Even Newer 4 to 24-cell Battery Management System (BMS)

[fechter]

I hope Fechter could help me with my BMS 4 cell v.2.2 board from TP-packs I have built.
There is one big Problem with the shutdown process.
All leds lit and I am reading from ground to pin 7 of the opto's 12V and on pin1/ pin2 ~1V. But the opts don't turn the leds off nor the red led goes to green!
Shorting pin 7 and pin 8 of one of the optos the main led turns green. It seems that the optocooplers don't work as they should!
I also checked the auto shutoff feature, when I bridged the control ground to the all shunts' line - it works! :?

Sounds like you have it mostly working.

ejonesss is correct that you may need to let it 'cook' for a while to get all the cells balanced the first time. My pack took several hours on the first cycle because one cell was much lower than the rest. If you get all the cell LEDs lit and let it go for an hour or so, that should be enough time to get them all balanced depending on the cells.

You can get the LEDs lit, then measure the voltage across each cell to see if there is one that is a bit low that's keeping it from turning off. Another way to check this is to measure the voltage across each power resistor. Should be around 3v on each one when the LEDs are lit. Look for one that is lower than the rest.

Another possibility is your charging source is just a little too low on voltage.
How many cells are you using and what is the charger voltage? Do you have a way to increase the charger voltage? The charger voltage should be several volts higher than the pack voltage at full charge. The circuit will regulate the voltage.

Another possibility is an accidental solder bridge on one of the optos or the diodes. Hold the board up to a light and look at the solder side carefully for bridges. Also double check to see that all the optos and diodes are installed in the correct direction.

 

[GGoodrum]

...and the increase in the diameter of the shunt transistor holes so that the larger parts methods is using in his version can be used. ...

How many volts over HVC do you generally have to set the power supply with the 500ma shunts to get consistent shut-off?

-methods

Usually, no more than a few hundred millivolts over the sum of the shunt cutoff voltages, I think. I ususally use about 1/2V

 

[GGoodrum]

I still didn't got anymore 28V dewalt for you, but i keep an eyes on them for you like you asked for! :wink:

Doc

No rush, Doc. Just keep looking.

Thanks -- Gary

 

[GGoodrum]

i was thinking somewhere there was a mod where you changed a couple resistors to the existing bms to make it handle a charge of 4.1 to 4.2 volt range and an lvc of 2.5 v.

Yes, that is correct. If you replace the 2.1V version of the TC54 chips with the 3.0V version, which you can get from Digikey, all you then have to do to make this work with your eMoli cells is change the two voltage divider resistors in each channel that feeds the LM431 detector. I don't remember what values methods used, but these are easy enough to calculate.

-- Gary

 

[ejonesss]

you mean r 101 and r102?

do i also have to replace r 103?

also i think we need a sticky or someplace to keep the basic instructions.


not to sound redundant here but i was thinking instead of re designing a new board for the cheaper opto on the pc mount version you could just design a small adaptor board with wires to adapt the existing board to the new opto (yes it is going to dangle off the board like the mod chips on the game consoles do but it is cheaper that redesigning the board).

i am suggesting above because it can potentially save production time too and you can re stock the boards.

also would sending raw dc from a simple transformer,rectifier and capacitor setup like used in some cheap and older automotive battery chargers work meaning does the bms board do all the cccv work?

i was thinking somewhere there was a mod where you changed a couple resistors to the existing bms to make it handle a charge of 4.1 to 4.2 volt range and an lvc of 2.5 v.

Yes, that is correct. If you replace the 2.1V version of the TC54 chips with the 3.0V version, which you can get from Digikey, all you then have to do to make this work with your eMoli cells is change the two voltage divider resistors in each channel that feeds the LM431 detector. I don't remember what values methods used, but these are easy enough to calculate.

-- Gary

 

[ejonesss]

i did a search and found the methods's post

http://endless-sphere.com/forums/viewtopic.php?f=14&t=5416&p=103609&hilit=r102#p103609

it turns out you really need to change the 75k r101 to 111k witch is a bit low (better low than too high (undercharge the cell not explode or light a fire or destroy the cell)

107k a bit high may or may not be safe for the cell or you and property (i say that because i am not sure how much over charge you can take before thermal runaway happens with normal lithium ion cells).

maybe someone who knows more about charging 4.x volt lithium ion cells would know how much tolerance there is for safely charging.

also is it mostly the first generation of li ion cells that are a problem?

 

[methods]

I have later posts with different numbers (Due to availability of resistors).
Scan backwards from this post to the post where I stated that I was ordering parts.

You do understand that you are not trying to make 4.2V with that voltage divider right?
You are trying to create ~2.6V with that divider to turn on the LM431

There are no issues of anything blowing up here. . .

My circuit takes the cells right to 4.19V

-methods

 

[ejonesss]

gary i mentioned before that you can some changes without redesigning the board.

i think it is better to redesign the board because then you can add a perforated line so the regulator section on segment 1 can be snapped off or design a separate regulator section so it can be used with the unused segments.

 

[ejonesss]

let me see if i got this right.

the divider putting out 2.6v it causes the lm431 come on at a higher voltage and not stop the charging until the cell reaches 4.19v?

I have later posts with different numbers (Due to availability of resistors).
Scan backwards from this post to the post where I stated that I was ordering parts.

You do understand that you are not trying to make 4.2V with that voltage divider right?
You are trying to create ~2.6V with that divider to turn on the LM431

There are no issues of anything blowing up here. . .

My circuit takes the cells right to 4.19V

-methods

 

[methods]

Go back and read the thread.
If you do not clearly understand how the circuit works then do not attempt to modify the circuit.
I have already posted everything you would need to know...

-methods

 

[ejonesss]

check your pm box

Go back and read the thread.
If you do not clearly understand how the circuit works then do not attempt to modify the circuit.
I have already posted everything you would need to know...

-methods

 

[GGoodrum]

I agree with methods, you should not be attempting to change this without understanding the way it is suppose to work.

The changes in resistors will simply raise the input voltage required to have the LM431 start conducting. As he points out, this is around 2.6V, period.

I would also use exactly the values he did, as 4.19V is perfect. There is very little headroom above this for LiCo-based cells, before they can, and will, explode. LiMn won't explode, but you can damage them with overcharging.

-- Gary

 

[ejonesss]

the second line is what i thought but wanted to make sure and sorry if i worded it slightly different so the do not agree.

maybe now on to make the question better i will start using more "correct me if i am wrongs"

i took a measurement on the bms and i do not need a cell in place to verify the correct voltage thereby making it easier and safer to verify before connecting the cells.

I agree with methods, you should not be attempting to change this without understanding the way it is suppose to work.

The changes in resistors will simply raise the input voltage required to have the LM431 start conducting. As he points out, this is around 2.6V, period.

I would also use exactly the values he did, as 4.19V is perfect. There is very little headroom above this for LiCo-based cells, before they can, and will, explode. LiMn won't explode, but you can damage them with overcharging.

-- Gary

 

[manfred59]

richard wrote:

You can get the LEDs lit, then measure the voltage across each cell to see if there is one that is a bit low that's keeping it from turning off.

I built two boards for 4 cells BMS and am using an adjustable power supply at 15,5V and no battery connected so far.
One of the boards (I think) works like this: Connected to the power supply at 15,5V first the orange leds light up a short moment an then the main led goes green the orange leds off! The second board does not the same
ejonesss wrote:

minute 20 for the leds to come on fully and another 40 seconds or so to turn off. (may vary board to board (mine is 16 cell 48 volt))

Would this happen also if no battery is connected? I cannot believe it!

 

[ejonesss]

the minute 20 is for no cells connected from a cold board.

once the board is warmed up it will light up sooner.

since cells are expensive and no one wants to ruin cells it is always good to do dry tests (especially when modifying anything) with no cells for 2 reasons.

1. allows testing in a shorter time.

2. prevents risking damaging the cells (it is better to find out that the charging voltage is too high for example with no cells connected than to find out when a cell is damaged).

richard wrote:

You can get the LEDs lit, then measure the voltage across each cell to see if there is one that is a bit low that's keeping it from turning off.

I built two boards for 4 cells BMS and am using an adjustable power supply at 15,5V and no battery connected so far.
One of the boards (I think) works like this: Connected to the power supply at 15,5V first the orange leds light up a short moment an then the main led goes green the orange leds off! The second board does not the same
ejonesss wrote:

minute 20 for the leds to come on fully and another 40 seconds or so to turn off. (may vary board to board (mine is 16 cell 48 volt))

Would this happen also if no battery is connected? I cannot believe it!

 

[fechter]

...
One of the boards (I think) works like this: Connected to the power supply at 15,5V first the orange leds light up a short moment an then the main led goes green the orange leds off! The second board does not the same

The first board is behaving properly. The second board has a problem.

With all the orange LEDs lit, you can measure the voltage across the input (emitter) side of each opto (sorry, I forget the pin numbers). It should read about 1.7v. If one reads lower, then it may not be turning on. Then you can measure the voltage across each of the opto outputs. When on, they should be something less than .5v or so. Look for one that is not the same as the others.
The problem has to be with one of the opto circuits.

Someone else had a similar problem that turned out to be a tiny solder short to one of the diode connections. Use a magnifier to examine the board if your eyes are like mine.

 

[manfred59]

fechter wrote:
With all the orange LEDs lit, you can measure the voltage across the input (emitter) side of each opto (sorry, I forget the pin numbers). It should read about 1.7v. If one reads lower, then it may not be turning on.

I desoldered the optcoopler #3 because the voltage between pin 1 and pin 2 has been about 0,5V and the other optos read about 1V. There should be no solder bridge I did check this - could it be a a defective component like U102 or Q101? How could I go on?

 

[fechter]

I desoldered the optcoopler #3 because the voltage between pin 1 and pin 2 has been about 0,5V and the other optos read about 1V. There should be no solder bridge I did check this - could it be a a defective component like U102 or Q101? How could I go on?

OK, you're making progress. At least you know which channel is the problem.

Next try measuring the voltage across the big resistors when the LEDs are lit.
Also try measuring the voltage across the channels (cell voltage).

It's possible the opto was bad, and I assume you have replaced it.

It doesn't sound like Q101 is bad, or the LED would not light up. If U102 was bad or if there was a problem with the divider resistors, the shunt voltage could be higher than it's supposed to be.
You can use an ohmmeter to measure the values of R101 and R102 (power off) to make sure they look right.

 

[manfred59]

thanks for helping @ fechter!!
After changing KSA 931 and LM431 same fault. The opto # 3 is out( desoldered)
Power source is 15,5V/0,48A all four leds lit. On pin1/2 I am reading on opto #1: 1,01V opto #2: 0,93V opto #3: 0,62V opto #4: 1,48V

fechter wrote:
Next try measuring the voltage across the big resistors when the LEDs are lit.
Also try measuring the voltage across the channels (cell voltage).

first across the big resistors: # 1:3,71V, #2: 3,77V , #3: 3,73V , #4: 3,60V
cell #1: 3,84V cell #2: 3,90V cell #3: 3,84V and cell #4: 3,80V

You can use an ohmmeter to measure the values of R101 and R102 (power off) to make sure they look right.

R101 and R102 are correct!

 

[ejonesss]

when semi conductors are defective can they be shorted or are they usually open?

I desoldered the optcoopler #3 because the voltage between pin 1 and pin 2 has been about 0,5V and the other optos read about 1V. There should be no solder bridge I did check this - could it be a a defective component like U102 or Q101? How could I go on?

OK, you're making progress. At least you know which channel is the problem.

Next try measuring the voltage across the big resistors when the LEDs are lit.
Also try measuring the voltage across the channels (cell voltage).

It's possible the opto was bad, and I assume you have replaced it.

It doesn't sound like Q101 is bad, or the LED would not light up. If U102 was bad or if there was a problem with the divider resistors, the shunt voltage could be higher than it's supposed to be.
You can use an ohmmeter to measure the values of R101 and R102 (power off) to make sure they look right.

 

[flashedarling]

I think would sure help if you could think up a name for your BMS system. That way when asking for advice about it on other threads we could refer to it as something else than Gary's BMS. Maybe even start adding in some v1 and v2's to your iterations?

Just my 2 cents.

 

[dnmun]

i always refer to it as the open source design BMS, since gary has not done anything to try protect his work besides trademarking the pcb. so much of richard is in the design, and gary really made it happen, and i always bring up bob's idea about how to do the LVC. but no doubt it is gary and richard who made it happen.

GRB open source? grobos? or maybe just GROS? so version one is GROS1? or the first turn of the the pcb with the reversed leads could be a NOGOGROS? maybe include the number of channels instead so a a full pcb is a 24GROS. a 48 volt with 16 channels populated would be a 16GROS? or refer to the voltage, and call it a 48VGROS?

i wonder how long until we see copies appear from china, but i doubt if many battery manufacturers will be willing to use it since there is no protection for the pack built into the BMS itself and the LVC is lower than they would want to allow because of warranty concerns.

i really think the idea of such a simple and clean design with built in charging diagnostics in the form of the LEDs and fact that there are no output fets to drag down the power and the LVC control over the controller through the brake line will be ubiquitous eventually, and i expect the next round will be to reduce costs and assembly hassle by eliminating the LEDs and maybe something else, but this really seems so neat, and open source is really where it's at, all across the technology spectrum, imo. something these guys should be proud of.

 

[fechter]

thanks for helping @ fechter!!
After changing KSA 931 and LM431 same fault. The opto # 3 is out( desoldered)
Power source is 15,5V/0,48A all four leds lit. On pin1/2 I am reading on opto #1: 1,01V opto #2: 0,93V opto #3: 0,62V opto #4: 1,48V

fechter wrote:
Next try measuring the voltage across the big resistors when the LEDs are lit.
Also try measuring the voltage across the channels (cell voltage).

first across the big resistors: # 1:3,71V, #2: 3,77V , #3: 3,73V , #4: 3,60V
cell #1: 3,84V cell #2: 3,90V cell #3: 3,84V and cell #4: 3,80V

You can use an ohmmeter to measure the values of R101 and R102 (power off) to make sure they look right.

R101 and R102 are correct!

Hmmm.... sounds almost like the C101 on cell#3 is bad (or reversed). Try removing it and test without it. Re-install the opto. I won't trip off without it.

ejonesss, most semiconductors fail shorted, but they can also fail open. It sort of depends on the device. Sometimes they short first, then fuse open due to high current.

 

[fechter]

i always refer to it as the open source design BMS, since gary has not done anything to try protect his work besides trademarking the pcb. so much of richard is in the design, and gary really made it happen, and i always bring up bob's idea about how to do the LVC. but no doubt it is gary and richard who made it happen.

Yes, I owe credit to a number of people, especially Bob Mcree, Silicium and Randomly who contributed significant portions of the design. Gary did all the board layout work, which is an art in itself.

I agree a catchy name would be good.

 

[manfred59]

fechter wrote:
Hmmm.... sounds almost like the C101 on cell#3 is bad (or reversed). Try removing it and test without it. Re-install the opto. I won't trip off without it.

Now I have reinstalled the optpcoopler and took out the C101 , but nothing changed!
I did some measurements on my board - perhaps it helps to find the fault.
Please have a look at the picture what I have measured. I seems to me curious that if I have shorted the power transistor # 1 the current increases to 0,57A!