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

[dnmun]

going back to your original argument that the over discharge would happen during an event when the battery is not supervised, i think that was a very valid argument, but still think the damage could be minimal if the discharge current is in the milliamp range. but i am not an expert so i cannot be certain of that especially if the drawdown is significant.

however i do think that 300mV of sag or internal resistance beyond your 3V cutoff selection would not imply that the cell was overdischarged or reversed at that point in time, and that an immediate cessation of current draw would save the cell from damage.

again, this is all my opinion, not meant to diss you or your concerns. it would be neat if there was more knowledge from rcgroups.com about lipo over discharge here on ES as well. maybe some people who spend a lot of time there like gary or holmes or anyone else could drag in a good link to provide people an opportunity to think it through themselves.

i agree it is best to be safe than sorry however, especially where the cells are so expensive and can be totally ruined by a single over discharge event. everyone will have an opinion, and maybe if those who have ruined cells from over discharge could recount how it happened, or who actually have a good grasp of the dynamics in lipo over discharge could explain more, it would be less subject to anxiety and more based on real experiences from those who have already lost cells this way.

but i think you are right to be very careful when getting down there, and maybe the CA could give you that 3V warning, and let you leave the 2.7V part in the BMS.

really, not intended to diss method's argument, just to discuss. he actually appears to be the one pushing the power envelope for this BMS to the max and i think will be the touchstone everybody will have to follow whether with lipo or lifepo4. 24 cells is the new 48V because of this board development, imo.

 

[methods]

That would be great if someone could bring some of the RCGroups super lipo experience over here. I pretty much just regurgitate what I have learned and am by no means a Lipo expert. I even thought that you still had to charge at 1C till a month ago

Yep, until I am convinced otherwise, I am going to stick with the 3V rule. There is one particular guy on RCGroups that I have been following for several years (has some sort of confederate flag as his avatar) and he seems to be the authority on lipo. . . He is super conservative and even warms up his packs before flying and never uses more than something like 60% capacity. Of course he has like 2000 cycles on his packs too.

Argument (the logical type, not the hostile type) is always good. I always invite someone to argue a point, especially if I may be wrong. It is better to be corrected and embarrassed than ignorant and wrong.

If that person shows up from RCGroups, lets start a thread on it.

-methods

EDIT: Here he is: EveryDayFlyer http://www.rcgroups.com/forums/member.php?u=44974. Lipo is this guys life. Search through his posts, he is truly tune in and has been since the beginning. I think he is a little too conservative, but hey, others think I am too conservative. Anyway, this guy has graphs and charts and facts and tons of real life information.

 

[GGoodrum]

I probably have as much experience as anybody on RCGroups with LiPos (I have about 5,500 posts over there...), and I originally started my site (http://www.TPpacks.com) as a favor to Charlie Wang, to have a way to sell his original, 1st generation, Thunder Power LiPo packs. This was back in 2003.

Anyway, from my experience, LiPos act prretty much the same as our favorite LiFePO4 cells, in that the higher the "C" rating, the better the cells will hold up the voltage under load. At the end of the capacity, the cell voltage will drop quickly. With the latest 30C cells, I've seen a 5Ah pack hold 3.5V under 80A loads, but once the end is reached,it will drop fast. With an RC plane the loads are typically much lower, but usually more constant, then say, an electric helicopter, doing full-on 3D manuevers. Because of the constant load, but at a lower current, it is important to have the cutoff be at 3.0V, or a bit higher. This is because you can very easily fly a plane until the cells die. With a helicopter it is a bit different. The loads are stronger, but not continuous at all. You might have huge peak currents (100A+...), but they come at a frenetic pace. The average current used can be as little as 1/10th what the peaks will typically be, even in general "sport" flying profiles. With all but the highest C-rated packs, this can mean that the voltage drop a pack/cell might see could very well be a number that is below 3.0V per cell, so in a lot of cases, we use to disable the LVC cutoff. One reason this was done is that the "resting" voltage of a cell/pack needed to be significantly above about 3.3V per cell for the helicopter to be able to lift off the ground and maintain a hover. It worked out to be a self-governed LVC function. Even when we started flying a123 packs, we could always just fly until the power dropped, and then you had about 20-30 seconds to decide where you were going to land/auto-rotate. The resting voltage of the cells where still well above 3.0V per cell. I didn't actually kill my first a123 cell until I put it on an ebike. After I did kill the first ones, I started on my quest for an effective LVC circuit, which lead me to Bob Mcree.

What I found is that at the end of capacity, if the cells were all fairly well balanced, they would all dump at once. Usually, this happenen to me when I had the throttle wide open, going up a hill. The first time it happened, I thought the controller blew. It was like a kill switch was hit. In any case, if you immediately back off the power, the resting voltage of the cells will recover to about 3.10-3.15V per cell. With the early LVC circuits, I used the 2.7V version of the TC54s. With the circuits wired into the controller brake line, it would "hit" the throttle, like a big tuna, when one of the channels tripped. Power would come right back, but it would hit again in seconds, if I didn't let off the throttle. If I backed off the throttle, I found I could go about another 1-2 miles, and then even the slightest throttle would cause a constant "oscillation" at about a 2Hz rate. Later, I did LVC boards that I used with the LiFeBatt packs. For those, I worried that because the LiFeBatt cells had a much lower C-rating than the a123s (10-12C vs 30-50C...), they might hit the 2.7V value under load, too soon, so I switched and started using the 2.1V parts. My results were very similar. At the point of the first "tuna strike", if I stopped, the pack took back in about 90% (about 9050-9150 mAh...). If I continued until I got the max oscillation, with the slightest throttle, the packs would take their full rated capacity (9900-10100 mAh...) back in. Even with these drained packs, however, the resting voltage for each cell was always well above 3.0V. Later, I tried using LVC boards with the 2.1V parts on some of my a123 packs, and I couldn't tell any difference. Now, I just use the 2.1V chips for all LiFePO4-based packs.

What you really need to do is detect when the cell is right at that end of capacity, when the cell's voltage under load, even a moderate one, is starting to fall off. As long as you catch it before the cell gets so drained that they become damaged, I don't think it is all that critical whether you are looking for 2.1 or 2.7V, or whatever. The lower the C-rating, the less important the lower value number needs to be. That's because lower C-rated cells will have higher voltage drops, under load, so they are more likely to hit the cutoff first.

The bottomline is that I think that with TS cells, using the 2.1V parts is fine. Also, with "average" LiPo packs (5-10C...), I think using the 2.7V parts will work fine. I'd only worry about it if you were using the high-power, 30C+ variants that are the current state-of-the-art. In that case, I'd probably try and make the 3.0V SMT versions work, or add a voltage divider to the front end to dial in a higher cutoff point.

-- Gary

 

[ErikK]

Wow, I leave the computer for a couple days and miss my chance. O well, I'll wait to order in the next batch. Would have been happy to place a back order now, if the site would let me.

Great work on this project! Thanks!

 

[methods]

........ I originally started my site (http://www.TPpacks.com) as a favor to Charlie Wang, to have a way to sell his original, 1st generation, Thunder Power LiPo packs. This was back in 2003.

...

The bottomline is that I think that with TS cells, using the 2.1V parts is fine. Also, with "average" LiPo packs (5-10C...), I think using the 2.7V parts will work fine. I'd only worry about it if you were using the high-power, 30C+ variants that are the current state-of-the-art. In that case, I'd probably try and make the 3.0V SMT versions work, or add a voltage divider to the front end to dial in a higher cutoff point.

-- Gary

Wow, back in the day... I loved those Thunder Power cells. I am still running my original 1320 and 480 packs. Indestructible! I used to swear by then until I discovered Hobby City. The Zippy R and Zippy H packs perform amazingly well for a fraction of the cost. They are heavy, but who cares about an extra 28 grams on an ebike

Thanks for the knowledge base.
I am not running the state of the art but I am running 25C Chinese cells. It pains me but. . . I guess I will try the 2.7V parts. I suppose that it is a valid argument that I can let the CA or the Kelly cut off earlier if I am worried about LVC.

It is very hard to unlearn rules of thumb and "good practices". . .

-methods

 

[GGoodrum]

Wow, I leave the computer for a couple days and miss my chance. O well, I'll wait to order in the next batch. Would have been happy to place a back order now, if the site would let me.

Great work on this project! Thanks!

Actually, you haven't missed anything yet. . I ordered a second batch of boards last week.They should be here by Thursday or Friday. These continue to be available, but this next batch won't ship until Friday.

-- Gary

 

[dnmun]

i'll double up on the thanks for the "knowledge base" i had seen a few pages in xyster's design thread from last year where bob was talking about the LVC idea, but never knew how it got started.

what little i have read about the factors that determine longevity in lifepo4 cells is the concept of something resembling super cooling in liquids which causes the lithium to precipitate in the solution adjacent the anode, but not on it or in contact with the lithium in the anode. as these globs of lithium condense out of solution adjacent the anode, they grow to sizes that tear apart the crystalline structure of the anode and by doing so reduce the continuity of the anode and reduce it's conductivity, or increase the internal resistance. or that's the way i interpreted it.

the idea behind the new silicon nanotube anode structure that is covered in the stanford patent involves the idea that the nanotubes form in such a way that the anode can be very large in area extent because of their filamentous structure, but are not damaged by being 'hung up' on these growing globs of lithium as they precipitate out of this 'super cooled' state adjacent the anode, so the anode damage is minor and the areal extent can be large so the anode can conduct large amounts of current.

separately and originally, this is why the goodenough patent behind the a123 cell is so effective because he developed a way to make this anode structure with a large dendritic surface capable of conducting large amounts of current by comparison to simple anode structures.

goodenough's patent involved a specific heating regime to grow the dendrites as i understand, but one of his students has developed a technique to grow structures on the anode using microwave energy. i think that was at arizona technical institute, can't remember now. don knows i bet.

but i think in overdischarge the damage caused to the lipo is by the collapse of these layers of cathode polymer 'plates', which is then difficult to get the lithium ions to flow back into, but with the lifepo4, the crystals of lifepo4 have a rigid but open structure which allows the lithium ions to move in and out of these micropores inside the cathode structure and the structure does not collapse when the lithium ions are stripped out of the cathode structure where they are 'hiding' in these micropores and transported to the anode.

i liked gary's rational behind the new/old lipo and the hiC/hiLVC logic.

methods was right about that guy with the confederate halter top being conservative, he swore on 3.2V-3.3V.

 

[boostjuice]

If you can compile a Generic/Mouser Part no. BOM then from this i can compile a Farnell/RS Part.no. BOM for Australians ordering the BMS board. But for this to happen certain things like component footprints must be known. Manufacturers often provide the same Part in different types of physical packages (footprints).

So i started compiling a Farnell BOM - based off the Mouser BOM - for Australians wanting to avoid the International postage cost.

-Comparing Manufacturers part numbers, many components that Mouser lists are not in the Farnell catalogue, and several didnt even have alternatives that could be used

-Some parts had minimum order quantities that would see wasted components and cost more.

- About 80% of the parts would have to be delivered from Farnell's sister company Newark In-One USA to Farnell Australia before local dispatch. They arent stocked locally.

-And of course the real killer, it cost about twice as much (allowing for conversion rate/postage of course)

Bottom line.... Just order from Mouser. You cant beat them on single quantity component availablity and price. If you order over US$100 worth of components (2x16channel boards in my case), then you recieve an international shipping discount at a flat US$30 rate for express airmail via fedex or USPS.

 

[Patriot]

I know this may seem very amateurish, but I don't see a problem with just using plain Isopropyl to clean the board. Then, Use a fine tip soldering iron with flux and fine solder to attach everything. Take the time, and be careful not to overheat components.

1. This is how I've always built every circuit board in the past, though never a BMS. But still, why would this be a major problem?

2. Also, is it really necessary to spray the whole thing when done to seal it? if it is kept inside a dry compartment or container, will it still corrode or get damaged from the elements and use?


---------------------------------------------------

Oh, and I assume I can place an order for a board, even though GGoodrum hasn't gotten the next batch in yet?

 

[methods]

The cleaning with the exotic alcohol was just for the conformal coating, not for the soldering.
The conformal coating is to avoid moisture problems

You may think your box is warm and dry but often it is not. Take the CA for instance. Water gets in there but it does not matter since the board is coated.

Coating can save your butt to if you do something stupid like drop a wrench on the board. What would be a serious short circuit is only a few bent leads.

Conformal coating is a luxury that I would like to have but is by no means necessary, as you say.

-methods

 

[ejonesss]

is this mean that a battery can be cooked in a microwave over for a couple seconds to revive the plates?

goodenough's patent involved a specific heating regime to grow the dendrites as i understand, but one of his students has developed a technique to grow structures on the anode using microwave energy. i think that was at arizona technical institute, can't remember now. don knows i bet

i'll double up on the thanks for the "knowledge base" i had seen a few pages in xyster's design thread from last year where bob was talking about the LVC idea, but never knew how it got started.

what little i have read about the factors that determine longevity in lifepo4 cells is the concept of something resembling super cooling in liquids which causes the lithium to precipitate in the solution adjacent the anode, but not on it or in contact with the lithium in the anode. as these globs of lithium condense out of solution adjacent the anode, they grow to sizes that tear apart the crystalline structure of the anode and by doing so reduce the continuity of the anode and reduce it's conductivity, or increase the internal resistance. or that's the way i interpreted it.

the idea behind the new silicon nanotube anode structure that is covered in the stanford patent involves the idea that the nanotubes form in such a way that the anode can be very large in area extent because of their filamentous structure, but are not damaged by being 'hung up' on these growing globs of lithium as they precipitate out of this 'super cooled' state adjacent the anode, so the anode damage is minor and the areal extent can be large so the anode can conduct large amounts of current.

separately and originally, this is why the goodenough patent behind the a123 cell is so effective because he developed a way to make this anode structure with a large dendritic surface capable of conducting large amounts of current by comparison to simple anode structures.

goodenough's patent involved a specific heating regime to grow the dendrites as i understand, but one of his students has developed a technique to grow structures on the anode using microwave energy. i think that was at arizona technical institute, can't remember now. don knows i bet.

but i think in overdischarge the damage caused to the lipo is by the collapse of these layers of cathode polymer 'plates', which is then difficult to get the lithium ions to flow back into, but with the lifepo4, the crystals of lifepo4 have a rigid but open structure which allows the lithium ions to move in and out of these micropores inside the cathode structure and the structure does not collapse when the lithium ions are stripped out of the cathode structure where they are 'hiding' in these micropores and transported to the anode.

i liked gary's rational behind the new/old lipo and the hiC/hiLVC logic.

methods was right about that guy with the confederate halter top being conservative, he swore on 3.2V-3.3V.

 

[dnmun]

mr jones, the microwave technique was developed as a way to bypass the goodenough patent and make the anode grow dendrites with less energy consumed.imo.

almost fitting it was one of his students.

if you wanna nuke your lipo, it would be interesting to try, maybe you should take the microwave out in the parking lot of your apartment building first.

i'm curious. try to evaluate just how low the capacity is beforehand, and do the same test afterward.

 

[Patriot]

Just bought a board on TP Packs website. He still has it selling.

I guess because he's getting more in this week?

Anyway, I'd like to find a small plastic box for this thing.

Actually, what would be really cool, is to figure out a way to install it inside an SLA charger, and integrate the thing into the charger as one unit. That would sure make carrying it around easier.

 

[methods]

I think you should just microwave it in your apartment
I recommend sitting on the microwave during the process
If you stick a piece of cardboard in the latch hole you can even do it with the door open :wink:

-methods

 

[ejonesss]

dnmun is right about doing it outdoors for a couple reasons.

1. if it does not work you dont want to be breathing the smoke.

2. first generation lithium batteries can explode and burn (fire) and you dont want that in your house.

3 if you are going to microwave anything other than food stuff then you should use an old junker microwave.


bypassing the door switches is never a good thing as microwaves can be dangerous.

I think you should just microwave it in your apartment
I recommend sitting on the microwave during the process
If you stick a piece of cardboard in the latch hole you can even do it with the door open :wink:

-methods

 

[wybornd]

Hi Guys,

Thank you so much for all your work on this. It truely is a work of art, this global communication and collaboration.

I have 2 boards ordered for the next batch, however when I tried to order the components, 2 of them are unavailable for up to 8 weeks

747-IXDF404PI "Low Side Gate Drivers 40V 4A"
and
512-KSA931YTA "Power Transistors PNP Epitaxial Sil"

Knowing I am a novice at this (hell I am even scared just to solder it all up) what I couldnt pretend to know is, is there a comparible part I could use?

8 weeks is a long time and they will charge me another round of shipping to get it to me it's really expensive to ship stuff to New Zealand ......

So to anyone watching this rather than the election, HELP me please so I can order the components

Cheers Dean

 

[wybornd]

dnmun is right about doing it outdoors for a couple reasons.

1. if it does not work you dont want to be breathing the smoke.

2. first generation lithium batteries can explode and burn (fire) and you dont want that in your house.

3 if you are going to microwave anything other than food stuff then you should use an old junker microwave.


bypassing the door switches is never a good thing as microwaves can be dangerous.

I think you should just microwave it in your apartment
I recommend sitting on the microwave during the process
If you stick a piece of cardboard in the latch hole you can even do it with the door open :wink:

-methods

Na, you'll be OK. My suggestion though is you put some tin foil on your head like a hat to prevent microwaves getting ya :lol:

 

[jwpower]

First time poster, long time lurker. Sorry, I’m starting with questions rather than answers. Hopefully I can contribute more in the future.

I hope this isn’t too far off topic, but I’m sure others out there would like to use a Soneil 4808SRF charger with the Even Newer… BMS. From prior posts, I think maybe Gary already has. I just ordered the BMS board and I want to check on a couple of things before spending $170 for the charger. I’ll be using it with a 16s3p a123 pack and adjusting the Soneil’s CV setting up to 59.2 volts accordingly.

I read the skimpy specs at: http://soneil.com/Completesets/4808SRF.pdf and it looks like this charger behaves exactly like a CC/CV bench supply except after it hits CV of 59.2 V it then waits for the current to fall to some predetermined value, at which point it lowers the CV setting to 55.2 volts, the “Stage 3 Float Mode”. Sounds like that ought to work, but here’s a potential problem, as I see it. When the BMS starts throttling back the current as the shunts kick in, the charger would run the voltage up to 59.2 right away, because the current can no longer stay at CC. At this point the current may be low enough to cause the charger to switch into its float mode of 55.2 volts and consequently all the cells would not get a full charge. Perhaps a capacitor on across the charger output leads would help ride out the fluctuations as the BMS FET switches on and off, but I still worry that the average current could be low enough to trip the charger into float mode.

As fechter says, “One test is worth a thousand opinions”, so here’s my first question. Has anyone tested the Soneil 4808SRF with this BMS to see if it really works and all cells get a full charge?

The Soneil specs also state, “Increases battery life by de-sulfating the battery”, but no clue as to what it’s really doing to de-sulfate. That sounds like another potential source of problems to me. Does anyone have an opinion on that? I hope it’s not going to pulse current through the battery while it’s in charge mode. I noticed that one of the other Soneil chargers does it’s de-sulfating current pulsing only when bringing a dead battery up from 0.5 V to 5 V, so that wouldn’t be an issue, but the 4808 specs don’t say.

Many many thanks to fechter and GGoodrum, and all the other contributors who had a hand in developing this BMS.

-James

 

[methods]

Na, you'll be OK. My suggestion though is you put some tin foil on your head like a hat to prevent microwaves getting ya :lol:

AFDB
http://zapatopi.net/afdb/
http://zapatopi.net/afdb/build.html

I wear mine every day.

-methods

 

[Patriot]

Hi Guys,

Thank you so much for all your work on this. It truely is a work of art, this global communication and collaboration.

I have 2 boards ordered for the next batch, however when I tried to order the components, 2 of them are unavailable for up to 8 weeks

747-IXDF404PI "Low Side Gate Drivers 40V 4A"
and
512-KSA931YTA "Power Transistors PNP Epitaxial Sil"

Knowing I am a novice at this (hell I am even scared just to solder it all up) what I couldnt pretend to know is, is there a comparible part I could use?

8 weeks is a long time and they will charge me another round of shipping to get it to me it's really expensive to ship stuff to New Zealand ......

So to anyone watching this rather than the election, HELP me please so I can order the components

Cheers Dean

I myself, am waiting on the parts, since I won't be getting the LiFePO4 cells 'til after Christmas. For now, I can at least get the board I need, and tuck it away safe until I need the rest.

 

[fechter]

I read the skimpy specs at: http://soneil.com/Completesets/4808SRF.pdf and it looks like this charger behaves exactly like a CC/CV bench supply except after it hits CV of 59.2 V it then waits for the current to fall to some predetermined value, at which point it lowers the CV setting to 55.2 volts, the “Stage 3 Float Mode”. Sounds like that ought to work, but here’s a potential problem, as I see it. When the BMS starts throttling back the current as the shunts kick in, the charger would run the voltage up to 59.2 right away, because the current can no longer stay at CC. At this point the current may be low enough to cause the charger to switch into its float mode of 55.2 volts and consequently all the cells would not get a full charge. Perhaps a capacitor on across the charger output leads would help ride out the fluctuations as the BMS FET switches on and off, but I still worry that the average current could be low enough to trip the charger into float mode.

I think it will work if you crank the CV voltage high enough to keep all the shunts on. As the cells charge, the shunts will take over and keep the current high enough to keep the soniel from tripping into float mode. The Soniel I have trips at around 100ma. Once all the cells are full and the BMS trips, the current will be near 20ma and the soniel can go to float with no problems.

I have not tried it with a Soniel and this circuit.

 

[fechter]

Hi Guys,

Thank you so much for all your work on this. It truely is a work of art, this global communication and collaboration.

I have 2 boards ordered for the next batch, however when I tried to order the components, 2 of them are unavailable for up to 8 weeks

747-IXDF404PI "Low Side Gate Drivers 40V 4A"
and
512-KSA931YTA "Power Transistors PNP Epitaxial Sil"

Knowing I am a novice at this (hell I am even scared just to solder it all up) what I couldnt pretend to know is, is there a comparible part I could use?

8 weeks is a long time and they will charge me another round of shipping to get it to me it's really expensive to ship stuff to New Zealand ......

So to anyone watching this rather than the election, HELP me please so I can order the components

Cheers Dean

Yes, there is one other part that is interchangable for the gate driver. I'll have to look it up and get back to you.
There are a number of alternate parts that would work for the KSA931. Again, I'll have to look it up.

 

[velias]

It's amazing that everyone has already caused Mouser to run low and out of some of the parts.
and that the first batch of boards were sold out in less than an hour last saturday.
I'm surprised that there are so many people who have lifepo batts and need a BMS.
It also says a lot that it was a good solution for everyone especially the break-away board scheme.
I'm curious, Gary, if you dont mind, roughly how many have you been selling? Has it been distributed all over geographically, or just in the U.S. and Canada?
It looks like theres a big market for Batterys and BMS stuff or lack of companies that provide a good/inexpensive solution.
I guess it tells a lot about the lack of decent BMS's and the demand for raw cells on the market

 

[boostjuice]

As a replacement part for the 747-IXDF404PI i ordered the 747-IXDF404SI which is the SOIC (surface mount) version. These are still in stock and have the same pin configuration, but different pin spacings and dimensions. It will require a bit of craftiness to get the legs connected to the PDIP (Dual inline package) footprint, but if your good with a soldering iron and a scalpel will be very achievable.
The transistor will have many substitutes to choose from. Just compare its datasheet with the others using the "find similar" button on mouser.

It wont be long before Mouser sells out of other parts on the BOM list due to the enormous demand for these BMS units.
There is going to be long delays for those who get in late....

 

[Patriot]

So, exactly where do I hard mount a 18 pin connector to the board?