Ayaa 7 cell LiPo BMS schematic

[Randomly]

This is a schematic I did as a learning exercise for a new schematic capture program. This is from a BMS I reverse engineered that is manufactured by Ayaa out of china, and remarketed under the Powerizer brand by Batteryspace.com. The PCB was laid out to handle a maximum of 10 series cells, but only populated with parts in this case for 7s. The pack was a LiPo 7s 8AH 25.9V nominal voltage made from Wanma cells.

The S-8261 and S-8241 are Seiko Instruments battery management ICs. Here they are used only as low power voltage detectors.

The S-8261 handle the overcharge and overdischarge detection functions. Over discharge is triggered at 2.5V and turns the Discharge FET off (Q40), Over discharge is released when the cell voltages go above 2.9V. Overcharge is detected at 4.325 V and turns the Charge FET off (Q52). Overcharge is released at a cell voltage of 4.1V. There are internal time delays in the ICs. An overdischarge condition must exist for 0.15 seconds to be recognized, An overcharge condition must exist for 1.5sec to be recognized. Release times have no delay.

If any cell drops below 2.5V, an overdischarge is detected and the S-8261 output turns on an associated PNP transistor that then turns on Q3 which pulls the gate of Q40 low and disconnects the battery from the load.
Likewise if any Overcharge is detected, the S-8261 turns on another PNP which turns on Q4 which pulls the gate of Q52 low and turns off charging to the battery. The gate drive voltage is supplied by tapping off the 3rd cell in the stack so the gate drive varies from 7.5v - 12.9V depending on the state of charge.

Q2 is the overcurrent cutoff circuit. If the current through Q40 and R46 is high enough to create a voltage around 0.46 V on the base of Q2, Q2 turns on and pulls the gate of Q40 Low, disconnecting the battery from the load. The Rdson of the 11 paralleled FETS that make up Q40 is about 1 milliohm, R46 is about 2 milliohms. It's not a very precise system, and it's quite temperature dependent. The overcurrent trip point would be in the 130-160 amp range at 25C.

Q1 is the Load removed detection circuit. If an overcurrent trip turns off the FETS, the load will pull the negative battery terminal up. As long as that negative terminal is pulled up by more than about 800mv (if the load is drawing more than about 1 microamp through R6,R8,R9) Q1 remains on, and keeps Q40 Off.

U8-U14 (S-8241) do the charge balancing. When a cell voltage reaches 4.2V they turn on a PFET which puts a 68 ohm resistor across the cell. This reduces the charging current into the cell by about 68 milliamps. All cells continue to charge. When any cell reaches 4.325V the Overcharge detection goes off on one of the S-8261 (U1-U7) and stops all charging. U8-U14 continue to discharge the cells until their cell voltage goes below 4.2V and then they turn off.

That about covers it.

For you circuit types I have a few posers for your entertainment
1) What do R22 and C1 do?
2) What is D1 for?
3) Why is R46 needed?
4) How can you reduce the cost of this?

 

[fechter]

That's great! Thanks for posting it.

For you circuit types I have a few posers for your entertainment
1) What do R22 and C1 do?
2) What is D1 for?
3) Why is R46 needed?
4) How can you reduce the cost of this?

1) I have no clue. why would cell 1 be different than the rest? It looks like a snubber to prevent some kind of oscillation.
2) When Q40 opens to drop the load, the pack negative lead will go high. Diode prevents reverse current into the transistors.
3) I think that's part of the overcurrent protect circuit. Higher resistance lowers trip current.
4) ? It looks pretty 'bare bones' to me. You might substitute fewer, larger FETs for many small ones, but I don't know how the pricing compares. I would guess they already looked at that.

 

[Randomly]

Here's a picture of the BMS. The balancing circuits with U8-U14 are on a separate smaller red board seen in front connected by ribbon cable to the rest of the circuit on the larger green board in back.
The data sheets for the ICs are here
http://dl.sii-ic.com.edgesuite.net/spd_dtst/dt_sht_e/lithium/S8261_E.pdf
http://dl.sii-ic.com.edgesuite.net/spd_dtst/dt_sht_e/lithium/S8241_E.pdf

Hints for questions
1) You'll need to read the S-8261 data sheet about the VM pin function.
2) You're right Fechter about the negative pack lead going high, but how much reverse current are you going to get through that 6.8M resistor? and where does it go?
3) Again you are right about R46 being used to help determine the overcurrent trip point. But at 0.002 ohm it's generates as much heat dissipation as the 22 TO-220 Fets on the board. Is there some way to get rid of it?
4) Reducing the cost is related to 3)

 

[fechter]

Holy crap! I thought the FETs were going to be tiny little surface mounted things.
I don't know what the pricing is on a S-8261, but there's a crapload of circuitry inside there that is not being used. A simpler voltage monitor might be cheaper.

Back to the FETs, the function of them is to drop the load when the voltage gets too low on any cell or when the current exceeds a certain point. The other half of them is to drop the charger when the voltage gets too high. Since the body diodes in the FETs conduct backwards, it was necessary to put two FET switches in series so it could open in both directions.

If the charger input was on a separate wire, you could have a separate (presumably much smaller) FET switch on the charger line. What is the maximum charging current?

For overcurrent protection, you could ditch the shunt and associated circuitry, and just use a fuse. Your load device should have some kind of current limiting. The fuse will protect against shorts or accidents.

Likewise if you could interface the load device to drop power when the low voltage threshold was reached, you could eliminate the switch.

 

[Randomly]

Sorry I got distracted and left your post unanswered.

I think the S-8261 and S-8241 are in $0.20-$0.30 range each in quantity. The S-8261 is essentially two voltage detectors in one. probably not going to save much money there. The S-8241 you might find something cheaper that would work, but these guys probably have some pretty good volume pricing going on.

You're correct about the FETs.
Yes you could get rid of most of one half the FETs with a separate charger, but you would lose the simplicity of the battery pack with just two big leads coming out, no special connections required. You are correct again though that this would be a considerable cost savings if the separate battery charging leads is acceptable.

Yes you could use a fuse, but that would make it a one shot deal. Or you could interface it to the load and have it shut off the load. But the goal here I think was to have a self contained battery pack that just protected itself no matter what you did on the outside, charge it, discharge it, short it out, whatever.
on to the posers
1) What do R22 and C1 do?......Absolutely Nothing!
I'm pretty sure this circuit is just one of a long line of adaptations and revisions of some prehistoric design. They keep adapting and scaling it to fill their different battery pack needs. Probably the original designer has long since moved on, and R22 and C1 are just left over parts from when they once used the over current functions in the S-8261, but nobody realizes they are no longer needed.

2) What is D1 for?......Again Nothing.
It does block reverse current from flowing in this point when the FETs turn off and the load pulls the output high, but it's in series with a 6.8 Meg resistor, so extremely little current is flowing and it just flows into collector of Q8 to the base and into the 10M resistor connected to the output driver of the S8261. If the pack voltage is high enough it may pull the base of Q8 high enough so the BE junction avalanches and the few microamps will flow into the top of the first cell. Nothing anybody cares about. Again this is probably left over from when the S-8261 was used for the overcurrent control and the open drain PFET needed protection since there was no series resistor to limit current.

3. R46 is part of the over current sensing. When the voltage drop across R46 and the Q40 FETs get high enough to turn on Q2 it pulls the Gates of Q40 low and turns them off. With Q40 off the load pulls the drain of Q40 even higher and this keeps Q2 on and the FETs off till the load goes away. In fact the Load Removed circuit of Q1 does exactly the same thing except it connects to the Source of Q52 instead of the drain. But if the output is pulled up current just flows through the body diode of Q52 to the drain of Q40 and Q2 senses that. So the circuit of Q1 is completely redundant as it's already performed by Q2 except with an extra diode drop in there. I can't think of any situation where it would make a difference. sheesh these guys....
back to R46 - One reason to use R46 is to set the trip current point. the larger R46 the lower the trip current. The big drawback is power dissipation. This pack can handle 100A loads. With R46 at 0.002 ohms that's 20 watts of heat being generated. With Q40 and Q52 each at about 0.001 ohm they each dissipate 10 watts. So 40 watts total. Since they are turning on a transistor, the trip voltage needed has to be about 0.5V, So they need the 0.003 ohms of R46 plus Q40 to achieve that desired trip current. Why not get rid of R46 completely, and use only 1/3 of the FETs to make Q40. That would give you your trip voltage and save you a bunch of money!
The one drawback to this is the high thermal coefficient of resistance of the FETs. From -40C to +100C the resistance almost doubles. This makes for a pretty inaccurate trip point (not to mention as the transistor Q2 gets hotter it's on threshold voltage gets lower further reducing the trip point).
So maybe they put R46 in there to reduce the variation of the trip point with temperature. Somehow I doubt it though.

So how to reduce cost without changing any of the function of this at all? The cost is in those acres of FETs.
Throw out half the FETs for Q40 and Q52 and get rid of R46. You now still have 0.004 ohms through the whole thing and the same power dissipation as before. You fix the current trip point by using a cheap ultra low power op-amp instead of a transistor to do the current sensing across Q40 and temperature compensate it. You save the cost of 12 expensive TO-220 power FETS at the price of 30 cents worth of opamp and some penny parts, and you've improved the current trip point accuracy over temperature.
Another place to save money is in the type of FETs. They are using 75V parts on a 30V pack. Changing to 40V parts you would get 1/2 the on resistance for about the same price. You could take Q40 and Q52 down to 3 parts each, maybe even 2, a total savings of 18 FETS!

If you added in your suggestion of a separate charger connection you can get rid of another couple FETs in Q52.

Well so much for chinese BMS circuits.
I think that about wraps it up for me. I hope you enjoyed the romp through this circuit, I had fun with it anyway.