GGoodrum
1 MW
PJD said:
No, this is not required, or advisable with LiFePO4-type cells. This is true with pretty much any Lithium-based chemistry. The self-discharge rate is very low, on the order of a couple of mA.
New 16-cell Battery Management System (BMS)
- Thread starter GGoodrum
- Start date
No, this is not required, or advisable with LiFePO4-type cells. This is true with pretty much any Lithium-based chemistry. The self-discharge rate is very low, on the order of a couple of mA.
PJD
100 W
Gary and/or Bob,
thanks for the fast answer - another question...
I was a little confused at bob's remark about a charger detecting an end of charge condition by a voltage rise.
I'm not sure but don't most CC/CV/float chargers hold a constant voltage, then detect the completion of charging when the charge amperage decreases to a certain low value?
I assume such a charger will properly work with your BMS - although the float function will become inoperative. Is this correct?
thanks for the fast answer - another question...
I was a little confused at bob's remark about a charger detecting an end of charge condition by a voltage rise.
I'm not sure but don't most CC/CV/float chargers hold a constant voltage, then detect the completion of charging when the charge amperage decreases to a certain low value?
I assume such a charger will properly work with your BMS - although the float function will become inoperative. Is this correct?
GGoodrum
1 MW
Actually, what a normal CC/CV charger does is to start in a current limit mode, where the cell actually determines the voltage level. As the cell gets fuller, it becomes harder for the cell to accept the same current level, so the voltage rises. When it gets to about 3.65V, the cell is almost full (85-90%...). After that, the voltage starts rising at a much quicker rate. To get the last 10-15% in, you need to hold the voltage at the cutoff, and let the cell gradually taper off the current.
With SLAs, you can do this at the pack level because the cells can chemically aborb extra current at the cutoff, so they self-balance. With Lithium-based chemistries, you can't let them overcharge, especially high-powered LiCo cells, which tend to blow up when over-charged. LiFePO4 cells won't blow up but continued overcharging will shorten the longevity of the cells.
In any case, what is needed is a proper CC/CV charging profile. If the cells were all exactly the same capacity, at exactly the same SOC and have exactly the same internal resistance and are all at exactly the same temperature, you could use a single CC/CV charger, set to have a crossover of about 3.65V x the number of cells. The reality is that the cells really need to be able to hit the crossover point at their own pace, and taper the current off at their own rate as well. One solution is to use a bunch of individual cell CC/CV chargers. I still have a setup of 16 of the Voltphreaks 2A CC/CV chargers, and they do the job. Two problems, though. First, you need to bring out all of the junctions for each cell to outside the pack. Secondly, the Voltphreaks chargers are only 2A, so it takes a long time to charge a fairly drained pack.
The cheap Chinese BMS units that come with most of the duct tape offerings leave the CC/CV control to the single bulk charger, and then afterwards tries to bleed off the higher voltage cells to the level of the lowest. The problem is that without individual CV control for each cell, the high cells will start the quick voltage rise before the low ones can get to the cutoff point. The charger just sees the total voltage so as soon as the total hits the crossover point, it holds it there which will cause the high cells to start dropping the current. Since all the current has to go through all the cells, the low cells don't ever get a full charge. Then, the balancer will then start pulling down the the level of the high cells to that of the lowest one. Sure, they might eventually get balanced, but it will be to the lowest common denominator. What you really want is for the cells to just get as full as they can, whatever that level happens to be.
With the shunts, you get the best of both worlds, individual CV control, but a simple bulk charger. The way it works is that once the cell's voltage gets to 3.65V, it is held there and the cell gradually reduces the current it takes in. The trick is that whatever the cell doesn't take in, is bypassed by a big Darligton pair power transistor. If the cell is almost full, most of the current gets bypassed. This way the next cell in series has all the current available to it that it needs, and so on, down the line. The charger just keeps pumping out the max current, and the cells use as much or as little of it as they need. Since the current never reduces, the charger's shutoff feature, it it has one, is not used. Instead, the BMS logic detects when all the shunts are in full bypass, and then cuts off the charge current completely, which then will trip the charger's low current end-of-charge detection logic. On the BMS, you will see all the orange LEDs go off, and the green one come on.
Anyway, because the shunts are really controlling the CV mode(s), a regulated supply could be used instead of a charger. What Bob was refering to is that the signal that cuts off the charge current, and turns on the green LED, could be used by some chargers, or suppies, to stop the charge process. The only charger I'm aware of with this capability is the $450 Zivan NG1.
-- Gary
With SLAs, you can do this at the pack level because the cells can chemically aborb extra current at the cutoff, so they self-balance. With Lithium-based chemistries, you can't let them overcharge, especially high-powered LiCo cells, which tend to blow up when over-charged. LiFePO4 cells won't blow up but continued overcharging will shorten the longevity of the cells.
In any case, what is needed is a proper CC/CV charging profile. If the cells were all exactly the same capacity, at exactly the same SOC and have exactly the same internal resistance and are all at exactly the same temperature, you could use a single CC/CV charger, set to have a crossover of about 3.65V x the number of cells. The reality is that the cells really need to be able to hit the crossover point at their own pace, and taper the current off at their own rate as well. One solution is to use a bunch of individual cell CC/CV chargers. I still have a setup of 16 of the Voltphreaks 2A CC/CV chargers, and they do the job. Two problems, though. First, you need to bring out all of the junctions for each cell to outside the pack. Secondly, the Voltphreaks chargers are only 2A, so it takes a long time to charge a fairly drained pack.
The cheap Chinese BMS units that come with most of the duct tape offerings leave the CC/CV control to the single bulk charger, and then afterwards tries to bleed off the higher voltage cells to the level of the lowest. The problem is that without individual CV control for each cell, the high cells will start the quick voltage rise before the low ones can get to the cutoff point. The charger just sees the total voltage so as soon as the total hits the crossover point, it holds it there which will cause the high cells to start dropping the current. Since all the current has to go through all the cells, the low cells don't ever get a full charge. Then, the balancer will then start pulling down the the level of the high cells to that of the lowest one. Sure, they might eventually get balanced, but it will be to the lowest common denominator. What you really want is for the cells to just get as full as they can, whatever that level happens to be.
With the shunts, you get the best of both worlds, individual CV control, but a simple bulk charger. The way it works is that once the cell's voltage gets to 3.65V, it is held there and the cell gradually reduces the current it takes in. The trick is that whatever the cell doesn't take in, is bypassed by a big Darligton pair power transistor. If the cell is almost full, most of the current gets bypassed. This way the next cell in series has all the current available to it that it needs, and so on, down the line. The charger just keeps pumping out the max current, and the cells use as much or as little of it as they need. Since the current never reduces, the charger's shutoff feature, it it has one, is not used. Instead, the BMS logic detects when all the shunts are in full bypass, and then cuts off the charge current completely, which then will trip the charger's low current end-of-charge detection logic. On the BMS, you will see all the orange LEDs go off, and the green one come on.
Anyway, because the shunts are really controlling the CV mode(s), a regulated supply could be used instead of a charger. What Bob was refering to is that the signal that cuts off the charge current, and turns on the green LED, could be used by some chargers, or suppies, to stop the charge process. The only charger I'm aware of with this capability is the $450 Zivan NG1.
-- Gary
karma
10 kW
nice im getting a better understanding on how a bms works, your system is geting more and more tempting 
GGoodrum said:
Nice to see a packaged charge/discharge BMS being assembled.
Something to consider:
Your voltage clamp appears to be remotely mounted from the parallel cell cluster it controls. It's nice for installation and appearance, but will not work really well, especially at higher charge currents. There will be losses in your interconnects, depending on how much current you are clamping, and would be difficult to maintain an accurate clamp voltage. The differential would manifest itself through variations in state of charge in any given parallel cluster. The only way to reduce the effect would be to use a longer period of low current finishing charge before termination. Of course this only matters if you need to maintain a repeatable clamp voltage under varying charge currents, or the bulk charge current is very low.
In the image attachment below, the voltage clamp circuits are located close to the cluster to address this particular issue. Bulk charge is 12A, tapers until clamp release or 1A (whichever is greater), terminates when all clamps are active with 100MA + 5 minutes finish.
Anyway, great to see what you've done here. Regards, jeff
chas_stevenson
10 mW
- Joined
- Feb 23, 2008
- Messages
- 20
I don't know about anyone else but I am excited about these BMS boards. I am ready for at least 6 kits as soon as they are available.
So do you guys have an estimated time for production kits or early adopter kits?
Chas S.
So do you guys have an estimated time for production kits or early adopter kits?
Chas S.
Jeremy Harris
100 MW
Jeff,
Are those neat little clamp PCBs available to buy anywhere, or were they just a one-off for your own project?
Jeremy
Are those neat little clamp PCBs available to buy anywhere, or were they just a one-off for your own project?
Jeremy
Jeremy Harris said:
Jeremy, I'll sell the unstuffed boards at $4 each, plus a little for postage. The price includes a BOM and and assembly notes. I'd rather not sell fewer than 5 at a time though. Component cost will be about $10 per channel, including the board.
I have about 50 unstuffed boards remaining. These are designed for a sharp cutoff (20mv), calibrated using a trim resistor that you determine.
They're mostly 0603 surface mount, so it requires some skill to assemble. The results are excellent. An opto isolator on each board can be used to flag your charger to step down the current, and/or terminate the charge. You have to decide how you want to flag a charging system of your choice.
These are a part of a larger BMS project for a full size EV, and I've not developed a charge controller for an ebike application.
Regards, jeff
PJD
100 W
Gary and/or Bob,
One more question (well, actually two):
What is the maximum allowable running amperage that the LVC part of your BMS can handle? I am interested your BMS's for a scooter with a maximum current to the controller of 95 amps.
And one thing I might point out is that imbalance is also a problem with SLA packs. The early-finishing batteries can see 17 volts and be gassing badly while the others are at just 13.5 volts. The result is a couple batteries always fail prematurely. In fact this is one of the biggest reason for the failure of the various Chinese scooters like the EVT's to catch on.
I am currently using "Powercheqs" battery balancers on my current SLA packs.
http://www.powerdesigners.com/powercheq.htm
Ever heard of them or do you have any opinions?
One more question (well, actually two):
What is the maximum allowable running amperage that the LVC part of your BMS can handle? I am interested your BMS's for a scooter with a maximum current to the controller of 95 amps.
And one thing I might point out is that imbalance is also a problem with SLA packs. The early-finishing batteries can see 17 volts and be gassing badly while the others are at just 13.5 volts. The result is a couple batteries always fail prematurely. In fact this is one of the biggest reason for the failure of the various Chinese scooters like the EVT's to catch on.
I am currently using "Powercheqs" battery balancers on my current SLA packs.
http://www.powerdesigners.com/powercheq.htm
Ever heard of them or do you have any opinions?
GGoodrum
1 MW
Jeff said:
Actually, this board is designed to te mounted on the front of a LiFeBatt pack, similar to the way the LVC boards were mounted befire:
There are equal length 16-gauge wires that go from the edges of the board to each cell junction. We don't have parallel groups, per se, just big cells. In addition, each cell's shunt can have the clamp voltage adjusted, via a small pot for each channel. Also, the way this one works, the bulk charger charges at it's max current rate through the main pack + and - terminals. The shunts do nothing until one of the cells hits 3.65V, and then it's shunt will just start to conduct. You will see the LED for that channel just start to come on. As soon as the first shunt starts conducting, logic gets triggered that throttles back the charge current, via a FET, to about 2A. This value is also adjustable, from about 1/2A up to a max of about 10A. This is really a function of how big a heatsink you use. With the one I'm using with the LiFeBatt packs, 2A is a good number. It really doesn't matter all that much becuase unless the cells are really out-of-balance, this last charge phase goes pretty quick, like 10-15 minutes. When all the shunts are in full bypass, and the cells are full, the current limit FET is shut off completely, all the individual shunt LEDS go off and a single green LED is lit. Disconnecting the charger will reset everything and disable all the shunt logic until the next time the charger is connected. With the Soneil chargers, that I will be offering with these packs as an option, the charger will reset when it sees the charge current shut off at the BMS.
-- Gary
GGoodrum
1 MW
PJD said:
There is really no limit on current for the LVC circuits because the current does not go through them, so 95A is not a problem. I've hit 88A on one of my bikes with a controller Bob modded for me.
There was a thread here recently, on the Powercheqs. The idea is not bad, but these things work on 12V "chunks", which in the case of LiFePO4 cells is 4 cells. The concept is interesting though.
-- Gary
PJD
100 W
I hope this this doesn't sound too e-illiterate, but how does the LVC work if it doesn't open the pack circuit somehow?
I had this notion that it opened the pack circuit, (or it could open just the contactor relay circuit or separate power circuit to the controller) when any cell in the series went below 2.1 volts. Not correct?
GGoodrum
1 MW
PJD said:
My bad, I was forgetting that we now have an active cutoff of pack power if an LVC circuit is tripped. The basic LVC circuits we started out with used the controller's ebrake line to have the throttle cut if a LVC circuit tripped.
For the cutoff, we are using a pair of 4110 FETs, which are the same ones used in the controller upgrades. The pair are easily good for 100A, but to go that high the traces should be beefed up a bit with some extra solder.
You can also simply use the LVC opto signal and connect it to the ebrake line, like before.
-- Gary
i trust gary has answered all the questions. any that remain we will try to answer. as gary pointed out the shunts never operate at full charger current, so voltage drop in the wires is not really a problem, and in any case any voltage drop is adjusted out by setting the voltage at the cell with the bms mounted pot. The board has space for 2 irfb4110's for the lvc cutoff so 100A peak is not a problem, but i would advise anyone who can to just use the ebrake signal and eliminate the cutoff fets on our board. no point throwing away power, even if it is only a few watts. at 100A the 2 fets resistance of about .002 ohms will result in a power dissipation of 20W in the fets. this will definitely heat things up. we are recommending 50A which would only be about 4.5W power dissipation in the fets.
the lvc draws only a microamp when not active and a milliamp or two when asserted. the charge cycle circuit is powered by the charger.
we are currently beta testing the latest pcb version. we do not want to release the kit or packs prematurely and let our customers do the beta testing. we intend to deliver a product that will function reliably for years.
the lvc draws only a microamp when not active and a milliamp or two when asserted. the charge cycle circuit is powered by the charger.
we are currently beta testing the latest pcb version. we do not want to release the kit or packs prematurely and let our customers do the beta testing. we intend to deliver a product that will function reliably for years.
bobmcree said:
Now.. THAT i like to hear !
Thanks Bob !!
GGoodrum
1 MW
As Bob says, we want to get it right, which is the reason for the delay. We've been doing a lot of testing, which is a good thing. One thing we've determined is that once the cells get full enough that the shunts start to activate, it doesn't make much of a difference,time-wise, whether you let the shunts run at 1A, or 3-4A. It makes a huge difference, though, in how much heat is generated. On my test setup, shown below, I'm using two 1/4" x 3/4" bars that attach to the shunt darlington power transistors, and then a 3/16" thick plate that mounts on top of these.
In testing this with all shunts fully on, 2A is about the limit that can be run and still have the heat down under about 160F. Dropping down to about 1A drops the temps down to under 120F. Once the cells get to 3.65V, the current that the cell can accept starts dropping anyway, so the amount of time it takes for the cell to get full doesn't change much if the current is limited to 1A vs 2A. It takes about 15 minutes, or so, in either case.
The last thing we are working on right now is the current limit logic. We've made some changes, as a result of all the testing, and I've had new boards made. They are arriving today, in fact. We just need to verify that changes work as tested, and then we are done.
Bob is working on a "theory of operation" document and I'm putting together assembly instructions. We also need to finalize a set of test procedures. We're trying to make this as complete as possible and not require any exotic test equipment or supplies.
-- Gary
In testing this with all shunts fully on, 2A is about the limit that can be run and still have the heat down under about 160F. Dropping down to about 1A drops the temps down to under 120F. Once the cells get to 3.65V, the current that the cell can accept starts dropping anyway, so the amount of time it takes for the cell to get full doesn't change much if the current is limited to 1A vs 2A. It takes about 15 minutes, or so, in either case.
The last thing we are working on right now is the current limit logic. We've made some changes, as a result of all the testing, and I've had new boards made. They are arriving today, in fact. We just need to verify that changes work as tested, and then we are done.
Bob is working on a "theory of operation" document and I'm putting together assembly instructions. We also need to finalize a set of test procedures. We're trying to make this as complete as possible and not require any exotic test equipment or supplies.
-- Gary
slayer
1 kW
So this is the 16 cells bms ...how about the 12 cells bms ...is it going to be this one but we will connect only 12 cells .
GGoodrum
1 MW
slayer said:
The same board can be used with any number of cells, up to 16. You just populate the number of channels that you need, and run one jumper across the unused channels. You can also go up beyond 16 cells, and use the same charger control and LVC cutoff circuits, by using two boards with a few interconnecting wires.
The initial kits I will offer will be for 12-cell and 16-cell versions, but I'm sure if somebody wants something different, a custom version can be supplied.
-- Gary
recumbent
100 kW
GGoodrum said:
8) Thank you, Not all of us have electronic test equipment, except of course a "multi-meter" and a "cycle Analyst".
:| I've already placed my order with Josh Goldberg in T.O. and waiting patiently.
I hope you guys get rewarded properly after all this work, and hassled in the forums.
I for one am very gratefull.
albie
100 W
i'm just wondeirng... would the BMS work on the ebay duct tape packs?
GGoodrum
1 MW
albie said:
Yes, it can work with any LiFePO4-based pack with no changes. You could als use it with Li-Mn or Li-Co chemistries, simply by adjusting the shunt voltages higher, using the pots that are on each channel.
albie
100 W
oh man that is way awesome...
I hope you complete testing really quickly! Cause i'm putting my name down for one!
I hope you complete testing really quickly! Cause i'm putting my name down for one!
GGoodrum said:
the voltage of the shunts is nominally adjustable from about 3-4v. by changing resistor values any shunt voltage from 2.5v - 5v can be achieved. the low voltage cutoff can be either 2.1v or 2.7v with standard parts. if a different low voltage cutoff level is desired any level above 2.1v can be achieved with a resistive divider on the input. because the TC54 draws a microamp the resistor pair should draw 10-100x that for good accuracy.
the TC54 is laser trimmed by Microchip so if you want 10,000 you can probably have any custom LVC voltage you like and still keep the drain at a microamp from that part of the circuit
albie
100 W
not sure if you saw what i posted in the other thread about the duct tape packs but...
would these bms be able to handle 2x 48v packs connected in series to make a 96v pack?
would these bms be able to handle 2x 48v packs connected in series to make a 96v pack?
Gary: one feature I've seen on the fancier BMS systems is a provision to drop the charging current low if any of the cells is below the low voltage threshold. If you had a discharged pack that sat around long enough to self-discharge, some cells could be that low. Apparently the cells can be damaged by full charging current if they are too low.
Any thoughts?
Any thoughts?
