How BMS Works (Gotway pack teardown)

[RagingGrandpa]

For future reference, below are my findings from teardown of a Gotway BMS.

DISCLAIMER: This BMS was a Gotway design from 2015, originally sold in the MSuper V2. Battery Management Systems from other models and manufacturers vary widely.

 

What can it do?

1. Disconnect the charger ("Charge stop")
   ... if any cell exceeds 4.25V ("overvoltage protection")
   ... or if any cell is between 2.0 and 2.80V ("undervoltage protection")
   
   Charge Stop can be linked to additional packs, so that a cell problem in one pack can trigger all parallel packs to Charge Stop.
   Overvoltage protection is a 'latching' behavior-- it does not reset until the charge input voltage is removed.
    
2. Top-balancing
   Balance resistors will be connected to any cell that exceeds 4.20V.
   (Normally this occurs at the very end of a charging event, when the pack is nearly full.)
    

What are its limitations?

• No overcurrent (short-circuit) protection.*
  * charging port short-circuit protection was added in newer Gotway BMS, and I confirmed it works in my 84V 2019-build MSX, for example.
  In this older BMS, overcurrent in the Charger input (e.g.: short at the charge port) will pass through the Charge Stop FET (rated 200 A pulsed) until some part of the circuit is burned open. Typically the 'weak spot' seems to be the PCB traces between the charging leads and the Charge Stop FET.
  
  Overcurrent in the pack output (in both older designs and today's products) is not borne by the BMS PCB at all- the high current will flow through the leads and cell straps. Bolted shorts will flow huge amounts of current, until something melts or the cells themselves become destroyed.
   
• No thermal protection.
  No thermostatic fuses nor temperature sensors are present.
   
• Vulnerable to corrosion.
  Corrosion is likely to cause open-circuits in the connections between the pack and the BMS. If this occurs, Balancing and Charge Stop are defeated. The pack may still charge and discharge normally, but cell imbalance will become progressively worse. Unmonitored, over-voltage cells may cause heating and fires during charging. Unmonitored, under-voltage cells may cause heating and fires during use.
   

How does it work?

• For each cell, a HY2213-BB3A ASIC is used to connect the balance resistance.
• For each cell, a HY2113-KB5B ASIC is used to detect over- and under-voltage, and drives the Charge Stop trigger signal.
  (This device can also monitor current, but that function is not used in the Gotway BMS.)
• Charge Stop is realized by a FQP50N06 FET, which disconnects the negative side of the charger from the pack.
  (This device has no built-in overload protection.)
   

How are the packs connected?

• (The packs inspected were from the Gotway MSuper V2 EUC, an 850wh 5p configuration using 3 packs: 16s1p, 16s2p, 16s2p.)
   
• According to my diagram below, only one pack is connected to the charger, and therefore Charge Stop is performed at this 'master' pack only. The two 'slave' packs can trigger Charge Stop using the trigger link connection. If any pack drives a positive voltage across these "HV" and "LV" Trigger Link terminals, all packs will disconnect their Charger input. The trigger link wires are redundant- two "HV" and two "LV" leads are used between each pack, but they are electrically the same circuit. This provides fault-tolerance to open-circuit conditions in the wiring. If the trigger link was broken or not connected, Charge Stop would not be possible, and therefore overvoltage and undervoltage protection will be defeated.
   

Discussion

• Fast-charging:
  Charging Input wires could be added to the two 'Slave' packs as well, in order to facilitate fast-charging with use of multiple charging connectors. 1C charging for this 5p configuration is about 13 amps, which exceeds the safe ratings of the single GX16 charging connector, but could be achieved easily by 3 charging connectors in parallel. Aggressive fast-charging at 1C or above is not recommended for these packs, because of the lack of thermal monitoring of cells.
   
• Under-voltage:
  In my tests, I was unable to trigger Charge Stop for the cell undervoltage condition. I had some cell groups disconnected from the BMS, and manually applied various voltages to the BMS cell monitor leads using a potentiometer and 9V battery. I did observe the "overdischarge" output of the HY2113 being driven when the voltage falls below 2.8V, but it did not propagate to the Charge Stop FET. I suspect this irregularity was due to the condition of the damaged and corroded pack and BMS that I was testing; I chose not to abuse a healthy pack to create a cell-undervoltage condition.
   
• Backstory:
  A friend had this old MSV2 that would run but would not charge. He reported a previous instance of a short-circuit and arcing when probing the charging port. I opened the packs and discovered the two 'slave' packs were healthy, but the 'master' pack (with charging input) was badly corroded, with entire balance leads rotted away. Cells were badly imbalanced (some 0V). And the copper trace between the charger input and the Charge Stop FET had burned and blown open.
  Since the other two 'slave' packs were in pristine condition and very well balanced, I was able to salvage the wheel by simply discarding the damaged pack, and moving the charging connector leads to a remaining pack. After this repair, I was able to simulate Charge Stop events by manually applying a voltage between the Trigger Link HV&LV using 9V batteries, and observed charging current being interrupted accordingly.

Cheers

Edited September 2, 2020 by RagingGrandpa

[svenomous]

37 minutes ago, RagingGrandpa said:

Overcurrent in the pack output is not borne by the BMS PCB at all- the high current will flow through the leads and cell straps. Bolted shorts will flow huge amounts of current, until the cells themselves become destroyed.

On my KS wheel I noted that there are fuses (mini ATX) on the mainboard to provide overcurrent protection on the battery outputs (one fuse per pack).  This implies that the packs themselves do not provide such protection. I can attest that the fuses saved one of my packs from self-destruction when I accidentally bridged connector pins while trying to measure voltage. There was a mighty arc, my voltmeter lead was melted/fused, and one of the fuses blew.  The pack was fine, and so was the wheel after I replaced the blown fuse.  Luckily I was measuring opposite-side-pack voltage through the connector coming from the mainboard (before connecting that to the other pack), instead of pack voltage directly from the pack's own connector.

[Chriull]

21 minutes ago, svenomous said:

On my KS wheel I noted that there are fuses (mini ATX) on the mainboard to provide overcurrent protection on the battery outputs (one fuse per pack).  This implies that the packs themselves do not provide such protection.

KS BMS have to my knowledge mosfets that can cut off the output. Normally there are about 1 (to 2?) Mosfets for charge input and 2-?4? Mosfets for output protection.

Output protection is only used for overcurrent protection (cell undervoltage protection was cancelled "long" ago(. Seems this is additional to the fuses. BMS overcurrent protection could be a fast, high current short circuit protection, the fuses some "slow", lower current mainboard , mosfet, wire protection

[alcatraz]

Very interesting RagingGrandpa. Thanks for taking the time.

Is the master pack different than the other packs or is it the only pack with a connected charge input? 

Do the other packs also have a charge input so that they can act as the master pack? I don't remember seeing unplugged connectors on my two tesla packs.

So all the current of the fast charger people use go through that single bms. I wonder if it can get fried by using high amperage chargers like 5 or 10A.

A point I was trying to make in another thread is that you can't combine gotway packs with generic BMS packs and expect any protection. There's zero protection for that additional pack as it is charged through the output terminals even if it does decide it wants to cut charging. (It assumes that bms design is the same on current gotway wheels.)

Edited May 19, 2020 by alcatraz

[RagingGrandpa]

11 hours ago, alcatraz said:

Is the master pack different than the other packs?

No. Every pack has the charging input hardware; the only difference is the 'master' pack has the charging input wires. (The Charge Stop FET in each 'slave' pack is idle and not being used.)

11 hours ago, alcatraz said:

So all the current of the fast charger people use go through that single bms. I wonder if it can get fried by using high amperage chargers like 5 or 10A.

The charging input wires are 20AWG. It includes an inline connector with small contacts- I suspect the contacts are the lowest-rated component in the chain. The Charge Stop FET is huge (50A nominal).

Practically speaking, I would attempt 5A charging without modification, but beyond 5A it would make sense to connect an additional charging input of a 'slave' pack.

11 hours ago, alcatraz said:

A point I was trying to make in another thread is that you can't combine gotway packs with generic BMS packs and expect any protection.

Some generic BMS have an interruptible high-current output. This has pros and cons, but can be compatible with Charge Stop as long as the charging input is only going to a Gotway pack.

Edited May 19, 2020 by RagingGrandpa

[alcatraz]

Theoretically, how would that interruptible output work in combination with a gotway pack? Generic bms set up for euc use won't really use the onboard output terminal out of fear of a cutout. So there's no discharge protection. Charging protection is what's left and I don't see how there could be any unless it's the only pack with the charge port connected to it. 

[RagingGrandpa]

1 hour ago, alcatraz said:

how would that interruptible output work in combination with a gotway pack?

The output from the auxiliary pack may turn off. If it does, you're left with the Gotway packs alone (which should carry you onward nicely). Resetting this type of "trip" could be quite challenging though, depending on the reset logic used by the auxiliary pack's BMS...

[alcatraz]

That assumes the output terminals on the bms are used which they aren't when building the pack for euc use.

And even if they were used, what's the chance of the bms cutting the output as an overcharge protection? That's for discharge or overcurrent protection I think.

And like you pointed out. How do you open that connection without getting a split second one hundred amps rushing? Hmm... 

[hyperair]

On 5/20/2020 at 12:19 AM, RagingGrandpa said:

The output from the auxiliary pack may turn off. If it does, you're left with the Gotway packs alone (which should carry you onward nicely). Resetting this type of "trip" could be quite challenging though, depending on the reset logic used by the auxiliary pack's BMS...

Wouldn't disconnecting it from the main pack and charging it independently usually reset the auxiliary BMS's charge and discharge stop circuits? 

[RagingGrandpa]

Also, while my teardown was a "old 84V BMS"... Gotway's design hasn't changed much- 2019 production 100V Nik+ BMS pictured below.

Finally a 2-layer PCB. And charging input reverse polarity / short circuit protection at C+ (diodes not populated in this example).

[Chriull]

1 hour ago, RagingGrandpa said:

And charging input reverse polarity / short circuit protection at C+ (diodes not populated in this example)

There seems to be still no short circuit protection - or charge and discharge wires are to be connected at B+ and C-?

The 2 pairs of mosfets should be n and p channel so they could perform as reverse polarity protection. And with the R005 as current shunt a short circuit protection?

[alcatraz]

14 hours ago, RagingGrandpa said:

Also, while my teardown was a "old 84V BMS"... Gotway's design hasn't changed much- 2019 production 100V Nik+ BMS pictured below.

Finally a 2-layer PCB. And charging input reverse polarity / short circuit protection at C+ (diodes not populated in this example).

Is this the side of the bms board that's against the shrinkwrap? I'm curious to cut my tesla packs open in a way so I can expose the measue points. I'm hoping that the group voltages are all measurable at the top of that array.

Is the bms on the pack side facing inward or outward?

[RagingGrandpa]

8 hours ago, alcatraz said:

Is this the side of the bms board that's against the shrinkwrap?

Yes, the image above is the 'exterior' surface of the pack.

It is normally straightforward to identify the layout of the board from the bulges in the shrink of the soldered connections and balance resistor array. Feel free to send a photo of your pack if you'd like collaboration.

I presume you intend to needle-probe the cell pads shown above? Sounds reasonable if done with care...

  

21 hours ago, Chriull said:

There seems to be still no short circuit protection

* protection from short circuit at the charging port (via diodes between C+ and B+).

No protection from overcurrent / short circuit of the pack output.

 

21 hours ago, Chriull said:

R005 as current shunt

I'm perplexed by that one too. 5-milliohm current sense resistor, perhaps...

Edited June 3, 2020 by RagingGrandpa

[Cody]

about that, i found on other scooter baterry packs with samsung cell leaking (brand new pack ...) and cell are still selling on the internet

[WI_Hedgehog]

What is the value (in ohms) of the balance resistors?

Edited June 8, 2020 by WI_Hedgehog

[Chriull]

5 minutes ago, WI_Hedgehog said:

What is the value of the balance resistors?

Here 

are two 150 Ohm resistors in parallel. 

[WI_Hedgehog]

Is the package rated at 1/16 watt? (What is the wattage?)

[Chriull]

56 minutes ago, WI_Hedgehog said:

Is the package rated at 1/16 watt? (What is the wattage?)

The mosfet above the resistors labeled 7002 is (most probably) in a sot-23 package measuring 3x1.4 mm. (2n7002 from the datasheet at digikey)

The resistors are about the same length and a bit wider. So i'd guess it's a 1206 package according to your link. There they mention 1/8 W typical. Should be just enough for 4.2V.

http://www.resistorguide.com/resistor-sizes-and-packages/ lists a bit bigger sizes for 1206 and 1/4 W...

Edit: this numbers of course don't have to mean too much in a plastic wrap if the copper tracks were not designed to distribute the heat...

Edited June 8, 2020 by Chriull

[alcatraz]

Does anyone know or would care to guess if the cell voltages are measureable at the top or at the bottom of these balancing circuits? I'd like to make as small a cut as possible. Thank you. 

Edited June 8, 2020 by alcatraz

[Chriull]

6 hours ago, alcatraz said:

Does anyone know or would care to guess if the cell voltages are measureable at the top or at the bottom of these balancing circuits? I'd like to make as small a cut as possible. Thank you. 

 

The cell voltages have to be on both rows of the 6 pinners (comperators).

And also from on pin of the balancing resitors (121) to one pin of the mosfets (3 legged) directly above.

As the cells are all in series and as the pcb traces seem to be the cell voltages should be measurable between the lower pins of the balancing resistors. Just for the first or last cell you have measure against another point (B+ or B-).

Seems to be B- as there is a track from B1 to the first resistor on the left...

Follow the tracks!

[Planemo]

I like the idea, but if I am going to the bother of opening up the heatshrink, it wouldnt be too much of an effort to solder on 24 small diameter leads and run them to an onboard voltmeter. Does anyone know such a meter that would do the job? Maybe one that has an onboard button to scroll through the inputs, negating the need for a big display to display all voltages at the same time?

Edit: best I can come up with is 3x 8 cell meters. Not as neat or discreet as I would like, but they do have the benefit of being able to set alarms which would be handy:

https://www.google.com/amp/s/zenid10.wordpress.com/2011/09/05/monitoring-pack-cell-voltages/amp/

Edited June 9, 2020 by Planemo

[Chriull]

35 minutes ago, Planemo said:

I like the idea, but if I am going to the bother of opening up the heatshrink, it wouldnt be too much of an effort to solder on 24 small diameter leads and run them to an onboard voltmeter. Does anyone know such a meter that would do the job? Maybe one that has an onboard button to scroll through the inputs, negating the need for a big display to display all voltages at the same time?

Edit: best I can come up with is 3x 8 cell meters. Not as neat or discreet as I would like, but they do have the benefit of being able to set alarms which would be handy:

https://www.google.com/amp/s/zenid10.wordpress.com/2011/09/05/monitoring-pack-cell-voltages/amp/

Did not find anything real better

and i could not find any specification about accuracy, just the display resolution

For the price asked one could consider switching to a BMS with bluetooth connection showing all the data on the mobile phone...

Other solution would be a (weatherproof! and safe) connector brought to the outside and just measuring with the DMM or just some cheapo 8s li ion cell tester.

[RagingGrandpa]

16 hours ago, alcatraz said:

Does anyone know or would care to guess if the cell voltages are measureable at the top or at the bottom of these balancing circuits? I'd like to make as small a cut as possible.

Yes, you can easily probe them all per below.

The spacing is tight- take care not to short them! Especially when cutting the shrink!
If soldering, tapemask one at a time... risky business with cells connected.
Soldering will be difficult- most boards have a conformal lacquer painted over this area, for corrosion resistance.

The only part that's missing is B-, which is easy to probe manually at the pack output, but challenging to find a solder spot that's small and safe... if it's this exact board, I suggest the little unused pad near 'R005'.

 

5 hours ago, Planemo said:

solder on 24 small diameter leads and run them to an onboard voltmeter

I don't see a need for inspecting per-cell voltages frequently. Maybe once a year, or whenever you have the shell open, or whenever you notice an abnormality during charging... just bringing all 25 pins to an internal connector would be awesome.

(As before, take care with insulation protection for this new pigtail. It must be well-secured inside the wheel in a way that cannot chafe or wear. Smaller wire is better- they'll melt at a lower current if they short out... but you'll likely burn the balancing traces off the BMS too, destroying it.)

Edited June 9, 2020 by RagingGrandpa

[alcatraz]

Thank you very much. I'll try at the bottom first then.

I believe the balancing works and my packs are small @425Wh each. Considering they are 3-4y old I just want to be sure and to kind of see how the bms performs. If I am to rebuild with Tesla 21700 cells that are much larger capacity I probably would attach balance leads, and in that case I'd solder to the large connection points spread all around and not the balancing circuit.

[Cody]

The better is to attach Small schottky Diodes to prevents cell reversal polarity, during deep discharge, this save a lot my pack of 14d, they are still alive, and not so balanced...
try smart BMS, with bluetooth, it's nice to see how deep we are abusing battries, i set cutoff to 2.7, but it's Hardware disabled, for safety ...