New 16-cell Battery Management System (BMS)

[fechter]

Somehow one of my posts went into the ozone somewhere.

Anyway, if you wanted to run multiple packs with BMS in series, you would put diodes across each one like this:

BMS in series config.jpg (19.66 KiB) Viewed 8182 times

This way, if any BMS goes open circuit, the entire pack voltage is not across the switch, just the normal pack voltage (plus one diode drop). This scheme would be useful if you wanted to run voltages higher than the BMS was rated for. This will also work with other styles of BMS.

[GGoodrum]

Okay, we finally finished the testing, and everything seems pretty rock solid. I've built up two new boards, using the newest assembly and test instructions, and with a couple more tweaks, it will be ready-to-go.

Here is the final schematic:




Bob wrote up a very good "theory of operation", which also includes a more easily readable schematic. It can be found here LiFeCycle BMS Description

One of the last changes we got in was Richard's suggestion about throttling back the charge current if any of the cells are too low (i.e. -- below 2.1V...). It turns out this simply required adding one more inexpensive NPN transistor.

We made a couple of changes to the layout, so I'm having another run done. I will have these back by Thursday. For the assembled and tested versions, we can use the existing boards, as it just requires a couple additions, mainly that make use of existing vias. The bottomline is we can probably make the assembled versions available starting tomorrow or Wendesday, and the kits by Friday. It will take about a week for the assembly and test for those versions

Here's what the PCB looks like:




It is pretty busy, mainly because we weren't ready to switch to surface mount, just yet, and because I wanted to keep this to the same 3" x 6" form factor, so that it will fit on front of a LiFeBatt pack.

Once I get the assembly and test procedures done we should be pretty much done. I still have some work to do on the heat sinks, but I'm hoping to finish up these tomorrow. These will be optional for the kits and the assembled versions, as there are lots of choices for these, depending on the installations.

Anyway, once we get these going, I will then also start offering LiFeBatt packs with integrated BMS units. I also just worked a deal with Soneil, so we will also be able to offer 36V and 48V versions of their excellent SRF series of chargers, with "tweaked" outputs to perfectly match the BMS.

-- Gary

[EMF]

Yep, that there scheematic looks right to me! j/k

Wow! I cannot believe what you guys are able to come up with. This is very impressive indeed. It's way over my head, but still, I can be impressed by it.

I think you guys should take your wives out to dinner and a glass or two of wine since you are tinkering too much!

[jeffkay]

Great job Bob and Gary! My pack is A123 M1, 16s12p, at 8 amps, how long do you think it would charge from complete depletion? What end of charge voltage measurement with no load should I expect to set it for?

Jeff K./Deep Cycle

[GGoodrum]

Great job Bob and Gary! My pack is A123 M1, 16s12p, at 8 amps, how long do you think it would charge from complete depletion? What end of charge voltage measurement with no load should I expect to set it for?

Jeff K./Deep Cycle

Hey Jeff --

16s12p? Yipes! That is some serious Wh.

Let's see, 12 x 2.3 Ah is 27.6 Ah. Charging at a 1C rate would require 27.6A of charge current to recharge a depleted pack in one hour, so at an 8A rate, it would take about 3-1/2 hours.

I'm not sure what you mean in your second question.

-- Gary

[bobmcree]

i believe he meant what "peak voltage do i use for a123 charging?" and the answer is any voltage you want between 3-4v most people use 3.65v

[bobmcree]

Yep, that there scheematic looks right to me! j/k

Wow! I cannot believe what you guys are able to come up with. This is very impressive indeed. It's way over my head, but still, I can be impressed by it.

I think you guys should take your wives out to dinner and a glass or two of wine since you are tinkering too much!

this kind of stuff comes pretty easy when you don't sleep anyway we have a 2 lb 100v 100Ah battery in the works next but it may take a bit longer...

[Deepkimchi]

Hello Bob,

What kind of sleep problems are you having? Remember Xyster seemed to have it too.

My brother swears by Melatonin. I try to think about ebikes when I have trouble sleeping. I'm almost 51, and have PTSD from something I can't talk about (things that went boom). A shot of Cognac helps too about 1/2 hr before bed. Working out at the gym helps relieve stress. Going to bed at a fixed time, not doing anything else but sleeping in bed (besides that other thing), watching caffeine intake such as coffee/sodas (avoid late afternoons), may be some things that help.

DK (an American in S Korea, not Paris)

[jeffkay]

What I meant was this: The cutoff voltage is say 2.7 or 2.5v. After the BMS/CMS device does its cutoff, if the load is removed, the cells should be at some voltage of rest--what is that? It is probably in a range of 3.00v?

"Serious watt hours"... Yes, "Deep Cycle" is a full size commuter motorcycle conversion. Originally a 1973 Yamaha 750cc, it uses a first gen Etek and an Alltrax set for 150 amps. I have been reconfiguring the packs and now have 4 packs with 4 sub-packs of 16s12p. I drove the heavy bike and heavy rider (!) the other day with 3 packs (36v nominal) and it went over 14 miles on very hilly route. It is geared to be about 7:1. The 4 pack estimate should get me 19.5 miles or more on the flat. Speeds with that pack have reached 50mph but realism dictates about 40-45mph. It is a blast.

I have charged the 12p sub-packs in under 20 minutes using my 80 amp Vicor supply! Yes the charge wires (6ga.) get hot!!!

Jeff K./ D"Deep Cycle"

[bobmcree]

What I meant was this: The cutoff voltage is say 2.7 or 2.5v. After the BMS/CMS device does its cutoff, if the load is removed, the cells should be at some voltage of rest--what is that? It is probably in a range of 3.00v?

that voltage or higher, yes. what typically happens is that you will hit the lvc under heavy load and the system will cut out, then when the load current drops and the voltage rises above cutoff + .150V the low voltage signal goes high and either releases the ebrake signal or cuts off current to the load, which immediately clears the lvc and restores power, then if you reduce the throttle you usually have 10% or so remaining capacity you can still use at a lower throttle setting.


if that doesn't answer your question i will try again...

I have charged the 12p sub-packs in under 20 minutes using my 80 amp Vicor supply! Yes the charge wires (6ga.) get hot!!!

Jeff K./ D"Deep Cycle"

that's the reason we put the 8A current limit in the spec. when the first shunt starts to conduct the current is reduced,but the full charge current still needs to go through the cutoff fets, and with .002 ohms from a pair of 4110s it should be possible to use higher currents, but we need to put a limit somewhere or somebody will try to run 80A which would be 6.4W in the cutoff fets, still manageable but a fair amount of heat.

[GGoodrum]

What I meant was this: The cutoff voltage is say 2.7 or 2.5v. After the BMS/CMS device does its cutoff, if the load is removed, the cells should be at some voltage of rest--what is that? It is probably in a range of 3.00v?

"Serious watt hours"... Yes, "Deep Cycle" is a full size commuter motorcycle conversion. Originally a 1973 Yamaha 750cc, it uses a first gen Etek and an Alltrax set for 150 amps. I have been reconfiguring the packs and now have 4 packs with 4 sub-packs of 16s12p. I drove the heavy bike and heavy rider (!) the other day with 3 packs (36v nominal) and it went over 14 miles on very hilly route. It is geared to be about 7:1. The 4 pack estimate should get me 19.5 miles or more on the flat. Speeds with that pack have reached 50mph but realism dictates about 40-45mph. It is a blast.

I have charged the 12p sub-packs in under 20 minutes using my 80 amp Vicor supply! Yes the charge wires (6ga.) get hot!!!

Jeff K./ D"Deep Cycle"

Ah, okay, now I know what you mean. With so many a123 cells in parallel, the voltage drop is minimal, all the way up to when the cells are pretty close to being fully discharged. If the cells are pretty close, they tend to all dump at once. You don't get much notice. Because of this, the cells will have resting voltages up close to 3.3V per cell, even up to when there's less than 10% capacity left. When they do finally dump, the voltage drops off pretty quick. Because of this, what the LVC need to do is detect this drop as soon as possible, which is why we used the 2.7V TC54 versions in the LVC circuits that are used with a123-based packs. With LiFeBatt, and other LiFePO4-based cells, you can actually feel a slight difference towards the end of the capacity, so the 2.1V version of the TC54 seems to be perfect. I have driven my LiFeBatt packs down to when the cutoff trips, under a big load, and what happens is that you will feel a quick "hit". If you keep the load on, it will hit again, about a 1/2 second later. If you back off the throttle, this oscillation will stop, and you can go about another couple of miles, with a lower load. Eventually, any throttle at all will cause this oscillation. At this point, the WattsUp says I've used about 9.8Ah, which is about what it takes back in.

-- Gary

[bobmcree]

Hello Bob,

What kind of sleep problems are you having? Remember Xyster seemed to have it too.

My brother swears by Melatonin. I try to think about ebikes when I have trouble sleeping. I'm almost 51, and have PTSD from something I can't talk about (things that went boom). A shot of Cognac helps too about 1/2 hr before bed. Working out at the gym helps relieve stress. Going to bed at a fixed time, not doing anything else but sleeping in bed (besides that other thing), watching caffeine intake such as coffee/sodas (avoid late afternoons), may be some things that help.

DK (an American in S Korea, not Paris)

Thanks DK, the major factor in my sleeplessness is pain from a severe back injury. i already take enough narcotics to kill a horse, and i can't just keep increasing the dose because of the side effects. i agree about the melatonin, it helps a lot. I had a spinal fracture from falling off the roof of my tour bus when i was up there tying off some luggage after river rafting in the snake river in idaho. i could make it more interesting by saying i got hurt river rafting, but the truth is not quite so romantic.

[rarebear]

Hello Bob and Gary!!

Congratulations !!!!!!!!!!!!! very impressive!

My question: (12 cells LiFeBatt)

Is the LVC realy needed when i set my programmable controller to a safety LVC (28V?)

I would like to use your:

"Charger Management System (CMS) as an external board/unit that would sit between the output of a standard SLA charger, or a regulated power supply, and the pack, connected to the latter via a special 18-pin plug."

Is your 12 Cell CMS (external board) also available? or only the BMS board with LVC?

I will use two of yours LifeBatt 12 Cells Packs in a high speed application (wired in series)
and in a long range application (wired parallel)

I also want to quickcharge it with a 20 amps output SLA charger. (Is it possible?)

Any suggestions?

Please help!

Thank you!
Albert

[bobmcree]

Hello Bob and Gary!!

Congratulations !!!!!!!!!!!!! very impressive!

My question: (12 cells LiFeBatt)

Is the LVC realy needed when i set my programmable controller to a safety LVC (28V?)

depends on whether your cells stay balanced. i feel it is cheap insurance on expensive cells to watch each one or each parallel group

I would like to use your:

"Charger Management System (CMS) as an external board/unit that would sit between the output of a standard SLA charger, or a regulated power supply, and the pack, connected to the latter via a special 18-pin plug."

you can use this system that way just with longer wires and an 18 pin molex connector which i believe gary has

Is your 12 Cell CMS (external board) also available? or only the BMS board with LVC?

you would have to use the complete board and just not stuff the lvc parts

I will use two of yours LifeBatt 12 Cells Packs in a high speed application (wired in series)
and in a long range application (wired parallel)

I also want to quickcharge it with a 20 amps output SLA charger. (Is it possible?)

the shunts can only handle 8A , but when the shunts activate the charge current is reduced, so the shunts do not have to carry all the current, the lvc circuit does if it is used, and if the charger has the proper output voltage of 3.65v per cell plus a half volt or so for fudge factor, and not so much that there is excessive voltage across the fet that limits the charge current. we really do not intend the system to handle that high current, but it is possible since the high current only goes through the cells if the low voltage disconnect function is omitted and the ebrake signal used instead. even at 20A the dissipation in the fets would be just a watt or two, but the pcb traces are not intended to carry that much current. beefing them up with extra solder would probably make it possible to run at 20A but we have to draw the line somewhere for warranty purposes.

[bobmcree]

As i said before, the LifeCycle BMS (my idea for the name - not connected with LifeBatt except that i am also US distributor for their cells for e-bike applications) is not perfect. I felt that we could not bring a PIC based product online that was reliable for at least another month or two, and every day we have been getting requests for complete systems including bms and charger. Lifebatt has now finally come out with their own system, and i am about to start testing it. I will be using that vms system and the LifeBatt 8A charger on the system i am putting together for Jim Parker of Cruzbike, as that is the only way to get him the promised 3 year warranty. The LifeBatt system is certainly more sophisticated than mine, but does not have exactly the same functions, so it is up to the user whether to buy a complete LifeBatt "hard pack" system and get 3 years, get raw cells and get 90 days, or somewhere in between by using our system and or charger. Gary has secured an agreement with Soneil to customize some 4.5A chargers that he will be able to offer at a good price.

A few notes on the design; it has been described as "brute force", and i suppose that is valid from the viewpoint of the wasted power, simple minded control system, and very high parts count, but i would describe it as just the opposite. We are treating the cells as gently as we can by assuring that the cell voltage never exceeds 3.65v or another user-variable setpoint from 3-4v, and we are reducing the charge current when near the endpoint to assure accurate measurement without the voltage drop of the wiring interfering. It has been said that the component count is too high. The shunt circuit could be a simpler if the pot was replaced by a resistor, but there is no simpler method i have found for achieving the degree of control over cell voltage this provides. If the logic section is left out it can be used with just the shunts and a charger that cuts off at the total shunt voltage, as long as the heat sink is big enough to handle the power dissipation. The LEDs and associated drive circuitry can be eliminated if desired.

The current reduction circuit could be changed to pwm, but i wanted to maintain the capability to use existing sla chargers insofar as is possible, so i wanted almost no voltage drop in the charge path, and .004 ohm drop of the fet used will be pretty close to none with the circuit the way it is. anybody who wants to could replace the fet and op amp circuit with any other type of current regulator that can handle the task, provided you provide the required control inputs to drive it on or off or regulate. Generating the logic signals from 16 channels to tell when none, any, or all are on could have been complicated and would require a lot of i/o pins on the PIC, but instead i did it with 32 2 cent diodes. (1 cent in qty 1000). the one shot that generates the delays is accurate enough to provide the delays required, though of course a PIC could be more accurate.

To achieve the kind of low voltage protection the system provides with a dedicated TC54 for each cell costs less than a dollar a cell in parts, but permits us to perform this vital function drawing only 1 microamp from the cell while not charging. This real-time system also provides warning by cutting power at full throttle then immediately restoring it when the throttle setting is reduced. This provides ample warning to get to a safe place to stop without abruptly cutting power, while still providing excellent protection against over-discharge. There is no way a microprocessor based system could sample each cell at an adequate rate and resolution to provide a matching function without capability that would make it too expensive.

The LM431 shunt regulator IC is comprised of an accurate voltage source, a low current comparator, and drive circuitry so it is pretty well integrated, as is the darlington output stage. both of these are common 50 cent parts. The pot setting the clamp voltage could be replaced with a fixed value but that would result in about a 5% possible error which can be reduced to the accuracy of the meter used to measure it and he setting will drift of <1%. over the expected temperature range.

Every part in the design was carefully chosen with cost and reliability weighed along with simplicity and the idea of creating a kit that the user could understand and learn from while they are building it. Hopefully putting together the bms will help users become better informed about the use of their expensive battery systems, and we can help them to get optimum performance. The design is too complicated to make money on it but that does not mean it is not a valid, useful design. If it makes it possible for us to offer a complete solution to our users the profit on the bms is secondary to its preventing a landslide of battery failures while still under warranty a couple of years down the road.

We made the schematic public so the tinkerers could play around with it and customize it to suit individual needs. If the guys working on a PIC based BMS come up with something workable that is less expensive i am all for it. Right now i just don't have time to follow that thread. Spring is coming and it's time to get to work if we are going to have a garden.