"Ideal" Lithium charger in-design... looking for suggestions

[ronbot]

How to start this...
I've had 25+ years in electronics design engineering, along with being a lifelong hobbyist. Rechargeable battery technologies has been a keen focus of mine for many of those years (along with alternative energy systems).

I'm in the middle of a personal Lithium conversion project (golf cart), and decided I didn't want to take the old "bulk charging / BMS balancing" route... I wanted better.

No... I actually wanted the best.

So I'm designing a modular system that (similar to what others have tried) utilizes a large 36-48V DC power supply (aka: "rectifier supply" as some are called for the telecom market) to provide input power. These are best purchased in the surplus market (such as eBay) to suit your local Mains grid. I could provide them, but for my 14p-16p project, I'm talking about a 1kW power supply, and they aren't cheap. (however, since these are totally isolated "per cell" modules, you could buy multiple smaller/cheaper power supplies to power them with, and just use as many as each of your 48V power supply will handle) This "Input Power Supply" could be left off of the vehicle, and connected using large Anderson PowerPole SBS connectors (or others of your choice).

My present design is fully protected for "walk away" confidence, and utilizes the following charge profile:
NOTES: "CC" = Constant Current charge rate = 10A for initial model (easily reduced)
. "CV" = Constant Voltage charge = 4.10V for the initial model

START CHARGE CYCLE:
1) Detect for over-discharged cell (<3.0V), if TRUE then proceed with "precharge" (slow charge @ CC/10) conditioning until cell > 3.0V
2) Charge at Full "CC" charge current, until "CV" cell Voltage is reached
3) Switch over to Constant Voltage (CV) charge mode
4) CV charge mode terminates when actual charge current drops to CC/10
5) End charge, until AC power is cycled

FEATURES:
ISOLATED... fully... in all regards.
Each and Every cell is ALWAYS balanced, and charged to the ideal voltage for long-life
Designed to be left connected to battery pack permanently, with leakage current of FAR less than 1mA per cell.
Each module is DC-Input fuse protected.
Each module has input power Reverse-Polarity Protection.
Each module is expected to have 90% efficiency, so very little power (heat) will be dissipated.
Each module is expected to be about 2" x 3"
Each module will have on-board status LED's

I'm still fairly early into this... but each "module" is expected to be splash-proof (conformal coated, at least), and have 20A terminal barriers as Input/Output connections, with redundant connections on DC-IN and Battery-Out. I'm also planning on an "open-collector" status output, so that every module can be "daisy-chain" inter-connected and give an "All Cells Charged" indication via a single remote LED.

Obviously a LiFePo4 version would also be needed, but my initial personal project is using Leaf battery packs, ergo my choice of voltages.

I actually have some 2nd generation model ideas in the works as well... a 100% fully custom version that could reduce cost a fair amount, as well as increase functionality and versatility (one size fits all)... but that is a topic for a later date... I have to get this one done first.

Shoot me some feedback!

How many cells do you need to charge?

What Chemistry?

What Charge Rate?

What price? (be realistic)
I realize that as DIYers, we're extremely sensitive to cost... and this 1st gen may be out of the reach of super cost sensitive projects... but for the functionality and ability to eliminate a costly Balancing BMS system, it's worth thinking about what this would be worth to you... on a "per-cell" cost. This will help me determine if it's even feasible for me to try to produce them in quantity for sale... or just for myself. I have PC boards to get designed and made, many parts to order, but I'm moving ahead with mine anyway.

$30/cell?
$40/cell?
$50/cell?
or?

Even if you're not in the market for such right now, I'd appreciate your input!

Ron



edit 4/29/15
Adding another simple but very useful feature... possibly even two...
(As I said, I'm still in the design process, determining what I want, and what others might want as well.)

Cell low voltage detector(s)

1) Cell DEAD: When the cell drops to 3.2V (probably adjustable) it will pull an Open Collector output "Low'. It's meant as a "last line of defense" to protect the cells. This will allow the cell to immediately enter the CC phase of recharge, versus waiting for the CC/10 slow charge. This output, when "Wire-OR'ed" (summed) with all other Cell Charger modules, will be used to drive the coil of a relay that will disable the Motor controller.

On the cell-charger PC Boards there will be another status LED that is lit when the battery voltage is above the threshold, and extinguishes when it's below that point. This will give "eyes-on" way to determine "which cell is it that just caused my vehicle to stop running", if you waited too long.

(If you really want to know the SOC condition of each cell, use a battery monitor. Many people won't need/want to know the cell condition, but only to protect them from damage.)

2) Cell "almost dead": See above, but the voltage is set a little higher... (probably about 3.60V for me). Also an Open-Collector output, but for another "dash mounted" LED. It's conceivable this could be utilized to enable a "low power mode" in a motor controller, to extend the last few miles... if the operator can't figure that out. Will also likely have an LED on-board for this as well... it's good for quick diagnostics if you don't have a BMS.

These are simple things to add-in, and even with the additional cost for each module, it's still far less than a BMS, and gives the essential protection functions.

 

[ronbot]

Quite a few "reads", but no replies... I take it this means no interest?

 

[Nobuo]

Each and Every cell is ALWAYS balanced, and charged to the ideal voltage for long-life
Designed to be left connected to battery pack permanently, with leakage current of FAR less than 1mA per cell.

So this would be a BMS and charger at once? (the charger is installed along with the battery pack)

The balancing will be done only at CV phase or also when LVC hits?

 

[ronbot]

It takes a bit different viewpoint to see the full advantage, and can't be compared to a Charger with BMS.

Yes, the charger modules are able to be permanently installed with the battery pack, recommended because otherwise you would need every cell-to-cell junction brought out of the pack (along with end-connections) to make connections with. That gets very messy when packs of more than a few cells in series. The Mains power supply does not need to be on-board with the cell charger modules, since it only requires 2 connections (+ and -) to the entire "set" of cell-charges, a single large Anderson 2-position connector can handle that disconnect.

Each cell is treated 100% individually... the cell-charger only knows that the one cell it is mated to exists. It is charged according to its own ability to accept the charge. If 20 cells start out at above the low voltage condition, they all start with full CC charge current... and if you have 2 cells below the low voltage limit, they only get CC/10 until they reach the minimum CC point, then they each transition on their own.

Each cell charger will terminate charge when it drops to CC/10 rate after CV phase.

If a single cell shorts out, that module reverts back to Low Voltage Condition charge, which is CC/10 charge rate... pretty safe.

There is no "overall" brain, each cell-charger contains its own controller.

Another advantage with it being modular is in future maintenance... if one cell-charger ever has a problem, that's all you need to replace... one little module... not the entire charger system.

Yet another advantage is for those who like to make future upgrades. Maybe your project started with 20 cells, and you decide to move to 24... just add 4 little modules. (as long as your Mains 48V power supply has adequate capacity)

If you "outgrow" your Mains power supply, just add another... there's nothing that prevents you from using multiple Mains supplies... just split your 48V DC load to match your Mains power supplies.

 

[Nobuo]

When you say each cell it also count for a paralleled cells group?

As a paralleled group is balanced itself inside if all the cells are connected each other,
the modules could be connected to a paralleled group of any cell quantity?

 

[Willow]

Adapto already have the ultimate charging solution... high power, fully integrated BMS and programmable.

 

[ronbot]

When you say each cell it also count for a paralleled cells group?

As a paralleled group is balanced itself inside if all the cells are connected each other,
the modules could be connected to a paralleled group of any cell quantity?

YES... but as one who has done study on battery technologies, that's not normally a good idea. Over time, each cell will develop their own, "independent personality"... much like twin siblings. Born together, they are very compatible in early years, but as they mature - each develops their own traits... and forcing them to co-exist won't always result in the best care (charging) for each.

IF they are exactly the same (exact same cell, same mfr date), then they will likely not "mature" differently until very late in life. (talking about battery cells now) In this case it can work out pretty good, as each "paired twin" will reduce the impedance to a value lower than the combined individual values. Once they begin to weaken, then the weaker one will not be sharing the load well, and the better one will be overloaded, making it weaker... cascade effect.

One large cell won't have the same problem, since it's a singularity in itself... so to speak.

The initial model will only have a 10A charge rate, so if the paralleled capacity grows over 50Ah or so, the CC/10 charge termination detection will have problems. The 10A charge rate is good for the range of about 10Ah to 50-60Ah without stretching it too far.

As mentioned, that rate can be lowered (during manufacturing), but not increased at this point.

 

[Nobuo]

I understand.
I have some question. if the number of twin siblings is enough, wouldn't be that create a "overall" personality on each paralleled group? That average personality would help the balancing purpose to have the serial formed with similar personality paralleled groups..

I mean for example having 20 2.5Ah cells in each paralleled group would make 50Ah. with 20 cells per group the weaks, strong cells rate would make a high probability to have all the groups with an average close personality.

Would your modular charger do a good job in that case?

I think apart from LiFePo4 batteries, lithium-ion are always under the 3Ah range per cell, so that would be great to improve

 

[ronbot]

True! The more identical cells you parallel, the more the load gets spread out, the less stress (load) that any one member of that paralleled group would be asked to bear.

This is exactly how Tesla configured their Model S battery packs... many cells in parallel for each "4.1v group"... but every single cell has its own fuse in case of an individual cell short-circuit.

The Nissan Leaf "packs" are 2s2p (16S2P total, 8.2v 60Ah packs), and this is what I'm initially designing it for. Yes, it's a slow charge rate (est. 8 hours), but adequate for my needs on a 120V Mains outlet (around 900W AC input).
My initial 10A CC charge rate is a conservative limit... once I get them built, I can modify the current sense resistors and push them to 12-14A (maximum for now) and see how things go.

 

[ronbot]

Adding another simple but very useful feature... possibly even two...
(As I said, I'm still in the design process, determining what I want, and what others might want as well.)

Cell low voltage detector(s)

1) Cell DEAD: When the cell drops to 3.2V (probably adjustable) it will pull an Open Collector output "Low'. It's meant as a "last line of defense" to protect the cells. This will allow the cell to immediately enter the CC phase of recharge, versus waiting for the CC/10 slow charge. This output, when "Wire-OR'ed" (summed) with all other Cell Charger modules, will be used to drive the coil of a relay that will disable the Motor controller.

On the cell-charger PC Boards there will be another status LED that is lit when the battery voltage is above the threshold, and extinguishes when it's below that point. This will give "eyes-on" way to determine "which cell is it that just caused my vehicle to stop running", if you waited too long.

(If you really want to know the SOC condition of each cell, use a battery monitor. Many people won't need/want to know the cell condition, but only to protect them from damage.)

2) Cell "almost dead": See above, but the voltage is set a little higher... (probably about 3.60V for me). Also an Open-Collector output, but for another "dash mounted" LED. It's conceivable this could be utilized to enable a "low power mode" in a motor controller, to extend the last few miles... if the operator can't figure that out. Will also likely have an LED on-board for this as well... it's good for quick diagnostics if you don't have a BMS.

These are simple things to add-in, and even with the additional cost for each module, it's still far less than a BMS, and gives the essential protection functions.

 

[dmwahl]

Personally what would be most useful is a programmable charger that you can create your own profile and have some simple logic. For example, if the pack voltage is below 40V, charge at 1A. Once above 40V charge at 5A. At 56V, reduce charge to 1A. Also helpful would be one or more logic level inputs to interface with a BMS and allow additional control of the charger. A method to switch between multiple stored profiles (and display which is active) would also help.

I doubt many people here would find modules that attach to each cell useful. It would add weight, cost more, and I don't really see the advantage. Maybe for larger vehicles where space is less of a premium, but it seems overly complicated compared to a bulk charger and BMS.

 

[dnmun]

i cannot imagine people will pay more for a BMS than the battery itself costs.

if you can get the cost down to $2/channel then it would be able to compete with the best BMS we have access to now for 24S packs.

most of these people think that is too much so they build a battery without a BMS and talk about how they will manually balance the pack so they don't have to waste $50 on a BMS.

 

[John in CR]

Ideal means to me a battery charger that will accept standard AC or variable DC from solar panels or a wind turbine. How can a charger that tethers your ebike to a wall be considered ideal?

 

[peters]

If I understand well, your design is basically isolated 10A DC-DC converters with CC-CV for each cell. I think it is a good idea, but cannot be made in a small size.
To reduce cost and size, what about a combination of a bulk charger and smaller isolated chargers? The bulk charger can be very high current, let's say 1C (or the battery limit), and the isolated ones only 1..2A amps instead of 10A. If one of the cells is at its top voltage, then bulk charger switches off and the isolated chargers switch on. Similar to a resistive BMS balancer, but at least 10 times higher balancing current. No need to wait hours for balancing and would be less expensive than many 10A DC-DC-s.

 

[ronbot]

"Personally what would be most useful is a programmable charger that you can create your own profile and have some simple logic. For example, if the pack voltage is below 40V, charge at 1A. Once above 40V charge at 5A. At 56V, reduce charge to 1A. Also helpful would be one or more logic level inputs to interface with a BMS and allow additional control of the charger. A method to switch between multiple stored profiles (and display which is active) would also help.

I doubt many people here would find modules that attach to each cell useful. It would add weight, cost more, and I don't really see the advantage. Maybe for larger vehicles where space is less of a premium, but it seems overly complicated compared to a bulk charger and BMS."

The whole premise is to to away with bulk charging, in favor of precise cell charging. Keeping a CPU/MCU out of the heart of the charger increases the reliability in a huge way. Any time a CPU hiccups, the possibility of life-(and property)-threatening consequences arises, due to over-charging. This design utilizes a state-machine, which suffers none of the CPU-related issues.

I think you hit the nail on the head though... that for E-bikes, this may well be too "upscale", since most seem to be dyed-in-the-wool DIYers, and don't care if they have to do extra work in order to save their hard-earned $$. This design is really for people who want absolute "walk away for forget it" operation... like a production EV automobile.

When I looked up BMS systems, and proper profile chargers, I didn't find any that were "cheap" (not counting RC chargers, where you have to go through several steps every time to start a charge, and still require a high-power input power supply). If an e-machine owner wants to have a fully self-contained system, where you arrive at your destination and just plug an AC cord it to charge... Plug-in EV style... that's what I was aiming at.

"i cannot imagine people will pay more for a BMS than the battery itself costs. if you can get the cost down to $2/channel then it would be able to compete with the best BMS we have access to now for 24S packs. most of these people think that is too much so they build a battery without a BMS and talk about how they will manually balance the pack so they don't have to waste $50 on a BMS."

Probably the same answer as above... I didn't know many e-bikes were using low-end RC market parts like batteries, chargers, BMS systems. I thought this forum was about all kinds of e-vehicles... where many are major financial investments.

"Ideal means to me a battery charger that will accept standard AC or variable DC from solar panels or a wind turbine. How can a charger that tethers your ebike to a wall be considered ideal?:

Standard AC voltage... which standard? That's what the 48V power supply requires, which is what powers these modules. I left procuring that as an option, to allow the end user to buy a 48V supply according to their systems needs, at drastically reduced prices and designed for the Mains supply in their own country of use. But you mentioned Solar or Wind... and many of those are 48V DC storage-battery-based systems... which would eliminate the 48V DC power supply requirements. The DC input isn't that particular... even the 48V does not "have" to be regulated, just rectified and well-filtered, and actually from about 36V up to 70V... but 48V is the most common "Distributed DC Mains" voltage available, and easy to get... cheap.

"If I understand well, your design is basically isolated 10A DC-DC converters with CC-CV for each cell. I think it is a good idea, but cannot be made in a small size. To reduce cost and size, what about a combination of a bulk charger and smaller isolated chargers? The bulk charger can be very high current, let's say 1C (or the battery limit), and the isolated ones only 1..2A amps instead of 10A. If one of the cells is at its top voltage, then bulk charger switches off and the isolated chargers switch on. Similar to a resistive BMS balancer, but at least 10 times higher balancing current. No need to wait hours for balancing and would be less expensive than many 10A DC-DC-s."

Small size? I believe you mean total size, right? So for a 6S pack, the total volume for 6 (six) cell-charger modules would occupy 'maybe' 27cu.in., which is 6" x 6" x 0.75" . That only needs a 300W 48V power supply, which for a nominal priced unit, is about 62 cu.in. (2x4x7.8). A complete 10A charging system, with battery protection during charge and discharge, and a low battery indication consuming around 89cu.in. I'd like to see the competition... and see what they provide so I can know what I can do better. I believe this will charge faster than any 10A balancing bulk-charger, because there is no "balancing act" to perform... every cell (or paralleled cells) get exactly a full charge, as fast as it can (with 10A+)... nothing less, and nothing excess to bleed-off to its neighbors. Cell balancing on a bulk charger, by definition, can't match individual cell charging performance, reliability, and cell longevity... when comparing equal charge rates.

Any product that's too specialized for one specific small market, may win the small low-end market, but lose the ability to appeal to many others. With the modular scalability approach, I was hoping to make it useable for many markets... especially the broad DIY market, where "plans change", and often get revised along the way... and adding more modules is cheap (-er than replacing an entire charger), easy, quick. The only other thing that might need to change is the 48V DC supply... and as mentioned, multiple supplies can be used and split the load accordingly (they don't need to all share a common input DC power supply)

I'll just leave it at this, but it seems I need to look for a forum that's got a broader EV-DIY appeal than just e-bikers

Thanks for your feedback guys!

 

[dnmun]

i use 4 of the 24S D131 in my ZENN car (131Ah @ 82V) so that is not just ebikes. totally reliable. cost $53 each before shipping.

 

[ronbot]

i use 4 of the 24S D131 in my ZENN car (131Ah @ 82V) so that is not just ebikes. totally reliable. cost $53 each before shipping.

Cool! Looks like wholesale pricing... correct?
But that's only the BMS... what CC/CV charger do you use?
How long does it take to charge & balance, if there's any to do?
Using R/C battery packs, is that right?

 

[dmwahl]

I'll just leave it at this, but it seems I need to look for a forum that's got a broader EV-DIY appeal than just e-bikers

I'd try the DIYElectricCar forum, the size/cost may be more palatable to them. Post a link here if you do, I'd still be interested in following this.

 

[dnmun]

not wholesale price but close since i order a bunch at the same time to help get a discount.

for charging i allow the Elcon charger that maintains the SLA pack to recharge the entire pack up to the SLA sustaining level 79V and then when i use the car i will charge with a pair of tin can chargers to push about 17A up to 87.6V because i have 87Ah of 24S lifepo4 pouches and 44Ah of 21S lipo pouches connected in parallel through the drains on the BMSs. i use a parallel array of diodes to isolate the SLA and lithium packs. the drains of the two types of lithium chemistry BMS are isolated by diodes also.

i use 10 10A 45V axial diodes soldered in parallel to make up the 100A diode i use for isolation.

 

[ronbot]

not wholesale price but close since i order a bunch at the same time to help get a discount.

for charging i allow the Elcon charger that maintains the SLA pack to recharge the entire pack up to the SLA sustaining level 79V and then when i use the car i will charge with a pair of tin can chargers to push about 17A up to 87.6V because i have 87Ah of 24S lifepo4 pouches and 44Ah of 21S lipo pouches connected in parallel through the drains on the BMSs. i use a parallel array of diodes to isolate the SLA and lithium packs. the drains of the two types of lithium chemistry BMS are isolated by diodes also.

i use 10 10A 45V axial diodes soldered in parallel to make up the 100A diode i use for isolation.

Truly, a master DIYer at heart! You have skills above most, so you are able & willing to put up with the inconveniences that allow you to save $$ in exchange for some additional work... that's what DIY is all about!
But "tin can chargers"... I'm lost on that term... care to explain? I assume they back off to CV after reaching 87.6V ? What component is responsible for terminating the charge cycle, once the cells are balanced out?
From a designer viewpoint, I get concerned about paralleling diodes... without balancing them with a very small series resistor, they will never share current perfect... but then again, as someone with the ability to repair, you can always take care of issues like that. Normal silicon diodes, or schottky?

What grabbed my attention most was when you said you "sustain" the LiPo/LiFeO4 packs at 79V... but as long as no individual cells creep up above their nominal operating voltage, you're good. The papers I read on this "maintenance" mode say full-time per-cell balancing is required to prevent any single cell from creeping too high.

I also have to keep reminding myself that DIY also means you're willing to put up with the occasional repairs, cell replacements, etc. This is a transition for me, as I come from a design background where every product, every circuit is designed for ZERO failures... because every failure reflects on the reputation of the company... and when it's a multi-billion dollar multi-national corporation like GARMIN... nobody takes risks. Even after I left Garmin (was one of the original 12 employees), and ran the R&D department at a contract design and manufacturing company... that design approach was maintained because we wanted to succeed and be the best.

I guess if I'm going to design for the DIY field, I'm going to have to come at it with a different approach than "do it once, do it right", because that will always carry with it a certain amount of cost.

 

[dnmun]

it is just impossible to compete with the chinese manufacturers making high quality devices so cheaply.

i could not even buy the surface mount components that are on the balance board for the money it costs to buy an entire high quality BMS. then there is the cost of the pcbs and the placeement machines for the surface mount and reflow and the power board itself has 20 high power mosfets as well as heatsink so there is just no way to compete.

if you look at the active pack management schemes that tesla has developed it is a hugely different game.

tin can refers to what people call alloy case chargers. i use a 600W kingpan and a EMC-1000 in parallel.