Stack-able Low Cost Balancing BMS

methods

1 GW
Joined
Aug 8, 2008
Messages
5,555
Location
Santa Cruz CA
Sigh... has someone got it right yet?
We gave up on this years ago on the assumption that no matter how good of a job we did someone would beat us to the punch, under cut us, and we would eat the development costs.

Assuming the average high power ebike battery would look something like:
13S to 24S
10Ah to 20Ah
Discharge Currents from 20A - 80A nominal with bursts upward of 300A
3.0V to 3.6V chemistry
Balance needs in the hundreds of mA or less
Quiescent current that would allow a year + of shelf life
Ability to terminate Charge or Discharge based on programmable limits (Voltage, Current, Time, Temperature)
Thermal inputs - 5 minimum
Ability to detect abnormal starting conditions (like a pack that has been abused)
Very small, IP67, mass produce-able
Ability to stack and accommodate any combination of cells in chunks of 4S or greater - such that any two packs can be hooked in parallel or series to meet requirements
Noise immune
Wireless connectivity via BlueTooth or other for logging, phone apps, etc - or some other method of display and programming other than a USB port

Something like that?

-methods
 
I'm not aware of anything in existence that meets all of your requirements.

Best thing I've used to date is the Adaptto BMS, though most of the smarts are upstream in the controller:

hqdefault.jpg

display_bms.png


Those are 4s boards, you can score and break off excess boards you don't require or you can purchase additional and extend the string. Balances and displays cell status on the display, all thresholds being configurable so doesn't care about specific cell chemistry.

I'm dead keen for something along similar lines that's controller independent for my next build, but it's proving very difficult to find something that's compact, reasonably affordable (I'm not talking about $30 budget here), can log/output status in some useful format, can drive external contactor, configurable thermal inputs/thresholds etc.

I guess what I'm really after is a battery MONITORING system, a guardian angel rather than a gatekeeper.
 
I guess what I'm really after is a battery MONITORING system, a guardian angel rather than a gatekeeper.

I think speedict has a monitoring system available. called Neptune? I guess it relies on a cellphone/bluetooth. I think they are trying to do some balance ones in the future.
 
velias said:
I guess what I'm really after is a battery MONITORING system, a guardian angel rather than a gatekeeper.

I think speedict has a monitoring system available. called Neptune? I guess it relies on a cellphone/bluetooth. I think they are trying to do some balance ones in the future.
Yup, I am recently waiting for model 4 or 5 to come out. https://endless-sphere.com/forums/viewtopic.php?f=14&t=73354#p1135558
 
Methods,

Look like you rassembled all the most wanted requirement of a BMS for the ebike community.

I would certainly buy one if these could be deverlopped and sold!.. I like the Adaptto BMS that are programmable and have a monitor... But they only work with the adaptto!

Doc
 
Yea....

I am trying to find some contract work where I can rapidly develop a BMS that would meet the needs of a customer and the community. There are a lot of competing requirements.... and an Ebike is about the toughest case... as it has the harshest environment, greatest number of configurations, highest probability of user error, lowest cost target, lowest volume... sigh :)

If I were a "kept man" (rich guy who could choose what he did with his time) this would be the project I would nail down.

Driving an external contactor is super easy - just a mostfet and a few 0805's. My favorite contactor are these for $30

MiniTACTOR

They are tiny, very resistant to the type of sparks we are creating, cheap, good for at least 50A continuous....
The 105 (datasheet) draws only 45mA when in the 48V coil configuration

For temperature pickups... I spent a lot of time thinking about what a BMS really needs to do... in the worst case situations... if the battery is instrumented well for temperature, the controller is smart, and the main power switch is reliable - they can be made to be pretty damn safe for cheap.

-methods
 
I think the problem is cheap.

I have looked and specced out a possible BMS designed that would work, but cost would be high, unless I can get huge quantities.
 
Let's turn those specs into engineering terms for the fun of it:
methods said:
13S to 24S
BMS would probably need to be divisible by 12 or at least 4.
My take is a stackable BMS of 4s or 6s.

methods said:
10Ah to 20Ah
Does not really matter. Though it will drive your design specs below

methods said:
Discharge Currents from 20A - 80A nominal with bursts upward of 300A
This is where the crux of it is. 80A probably requires at least 4 reasonably sized FETs to be safe and with burst of 300A, you are looking at more. The way adaptto solved this is just to shut down the controller, the only thing that is drawing current anyways. The FETs are the\ large cost here.

methods said:
3.0V to 3.6V chemistry
easy enough to get. But I think most battery tech are going towards the 3.6V/3.7V/3.8V nominal side. Most Off the shelf (OTC) BMS are supporting only the li-ion type batteries.

methods said:
Balance needs in the hundreds of mA or less
Most OTC BMS are 100mA or less anyways.

methods said:
Quiescent current that would allow a year + of shelf life
This is another major problem. There is 365.25*24 = 8766 hours in 1 year. This means for a 10Ah pack, we need 0.00114A Quiescent current at max average current draw assuming fully charged before storage and checked back in exactly 1 year. Adding buffers, i.e. not fully charged or balancing and a bit longer than 1 year, then really you are looking at 8Ah and 1.5 years. So you are looking at 608uA average Quiescent current. This is possible, but would probably require some designing to get this average quiescent current without users input, i.e. a switch to say we are going into storage mode.

methods said:
Ability to terminate Charge or Discharge based on programmable limits (Voltage, Current, Time, Temperature)
Easy enough. Build an interface and load the new values.

methods said:
Thermal inputs - 5 minimum
Easy enough, more uCs or uCs with more ADC

methods said:
Ability to detect abnormal starting conditions (like a pack that has been abused)
Not sure what you mean, but if you mean by weird voltage and such, then sure easy enough.

methods said:
Very small, IP67, mass produce-able
Easy enough. Just some designing to make it compact and cheap.

methods said:
Noise immune
I am not an RF engineer, but I don't see how this can't be more than just putting in a few RF shields and having good grounds (sinks).

methods said:
Wireless connectivity via BlueTooth or other for logging, phone apps, etc - or some other method of display and programming other than a USB port
Seems easy enough. Just software at this point. I would recommend a smartphone because everyone has one of those these days.
 
The $20 smart bms does nearly all this. Just 3 thermal inputs falls short but seems ample to me. The hibernation current is years not months. Voltage range greater. The programming needs the $100-$200 software. There is data available but you would need the software again. Adapting to wifi and smart phone seems a reasonable step compared to starting from scratch. I can about figure stacking them myself and have seen it done commercially.

The oz890 is cheap as chips and I don't understand why nobody just starts with this readily available design and polishes it up. Some seem to think it has issue's but I always think they are just confused, and if it has, the manufacturer would surely welcome our input if we take the corrective measures far enough for them to utilise them easily.

I know a trusted member did set to work on this but such a project is huge. Much better than basically writing the analogue approach to a chip though, and coming out with 5 times the price.


I have no desire to work with anything else. Though it's time I coughed up for the software.
 
friendly1uk said:
The $20 smart bms does nearly all this.

Do you mean this one?
https://bmsbattery.com/bmspcm/330-smart-bms-513-cells-in-series-bms-pcm.html

friendly1uk said:
Some seem to think it has issue's
I am 1 of those, mine 2 just switched off moments after connected to the battery, leaving battery unprotected. After resetting they still shut-down after a while, totally random and living Fets open (I do not remember now, was it charge or discharge side, or both). I was out of Karma after few intents. :lol:
 
To the "Quiescent current" part - it is a more thoughtful approach to introduce a configurable disconnect timer, rather than than pinch pennies trying to extend a shelf-life of a battery with a hooked-up BMS, IMHO.
What is BMS going to do if it senses a battery voltage going down ? Zilch, because it can't do jack without external power (except to cannibalize better cells to save the weak ones) . So what is the reason for it to stay connected to the batteries then?
 
Methods,

I suggest developping a BMS that will be the ONLY one that people will buy for all next years project!

Modular as well and one that can work with any ebike voltage and chemistry...

one that people will accept paying a bit more but that also understand that they will keep it for all projects and will never need to replace it due to power requirement change, voltage requirement change... chemistry requirememnt change...

Just ONE that you buy one time and keep it forever and that will follow you for all your projects! 8)

one that people will be proud of and spread how they like it!!

Doc
 
sharkmobil said:
To the "Quiescent current" part - it is a more thoughtful approach to introduce a configurable disconnect timer, rather than than pinch pennies trying to extend a shelf-life of a battery with a hooked-up BMS, IMHO.
What is BMS going to do if it senses a battery voltage going down ? Zilch, because it can't do jack without external power (except to cannibalize better cells to save the weak ones) . So what is the reason for it to stay connected to the batteries then?

Not specifically a ebike use, but i am working on a custom bms that will periodically read the voltage to a server just record and monitor the battery. So yes it needs to be on but not all the time.
 
methods said:
There are a lot of competing requirements.... and an Ebike is about the toughest case... as it has the harshest environment, greatest number of configurations, highest probability of user error, lowest cost target, lowest volume... sigh :)
I did a few modular BMS designs (3-6S per module), and had the same experience you're describing. Cost is the biggest hurdle it seems, why pay $100-200 when you can get a chinese bms for a fraction of that. I ended up going with a non-modular approach, which I'm currently using (see video of it in action here). At least for that design, making the modules plug and play is difficult to say the least.

sharkmobil said:
To the "Quiescent current" part - it is a more thoughtful approach to introduce a configurable disconnect timer, rather than than pinch pennies trying to extend a shelf-life of a battery with a hooked-up BMS, IMHO.
What is BMS going to do if it senses a battery voltage going down ? Zilch, because it can't do jack without external power (except to cannibalize better cells to save the weak ones) . So what is the reason for it to stay connected to the batteries then?
For the commercial stuff I've worked on, quiescent current actually turned out not to be a big factor. It's around 500uA for the design above if I disable the status LEDs, which are for debug only. I may piss off some people by saying this, but if you store a 10Ah battery for a year with the BMS attached without charging it, then you should just plan on damage. Disconnect the BMS from the pack, or charge it every few months. Or configure a charger to keep it at 3.0V/cell or something low so that once the self-discharge takes it that low the charger keeps it from going lower. Most commercial packs are not stored for more than a few months at a time, and even it they are, they tend to be much larger than 10Ah so even a few mA of quiescent draw is not a big deal.
 
The Iq comes into play when users ride until empty and store. This is surprisingly common among noobs.
 
Does anyone have a datasheet on the chip used in the SmartBMS? (yea I am being lazy)

To a T... nearly every single one of those Chinese BMS's sucks for the fact that it only balances at the very end of charge. It should be a simple algorithmic change ... but they all seem to emulate the old school method of detecting an over voltage and turning on a bleed resistor. My experience is that packs are often short cycled somewhere well below the HVC trigger point so for a BMS to get ahead of an imbalance situation it must be able to balance at voltages well below HVC.

Basically... der' needs to be sum smarts in der'.

2 - Copy.jpg

-methods
 
I thought the OZ890 was a so called "continuous" balance algorithm, but I don't recall. I haven't seen it at any of the normal distributors lately. App note attached below for reference.

For LiFePO4, I've found that starting balance above 3.38V is about right. The algorithm is basically to calculate minimum cell voltage, then any that exceed it by 10mV and are over 3.38V have the shunt turn on. Only a few lines of code to implement, nothing complicated. Compared to the standard method, it takes less time to balance and all the cell voltages just kind of float towards each other when they get up above 3.5V or so. A secondary benefit is that you don't need nearly as large of a shunt resistor to get the same effect, so power dissipation is much lower and easier to dissipate passively.
 

Attachments

  • OZ890.pdf
    485.2 KB · Views: 92
Hmmmm... thanks for posting.
I did my search routine and came up empty handed. Appears to be a one-off.

Pricing on the Linear chips (especially the newer ones...) is still out of reach. Too bad... as I really like Linear Technology and want to support them in a volume design. Linear Technology is a company that is about as legit as they get.

They told me: "We expect 12 hours of work in an 8 hour day - can you do that?"

-methods
 
okashira said:
The neptune draws power from the top cell. Total fail
I interpreted "top cell" actually as powering from pack terminals, as ground is common usually and 3V will not be best voltage to work with. :) But we can ask again, more specifically.
 
parabellum said:
okashira said:
The neptune draws power from the top cell. Total fail
I interpreted "top cell" actually as powering from pack terminals, as ground is common usually and 3V will not be best voltage to work with. :) But we can ask again, more specifically.


yes sir, Neptune Lite or next coming Neptune 1 powered from pack terminals rather than middle cell to prevent long term cell stress, if we go for higher cell support e.g. 24s we will also take power from terminals since 100v step down to 3.3v is not that hard with modern technologies now, however if design approach is separate e.g. each bms board support 8s, and 3 stack together for 24s, we will consider external DC 5v power supply for master board's mcu, bluetooth and charge discharge control circuit ...
 
Would love to be able to program in multiple balance points for larger monster packs.... or to simply push a button to enter balance mode regardless of voltage would be sweeeeeet
 
Lurkin said:
Would love to be able to program in multiple balance points for larger monster packs.... or to simply push a button to enter balance mode regardless of voltage would be sweeeeeet

Curious why you want a way to manually start balancing? Voltage isn't a good indicator of SOC until you're either nearly fully charged or discharged. If you push the "balance button" at 50% SOC you're more likely going to imbalance the pack further.
 
Back
Top