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

Batteries, Chargers, and Battery Management Systems.
ejonesss   10 kW

10 kW
Posts: 698
Joined: Aug 31 2008 4:21pm

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

Post by ejonesss » Oct 01 2010 5:38pm

dead bugging is a way to prototype without a board.

having a chip dangling with wires poked into the socket i think qualifies as dead bug.

julesa wrote:
ejonesss wrote:looks like dead bugging.
He's not dead, 'e's just resting.
stuff for Even Newer 4 to 24-cell Battery Management System (BMS)
32 lifepo4 cells 16s2p 40138 green cells √
16s2p 40138 based pack built assembled (building electrical equiv to $2000 eonyx system.*

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Oct 01 2010 9:12pm

He's not dead, for sure. Love's show tunes, though. :roll: :mrgreen:

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Oct 09 2010 4:39pm

Another update...

I'm going to do a new run of boards on Monday. I'm hopeful these can be a "release candidate". :) This latest variant has a number of "refinements". One addition to the shunt circuits is a 1.1A PTC fuse, which will keep the shunts from exceeding 1A and swamping. We are also moving away from the 3.5mm mini-terminal blocks, to an MTA 156-type header/connector arrangement. This will save assembly time because the 18-gauge wires can be attached to the MTA connector without stripping the wire, using a $15 "T-bar" tool.

We have three different charger/supply setups that need to be supported, a basic MeanWell-type supply, with nothing but a "hiccup" mode for current control, A regular CC/CV charger, which may or may not have a low current/end-of-charge shutoff/float function (like a Soneil...), and a full-featured, "smart" charger that not only has an auto float/shutoff function, but also has a temp sensor input (Elcon/Zivan...). In the case of the latter, all that is required in terms of a charge controller, is a current limited connection straight to the opto bus that interfaces with the temp sensor input. If the HVC trips, it will throttle the charge current.

For a simple MW setup, the charge controller can be the small one we've been testing that mounts to the front of the MW's terminal block. This board would connect directly to theHVC/alarm opto bus. For everything else, either all, or part, of the "standard" charge controller is required, which includes HVC limiting and a low current/end-of-charge shutoff function.

Here's what the new board looks like:
3x8s BMS-v4.2.4f.png
3x8s BMS-v4.2.4f.png (249.57 KiB) Viewed 148 times
This version is designed so that 16 channels fit in the 1.2" x 4.1" b 8.7"-sized Hammond box. The full 24-channel version will fit in two of the 6.3" boxes in an end-to-end configuration. In either "case", there is a custom lid provided that has holes for the LEDs. Not shown is a custom end plate with holes for all the connections.

To control the heat in this"fanless" variant, there is a 1/4" x 3/4" aluminum bar that fits snugly between the bottom of the PCB, which is mounted in the lowest slot. and the bottom of the case. Not only does the bar act as a heatsink, it turns the whole case into a heatsink. This technique, which is quite effective, will save costs, and reduce assembly time.

Anyway, this is about as close as I can come to a "one size fits all" setup for the full BMS.

-- Gary

User avatar
julesa   10 kW

10 kW
Posts: 948
Joined: Jun 23 2008 2:29pm

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

Post by julesa » Oct 10 2010 12:56am

ejonesss wrote:dead bugging is a way to prototype without a board.

having a chip dangling with wires poked into the socket i think qualifies as dead bug.
Perhaps he's just stunned.


rossasaurus   1 mW

1 mW
Posts: 14
Joined: Jul 27 2008 10:12pm
Location: Northern California
Contact:

Re: Testers

Post by rossasaurus » Oct 23 2010 1:13am

Hi guys,
I'd be up for some testing of the V4's.
I'm still using a V2 I assembled.
I haven't followed posts for a while, but I'm putting together an 18-cell Hipower LiFePo pack for which I need a BMS.
I'd cycle it daily during my daylight shop hours, and not at night; our shop policy for any charging.
I'm in the North Bay. How to proceed guys?

thanks,
Ross
eMotors
em = ross at emotors dot biz
EV Retailer; e-Recumbents, e-Cargo Bikes, Mountain-Trikes, e-Scooters, e-Motos, LSV's, Ag e-Vehicles, Components, e-Social Club, e-Asylum;
http://emotors.biz
http://veloverde.net

Indubitably   100 W

100 W
Posts: 125
Joined: Jan 09 2010 9:26pm

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

Post by Indubitably » Nov 05 2010 1:35pm

I'm seeing a lot of new products on the tppacks website, and while I am excited about the fact that the bms finally seems to be coming together, I'm also more confused than ever about exactly which products I will need to make my project go. So, I'd like to restate my previous request for some sort of an "idiot's guide".

For instance, in my case, I'm looking to string together 16 thunder sky batteries in series for a scooter class project, and I'd like something as fool proof and simple as possible. I suppose it be nice if I could at least pry the thing open and tweak a couple of resisters / swap out a couple of components or what ever in order to get it working with a different chemistry if I need to replace the pack at some point, but otherwise I just want something I can strap on to my batteries, plug into a charger, and start charging.

I don't need the guide to be terribly extensive, just something to point me in the right direction, something that will say "oh, so you have this type of battery, and this type of application, well then, you probably want BMS option X".

rossasaurus   1 mW

1 mW
Posts: 14
Joined: Jul 27 2008 10:12pm
Location: Northern California
Contact:

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

Post by rossasaurus » Nov 05 2010 1:49pm

Hi Indub....

You could read this post from the beginning, and you'd end-up with quite an education in BMS's; you'd no longer need an "idiots guide"...
R
EV Retailer; e-Recumbents, e-Cargo Bikes, Mountain-Trikes, e-Scooters, e-Motos, LSV's, Ag e-Vehicles, Components, e-Social Club, e-Asylum;
http://emotors.biz
http://veloverde.net

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Nov 06 2010 5:16pm

I think a "buying guide" will definitely be in order, because it is very confusing, even for Richard and Andy and I. :) Hard to keep them all straight. Anyway, I will do that, once we "fill in" all the options. In the meantime, I will try and clarify things a bit.

There are basically two product "paths" that we've been following. One is the more "traditional" BMS unit that we have been calling the "full BMS". It is the outgrowth of the original v2.x BMS design that we've been doing for some time. Basically, it has three functions, cell level over-discharge protection (i.e. -- LVC...), cell level overcharge protection (HVC...) and cell balancing. The LVC function is implemented via an optoisolated connection to either a controller's brake input, or to the throttle signal directly, to basically have the throttle cut if a cell tries to go too low. This will remove the load, which causes the cell voltage to recover above the cutoff point. This opto signal can also be used to drive an alarm/buzzer, or some other device, to alert the user that something is wrong.

The HVC signal, which can also be an optocoupled output, is used by a Charge Controller to throttle back the charge current so that no one cell voltage is allowed to go over a safe limit. This Charge Controller, which has been an integral part of the BMS PCB, also includes a shutdown function that cuts off the charge current at the end-of-charge. We now also have an option where the charge controller functions can be integrated in with Richard's MeanWell current limiter widget, which adds an adjustable current limiter function to these simple power supplies, turning them into full-fledged CC/CV battery chargers. This combo MW Charge Controller mounts right to the terminal block on the MW, and has a simple connection to the HVC signal coming from the BMS. The CC mode limit is adjustable from about 1-15A, and the low current shutoff feature can be adjusted so that charge power is cut when the current drops below the set point, which has a range of 0-1A.

Balancing is handled via shunt-based bypass circuits on each channel. The shunts are set to come on at the desired "charge to" point, which is about 3.60-3.65V for LiFePO4 cells, and about 4.15-4.20V for LiPo. When a cell reaches this point, it is close to being full, and its voltage will suddenly start rising at a much higher rate. If all the cells reach this point at the same time, the charger's CV mode will kick in, which then starts lowering the charge current. With perfectly balanced cells, they will all arrive at this point at the same time, so this is fine. With onbalanced cells, however, some cells will get full before others have a chance to, so the low cells don't get enough current to finish charging, and end up at a lower state of charge. What the shunts do is "bypass" current around the full cells, so that the low cells can catch up. This is basically the same idea as with the v2.x design, but with two significant differences. The v4.x shunt circuits can handle about twice the bypass current as the v2.x circuits, or about 1A each. The other difference is the basic bypass philosophy. With the old design (and virtually all other BMS shunt circuit designs I'm aware of...), the "charge to" voltage (i.e. -- the CV mode set point...) was set at a point about 1/2-to-1V higher than the sum of the fully charged cell voltages. The HVC signal and the shunts were used in order to keep the cell voltages right at the point each circuit was in full bypass, or in other words, each shunt circuit was cooking away at the end. This generates quite a bit of unnecessary heat, even with the lower shunt currents of the v2.x setups. The other problem is that each cell circuit is consuming the full amount, but there's no way to know if the cell itself is still taking in some, or all of the current, or if the cell is completely full, and the shunt is bypassing all of the current. With the v2.x design, the charge controller section simply waited until all the shunt circuits were active, and then shut things down. The last cell to come on didn't get as full as the rest, though.

With the v4.x design, the "charge to" voltage is set exactly to the sum of the individual cell shunt turn-on points, so for example, with a 16-cell LiFePO4 setup, the charger's CV point is set to 16 x 3.60V, or 57.6V. The shunt bypass circuits only come on when needed this way. If the cells are perfectly balanced, they'd all arrive at the voltage "knee" of 3.60v, where the voltage starts to rise quickly. The charger itself will keep all the cells at 3.60V and the current will start dropping. The shunts don't come on at all. In reality, though, the cells are never perfectly balanced, so some will get to the 3.60V point sooner. With the new design, the shunts will come on just as strong as they needs to in order to keep the voltage at the 3.60V point. This allows the low cells to have as much current as they need to "catch up". The 1A shunt circuits can handle pretty significant imbalance conditions, as long as the cells are still healthy. A dying cell, however, will hit the "full" point way too early, which will cause the shunt circuit to be swamped, or overloaded, as it struggles to keep the cell voltage in check. When it does swamp, the cell voltage will start rising again. To keep the cell from killing itself, the HVC circuit is used as a last stand/safety valve, to tell the charge controller to adjust accordingly. Since the HVC is now being used as a failsafe, "don't go over this point" limit, the set point can be higher, like 3.70V for LiFePO4, and 4.25V for LiPo.

Anyway, v4.x of the "full BMS" has the 1A shunt circuits, the LVC and HVC circuits and options for either the "embedded" charge controller, or the MW Charge Controller board. It is also designed to fit in an extruded aluminum case that ends up doubling as a heatsink. We are also going to have an option for active cutoff, for the LVC function. This would be used to cut the pack power directly, instead of relying on a controller. The circuit is designed to handle peak loads up to about 250A, or so. We are still "configuring" the layout options, but we will have versions available in the popular sizes, like 12, 16, 18 and 24 channels. As always, there will be LiFePO4 and LiPo variants.


The second product "path" we are following came about as we all got more into LiPo-based setups, made from RC hobby packs. These usually come pre-packaged in 5 or 6-cell 5Ah packs, with prewired balancing plugs attached. These balancing plugs made it fairly easy to combine packs and cell connections, which means less wire errors when hooking up a BMS. What we also found using these packs is that they just don't need balancing anywhere near as much as other commonly used chemistries, like LiFePO4. Part of the reason for this is because these cells have extremely high "C" ratings, so we aren't really taking them all that much in a typical ebike application, especially when multiple 5Ah packs are used in parallel. Anyway, what this all led to was a breaking apart of the typical BMS functions into what I have been calling "BMS elements". The three "elements" again are LVC protection, HVC protection and cell balancing. Of the three, only LVC protection absolutely needs to be "pack resident", to allow protection during discharging.

With LiPo-based packs, I found it is easier/better to make the packs compact and somewhat "sealed", so it doesn't make sense to include the heat-generating balancing shunt circuits inside a shrink-wrapped pack, so I started splitting the BMS "elements" apart, and only put the non-heat generating portions in the packs, and made the balancing as an "external" function/unit as balancing only needs to be done once every 5-10 charges anyway. Initially what I did was to make combo boards for the packs that included the LVC and HVC functions, plus a 4p parallel adapter. The charge controller was a separate small box. I then did separate balancer boxes that plugged into the packs when balancing was needed. This worked okay, but I found I always was "Y"-ing the balancer leads so that I could add a CellLog to monitor the balancing process. Again, this worked okay, but the wiring was cumbersome.

Next, I started using the Battery Medics, but found that it took forever to balance a 10-15Ah pack, so I came up with a "booster" that would add more balancing current. This worked quite well, but I found a bit too much variance between Battery Medic units. That, and the fact that nobody else seemed too interested in this combo (I only sold a few units...), caused me to abandon this path.

Next I decided that since the CellLogs already contained both LVC and HVC functions, I could make use of these to do a complete CellLog-based BMS. In theory, this is logical, but in practice it didn't make a whole lot of sense, mainly for the same reasons I went to the "split" elements in the first place. It was also hard to get everything to fit together in one unit. So, what I ended up concluding is that it still makes sense to keep the LVC/parallel adapter boards embedded in the packs, and then use the CellLogs to create the HVC signal, used during charging, and marry these up with the balancer circuits. The idea here is that what stays on the bike are the packs, with LVC boards, and then the balancer/CellLog units are plugged in when it is time to charge/balance. This is where we are right now. There are 6 and 8-channel LVC/parallel adapter boards available, along with a two-CellLog/12-channel HVC/Balancer unit. The MW version of the Charge Controller is also available. Soon I will also make the standalone Charge Controller available, for those with non-MW charging solutions. Finally, I am working on a 6/8-channel unit and single box varinats for 16 and 18 channels.

There's a couple more options that may be available at some point as well. One thing I'm still not completely satisfied about the current setup is that since the HVC function is in the CellLogs, the CellLog/Balancer units have to be connected in order to have the HVC protection during charging. While testing all this stuff, it was always handy to have the CellLogs connected, as it was useful to monitor whatthe cells were doing. When I'm not in the "test" mode, though, it is still sometimes desirable to just throw on the charger, and not worry about balancing, etc. The only way I'd be comfortable doing this, however, is if I still have the HVC protection. For this reason, I may eventually go back to doing combo LVC/HVC/parallel adapter boards in the packs. What I would do is have the set points a bit higher, like 3.73-3.75, for LiFePO4, and 4.28-4.30V for LiPo. The reason for the higher limits is that without the shunts to keep the high cell voltages at the set point, the voltage will rise more. If the cells are fairly well balanced, this higher trip point won't be hit, and if it is being hit, you will know it is time to balance. I'd still keep the lower HVC point in the CellLogs because the shunts will keep the cell voltages down. The nice thing about the CellLogs is that the HVC set point is settable, so it can be tailored to a particular setup. In any case, having a "backup" HVC function, pack resident, will allow simple charges, but still protected.

The second possible new option that might be offered is active cutoff for LVC. The reason for this is to provide absolute protection from any cell be discharged too far, no matter what the load is doing. The opto/"cut-the-throttle" technique is simple and very effective, but it won't stop a cell from draining dead if you leave the controller on and don't ride the bike for a couple months. :roll: What we are looking at is a modified version of a circuit first proposed by a "brief" member, Randomly, back on about page 23 of this thread. Instead of pwer-hungry optos, this circuit uses a "normally connected" ladder-type circuit that only draws about 100uA (.0001A...) while keeping four big FETs turned on, passing the main discharge current on through to the controller. When one of the LVC detectors trip, this ladder chain is broken, and the FETs are slammed off (to minimize heat during the transition...). Once tripped, the current drain on that low cell drops to 1uA (.000001A). Even if the charge state of that cell was down to 100mAh (.1Ah...), it is going to take 100,000 hours, or about 11 years for the cell to be drained dead, at a 1uA rate. Anyway, it is likely in the near future we might have a new LVC/HVC/Parallal Adapter board, and a small active cutoff module that can be used inline with the main pack connections.

Hope this helps clear some of the confusion. :) More soon...

-- Gary

dbaker   100 kW

100 kW
Posts: 1021
Joined: Oct 24 2008 2:32pm

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

Post by dbaker » Nov 06 2010 7:08pm

Great post :D

deVries   1 MW

1 MW
Posts: 1545
Joined: Jul 10 2008 11:29pm
Location: Solar Powered 3000w Texas

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

Post by deVries » Nov 06 2010 10:44pm

GGoodrum wrote:One thing I'm still not completely satisfied about the current setup is that since the HVC function is in the CellLogs, the CellLog/Balancer units have to be connected in order to have the HVC protection during charging. While testing all this stuff, it was always handy to have the CellLogs connected, as it was useful to monitor whatthe cells were doing. When I'm not in the "test" mode, though, it is still sometimes desirable to just throw on the charger, and not worry about balancing, etc.
Great! post! Gary :) especially to understand the development history (& future) of these designs! :mrgreen:

By using the CellLogs in your new design that is for sale now... is the "bulk charge" process being slowed down compared to what you might do with the upcoming "old/new" design that will come next at some point -the HVC/LVC parallel set-up w/o CellLogs?

I actually prefer the idea of charging to just under the full-cell point, i.e. 4.15v LiPo, to extend the lifespan of the batteries. Is there anyway to include that charge voltage adjustment feature in any upcoming design that puts the HVC/LVC back together on-board w/the batteries?

Also, have you or anyone used the CellLogs that can save the charge/discharge data using the Logview software w/usb port? Will these work with your set-up now, and do you think some useful data can be gathered by using the CellLogs with Logview software? :shock:

Thanks for doing a super job! :cool:

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Nov 06 2010 11:54pm

dbaker wrote:Great post :D
Thanks. :)

There's also a couple more special variants we are also working on in parallel with these efforts. Richard is working on a special 40-channel motorcycle BMS version that will end up being potted in an aluminum box. Should be interesting.

Andy and I are going the other direction, and are doing a special 4-channel BMS to be installed in a sealed 12V SLA replacement battery made up by eight 12 Ah PSI cells, connected in a 4s2p 12V/24Ah configuration. The cells use the "lego blocks" to hold together, and fit perfectly in a special ABS plastic case. The BMS board will go in the lid. What makes this unique is that the only outside connections at all are the two large brass battery posts. There's no special charge and/or balanced plugs. Inside will be four "heavy duty" shunt circuits (1-1/2A...) that will take care of balancing, and the low-power LVC "ladder" circuits with a 300A active cutoff circuit. These 12V batteries will be used as true replacements, and can be used in parallel and/or in series with each other. They are sized to be perfect as starter batteries for high performance motorcycles, marine battery application, such as genset starter batteries or even for use with electric trolling motors.

--Gary

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Nov 07 2010 2:39am

deVries wrote:By using the CellLogs in your new design that is for sale now... is the "bulk charge" process being slowed down compared to what you might do with the upcoming "old/new" design that will come next at some point -the HVC/LVC parallel set-up w/o CellLogs?
Let me clear, the two processes are basically identical, whether or not you use the CellLog-based external balancers, and the embedded LVC boards, or if your are using the v4.x "full BMS" unit. The only difference is where the HVC signal is generated. Both designs use the same 1A shunt circuits for balancing, and both make use of the identical Charge Controllers, either the standLone unit, or the MW version.

Both of these designs will actually balance a pack faster than the older v2.x BMS, mainly because it can make use of twice as much balance current, 1A vs. 500mA. The BMS doesn't come into play until the end, so it doesn't really affect anything during the bulk charging/CC phase. The max current the charger/supply can provide determines the overall time it takes to charge..
deVries wrote:I actually prefer the idea of charging to just under the full-cell point, i.e. 4.15v LiPo, to extend the lifespan of the batteries. Is there anyway to include that charge voltage adjustment feature in any upcoming design that puts the HVC/LVC back together on-board w/the batteries?
I'm not sure what you are asking, but the "standard" charge to/balance voltage we've been using is already 4.15V per cell, for Lipo, and 3.60V for LiFePo4. Both are a bit under what the normal voltages are for these. In any case, most all supplies and chargers have adjustable voltage settings, for tweaking the CV mode voltage.
deVries wrote:Also, have you or anyone used the CellLogs that can save the charge/discharge data using the Logview software w/usb port? Will these work with your set-up now, and do you think some useful data can be gathered by using the CellLogs with Logview software?
They are definitely compatible, you just set the record mode in a menu.

-- Gary

deVries   1 MW

1 MW
Posts: 1545
Joined: Jul 10 2008 11:29pm
Location: Solar Powered 3000w Texas

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

Post by deVries » Nov 07 2010 4:31am

GGoodrum wrote:
deVries wrote:I actually prefer the idea of charging to just under the full-cell point, i.e. 4.15v LiPo, to extend the lifespan of the batteries. Is there anyway to include that charge voltage adjustment feature in any upcoming design that puts the HVC/LVC back together on-board w/the batteries?
I'm not sure what you are asking, but the "standard" charge to/balance voltage we've been using is already 4.15V per cell, for Lipo, and 3.60V for LiFePo4. Both are a bit under what the normal voltages are for these. In any case, most all supplies and chargers have adjustable voltage settings, for tweaking the CV mode voltage.
I'm referring to where you wrote this... where you note the higher "top-off" set points below...
The only way I'd be comfortable doing this, however, is if I still have the HVC protection. For this reason, I may eventually go back to doing combo LVC/HVC/parallel adapter boards in the packs. What I would do is have the set points a bit higher, like 3.73-3.75, for LiFePO4, and 4.28-4.30V for LiPo. The reason for the higher limits is that without the shunts to keep the high cell voltages at the set point, the voltage will rise more.
In the bulk charge method you note above the cut-off or top-off point is set higher for cells that top-off "full" first.
GGoodrum wrote:Both of these designs will actually balance a pack faster than the older v2.x BMS, mainly because it can make use of twice as much balance current, 1A vs. 500mA. The BMS doesn't come into play until the end, so it doesn't really affect anything during the bulk charging/CC phase. The max current the charger/supply can provide determines the overall time it takes to charge...
What is the maximum amp charge limit that can be run through your design during the bulk or balance charge, since there has to be some reasonable amperage charge upper limits to prevent overheating and parts damage?

Thanks! :)

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Nov 07 2010 3:10pm

deVries wrote:I'm referring to where you wrote this... where you note the higher "top-off" set points below...
The only way I'd be comfortable doing this, however, is if I still have the HVC protection. For this reason, I may eventually go back to doing combo LVC/HVC/parallel adapter boards in the packs. What I would do is have the set points a bit higher, like 3.73-3.75, for LiFePO4, and 4.28-4.30V for LiPo. The reason for the higher limits is that without the shunts to keep the high cell voltages at the set point, the voltage will rise more.
In the bulk charge method you note above the cut-off or top-off point is set higher for cells that top-off "full" first.
Ah, okay, I see your confusion now. What I was talking about above is the HVC set point, which is not the same as the turn-on voltage for the shunts, which is what ends up being the balance or "charge to" voltage. With the old v2.x design, the HVC set point was very close to the same voltage as the turn-on point for the shunts. That's because the HVC signal was used to limit the cell's voltage to where the shunt was bypassing the max current it could handle (500mA...). It didn't really matter where the charger's voltage was set at as long as it was at least a little above the sum of the HVC set points. The difference between the HVC set point and the turn-on point for the shunts was maybe 10-15mV.

With the v4.x scheme, the charger voltage is set to exactly the sum of desired charge to/balance voltages, so either 3.60V or 4.15V x # of cells in series. The shunt turn-on point is also set to the same 3.60V/4.15V desired balance/charge to point. This way, if the cells are exactly balanced, they will all hit the 3.60V/4.15V point at the same time, and the charger's CV mode will kick in. When the current drops down to the cutoff point, all the cells will be completely full, and not one of the shunts come on at all. With the cells a little out-of-balance, one, or more, will hit the 3.60V/4.15V point first. This will cause the shunt to come on. How strong it comes on will be controlled by how far "ahead" the high cell is from the rest. With the shunt capable of bypassing a full 1A of current, the high cell can be pretty far ahead. :) It is only the case where you have a damaged cell that the shunt will end up overloading, and will let the cell voltage continue to rise. This is where the HVC signal comes back into play. We now simply use the HVC signal to make sure if all else fails (i.e. -- the shunt is unable to keep the cell voltage in check on its own...), the HVC signal will keep the cell from going any higher. It is not critical exactly what this value is, as it is just an absolute failsafe upper limit. During "normal" charges, the HVC never trips at all.

What I've described above applies equally to either the full BMS or CellLog-based variants. The only difference is that with the CellLogs, you can adjust/change where the HVC trip point is set. With the full BMS, you adjust at the time the board is built by selecting an appropriate resistor value. You can still make it whatever you want. With the current test systems, I've been setting the shunts to come on at 3.60V for LiFePO4 and 4.15V for LiPo, and then have the "failsafe" HVC trip points about 100mV higher, so 3.70V/4.25V.

What I was talking about in the reference above is the "special" case with LiPo, which usually don't need to be balanced any where near as often. Because of this, the balancer doesn't have to be connected for every charge. The cells will still not be perfectly balanced, though, so some are still going to hit the 4.15-4.20V "knee", where the voltage starts rising at a faster rate. When this happens, the charger's CV mode is still going to kick in but the difference is the "slower" cells are going to be close to the "knee" anyway, so the end result is the cells are still pretty close. What I'm worried about is not normal charges, or even the case where the cells are pretty out-of-whack. I'm concerned about when there is a damaged cell that hits the "knee" way before the rest, which is what happens. In extreme cases the rest of the pack can be empty, and the damaged cell voltage can rise immediately, as soon as the charge current is applied. In order to protect against this, assuming the CellLog balancer is not plugged in, is to add a "backup" HVC function to the LVC board that gets embedded with the pack. Again, the actual HVC trip point value is not critical. My though at making it a bit above what the CellLog HVC is set to is simply to give the cells a bit more "headroom" because the shunts are not connected, so they won't be working to keep the voltages in check.
deVries wrote:What is the maximum amp charge limit that can be run through your design during the bulk or balance charge, since there has to be some reasonable amperage charge upper limits to prevent overheating and parts damage?
There's no real limit in the cell shunt circuits because they don't really come into play until the end, and then they will only pass 1A, or a bit over if swamped, but the HVC "failsafe" will keep this in check. The rest of the circuit draws literally microamps.

Where charge current limits come into play is with the charge controllers, and even then it is more applicable for the standalone version. For those, there is a latching relay that is used. For the true standalone Charge Controller, that would be used with CellLog-based variants and standard CC/CV chargers/supplies, it is mounted in a small extruded aluminum box. In this case the latching relay used has a continuous rating of 16A. For the full BMS, a larger 60A latching relay can be used, as the charge controller is part of the BMS PCB, For the MW version of the charge controller, no relay is needed, so the limit is really controlled by the max continuous power rating of the supply.

-- Gary

deVries   1 MW

1 MW
Posts: 1545
Joined: Jul 10 2008 11:29pm
Location: Solar Powered 3000w Texas

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

Post by deVries » Nov 07 2010 3:51pm

GGoodrum wrote:I'm concerned about when there is a damaged cell that hits the "knee" way before the rest, which is what happens. In extreme cases the rest of the pack can be empty, and the damaged cell voltage can rise immediately, as soon as the charge current is applied. In order to protect against this, assuming the CellLog balancer is not plugged in, is to add a "backup" HVC function to the LVC board that gets embedded with the pack. Again, the actual HVC trip point value is not critical. My though at making it a bit above what the CellLog HVC is set to is simply to give the cells a bit more "headroom" because the shunts are not connected, so they won't be working to keep the voltages in check.
Ok, just one last check to make sure I understand this with your current CellLog set-up. Using the CellLogs now is set-up to work with an LVC board only... so, in order to protect against the worst case scenario with a bad cell going high and damaging or possibly worse, frying the whole battery pack, the present set-up requires the CellLogs to be used. The CellLogs now handle the HVC function, and there is no fail-safe or back-up HVC protection.

Only when you have the HVC/LVC parallel boards will there be this back-up, so one can bulk charge without having to use the CellLogs too.

Correct or ? :mrgreen:

Thanks! :)

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Nov 07 2010 6:22pm

deVries wrote:Ok, just one last check to make sure I understand this with your current CellLog set-up. Using the CellLogs now is set-up to work with an LVC board only... so, in order to protect against the worst case scenario with a bad cell going high and damaging or possibly worse, frying the whole battery pack, the present set-up requires the CellLogs to be used. The CellLogs now handle the HVC function, and there is no fail-safe or back-up HVC protection.

Only when you have the HVC/LVC parallel boards will there be this back-up, so one can bulk charge without having to use the CellLogs too.

Correct or ?
Just to be clear, the CellLog-based balancer can be used with a pack whether or not this pack has an LVC board. The LVC function is completely independent, and only affects the discharge of the pack. In order to have the HVC protection, a charge controller must be used, but can be either of the MW or standalone versions.

And, yes, you are correct, to get HVC protection without using the CellLog-based balancer you need to use either one of the charge controllers with a combined LVC/HVC board. The good news is that there won't have to be a wait for these. I'm turning in the following for a run tomorrow:
4x6s Active Cutoff LVC-HVC-v4.3.0k.png
The section on the left is the active cutoff module. With four IRFB4110 FETs, it is good for setups up to 24s LiPo, or 100V, and about 200-250A discharge peaks.

-- Gary

wookey   10 W

10 W
Posts: 83
Joined: Jul 13 2009 8:22am

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

Post by wookey » Nov 09 2010 9:41pm

A couple of queries about the 'HVC unit only connected at charge time' concept:

Does anything bad happen if you fit it to the bike and leave it there all the time? (e.g. are the cellogs still discharging two of the cells all the time, which would be bad for a bike left unused for too long)? Or does something isolate them when charging is terminated? I tried to work this out from the product description but failed.

If you fit it only at charge time then you need to bring out a 20-way connector rather than the 3-way(4way?) one needed for charging plus meanwell control, right? That seems like a definate disadvantage to me.

I'd prefer to leave the HVC/balancer unit connected permanently as I have a charger at work and another at home, and will have to carry it back and forth anyway. Might as well just live wired into the bike rather than flapping about loose in the top-box every day. But if it adds significant standby drain current then it needs to be removed when the bike is not used for a while.

It sounds like I really want the 'Full BMS' option, but that seems to be months away still, and my poor batts are suffering.

User avatar
fechter   100 GW

100 GW
Posts: 12766
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

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

Post by fechter » Nov 09 2010 10:03pm

The CellLog versions have a relay that cuts the power to the CellLogs when on standby. There is still some drain, but it is very low and you could probably keep it connected for 6 months or more if the battery started out charged. The CellLogs are enabled when charging or when the motor controller is turned on.

The full BMS version has a bit lower drain, but still puts some drain on the cells, around 200uA or so depending on voltage.
"One test is worth a thousand opinions"

deVries   1 MW

1 MW
Posts: 1545
Joined: Jul 10 2008 11:29pm
Location: Solar Powered 3000w Texas

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

Post by deVries » Nov 09 2010 10:45pm

GGoodrum wrote: And, yes, you are correct, to get HVC protection without using the CellLog-based balancer you need to use either one of the charge controllers with a combined LVC/HVC board. The good news is that there won't have to be a wait for these. I'm turning in the following for a run tomorrow:
http://endless-sphere.com/forums/download/file.php?id=48628&mode=view
The section on the left is the active cutoff module. With four IRFB4110 FETs, it is good for setups up to 24s LiPo, or 100V, and about 200-250A discharge peaks.
1) What will one need to parallel sub-packs, what parallel boards or ?, so one could do 12s3p without the Cell-Logs using this board instead? Use the same parallel LVC boards you have now -just w/o the CellLog connected?

2) And will you make smaller channel sizes in this version -12s/16s? Can the LVC voltage be tweaked similar to using the LVC boards for CellLogs -such as 3.2v or 3.3v or ?

Thanks! :mrgreen:

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Nov 10 2010 2:41am

wookey wrote:A couple of queries about the 'HVC unit only connected at charge time' concept:

Does anything bad happen if you fit it to the bike and leave it there all the time? (e.g. are the cellogs still discharging two of the cells all the time, which would be bad for a bike left unused for too long)? Or does something isolate them when charging is terminated? I tried to work this out from the product description but failed.
At one point, I did have a version that had relay-based logic for automatically turning the CellLogs on and off, based on whether the controller was on, or if the charge current is present, but frankly it was just too much to cram into one box. That's why these are called balancers and not a full BMS. Down the road, I might revisit this, but for my own setups, I couldn't see permanently mounting the whole unit to the bike, just to get the LVC function, which I have embedded in the packs already.
wookey wrote:If you fit it only at charge time then you need to bring out a 20-way connector rather than the 3-way(4way?) one needed for charging plus meanwell control, right? That seems like a definate disadvantage to me.

I'd prefer to leave the HVC/balancer unit connected permanently as I have a charger at work and another at home, and will have to carry it back and forth anyway. Might as well just live wired into the bike rather than flapping about loose in the top-box every day. But if it adds significant standby drain current then it needs to be removed when the bike is not used for a while.

It sounds like I really want the 'Full BMS' option, but that seems to be months away still, and my poor batts are suffering.
There is a "3rd" option that might suit your setup a bit more, and that's to use the new combo LVC/HVC/parallel adapter boards that I'm having made right now. As I said above, the reason I've decided to do another LVC/HVC board is so that you don't have to hook up the balancer if the cells aren't far enough out of whack yet. This would satisfy your "3-wire" charge desire. You'd still need a multi-pin connector for the balancer, but it only has to be connected whenever balancing is actually required. The rest of the time you just hook power and the HVC signal.

-- Gary

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Nov 10 2010 3:00am

deVries wrote:1) What will one need to parallel sub-packs, what parallel boards or ?, so one could do 12s3p without the Cell-Logs using this board instead? Use the same parallel LVC boards you have now -just w/o the CellLog connected?
Both the LVC-only, and the new LVC/HVC boards also have a 4p parallel adapter on the same board. This allows you to parallel up to four packs, at the cell level.
deVries wrote:2) And will you make smaller channel sizes in this version -12s/16s? Can the LVC voltage be tweaked similar to using the LVC boards for CellLogs -such as 3.2v or 3.3v or ?
The balancers will come in several variants, 6/8s, 12s, 16s and 18s. I submitted a new run of boards for all of these variants except for the 16s version, which I'm still finishing up. The LVC/HVC boards have four "tearoff" sections, that are 6-channels each, and a 5th section for the active cutoff function. These boards can be left connected, or can be separated. If separated, they are "daisy-chained" together, and connected to the active cutoff module.

-- Gary

NorCalTuna   1 W

1 W
Posts: 58
Joined: Mar 24 2010 2:52am

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

Post by NorCalTuna » Nov 10 2010 6:23am

I just wanted to chime in here with a voucher for gary's excellent kit quality. Part count exact (hey, a spare 1/8w resistor and a zener might be nice if there's a fault when testing...) and the boards are awesome, they came together perfectly.

I had all 24 channels of my LVC setup stuffed and soldered (and stuffed and soldered, etc...) in about an hour and a half- testing is on hold until after I perform a pair of Bach pieces in guitar tomorrow! I think it's 374 solder joints, for those keeping score. I see the rest of the stuff is up, are the kits now available?

Is my nerdmobile][ about to get finished? Will I have to breathe heavily on my fabricator this week? Stay Tuned!

User avatar
GGoodrum   100 MW

100 MW
Posts: 2908
Joined: Sep 20 2007 2:03pm
Location: South Orange County, CA
Contact:

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

Post by GGoodrum » Nov 10 2010 5:52pm

Okay, I think we are set on the balancers. I have orders in now for 6/8s, 12s, 16s, and 18s units. Here's what the PCBs, lids and end plates look like:
8s CellLog CMS-Balancer-v4.2.6.png
12s CellLog CMS-Balancer-v4.2.6.png
16s CellLog CMS-Balancer-v4.2.6.png
18s CellLog CMS-Balancer-v4.2.6.png
The 8, 12 and 18-channel boards will be here Friday, and the 16-channel variant on Monday. In the meantime, I will work on trying to do new BOMs, and see what I can do with the instructions.

I also have the new LVC/HVC/parallel adapter boards, with the active cutoff module, coming on Friday as well.

-- Gary

deVries   1 MW

1 MW
Posts: 1545
Joined: Jul 10 2008 11:29pm
Location: Solar Powered 3000w Texas

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

Post by deVries » Nov 10 2010 6:24pm

GGoodrum wrote:I also have the new LVC/HVC/parallel adapter boards, with the active cutoff module, coming on Friday as well.
Super! :)

What can the LVC LiPo voltage(s) be set at w/o using the CellLogs? 3.3v or 3.4 or ?

Thanks :mrgreen:

User avatar
Spacey   100 kW

100 kW
Posts: 1463
Joined: May 15 2010 9:38am
Location: United Kindom
Contact:

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

Post by Spacey » Nov 12 2010 6:47pm

So what's a guy got to do to get one of these (or perhaps many more) for a 16cell Lifepo4 Headway pack and how much do they cost :) ?
http://www.custom-ebikes.com is my new Ebike company serving the Uk and Europe
Free Uk shipping over £49.99

Post Reply