS). This would be like a precharge resistor. Downside of this is the controller could still drain something through the BMS. If the controller is the type with it's own switch or display unit, this might be fine. Resistor would be something like 1k, 5W.
Inserting a SPDT switch in series with the gate drive should work fine if the controller caps aren't too big. The BMS gate drive is kind of wimpy and might switch on slow enough to blow the FETs against a large capacitive load. In the off position, the FET gate can just be grounded.
fechter said:
good idea with hijacking thermistor.
in my opinion grounding the FET gate should be done only in conjunction with disconnecting the gate driver (otherwise you could keep the gate driver supplying current trying to keep the BMS on while you are trying to force it off)
vex_zg said:
Yes, that's the idea. It would look like this:
This would require finding and cutting the trace that feeds the FET gates. I think this will actually work quite well but there is still the risk of blowing up the FETs with a big controller.
izeman
1 GW
so you would not recommend to do it? "large" caps and "low enough" leaves a lot room, and i don't want to blow the bms.fechter said:
fechter said:
In the attachment I would add 2 resistors. When looking at your diagram let's name switch connections 1, 2 and 3, from top to down.
R1: 100kOhm, connected directly and always between all gates and ground. This will at all times ensure that the gate is never floating, not during switch transition and not if switch would fail. This might already be implemented at gate driver exit but you will disconnect that when switch is in off state
R2: 50 Ohm connected between ground and switch point 3, instead of now connecting directly point 3 to ground. This will ensure that the turning off is not too fast (thus creating high transients which can fry mosfets. This condition can only occur if you do a turnoff when BMS is under load).
I agree that placing a 50 or 100 ohm resistor in the ground leg of the switch will be helpful in the event you try turning it off under full load. I see no harm in adding this.
I also agree that some kind of gate bleeder resistor is good in the event the switch wires are disconnected or the switch fails. But this resistor would need to be very high value as the gates are normally powered all the time and drain from the bottom 3 cells. The supply for the gates might be going through a large resistance also, so we don't want it to pull down the gate voltage much. I've seen some designs where the gates are fed through a 4.7M resistor. I would suggest something more like 10M ohm, but even this might be too low for some designs.
How big are your controller caps?
What voltage?
If the main caps are 1,000uF or less, I think it won't have any issues. For more than that, it might be OK too, but I wouldn't want to try it without a bit of testing first. The key thing is how fast the gate voltage comes up. Too slow and it will blow.
I also agree that some kind of gate bleeder resistor is good in the event the switch wires are disconnected or the switch fails. But this resistor would need to be very high value as the gates are normally powered all the time and drain from the bottom 3 cells. The supply for the gates might be going through a large resistance also, so we don't want it to pull down the gate voltage much. I've seen some designs where the gates are fed through a 4.7M resistor. I would suggest something more like 10M ohm, but even this might be too low for some designs.
How big are your controller caps?
What voltage?
If the main caps are 1,000uF or less, I think it won't have any issues. For more than that, it might be OK too, but I wouldn't want to try it without a bit of testing first. The key thing is how fast the gate voltage comes up. Too slow and it will blow.
regarding the pull down resistor for which fechter is proposing 4.7MOhm. ..
...yeah this value is not that critical. Yes it will cause continuous drain of the gate driver. But 100kOhm @ typically max 20V for Vgs will only incurr 0.2mA current. So I don't thin it would be stressing gate driver. I personally would put something between 0.1MOhm and 1MOhm
...yeah this value is not that critical. Yes it will cause continuous drain of the gate driver. But 100kOhm @ typically max 20V for Vgs will only incurr 0.2mA current. So I don't thin it would be stressing gate driver. I personally would put something between 0.1MOhm and 1MOhm
izeman
1 GW
it's 3x 220u @50v. so way below your 1000u threshold.fechter said:
izeman said:
I think it will be no problem.
A while back, I tested a solid state switch that works in a similar manner. Sort of wimpy gate drive to turn on. It worked fine at 60V into 1000uF. I tried a gigantic 10,000uF cap and it blew up. Based on what I've seen with the spot welder circuits, I think if you can turn the FETs on hard enough they will survive a massive amount of peak current. Basically you want them to be nearly fully on before the current rises. There will always be some inductance in the wiring that will delay the current.
vex_zg said:
The problem is there is no "gate driver", just a 4.7M resistor that goes to cell 3. Any resistance to ground and it will act like a voltage divider and reduce the gate voltage. What might be better is to just place a 15v zener diode from gate to source to protect the gate from stray voltages in the event the wire breaks or gets disconnected and just live with the possibility the gate could be charged when the wire breaks. The leakage in the zener diode would eventually drain the gate.
fechter said:
Typically you want to turn mosfet off as fast as possible to keep it out of linear area most of the time, but sometimes with inductive load that will create transient spike Vds which will kill it.
I don't understand this: "The problem is there is no "gate driver", just a 4.7M resistor that goes to cell 3"
In the typical BMS, the gates get powered from a tap on cell #3, which is around 12v. No voltage regulator. From the cell 3 tap, it goes through a high value resistor to the gates and there are transistors which can pull the gates to ground.
izeman
1 GW
ha! now i understand #3fechter said:
seems strange as voltage per cell can go down to 2.8v which means 8.4v at #3. a little low for full FET turn on, isn't it?
Attached is a schematic of an AAYA BMS. This is slightly different than the Bestech, but similar in circuit design.
Full schematic:
The gates need 10-15v to turn on. There is a control circuit that gets the signal from the cell circuits or the over current detection circuit and has a latch to keep it off until reset. The control circuit is powered from the cell #3 tap to avoid needing a voltage regulator. If you look at the gate circuit for Q40 and Q52 at the lower right side, you can see how it looks. Q40 is the bank of discharge FETs and Q52 is the bank of charge FETs. R1 and R2 come off the #3 cell tap to supply power to the control circuit. When everything is on, there is practically no current passing through R1 or R2 as the gates don't really draw any current. Q2 or Q4 can pull down the respective gates to turn off charge or discharge.
Full schematic:
The gates need 10-15v to turn on. There is a control circuit that gets the signal from the cell circuits or the over current detection circuit and has a latch to keep it off until reset. The control circuit is powered from the cell #3 tap to avoid needing a voltage regulator. If you look at the gate circuit for Q40 and Q52 at the lower right side, you can see how it looks. Q40 is the bank of discharge FETs and Q52 is the bank of charge FETs. R1 and R2 come off the #3 cell tap to supply power to the control circuit. When everything is on, there is practically no current passing through R1 or R2 as the gates don't really draw any current. Q2 or Q4 can pull down the respective gates to turn off charge or discharge.
methods
1 GW
vex_zg said:velias said:
I guess most of ppl on this forum are more inclined towards learning and DIY vs. turnkey solutions. Repairing is good for environment
Vex is right.
Look at the mini-education that came out of this thread.
velias seems to be hot on the idea right now of popping up and making smart remarks.
Thats ok from time to time but it gets old...
Yes, velias is right... that if the goal is to get the job done then we are deep in the red here.
If the goal is to pass the time, learn stuff, help others learn stuff... well then...
We need to define success and then develop metrics for measuring our progress toward that success.
-methods
methods
1 GW
As for how to turn the BMS off... I suggest this:
1) Find the big red wire
2) Find the big black wire
3) Find a switch... any switch
4) Set switch to OPEN position (test this with Fluke DMM... open circuit)
5) Attach big red wire to switch
6) Attach big black wire to other leg of switch
Now...
When you are ready to test BMS....
7) Flip switch
-methods
1) Find the big red wire
2) Find the big black wire
3) Find a switch... any switch
4) Set switch to OPEN position (test this with Fluke DMM... open circuit)
5) Attach big red wire to switch
6) Attach big black wire to other leg of switch
Now...
When you are ready to test BMS....
7) Flip switch
-methods
izeman
1 GW
well methods this was a REALLY helpful one
Similar threads
