BMS mystery - how does it actually work?

[thunderstorm80]

Hi,
I wanted to buy a battery pack from:
http://em3ev.com/store/

They are very nice and seems very professional, but I couldn't get one thing right - Their BMS.
Because I need regen, and quite powerful one, I was told their BMS can't accept more than 12A of charge.
I offered to to buy two battery packs in parallel, to split the load, so theoretically my max charge current is 24A. (+-)
I was told parallel connection is not allowed as it will force drifting currents between the packs via it's discharge terminals, and it's also problematic with regen.
That's where I got lost - The BMS separates different battery terminals for discharge and different ones for charge?
They said some regen is possible via the discharge terminal but it's not recommended. (and that's where I (think) I understood the regen current would arrive at the discharge terminal so the 12A-charge limit is not relevant - but never the less it would destroy the BMS circuit)
They said I could put diodes to redirect the currents right, but diodes get hot and aren't long term reliable at those powers.
Shall I avoid that kind of BMS and buy from them the pack without it? (They said it's possible)
Anyone has experience with their BMS and batteries?

If someone could show me what kind of BMS they use, and a circuit diagram - that would really help me to understand what's going on there.
Also - what kind of BMS would you recommend that is meant to cope with powerful regen. (and I guess that means it only has a single battery terminal for discharge and charge)

All I want is a BMS that will be passively drift current between the cells with dedicated signal wires on each cell's terminal, so it doesn't interfere with discharge/charging.
It should also be doing this constantly - and not only at 100% SOC, because I will not use the battery at more than 80%-90% SOC so it doesn't have pathetic life cycle count.
If I had enough knowledge in electronics I could have built something like this myself - with capacitator - connecting to each cell for a fixed amount of time before going over to the next one - and therefore perform passive and simple voltage balancing.

 

[phild]

It should also be doing this constantly - and not only at 100% SOC, because I will not use the battery at more than 80%-90% SOC so it doesn't have pathetic life cycle count.

Curious about this part too... most I suspect will now be charging to 80% with BMS - so, does this mean we are no longer balancing our cells as they are never charged to 100% so the BMS is never having a crack at discharging the over voltage cells?

 

[thunderstorm80]

No one knows?

 

[dogman dan]

One possibility is bypassing the bms for the discharge.

Then you do have to be a human, manual bms for discharge. Lots of us do this, with hobby packs. Fallible yes, but only if you screw up. Conservative stopping point and a whole pack voltage monitor is all most of us ever need. For deeper discharge, add balance plugs so you can run low voltage alarms when running in the last 20% of capacity.

Charge and balance normally, through the charge plug.

Lastly,, you won't be seeing 24 amps of regen I think. regen back through your bms discharge ports. Worst thing can happen is you fry a bms. Order a spare with the battery.

 

[Alan B]

Most BMS circuits use N channel parallel FETs in the negative lead to control the power. Power FETs have an inherent body diode that parallels the FET in the opposite direction so they can only control current flowing in one direction, in the other direction the diode conducts so there is no control. The diode is capable of a LOT of current (as much as the FET), but voltage drop is about a volt, so it gets hot when the FET is off. If the FET is on it will "short" the diode and conduct the current at the lower FET "on" voltage drop. This is where regen current generally flows.

The BMS has two sets of (paralleled for high current) FETs, one for discharge and one for charge. If there is a single lead for charge/discharge then these FETs are in series to provide control in either direction. This is both more costly and adds more loss to the system.

Separating the charge and discharge leads allows fewer FETs to be used in the charge circuit since currents are much lower. It also takes half the loss out of the discharge circuit compared to single port designs.

Regen current is short term compared to charging current. In an average sense it is quite small, even if the brief peaks are large. If you want to handle a lot of regen current then use a large pack or use high charge current capacity cells or Lipo; or depend on the fact that it is not high current from an average perspective and let the cells take it.

Ideally don't put the main motor discharge and regen currents through the BMS, only pass the smaller control power through it. The problem with this is the location of the BMS switching FETs in the negative lead. The currents are combined in the negative lead, so that won't work with a standard BMS.

We need a BMS that controls discharge power with switching in the Positive lead, but it only needs to handle a few amps, enough for control logic and accessory power (not main motor power). The pack and improved BMS would have a main power connection, positive and negative, and a control power connection, positive or both, and a charge connection, both positive and negative. Charge switching could still occur on the negative lead on this separate connection.

The reason that BMS's don't have high current switching on the positive lead is cost. The N channel FETs are inherently much cheaper than the P channel FETs that are convenient to use on a positive switch. A positive side switch can be made from N channel FETs but it requires gate voltage above the pack voltage to turn it on, and this complicates the circuitry and raises cost.

Remember that regen braking is subject to several issues and should not be the only braking. It works very well (see my CroBorg link below for strong variable regen braking example), but if anything goes wrong with controller, pack, wiring, circuit breaker or BMS the regen braking may be lost, so a fully functional mechanical friction braking system is required for safety. I used dual disc on the front brake and variable regen on the rear Cromotor on the SuperCommuter.

A quality controller can be programmed to control regen current and maximum voltage to avoid overcharging the battery. If the regen current is flowing to the battery without BMS control there is no protection from cells getting out of balance unless the BMS is a single charge/discharge port type (or a separate control power type, if the imbalance turns off control power regen will stop (producing a braking failure)). With a separate charge lead type, the BMS attempting to "turn off" due to cell unbalance will only add a diode drop to the discharge port, and regen current will still flow.

The chance of cells getting unbalanced enough during regen to cause a problem is very low. It is most likely if bottom balancing is employed, but with a top balanced battery it is unlikely. If the pack is discharged much then there is no chance of this occurring except in an internal battery connection failure situation.

I'm not sure anyone makes a separate control power type BMS, but that would be the best setup, and probably the lowest cost because most of the high current parallel FETs are eliminated. A fuse or circuit breaker would be required to provide the high current "short" protection, but that is always a good idea anyway. Fuses at these high currents are problematic, so the AC/DC Solar circuit breakers are often the best choice.

 

[thunderstorm80]

Most BMS circuits use N channel parallel FETs in the negative lead to control the power. Power FETs have an inherent body diode that parallels the FET in the opposite direction so they can only control current flowing in one direction, in the other direction the diode conducts so there is no control. The diode is capable of a LOT of current (as much as the FET), but voltage drop is about a volt, so it gets hot when the FET is off. If the FET is on it will "short" the diode and conduct the current at the lower FET "on" voltage drop. This is where regen current generally flows.

The BMS has two sets of (paralleled for high current) FETs, one for discharge and one for charge. If there is a single lead for charge/discharge then these FETs are in series to provide control in either direction. This is both more costly and adds more loss to the system.

Separating the charge and discharge leads allows fewer FETs to be used in the charge circuit since currents are much lower. It also takes half the loss out of the discharge circuit compared to single port designs.

Regen current is short term compared to charging current. In an average sense it is quite small, even if the brief peaks are large. If you want to handle a lot of regen current then use a large pack or use high charge current capacity cells or Lipo; or depend on the fact that it is not high current from an average perspective and let the cells take it.

Ideally don't put the main motor discharge and regen currents through the BMS, only pass the smaller control power through it. The problem with this is the location of the BMS switching FETs in the negative lead. The currents are combined in the negative lead, so that won't work with a standard BMS.

We need a BMS that controls discharge power with switching in the Positive lead, but it only needs to handle a few amps, enough for control logic and accessory power (not main motor power). The pack and improved BMS would have a main power connection, positive and negative, and a control power connection, positive or both, and a charge connection, both positive and negative. Charge switching could still occur on the negative lead on this separate connection.

The reason that BMS's don't have high current switching on the positive lead is cost. The N channel FETs are inherently much cheaper than the P channel FETs that are convenient to use on a positive switch. A positive side switch can be made from N channel FETs but it requires gate voltage above the pack voltage to turn it on, and this complicates the circuitry and raises cost.

Remember that regen braking is subject to several issues and should not be the only braking. It works very well (see my CroBorg link below for strong variable regen braking example), but if anything goes wrong with controller, pack, wiring, circuit breaker or BMS the regen braking may be lost, so a fully functional mechanical friction braking system is required for safety. I used dual disc on the front brake and variable regen on the rear Cromotor on the SuperCommuter.

A quality controller can be programmed to control regen current and maximum voltage to avoid overcharging the battery. If the regen current is flowing to the battery without BMS control there is no protection from cells getting out of balance unless the BMS is a single charge/discharge port type (or a separate control power type, if the imbalance turns off control power regen will stop (producing a braking failure)). With a separate charge lead type, the BMS attempting to "turn off" due to cell unbalance will only add a diode drop to the discharge port, and regen current will still flow.

The chance of cells getting unbalanced enough during regen to cause a problem is very low. It is most likely if bottom balancing is employed, but with a top balanced battery it is unlikely. If the pack is discharged much then there is no chance of this occurring except in an internal battery connection failure situation.

I'm not sure anyone makes a separate control power type BMS, but that would be the best setup, and probably the lowest cost because most of the high current parallel FETs are eliminated. A fuse or circuit breaker would be required to provide the high current "short" protection, but that is always a good idea anyway. Fuses at these high currents are problematic, so the AC/DC Solar circuit breakers are often the best choice.

Wow! Thank you for the detailed answer!
It's amazing you can't find that kind of detailed info anywhere on the web, except here - on ES.

First, I must say that I do have proper mechanical braking on both wheels. I will never substitute them and relay only regen. Especially when I experience grades of 10-13% daily...
I am "obsessed" about regen, because I live in a very hilly place, and also because my bike tends to carry heavy loads sometimes: Imagine fully loading your front&rear panniers with grocery shopping, and then going down a 13% grade to your home... This adds not only to the safety but mostly to your range. And when your bike is heavy - the range is extended significantly!
What do mean when you say "brief" regen - In my city, I can go downhill from 500m to sealevel with aggressive grades, and see average continuous regen powers of 700-1000W (peaking at 1200W+).
This can last for 15min or so. Is it still considered brief?
I understand that the regen current is actually passing through the N diode in the conductive direction - causing heat, right? But since the discharge FETs are so much larger than the charge ones this shouldn't pose a problem?
So a regular 3-wire-BMS is controlling both the discharge terminal (and cutting off when a cell drops below it's cut-off level, and the charge terminal with a separate circuit? (that prevents overcharge)

The only problem is that cell balancing would only occur at 100% SOC, because the BMS only starts the balance once a cell reaches it's upper voltage limit.
I want to charge the battery to no more than 90% SOC. (80% most of the time)
Since 100% SOC is a death-spell for a non-LFP battery, I wonder if the benefit of cell-balancing outweight the rapid capacity loss of 100% SOC?

Is there a BMS that can do passive balancing at any SOC? Something like a capacitator that hooks up to a different cell each time, and carrying the voltage difference between them until they are even. This of course will force the user to monitor the battery voltage, but we all have CA's nowadays.

I think to buy this battery pack:
http://em3ev.com/store/index.php?route=product/product&path=35_52&product_id=134
(With 30Q cells for high power and 23.6Ah capacity)
Do you think it will suit my needs? (1200W regen through the discharge lead)
I can see that in the picture a dual anderson leads act as the charging terminal (up to 12A), and the discharge terminal has 3 of them. What is the third wire is for?

 

[Alan B]

Em3ev uses an extra pin for precharge, as I recall.

100% SOC isn't the big problem many make it out to be. Time spent sitting at 100% SOC is one wear factor. Elevated temperature is another. Truth for ebike batteries is don't worry about it, just use it and when the pack wears out in a few years there will be a much better battery available for lower cost.

In many cases the lifetime watt-hours through the pack is nearly constant regardless of what you do (depending on chemistry). Short cycling allows more cycles but if the amp-hour total to battery wearout is similar, to what advantage? Many other factors like temperature, charge and discharge rates, chemistry, and calendar time are also factors. Few ebike packs last as long as the ideal chemistry cycles would say is possible.

To get the best life, keep batteries at storage level most of the time (50-60%) and charge to 90% or 100% just before the ride, based on the needs of that ride (just in time charging).

Periodically charge and bake at 100% to balance, again just before a ride.

Balancing at other than the top or bottom of the SOC is very inaccurate. Balancing is only accurate at the ends where voltage is a good SOC indicator. In the middle the voltage differences are too small and temperature dependent.

 

[thunderstorm80]

In many cases the lifetime watt-hours through the pack is nearly constant regardless of what you do. Short cycling allows more cycles but if the amp-hour total to battery wearout is similar, to what advantage?

You're wrong:
If you limit your usable range from 0-100% to 10%-90% you double the life-cycle capacity.
I am neglecting the aging effect for the sake of the comparison, and assuming you are doing all those charge/discharges test within short time period.
I assume the aging effect has it's toll over the 100% penalty regarding the useful total energy over long time, but what I say (and correct me if I am wrong), that even just a small reduction in capacity spectrum will boost significantly the total available energy.
But it's obvious that if you limit much further, eventually you reach the limit of the total usable energy you can pull from the cells.
For example: 50%-10% will give 2000 cycles and 40-20% will give 4000 cycles. That's the same total energy storage.

 

[Alan B]

My source of this information is Justin of Ebikes.ca, he has reported this here on ES in the Satiator charger thread. It depends on the chemistry and many other things, but many people are fixating on a small bit of inappropriate or out of date testing and assuming it is generally true for all situations, which is not the case. You have to find data on the cells you are actually using, you cannot assume data from other cells applies. Cell chemistry is improving all the time, for many of them 4.2V is lower than they are capable of already.

The case you cite of 0-100% is not realistic, no one discharges to 0%, and if they needed that capacity then using a smaller loop would not meet their energy needs.

If you have data on the OP's cells, share it, and provide up to date sources.

It is fine to short cycle the cells, and most of us do it, but it defeats BMS balancing, so in many cases this has led to problems that have shortened the life of the pack far more than the expected benefits. Ideally you would end every ride at storage SOC and charge just enough for the trip and just in time to depart, but the practicality of ideal cell cycling is not a fit for actual use. Part of the regimen needs to insure that balance is maintained. Most simple BMS's don't tell you how much balancing they are doing so you have no idea how much is needed. You need to do at least that much. The less you fully charge the more risk you take of having a balance related problem. Having done a lot of manual pack management I can see that there is a huge range of variability in what the cells need.

 

[thunderstorm80]

Really?
On his Satiator page, Justin clearly shows one of the Satiator's benefit of partial charge - You will get many more cycles if you do a 90% or even 80% charge - and it looks as way more total energy capacity than the percentage you cut off. He claims you will pay off the Satiator's cost by being able to pull out more useful energy cycles from your battery packs.
If for example by cutting the useful spectrum to 50% I get double life-cycles, I didn't really extend my battery's cost/energy storage...

So why didn't he changed that info if the thread you talk about shows different findings?
Can you link to that thread?

It could be, that if you make those life-cycles over a long time (years), then by that time frame the aging of the 100% SOC becomes VERY significant.

 

[Alan B]

I was reading the manual for the charger and there is a line that I don't understand. It implies there is no benefit to limiting the charge level on LiFePO4, but everything I have read says otherwise?

If you can show me an actual paper and study to this effect I'd be happy to revisit that statement, but everything I have seen so far that is substantiated with firsthand data (ie not random people or websites spouting their "knowledge") suggests that with LiFePO4 cells you only get a cycle count increase in inverse proportion to the depth of the cycles so that the total watt-hours taken over the life of the pack is more or less the same. That is, you'll get roughly double the cycles if you only do 50% discharges, or 5 times the cycles if you just do 20% discharges. That's the same total energy output meaning you aren't gaining anything at the end of the day, just wasting more time connecting and disconnecting your pack to a charger.

Have a read of my posts on this thread here, which are based on a pretty detailed study of A123 LiFePO4 cells:
https://endless-sphere.com/forums/viewtopic.php?p=1142875#p1142875

I haven't done any firsthand testing myself on LiFePO4 cells, and will gladly sing a different tune if pointed to evidence that is contrary to to the paper above.

-Justin

I'm just relating what I read. Above is part of the discussion, from the Satiator thread referring to the manual. It depends on chemistry, with some benefiting less than others. I didn't recall what chemistry Justin was referring to, apparently it was A123M1 LiFePO4 cells which is probably not what the OP is looking at with em3ev packs. But you need to look at actual data on the cells being used to be certain of the advantages of a particular strategy. The calendar life factor cannot be overlooked either, most people don't cycle their ebike packs enough to escape the calendar being a major factor in their pack life. What's the point of a complex strategy when the calendar life is going to be dominant anyway?

My own experience is that ebike batteries generally are killed long before the "ideal life" is encountered, so all the effort to extend the life does not result in improvement when other issues kill the pack anyway. In fact the strategy itself may cause problems that promote early failure. Consistently undercharging and causing the cells to not be balanced is a perfect example of how this is implemented, I see a lot of cases of this causing problems resulting in packs underperforming or failing. To really take advantage of an undercharging strategy lengthening the life a standard BMS may not be desirable. Balancing at lower SOC is also problematic as indicated before. BMS's that support fully detailed cell block information are generally not the types deployed on ebikes, as they cost more than the ebike battery pack does. So implementing these strategies without full information about cell status may result in some cells being at higher SOC than others which defeats the strategy.

It is fine to have a charging strategy to lengthen the pack life. We all do that to some degree. Charge to a lower voltage, especially avoid deep discharges. But understand that there are many other factors including ambient temperature and letting the pack sit at voltages above storage voltage as well as calendar time, drain rate, charging rate and physical factors that may dominate in determining how long a pack lasts.

In most cases after a few years there are much better ebike packs available at lower weight, size and cost. At that point is extending the life of an old degraded heavy pack worthwhile or important, especially if that strategy is limiting your range and travel flexibility?

 

[999zip999]

What exact cell are we talking about ? Chemistry ? Different lap rat different results.

 

[thunderstorm80]

suggests that with LiFePO4 cells you only get a cycle count

That explains it - A123 LiFePO4 doesn't suffer from cycle life reduction at all with regard to it's SOC. I have four of these for 5 years, and they are always charged up to 100%. Needless to say, they still hold almost the same capacity as I bought them. One of the reasons they act so, is because they barely face the aging effect.

I was talking about non-LFP chemistry. In particular I was talking about NCA or NMC.

You are right that after several years there is a better technology, but when I look at my current A123 cells - which STILL surpass any spec of modern cells except the weight (100Wh/Kg), it's a really tough dilema what to do! (I am now purchasing some more "Wh", since with my new direct-drive motor the A123 capacity is too small)

 

[whatever]

will your motor actually deliver 12amps to the battery? The amount of current the regen can produce is dependent on the battery pack internal resistance ( and voltage: state of charge), if its just a standard ebike motor ( nothing to powerful) and a not too big ( in ahr) battery pack ( say 10-12ahr),
then it quite feasible the current your battery can absorb from regen might be well below 12amps and you wont have to worry.
You can bypass the bms ( probably better to remove it) and see how many amps you actually get delivered to the pack under high speed ( low battery charge) conditions. It might be alot less amps than you assume.
If you motor is 1200watt max, it doesn't mean anywhere near that amount of power will go back into the pack ( I note you say 1200watts of regen peak). If going down a extremely steep hill ( 13%) and creating regen for 15 minutes continuous that is alot of energy, but your batteries might only accept 5 to 10% of that energy due to their internal resistance.
How did you come up with 1200watts peak regen?
If its a 48v motor and 18amp controller, using lithium ( say cobalt type) of 12ahr capacity, you might peak at 5 or 6amps regen if your lucky, more likely 3 or 4amps. You dont really get much energy back using regen.

 

[thunderstorm80]

will your motor actually deliver 12amps to the battery? The amount of current the regen can produce is dependent on the battery pack internal resistance ( and voltage: state of charge), if its just a standard ebike motor ( nothing to powerful) and a not too big ( in ahr) battery pack ( say 10-12ahr),
then it quite feasible the current your battery can absorb from regen might be well below 12amps and you wont have to worry.
You can bypass the bms ( probably better to remove it) and see how many amps you actually get delivered to the pack under high speed ( low battery charge) conditions. It might be alot less amps than you assume.
If you motor is 1200watt max, it doesn't mean anywhere near that amount of power will go back into the pack ( I note you say 1200watts of regen peak). If going down a extremely steep hill ( 13%) and creating regen for 15 minutes continuous that is alot of energy, but your batteries might only accept 5 to 10% of that energy due to their internal resistance.
How did you come up with 1200watts peak regen?
If its a 48v motor and 18amp controller, using lithium ( say cobalt type) of 12ahr capacity, you might peak at 5 or 6amps regen if your lucky, more likely 3 or 4amps. You dont really get much energy back using regen.

I witness the numbers with my own eyes (and the CA)...
On downhill, if the speed is high enough, I see negative power exceeding 1000W easily... The highest I measured (so far) was 1200W, but I wasn't yet in a really steep downhill and applied full regen at the same time.
My A123 cells have very low resistance so they have no problem with that.

Remove the BMS? Why?
On my A123 I work without BMS for 5.5 years and the cells are fine, but I assume it's only because it's A123 and their superior nanophosphate tech.
On regular Li-Ion I don't dare to think to work without BMS, but am I wrong?

 

[whatever]

why not measure the amps rather than the power? can you measure regen amps on the CA?
If your measuring power of regen you dont really know how many amps, as you dont know what voltage the controller is creating for the regen ( the regen voltage has to be higher than the battery voltage).
Lets assume the controller is creating 60volts for the regen into the 48v battery pack ( more like 52volts or so),
using formula:
amps = power / volts
amps = 1200 / 60 = 20amps
So even at 1200watts is only 20amps or so, better to measure the actual amps being created.
If your getting 20amps into a123 20ahr cells, then will be even less with li=ion cells of equivalent or less ahr.

Sure you can use li-ion pack without bms, need to keep on eye on voltage of cells with cell log or similar and balance when required. I've been doing it for over 10years with lithium cobalt oxide cells ( more dangerous for fire risk than other types).

 

[The Stig]

Hi all, I have a general BMS question.

I currently have two typical china-made batteries in parallel and I use one charger with them. The charger plugs into the charge port on one of them and the second gets charged through the discharge leads at 4.2V per cell.

Would you expect the second battery to be balancing or is there a chance its not balance charging?

 

[thunderstorm80]

Hi all, I have a general BMS question.

I currently have two typical china-made batteries in parallel and I use one charger with them. The charger plugs into the charge port on one of them and the second gets charged through the discharge leads at 4.2V per cell.

Would you expect the second battery to be balancing or is there a chance its not balance charging?

Why not split the charging terminals so you can parallel charge both packs via their charger ports?
In general, for your question, if the second battery is only fed via the discharge leads and it has a separate charge plug as well - then it's almost 100% that it won't be balanced at all. The BMS would monitor the cells only when the current is transferred via the charge terminal.