New CellLog-based BMS "Lite"...

GGoodrum

1 MW
Joined
Sep 20, 2007
Messages
2,908
Location
South Orange County, CA
Richard came up with an idea recently for a "Lite" version of our BMS design that is extremely simple. Basically, it just uses the 431 shunt regulator as the actual shunt. Granted, you can't do 500mA of shunt current, but it will handle 100mA, which is as much, or more, than what the typical Chinese BMS circuits, and most RC balancing chargers, can do. With LiPo, 100mA will be fine, for healthy packs. For LiFePO4, I'm not so sure.

What I also wanted to do is try a surface mount version of the BMS, and we've been working on variants of this for awhile now. This is a much simpler task, using this "Lite" version. Anyway, I've been also playing around with a surface mount version of the CellLog logic that isolates and converts the CellLog's alarm outputs into LVC and HVC signals, and also the relay-based logic that will automatically turn the CellLogs on and off, if either the bike's controller is on, or if the charger is on. The idea was to do like I did before, and have the circuitry be "hidden", underneath the CellLogs. Doing this with surface mount parts freed up a lot more space than I thought it would, which then gave me the idea to also add the "Lite" balancing cell circuits. It all fit quite nicely. The resulting board would just be enough wider, horizontally, to have 8 LEDs on the right (to indicate the cells that are shunting...). I then went to work on doing a modified version of the Zephyr charge controller that I had already converted to surface mount. My original thought was that I'd just add the boards to the CellLogs, and then do the charge controller as a "lump" in the charger cable. I also thought it would be useful to have everything on one board, three 8-channel CellLogs, with the extra logic and Lite circuits, and then the charge controller at the bottom of the board. In the end, I decided to basically do the PCB in a way that would allow either option. This is what it looks like:

3x8 CellLog BMS Lite-v4.4.3o.png


I had some room left over, so I also added a 4x6-to-3x8 adapter, for mapping four 6s connections into three 8s, so that three 8-channel CellLogs could be used with a 24s configuration based on using 6s packs. The LVC and HVC trip points are settable via the CellLogs. The shunt turnon point is fixed, however, at 4.15V per channel. The CellLogs stay off unless either the charger/supply is connected, and on, or if the bike's controller is on. In the case of the latter, the +5V on the throttle connector will be on. If the logic sees this is on, it turns on the CellLogs. Each board also has a set of jumper pads that can be used with a switch, if manual operation is desired.

The charge process is pretty much the same as with the full Zephyr BMS. The combination of the CellLog-generated HVC signal, and the shunt circuits will allow the low cells to catch up/balance. Whenever a shunt is active, the LED for that channel will be on. The EOC and auto-shutoff functions work the same way they do on the Zephyr board.

I'm going to submit this tomorrow, to get a small run done. We want to try building a couple, mainly to verify the layouts, etc. I may look for a couple of volunteers, with younger eyes, to help in this test phase. :) I'll post something here, if that is going to happen.

More later...

-- Gary
 
Looks good - Better to use a decent shunt setup rather than the battery medics. Have you looked at the LTC4070's yet? Or too small for your average hobbyist to assemble.

A schematic might help check that the board is correct too!
 
This sounds like a great idea! I read through your other bms thread, and wile I'm sure (with good instructions) I could build your bms,this sure sounds like a much more simple solution.
 
heathyoung said:
Looks good - Better to use a decent shunt setup rather than the battery medics. Have you looked at the LTC4070's yet? Or too small for your average hobbyist to assemble.

A schematic might help check that the board is correct too!

I had a glance at the LTC4070 and 4071 parts, but It wasn't clear to me how this would be a lot better than just using the 431. I'll have to take a longer look. For the full Zephyr-sized BMS, Richard and I are still looking at options for a surface mount version. I did an equivalent layout, of the existing design, but there are just too many parts, I think. One option we're looking at is using the Seiko chip, which does both LVC and HVC functions. The only problem is that thee's a minimum quantity of about 1000, I think.

-- Gary
 
Pure said:
This sounds like a great idea! I read through your other bms thread, and wile I'm sure (with good instructions) I could build your bms,this sure sounds like a much more simple solution.

Yes, this definitely has a lot fewer parts. The plan is to build a couple by hand, and then look into getting some machine made. There is a related variant I've also been playing around with, and that is to have essentially a "kit" to do a complete plug'n play 12s2p pack, with this same sort of BMS functionality, but minus the CellLogs. All you do is add four 6s-5000 Turnigy/Zippy packs. :) Anyway, it worked out that it would be more cost effective to do the two sets of test boards at the same time. Here's what the PCB will look like:

12-24 Channel BMS Lite-v4.4.3s.png


The portion on the bottom right is the CellLog-based variant. For the 12s2p pack, there is one 3" x 12" board that will run along the bottom of the 12s2p pack. Two sets of 6s-5000 packs sit on top of this board, facing each other. There are balance plug connectors in the middle of the board, for each of the balance plugs. The two "tiki torch"-looking bits are actually 4mm parallel adapters for the main negative and positive connections. For the serial connections, between the 6s packs, the positive of one is simply plugged into the negative of the second one. These 4mm adapters are then bolted to the board, in the appropriate spots, using short standoffs (2-3 washers...). Two 8-gauge main discharge wires are also connected to these adapters/bolts.

The 12 cell circuits are on one end, with the charge controller in the middle. There are two special end plates, on with the main discharge, charger and throttle line connections (for the LVC function...), and a second one with holes for the LEDs. On the connection end plate, there's also a cutout for an 18-pin VAL-U-LOK connector that is mounted on the end of long board. This connector is wired just like the Hyperion 12s harness I've been doing, so that a 1420i can be used to cycle the packs, and/or check the health of the cells. Anyway, the end plates are stuck on the ends of the 6s-5000 packs, and then the whole assembly is covered with shrink wrap. You end up with a very clean-looking pack, that has a full BMS and a minimum number of external connections (main pack +/-, charger/supply +/- and throttle connections for the LVC function...). My intent is to get these boards made as well, and then offer complete kits that include the assembled/tested main board, the custom end plates, the shrink wrap and all the plugs and wires. All you would need are the four 6s-5000 packs. Everything else would be in the kit. You just stick the backs together, using double-sided tape, plug them into the main board, stick on the end plates, and then shrink wrap the whole assembly. :) That's the plan anyway. :roll: :)

-- Gary
 
Sounds awesome, sign me up.(When they are ready) My build plans from now until the new year/spring are such that I'm going to need 5 or 6 bms units. Hence my reluctance in haveing to assymble them all from scratch. This solution is much simpler and should prove to be somewhat more econimacal I saw in another thread that hobby king may not be selling the cell logs anymore. Any chance they may become hard to find?
 
Using cell-logs for a bms is a great idea. They're inexpensive, have proven to be reliable, and allow the users to record and see what's going on with their battery pack.

I love the on and off functionality of being switched on by the charger/controller. Much easier than unplugging.

Most importantly, being able to set the HVC/LVC makes this BMS useable with any battery. Sort of a holy grail IMHO.

I confess I don't understand how the HVC manages to balance the pack based on the HVC setting from the cell log.

Then again - I've been using my life pack daily for a month now with out balancing.

Any ways - on my next bike this was going to be my BMS.


Watt-meter is for monitoring the charge processes. HVC and LVC provided by isolated relays from the cell logs. Balancing done when needed with battery medics.
Looks like your new project is going to make my life a lot simpler!
 
auraslip said:
Using cell-logs for a bms is a great idea. They're inexpensive, have proven to be reliable, and allow the users to record and see what's going on with their battery pack.

I love the on and off functionality of being switched on by the charger/controller. Much easier than unplugging.

This is handled by using an "Enable" bus line that drives one half of an opto circuit for each CellLog circuit. The other half of this opto is used to isolate the alarm outputs of each CellLog into a combined LVC/HVC signal. Anyway, the opto-isolated "Enable" signal drives a simple relay circuit that controls the pin 0 input on the CellLogs.

LiPo CellLog BMS Logic.png

The relays used are small, very low profile versions that fit into a DIP10 footprint.

auraslip said:
Most importantly, being able to set the HVC/LVC makes this BMS useable with any battery. Sort of a holy grail IMHO.

I confess I don't understand how the HVC manages to balance the pack based on the HVC setting from the cell log.

The way it works is that when a cell's voltage reaches the set point of the HVC signal, it will cut the charge current. The shunt, which is set at 4.15V, will come on and pull the cell voltage below the cutoff point. That will cause the HVC signal to go off, and after a 1-2 second built-in delay, the charge current will come back on, and the cycle repeats.

LiPo CellLog BMS Balance Circuit.png

The zener is there primarily to protect against reverse wiring/inadvertent shorts. The zener will fail shorted, and will pop the trace fuse. These are easier to fix/replace than replacing a 431 chip. :)

-- Gary
 
Its a very well though out design, I'm interested in the throttling of the charge current you are doing - is it switchmode (PWM) or linear? I'd be betting its PWM as you don't have enough heatsinking for a linear reg. scheme.
 
I would be interested in testing this, if I can use it for a 66V Lipo pack and if the parts cost isn't too high. I already bought three CellLogs, intending to plug in one 6S4P pack into each. Presumably with this setup I'd only populate the first 18 channels of the board, use all 8 channels of the first two CellLogs and just two channels of the last one?

Edit: Just looked at the schematic. Nevermind, it looks like each section needs to be fully populated. :?
 
julesa said:
I would be interested in testing this, if I can use it for a 66V Lipo pack and if the parts cost isn't too high. I already bought three CellLogs, intending to plug in one 6S4P pack into each. Presumably with this setup I'd only populate the first 18 channels of the board, use all 8 channels of the first two CellLogs and just two channels of the last one?

Edit: Just looked at the schematic. Nevermind, it looks like each section needs to be fully populated. :?

No, you can use it with one 6s block per CellLog. The four diodes (Dx01-Dx04...) are there so that packs with either 5s, 6s, 7s or 8s configurations can be used. You just don't need to use the 4x6-to-3x8 adapter section.

-- Gary
 
heathyoung said:
Its a very well though out design, I'm interested in the throttling of the charge current you are doing - is it switchmode (PWM) or linear? I'd be betting its PWM as you don't have enough heatsinking for a linear reg. scheme.

It is the same as the full Zepher BMS control section. I'm not sure how to describe it but it is definitely not linear. The FETs are snapped on and off, like PWM, but at a really low rate. Basically, the current is cutoff when a cell reaches HVC, which is set a bit higher than the shunt turn on point, and stays off for about 1-1.5 seconds. The shunt pulls the voltage back down during this period. The current is then switched back on, and the cycle repeats. Eventually the cells are close enough that the shunts themselves will keep the voltage for the high cells in check, without the HVC tripping for any of the channels. At this point, as the low cells finally get full, the current will drop down below the set point for the EOC shutdown, and the charge current will then be cutoff.

-- Gary
 
Gary,
Do you want a few bucks to defray the costs of the test run?
 
julesa said:
Gary,
Do you want a few bucks to defray the costs of the test run?

No, but thanks for the offer. :)

What I'm going to do is a short run of the surface mount CellLog boards-only. I can't afford, right now, to order a bunch of surface mount parts right now, so what I will do is pick some people who would like to build/test one of these, and then I'll send them a board, along with the BOM file. The number will be flexible, but probably between 2 and 6. Jules, it sounds like you would be one, so there's room for 1-5 more. PM me if you are interested. I will cover the costs, including shipping, unless it is international. In that case, I might ask to split the shipping charge. Anyway, all I ask is that you try to build one of these, in a reasonable amount of time, and report the results here, good or bad.

-- Gary
 
Update...

Taking a page from methods' SMT-based LVC/HVC board, I took a look at reducing the parts count of the BMS Lite cell circuits by using resistor arrays instead of individual through-hole resistors. This allows for a significant reduction in build time, as you don't have all those resistor leads to bend and cutoff. This makes for a compact layout as well, close to the size of the surface mount variant.

Instead of the relay-based enable function this latest CellLog layout uses a single 3pst switch to enable up to three CellLogs. The alarm outputs are now used in the normally closed mode and are connected in series. This output is used to drive a small relay that has separate connections for the LVC.

This board, which actually becomes the top lid for a 1.2" x 3.07" by 6.3" Hammond box, also contains the BMS Lite cell circuits. The parts mount from the bottom of the board, with the CellLogs sitting on top. They plug into a 90 degree header receptacle. LEDs for the shunt circuits are mounted above each of the CellLogs. This same setup will also support a two-CellLog version, by simply cutting off the top portions of the boards. The trimmed boards will fit the next size down Hammond box (1.2" x 3.07" x 4.7").



The board on the left contains the charge controller section, plus connections for board-mounted 90-degree VAL-U-LOK connectors, 2 x 14 pin for 24 channels, and one 18-pin for 16-channels. The two-CellLog version can handle 12s and 16s setups, and the three-CellLog variant will work for 18s, 20s and 24s configurations. On the right are four 6s4p parallel adapters and two 8s4p adapters. These can be embedded in with the packs, with leads coming out with the matching VAL-U-LOK plugs.

The charge control section is basically the same as the one used with the full Zephyr BMS board except for one change. Instead of cutting power to everything, including the main LED, at the end-of-charge (EOC...), what will now happen is that the LED will stay on green and individual shunt LEDs will be off, at the EOC. During a normal charge, the main LED will be orange, during the CC phase, and will turn green at the CV point. The LEDs for the high cells will on whenever the shunt is active. For out-of-balance conditions where the shunts themselves can't hold the cell voltages for the high cells, the HVC will trip, and the charge current will be cut for a short period. When this happens, the main LED will blink between red and green, as the charge current cycles between on and off. Once the cells are close enough that the shunts can hold the cell voltages without tripping the HVC, the main LED will be green. Once all the red shunt LEDs go off, the cells are balanced.

The plan is to maybe try and offer assembled and tested versions. As usual, for those prefering to go the DIY route, a PCB-only version will also be available. We may also do full kits and maybe partially assembled versions. I'm not sure about pricing yet because I haven't figured out the costs yet. I'll post testing progress here, but this should be pretty straightforward. I've done a number of these CellLog variants in the past. Anyway, please don't ask when they will be available. I will post that here, after we've done the testing.

Later I will post more pics and the schematics.

-- Gary
 
There is a 200ma shunt regulator version - AP431 made by Diodes Inc (albeit for a somewhat higher price).
BTW am I correct that 431 cathode can not go below the reference input ?
 
I've been reading through this (and alot of other threads) and trying to understand (i guess I could call myself an educated noob on the subject). I want to build a 24s3p pack and fire safety for me is a great concern, thats why I think a bms like this is just the thing I need, the problem is, I have no knowledge in electronics (although I could manage just about anything with proper instructions). I have 2 questions
1- You say it is for 6s4p, can I do 3p instead?
2- when the kit is completed, how would i hook it up for a 24s3p pack?

I still have a lot to learn and i'm working very hard at it, any help would be appreciated :D
 
curious said:
There is a 200ma shunt regulator version - AP431 made by Diodes Inc (albeit for a somewhat higher price).
BTW am I correct that 431 cathode can not go below the reference input ?

Interesting. I'm going to take a look at those. We're running at around 140mA now, and this keeps things nice and cool. At about 150mA, the FAN431s start getting a bit toasty. At 200mA, there will be more heat generated than I would be comfortable with for a board buried in with the packs, like the new 12sp kit board I'm working on in parallel with this effort. Since the circuits for this CellLog variant are designed to go in a box, 200mA should be fine. Worth taking a look at, for sure. Thanks. :)

We use a divider on the input side, so we can dial in exactly when we need the shunts to come on. Anything below this point, even down to zero has never been an issue with these.

-- Gary
 
fractal said:
I've been reading through this (and alot of other threads) and trying to understand (i guess I could call myself an educated noob on the subject). I want to build a 24s3p pack and fire safety for me is a great concern, thats why I think a bms like this is just the thing I need, the problem is, I have no knowledge in electronics (although I could manage just about anything with proper instructions). I have 2 questions
1- You say it is for 6s4p, can I do 3p instead?
2- when the kit is completed, how would i hook it up for a 24s3p pack?

I still have a lot to learn and i'm working very hard at it, any help would be appreciated :D

Yes, the parallel adapters can be used with pack configurations from 2p to 4p. Actually, you could use two adapters and go up to 8p, if needed. These adapters would go inside your pack assembly, close to the 6s packs. They have to be close as the balance plugs coming from the packs are only a few inches long. The output of each parallel adapter is a single set of 7 wires. These wires are typically longer, as they have to go from the packs to wherever the CellLog box is located, so I like to use heavier gauge wire, like AWG 18. To make it easier to build these longer 7-wire 18-gauge cable sets, I put pads on the ends of the parallel adapter boards that are sized to fit MTA156 header connectors. I like using these because the matching plug for these doesn't require adding pins to the ends of the wire, and the wire doesn't have to be stripped first. The ends are simply pushed into the connector using a simple T-bar tool, or even a small screwdriver. The other ends of these 7-wire 18-gauge wire sets go into a VAL-U-LOK 4.2mm PE Series free-hanging plug. The end of the charge control board has matching connectors that will stick out of the end of the box, through a cutout in the custom end plate.

With the full-up, assembled version of this unit, I will have an option for pre-made 18" long cables sets with the MTA connectors on one end and the VAL-U-LOK plugs on the other. For the full three-CellLog version, there are two 14-pin VAL-U-LOK connectors on the board, so this would include two cable/harnesses, each having one 14-pin VAL-U-LOK plug on one end and two 7-pin MTA156 connectors on the other. For a two-CellLog variant that uses two 8s packs in a 16s configuration, there are two 8s4p parallel adapters, with 9-pin MTA156 connectors, and a single 18-pin VAL-U-LOK plug on the other. A 24s configuration, made using 8s packs, could also be supported by using a special harness that converts the two 14-pin plugs into three 9-wire cables with 9-pin MTA156 plugs.

This will all be a lot clearer when I get some pics done and posted. There will also be detailed connection drawings.

-- Gary
 
GGoodrum said:
curious said:
There is a 200ma shunt regulator version - AP431 made by Diodes Inc (albeit for a somewhat higher price).
BTW am I correct that 431 cathode can not go below the reference input ?

We use a divider on the input side, so we can dial in exactly when we need the shunts to come on. Anything below this point, even down to zero has never been an issue with these.
-- Gary

What I was wondering is what is the voltage at the 431 output when its reference input is held above the threshold? Is it equal to 2.5V (reference) or does go to near-zero like on TC54xxx ? This determines value of load resistor and power dissipation split.

BTW if you worry about individual package dissipation - you can use SOT89 SMT package, these can flow quite a bit of heat to the board. For aggregate power I think hard-cased cylindrical batteries like a123 can sink the heat quite well, lipo packs - not so sure. In general I am not decided myself if the balancer part belongs to the pack or to the charger.
 
GGoodrum said:
fractal said:
I've been reading through this (and alot of other threads) and trying to understand (i guess I could call myself an educated noob on the subject). I want to build a 24s3p pack and fire safety for me is a great concern, thats why I think a bms like this is just the thing I need, the problem is, I have no knowledge in electronics (although I could manage just about anything with proper instructions). I have 2 questions
1- You say it is for 6s4p, can I do 3p instead?
2- when the kit is completed, how would i hook it up for a 24s3p pack?

I still have a lot to learn and i'm working very hard at it, any help would be appreciated :D

Yes, the parallel adapters can be used with pack configurations from 2p to 4p. Actually, you could use two adapters and go up to 8p, if needed. These adapters would go inside your pack assembly, close to the 6s packs. They have to be close as the balance plugs coming from the packs are only a few inches long. The output of each parallel adapter is a single set of 7 wires. These wires are typically longer, as they have to go from the packs to wherever the CellLog box is located, so I like to use heavier gauge wire, like AWG 18. To make it easier to build these longer 7-wire 18-gauge cable sets, I put pads on the ends of the parallel adapter boards that are sized to fit MTA156 header connectors. I like using these because the matching plug for these doesn't require adding pins to the ends of the wire, and the wire doesn't have to be stripped first. The ends are simply pushed into the connector using a simple T-bar tool, or even a small screwdriver. The other ends of these 7-wire 18-gauge wire sets go into a VAL-U-LOK 4.2mm PE Series free-hanging plug. The end of the charge control board has matching connectors that will stick out of the end of the box, through a cutout in the custom end plate.

With the full-up, assembled version of this unit, I will have an option for pre-made 18" long cables sets with the MTA connectors on one end and the VAL-U-LOK plugs on the other. For the full three-CellLog version, there are two 14-pin VAL-U-LOK connectors on the board, so this would include two cable/harnesses, each having one 14-pin VAL-U-LOK plug on one end and two 7-pin MTA156 connectors on the other. For a two-CellLog variant that uses two 8s packs in a 16s configuration, there are two 8s4p parallel adapters, with 9-pin MTA156 connectors, and a single 18-pin VAL-U-LOK plug on the other. A 24s configuration, made using 8s packs, could also be supported by using a special harness that converts the two 14-pin plugs into three 9-wire cables with 9-pin MTA156 plugs.

This will all be a lot clearer when I get some pics done and posted. There will also be detailed connection drawings.

-- Gary

Gary, thanks for your help!!! it is greatly appreciated :D
 
Anyway, please don't ask when they will be available. I will post that here, after we've done the testing.

:D

Can't wait, sounds perfect. I love the functionality of being able to enable all three of the cell logs with one switch. Or even better, Using a stdp relay with a key switch to activate the controller ignition wire AND the cell logs.

Although, then I'd have to leave the bike on to charge. I suppose I could use two separate relays. One that activates the cell-logs and controller from the keyswitch, and second relay that activates the cell-logs when the bike is connected to the charger.
 
Back
Top