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

mimichris said:
In the note of the BMS "test instructions" it is written :

..........This oscillation has the net effect of keeping the shunts at full bypass, but not getting swamped,
which would cause the shunts to overheat. Like the LVC circuits, the shunt optos outputs are all ganged
together, so that if any one turns on, the FET will turn off, cutting power.........

On my BMS Tppacks with Ping battery 36V and 15Ah, 12 cells, I didn't notice an oscillation, when I plug the charger (charger Ping 36V and 2A), the bicolored led is orange, and orange remainder lasting all the time of the load, once the loaded cells, leds oranges ignite the then once some after others they are all alight, they die out enough quickly and the bicolored led becomes green.

Is what this working is normal on my BMS ?

The oscillation can affect a video camera. Here's the last two minutes of a charge thru the BMS. You can't see the red/green LED flash, but you can see the orange shunt LEDs reacting.

[youtube]GbA5wCA_M8A[/youtube]
 
My BMS now is working untill some weeks. I am charging a 15s6p A123 battery. Now I have this problem: When I plug the charger with no battery connected all 15 orange leds light a very short time, then the green led lights up. Everything o.k. I thought, but after connected the battery with the main + and - powerpoles ~56V only the first orange led lights and the main led lights red. The charging process is going on but the first cell is not charging! When all cells are fully charged the green led comes on, but the first orange led stays lit. I had this three days before and changed the U102 (LM431 2.5V linear regulator) and it worked again. Could something destroy this linear regulator again and again? Tomorrow I will change this U102 once more and test the BMS. Perhaps anybody had a similar problem?
 
By the first cell, do you mean the one closest to the charge control circuit?

About the only thing I can think of right off is some kind of inductive kick from the charger could send out a high voltage pulse. You could try placing a capacitor across the charger output (around 100uf).
 
I just wanted to say thanks for all the support thus far in this thread. My two ver 2.2 boards are fully assembled and LVC-tested. I'll be testing the charging side next. :)

Here is some twin-BMS eye candy for everyone.

Cheers,
--adam
 

Attachments

  • twin-bms-boards-2.jpg
    twin-bms-boards-2.jpg
    197.6 KB · Views: 1,985
@Manfred : Can be that cells are not connected respectable, for my BMS, here is how I connected cells:
Them + and - packs batteries are connected indicated how as "manual instructions" to the BMS, then:

the 1 (negative) of the first cells to the 1 of the BMS, either on the BMS:

1, 2, 3, 4...... ...6, 7, 8, 9,...... ..11, 12, 13, 14. For my 12 cells.
 
fechter said:
By the first cell, do you mean the one closest to the charge control circuit?

About the only thing I can think of right off is some kind of inductive kick from the charger could send out a high voltage pulse. You could try placing a capacitor across the charger output (around 100uf).
I think too that this could be a reason! I have changed the U102 and the U101 of channel #1 near the charge control circuit and mounted a capacitor across the charger output. I will see if this is the solution. Thanks to fechter! :D
 
Hi Folks,

I've made it past testing the LVC indicators for both of my 24 channel boards... and I've started testing charging. A 19.5v laptop power supply is all I have on hand right now so I've been testing with A123 and some Headway 18650 cells I have laying around. Everything looks good so far as far as charging - the channel LEDs come on at the 3.69-ish voltage that they are supposed to and the main LED starts morphing from red to an orange-ish-green. :D

I have two questions at this point. The first relates to charge cutoff latching and the second relates to the main LED being on a faint red when only cells are connected.

With regard to the cutoff latching I have 2 fully populated version 2.2 24-channel boards where I am testing charging on only the first 5 channels. I can short the opto outputs as shown earlier in this thread and can watch the main LED go green and feel the resistors start to cool down a bit. The problem is that when I remove the short on the opto outputs the BMS flips back to charging mode immediately with the glowing channel LEDs and an orange master LED. I am seeing the above identical behavior on both of my boards. Is it possible to get the SCR latch to lock when testing only 5 of 24 channels? As I understand it a "correctly" functioning should latch to a green main LED and stay that way until the connection charger, or really the small charger - lead, is disconnected and re-connected. In effect this "resets" the charger logic. If I have to have all 24 channels working for end-of-charge-cycle latching to work correctly then I'm happy - I just want to be sure.

With regard to the main LED being on slightly: When only the batteries are connected (on channels 1 through 4) and no connections made to the charger or the "small charger -" lead (that turns on the charging FET logic) I am seeing a faint red glow in the main LED. Again, this happens on both of my boards so whatever was done I've managed to reproduce with startling precision! This was mentioned in Patriot's build thread and now that I'm the second person to see this I thought it might warrant a comment in the main thread.

Like Patriot, I took rubbing alcohol and a toothbrush and rigorously scrubbed the solder side of the first channel segment of the BMS board. I also took a small flat-head screw driver and carefully etched the resin away from the traces that lead to Q2 and U1. After a couple of passes I feel like I have a significant amount of the cleaned away from the solder side of the board. When I connect the first 4 cells (with a 0.62", 5-pin polarized molex connector) the main LED starts to glow a faint red. I can disconnect the cells from my BMS board A and connect them to my BMS board B and get the same effect. I've scrubbed and scraped both boards.

I took some volt measurements on both boards and they were identical to +/- 0.01 volts on each board:

14.04v from "Pack -" to "Pack + after cell 4".
13.76v from "Charger -" to "Pack + after cell 4".
4.31v from "Charger -" to "Small Charger - Lead"
9.30v from "Small Charger - Lead" to "Pack + after cell 4".

Does anyone have any suggestions on what I might try to isolate the cause of LED staying mildly red with no charger connections? If I comprehended Fechter's notes in Patriot's thread it has to do with the sensitivity of Q2 and some voltage being picked up from somewhere causing the transistor to switch and turn on the charge-control logic (and thus the red LED signal). Is there some procedure I can trace with a multimeter to maybe isolate where the voltage is coming from?

Thanks,
--Adam
 
If you have all the channels populated, you won't be able to get the SCR to latch. For that to happen, all the shunt circuits have to be fully on. With all of them on, the ALL SHUNTS ACTIVE signal can finally be pulled low. If even one shunt is not on, the 20k pullup resistor will keep the ALL SHUNTS ACTIVE from going low, its active state.

You can test the proper SCR operation by temporarily geounding the ALL SHUNTS ACTIVE line. The SCR should trip, and the LED should be solid green, and stay that way until you reset.

-- Gary
 
Thanks Gary. I'll try specifically grounding that line to verify the latching behavior. And hopefully after a night's sleep I'll be able to figure out why I'm the faint red LED. :wink:

Cheers,
--Adam
 
It might be something about the transistor that switches the power for the control circuit.
What type of transistor are you using for that (Q3 I think)?

Try placing a 10k or 4.7k resistor across D2 and see if the glow goes away. If so, make the resistor permanent.
 
fechter said:
It might be something about the transistor that switches the power for the control circuit.
What type of transistor are you using for that (Q3 I think)?

Try placing a 10k or 4.7k resistor across D2 and see if the glow goes away. If so, make the resistor permanent.

On the version 2.2 PCB it's actually Q2 that switches the power for the circuit. Well - in the circuit diagram I have it's labeled Q3, but on the etched v2.2 PCB it's says Q2. I am using a KSP44TF for that transistor (mouser p/n: 512-KSP44TF). Placing a resistor (4.7k or 10k) across D2 actually makes the LED brighter.

Let me see if I understand what is going on here. In PNP transistors a small current leaving the base is amplified in the collector output. In this circuit the base is trace-connected to the anode side of D2 and R10 (10k) which traces to the "small charger -" lead, the emitter is connected to cell 4's positive to supply voltage and the collector drives the "In" pin of the LM78L12 voltage regulator.

So.. a small amount of current must be escaping the via R10, D2, or the traces around them. A small amount but enough to trigger the KSP44 to send enough current through the LM78L12 in order for the charger circuitry LED to come on. :eek: I've got a couple of other PNP transistors laying around - a couple of NTE 292 (General–Purpose Medium–Power). Would it be worth my time to try swapping in one of my NTE 292s to see if the problem goes away? Is putting in a "less sensitive" PNP transistor a good approach?

[edit: I think I might be an idiot. I just checked and I think I should've been using a KSP94 instead of a KSP44 in the Q2 position. I have a KSP44s soldered in place, which are NPN according to their data sheet. I'll swap in a KSP94, which is PNP, and see if this fixes it. Will post a second edit after testing.]

[edit 2: My idiocy is confirmed. I replaced the KSP44 (wrong part) with a KSP94 (right part). Everything works correctly now and I only get a red charger LED when the small charger - lead is connected. Its amazing how when you put the right part on the board it starts working correctly! :oops: Lesson learned: double check your part numbers. And if you build a GG/RF BMS board, and your charger active light stays on solid red with only the pack connected then you probably have an issue with the Q2/Q3 PNP transistor.]

Cheers,
--Adam E. Hampton
 
Hi Gary and all, I am going to be using the opto side of my BMS for another project of sorts... I am wondering what is the range of measurable voltage from the output side of the opto chip and how linear is it? So would it be able to report 2 volts to 4.5 volts and what kind of curve is there if any from the in to the out?
Jeff
 
jeffkay said:
Hi Gary and all, I am going to be using the opto side of my BMS for another project of sorts... I am wondering what is the range of measurable voltage from the output side of the opto chip and how linear is it? So would it be able to report 2 volts to 4.5 volts and what kind of curve is there if any from the in to the out?
Jeff

That's going to be tough. The LVC optos are going to be either on or off, not linear. The charge control optos are essentially reading the voltage across the shunt resistor, not the cell itself. Add to that the voltage drop across the LED, which is somewhat temperature dependent, and you will a hard time measuring.

If you used an optocoupler driven through a resistor directly from the cell, then you might be able to get a voltage measurement off it. See the datasheet for the ILD2. I guess this is possible in theory. I know the emitter is like an LED and does not begin to light up until it has around 1v, so I don't think the response is linear with input voltage. It will be more linear with current. The response will also be somewhat temperature dependent.
 
Yes, I meant in a different configuration. I will look upon the data sheet. Wonder what temp % tolerance that is. Might be in the data sheet as well. Thanks
 
i feel just a bit like old rip van winkle, waking up after his long nap to find the world changed in so many ways. i have not kept up with all the developments of the bms, but i am very pleased to see how my original basic ideas were expanded upon and turned into a workable system by gary and richard when health issues caused me to leave the project in their capable hands.

the latest question about using the optos to measure voltage is one i think i can help to answer. i have tried to do this in another application and i don't think it is possible. the optos i chose for the original bms design were selected on the basis of their high current transfer ratio, so that a small current through the input led could switch the output stage, creating the least possible drain on the cell. These optos are not particularly linear, by design, though the base of the output transistor is available for biasing to move the trigger point somewhat. This will let you set the trigger point more accurately but will do little for linearity. There are other optos that produce a much more linear output, but at the expense of higher drive current. In order to get the led in the opto into the linear range it usually takes at least 10 ma. and even with this current 5% linearity/resolution is about the best that could be expected, so you might get 20 steps from 2-4.5v. This would be at the expense of 10-20 ma.drain on the battery. If you try to operate at lower currents thermal errors will start to dominate and you will likely see even lower resolution. Measuring the battery voltage using the opto is just not the way to go. add a PIC or other micro.would be my best advice.
 
ok because I cant handle reading through 77 pages of what would be a very long read...

Is there anyone that makes a 50A BMS for 16 cells? or do I have to get one of these boards and make my own
 
pnet, mr li ping sells a 48V BMS, also Headway, and ecitypower all have cheap BMSes.

actually some of the better parts of this thread are in the previous ones, but maybe half of this thread is worthwhile. BOL, dm

great to see bob back here, the old bear is outa hibernation. looking forward to more meaningful discussions of electrical problems.
 
yeah I have 2 ping BMS but I need something that can put out more as they are both limited to 30A. its a shite BMS on a decent LIPO pack
 
justin at ebikes.ca has just come out with an improved switched capacitor bms for the lifepo4 packs he sells. he might be willing to sell the bms separately, i have not looked into it yet.

as for a 50A bms, this depends on whether you want a system that can equalize charge on a 50 Ah system or just a bms that has a switching fet capable of 50A. If the latter i think you just need to beef up the fets. If you have a higher capacity system you might need more balancing current, depending on the balance state of your cells.

i have been getting steroid injections in my spine, and i am going in tomorrow for a procedure where we will use RF energy to kill the nerves that feed the area that is crunched. If it works out well i may be up for a cross country ride next year ;) thanks to all for their support and for understanding when i get a bit grouchy.
 
just a 25ah system but I would prefer to get 50-70 amps out

peak would be set below the capacity of the BMS on a CA.
 
You could upgrade the FETs on one of the BMSs you already have and solder the shunt up a bit... or just solder the shunt, cross your fingers, and replace the FETs if they blow. :?

Either way though you'd need a way to test the new limit. I wonder if you can measure the shunt resistance and calculate exactly what the resistance needs to be to set the limit at 50A. Hmm...
 
Hi Guys, I can't get my 2.3 BMS to latch.
LVC is good. I'm connecting the charger w/o the battery. All LEDs come on evenly and the main is very red. I get 3.7v on each channel with my charger suppling 60.2v .5a. The shunts get very hot. Tested all resistors and they are fine except 1M and the 10Ks by Q1. The 1M reads .280 or .656 depending on lead +- placement and the 10Ks read 6.88. All diodes checkout except D1 and D2. I unsoldered them and they are ok, but when soldered seem to leak. Guessing that is ok... The back of the board is very clean. Lots of isopropyl and toothbrushing. I did have to replace one shunt resistor that read 3.4ohms...?

Thanks for your help!
 
You can't get an accurate measurement of most parts when they are in circuit due to the effects of the other components. Comparisons from one cell to the next may be valid.


First try testing the charge control circuit by jumpering pins 7-8 on any optocoupler. Power up the board and stick a small screwdriver or something between the pins and you should see the main LED change from red to solid green. It should go back to red when you remove the jumper.
View attachment 1

If the LED won't go green, then something is wrong with the bus line or the gate driver chip.

If the first test works, then try jumping the all shunts line to ground. This should cause the main LED to go solid green, but stay latched when you remove the jumper. The LED should reset back to red after you remove the power.

If this isn't working, try to measure the voltage from the ground bus to the any shunt line and the all shunts line.
Testing the Autoshutoff feature.jpg
 
Hi Fletcher. Sorry, I thought I posted the following but it was 3 or 4 in the morning and I rewrote the post four times. I forgot to add this:
I can get the main LED to go green using each opto, works correctly. The all shunts line latches correctly and a reset is required.

It is my understanding that without the battery hooked up the LEDs should flash then go off and the main goes green. All my LEDs stay on and the main is very red, no green tone at all.
 
Back
Top