[ruckus]

Hello,
I know the pros and cons of BMS are hotly debated. Please do not do that here.

I am starting this thread in hopes of avoiding WHETHER there should be a BMS and instead wish to discuss WHAT TYPE of BMS would work best, what are necessary features, and what are optional features. (I am assuming Lifepo4 as the current standard).

First, a little background to give a common starting point:

If EV's are to be mainstream, they need ABSOLUTE reliability. I am talking 10-20 years and 200,000-300,000 miles as is common with internal combustion engines.

My grandmother is NOT going to check cell voltages and "bring up" low cells. This is a ridiculous concept. Rickard (EVTV.ME) has proven the need for some type of BMS. After a mere 3,400 miles his pack already varied by ~.6 volts. Way too much to meet the long-term reliability function described above. This is worse than points ignition and carburetors. Unacceptable.

Here are the 3 things I think are absolutely necessary in a BMS:

1. High voltage cutoff (HVC). Right now, shunts seems to be the most popular and effective way to do this. They pass the charging current past a "full" cell to those still in need of charging.

2. Low voltage cutoff (LVC). Could be an alarm or total system shutdown.

3. Capacity balancing. (bleeding energy from the strong cells to the weak). A pack is limited by it's weakest cell. Capacity balancing seems like the only way to utilize the extra energy from the stronger cells and not be limited by the weakest cell. This could be a resistor system or a capacitor system not unlike the failed evassemble junk. This seems especially important as packs age and the capacity difference becomes greater.

I am building an EV using the EVpower bms. It only has #1 and #2. Is there a simple/inexpensive/compatible way to achieve #3? It seems like the key to getting maximum range AND max life from the pack. Any input/experiences on the evpower bms appreciated.

Cheers!

 

[Anaerin]

I don't think "Capacity balancing" is really going to be a problem. With all cells connected, the parallel strings (if left alone) will average themselves out, with weaker (lower voltage) strings being brought up and stronger (higher voltage) strings being depleted as the pack attempts to reach equilibrium. In thoery, you could connect every cell in parallel and let the entire pack balance itself out, but that would require a lot of switching, and definitely can't be done while the pack is in use.

As I talked about in another thread, I think each cell should be individually managed, with a smart controller. Then each cell can be charged to "Fully charged voltage" (Call it CVC) and switched out, easing the load on the charger and speeding up the charging rate. It would also top-balance all the cells (At CVC, rather than at HVC level, which would be above CVC). And it would be able to proactively switch out cells when they reach their LVC level. Along with thermal monitoring of the cells, to enable heating when necessary, and many other potential monitoring points, this would give the best protection for all cells, and the longest life.

 

[Jimdear2]

Hello,
I know the pros and cons of BMS are hotly debated. Please do not do that here.

I am starting this thread in hopes of avoiding WHETHER there should be a BMS and instead wish to discuss WHAT TYPE of BMS would work best, what are necessary features, and what are optional features. (I am assuming Lifepo4 as the current standard).

First, a little background to give a common starting point:

If EV's are to be mainstream, they need ABSOLUTE reliability. I am talking 10-20 years and 200,000-300,000 miles as is common with internal combustion engines.

My grandmother is NOT going to check cell voltages and "bring up" low cells. This is a ridiculous concept. Rickard (EVTV.ME) has proven the need for some type of BMS. After a mere 3,400 miles his pack already varied by ~.6 volts. Way too much to meet the long-term reliability function described above. This is worse than points ignition and carburetors. Unacceptable.

Here are the 3 things I think are absolutely necessary in a BMS:

1. High voltage cutoff (HVC). Right now, shunts seems to be the most popular and effective way to do this. They pass the charging current past a "full" cell to those still in need of charging.

2. Low voltage cutoff (LVC). Could be an alarm or total system shutdown.

3. Capacity balancing. (bleeding energy from the strong cells to the weak). A pack is limited by it's weakest cell. Capacity balancing seems like the only way to utilize the extra energy from the stronger cells and not be limited by the weakest cell. This could be a resistor system or a capacitor system not unlike the failed evassemble junk. This seems especially important as packs age and the capacity difference becomes greater.

I am building an EV using the EVpower bms. It only has #1 and #2. Is there a simple/inexpensive/compatible way to achieve #3? It seems like the key to getting maximum range AND max life from the pack. Any input/experiences on the evpower bms appreciated.

Cheers!

ruckus,

I agree in principle, if not in process. I think the best way to accomplish #3 is to combine it with #1.

If each cell has its own dedicated charging system and all the chargeing systems are set to a common HVC then each charge will not only charge the battery to the selected percentage but balance the battery as well.

We are in process of building individul charging boards for our 3P 50S Headway 38120S battery. Each 3 cell parallel buddy stack will have a dedicated charger attached. The heart of the charging boards is an isolated DC to DC converter board that is capible of very accurate voltage control, In our crude system we will have to manually monitor the output voltage. A properly designed commecial system all the manual monitoring would be automated.

The process of charging works as follows.
A 48 volt buss supplies the charging boards.
Each board will charge it's 3 cell stack (this could be a single prismatic cell as well) until the preset HVC point is reached. Once the voltage tops out the board just sits there witha trickle charge to the battery
During charging the 48 volt circuit is monitored for current.
When all cells reach the common preset voltage and current flow in the 48 volt buss drops to a maintainence point, the battery can be considered charged and balanced.

There are a couple of very nice positives to this system as well. The input voltage of all of the DC to DC boards I investigated was quite broad. The boards I chose have a 35 to 75 volt input tolerance. Another nice feature is the potential for fast charging. The boards I chose will output our chosen voltage (3.55 volts) at up to 25 amps if supplied with 48 volts at 1.8 amps so potentially I could supply each of 50 3 cell stacks at 3.55 volts and 25 amps with a 48 volt 100 amp supply. Reasonably easy to do with a decient charger and and PBA battery pack. That means I could recharge our 170 volt 24 amp hour battery from a 50% SOC in about 30 minutes.

I know some sloppy math in there but you get the idea.

Our pack is small because it is going into a small vehicle. I'm sure that a good EE could scale this up to any size.

 

[cyclops2]

I am new to your forum.

All the ideal charging features have been in use for many years in the electric powered airplanes.
Our LIFEPO4 chargers do every possible check before starting a charge. They check for out of average cells, balance, ramp up to a fast charge switchover to ramp down at 90%, then do a final peak charge to cutoff. Full shut off.

The motor controllers have a management system that is able to reduce power if 1 cell drops to much compared to a preset level.

The BMS chips are out there. Just have to web search several hours or be very lucky to find them quickly. Most are already in production by the Chinese electronic houses.

 

[karlos]

You guys are making a lot of sense; I've long thought individual cell chargers are the way to go and it's good to hear they are out there. Now to find a source....

 

[ruckus]

Thanks for the replies.

I love those blue Headway cells compared to the plastic-yellow of Thundersky.

I will lay out why I think "capacity balancing" is necessary (I made that term up). In any given group of cells, the capacity is never equal. Over time the difference can become greater. There are other causes of inequality such as replacement cells.

Top balancing is good because it fills the tank brim full, so to speak. But at the end of day, your range is still determined by the weakest cell. You might have 10% of your pack left, but one weak cell will trigger the low voltage alarm/shutoff.

To actually get the potential range out of the pack, the weaker cells need help from the strong cells. Otherwise, the extra capacity of the strong cells will never be used.

I realize batteries in parallel naturally do this, but mine are in series.

If the cell equalizing function was strong enough, it could replace the shunts normally used in top balancing.

It should be mentioned that neither top or bottom balancing allow the full capacity of the battery pack to be used. Only some kind of energy transfer will allow this.

Suggestions?

 

[Anaerin]

Top balancing is good because it fills the tank brim full, so to speak. But at the end of day, your range is still determined by the weakest cell. You might have 10% of your pack left, but one weak cell will trigger the low voltage alarm/shutoff.

Which is something else with my proposed BMS. An LVC won't disable the whole system, just drop the failing cell out.

 

[cyclops2]

Below are the pages of a VERY accurate & safe LIFEPO4 charger. Reason I am posting it is so everyone can read all the different checks and safety steps it AUTOMATICLY does.

The BMS chip...charging in this case....is old. But proven very reliable. Updates are available. I have never needed one. The unit is a brick. Have connected it to the power source in reversed polarity more than once.

This is for electric powered RC planes.

 

[cyclops2]

Below are the pages of a RC plane speed controller. Small. Yes. ONLY 6 amps.
The selectable features are very good for $15 . Load failure to start immediately, features are very good. Overheat protection is excellent.
The battery chips & processors are out there.
The " GUARD SERIES " does have Low Voltage Sensing of the battery & EACH CELL. 1 cell goes low. Reduced power to limp home.
This controller ranges up to 6 LIFEPO4 cells or 18 NICAD / NIMH cells.

The DIY electronic guys could daisey chain the chips to any # of cells.

 

[EVfun]

It should also be mentioned that you don't want to use the full capacity of the pack. Leaving some capacity on the table at both ends of the capacity will extend pack life. However, Lee Hart has already built such a system for his AGM lead acid battery pack. The short of it is to have an isolated charger than accepts DC input and outputs the voltage required for one cell. Then charge the lowest cell while driving, while parked, or while charging. Every X number of seconds stop, wait a few seconds, and then check for the lowest cell. If the cells are all close enough do nothing. If they are to far apart charge the lowest cell for X seconds.

I would suggest looking at cars out there using Lithium packs. See what kind of information is available on the the BMS in the Tesla or Leaf.

 

[tomofreno]

High voltage cutoff (HVC). Right now, shunts seems to be the most popular and effective way to do this. They pass the charging current past a "full" cell to those still in need of charging.

Shunts don't give an HVC signal. They are there to shunt current when cell voltage is above some threshold voltage until the HVC voltage is reached at which point the charger is shut down by the bms. You don't need shunts for HVC, it will work quite well without them - at least on the minbms I use. Same for evpower I believe, since I think they both use comparators to trigger a relay for HVC.

Low voltage cutoff (LVC). Could be an alarm or total system shutdown.

It can also be an alarm with a significant reduction in throttle to get your attention, but leave enough power to get out of the way of that truck you just pulled out in front of and floored your throttle to get out of the way because it is going faster than you thought, which caused your pack to sag and trigger LVC.

 

[GizmoEV]

It can also be an alarm with a significant reduction in throttle to get your attention, but leave enough power to get out of the way of that truck you just pulled out in front of and floored your throttle to get out of the way because it is going faster than you thought, which caused your pack to sag and trigger LVC.

This reminds me of another thing a BMS should do. It should keep track of the Ah in and out of the pack. I can get a LVC with cold batteries. The system should see that there is plenty of energy in the pack and delay the alarm. Maybe watch the individual cell voltage and if it continues to drop then give an alarm otherwise no alarm. If the voltage bounces back when the load is reduced then all is ok. Also, it could compare the voltage of one cell with the rest. If it is low compared to the others maybe alarm otherwise stay silent.

 

[ruckus]

Which is something else with my proposed BMS. An LVC won't disable the whole system, just drop the failing cell out.

I see problems with the switching system:

1. you are talking about buying 30+ contactor-sized switches. This is more cash, space and complexity than desired. This also means WAY more connections, lugs, nuts, etc. Plus, there is the added resistance of all those switches. Bus bars are bad enough.

2. Lack of balancing GUARANTEES that some cells will become deviant from the pack, causing more and more switching as the pack ages.

 

[ruckus]

It should also be mentioned that you don't want to use the full capacity of the pack. Leaving some capacity on the table at both ends of the capacity will extend pack life.

Right, but even if you are aiming for 70% dod, 1 low cell will only allow you to get, say, 65% out of the pack. You are still missing that 5% unless you want to push that one cell harder which means it will age faster and the problem will only get worse and worse.

However, Lee Hart has already built such a system for his AGM lead acid battery pack. The short of it is to have an isolated charger than accepts DC input and outputs the voltage required for one cell. Then charge the lowest cell while driving, while parked, or while charging. Every X number of seconds stop, wait a few seconds, and then check for the lowest cell. If the cells are all close enough do nothing. If they are to far apart charge the lowest cell for X seconds.

Now THIS is starting to sound good. Maybe not perfect, but good. I would only want it to do evaluation and charge while driving since they will start the drive equal from the shunt charging. It seems like the charger could be pretty small since the amount of energy needed is only the difference in capacity between the strongest and weakest. I would not want that on all the time as an errant cell could cause pack drainage if the system was constantly trying to "fix" the low cell.

Has anybody used a trickle resistor system in parallel for a series pack? I am not an electrician, so this may be a really stupid idea.

 

[Anaerin]

I see problems with the switching system:

1. you are talking about buying 30+ contactor-sized switches. This is more cash, space and complexity than desired. This also means WAY more connections, lugs, nuts, etc. Plus, there is the added resistance of all those switches. Bus bars are bad enough.

The way I saw it, you could use something like a pair of FETs to do the switching.

2. Lack of balancing GUARANTEES that some cells will become deviant from the pack, causing more and more switching as the pack ages.

What lack of balancing? I'm proposing that all the cells be charged to their individual "Fully Charged" voltage, rather than charging until only one cell trips HVC and stopping (Which will leave all the cells in an indeterminate state. Hopefully that state will be "close to full", but it's not guaranteed). As the cells reach "Full", they are switched out, and charging continues on the rest of the not-quite-charged cells until the entire pack has reached "Charged" status, fully and completely top-balancing the entire pack.
With the LVC isolation as well, it means that the pack can effectively be both top- and bottom- balanced, and if you put some Ah counting in there, you could even middle- balance the entire pack too.

 

[GerhardRP]

The short of it is to have an isolated charger than accepts DC input and outputs the voltage required for one cell.

I was thinking of using these VTM modules.
http://www.vicr.com/cms/home/products/vi-chip/vichip_VTM_transformers
They are actually bi-directional dc transformers.
I would have one on each cell with an intermediate bus at 24-48 volts. The only problem is that the operating voltage difference limit is not high enough, even though spec'd isolation is > 2kV. They could be disabled except near the top or bottom state-of-charge.
Gerhard.

 

[ruckus]

What lack of balancing? I'm proposing that all the cells be charged to their individual "Fully Charged" voltage... As the cells reach "Full", they are switched out, and charging continues on the rest of the not-quite-charged cells until the entire pack has reached "Charged" status, fully and completely top-balancing the entire pack.
With the LVC isolation as well, it means that the pack can effectively be both top- and bottom- balanced, and if you put some Ah counting in there, you could even middle- balance the entire pack too.

Ok, I see, sorry. Yes, top balanced through switching instead of shunting to produce the same effect. I wouldn't say the bottom is balanced, but protected from under-voltage.

How would pulling cells out affect other components like the dc-dc? If the nominal voltage is 96v, then when the cells reach 2.8v the total is already down to 84v. Switching out a couple of cells at that point could drop it into the 60-70v range and cause the DC-DC to shut down.

Also, is there currently available a cheap switch which can handle the 1000A?

 

[Jimdear2]

Ok, I see, sorry. Yes, top balanced through switching instead of shunting to produce the same effect. I wouldn't say the bottom is balanced, but protected from under-voltage.

Ruckus, GerhardRP, Anaerin,

Please reread my post, #3 in this thread. Help me out here. What did I not include in my post. I thought I had described exactually what you guys are discussing here.

In our system we have dedicated battery chargers mounted to each parallel stack of cells. We are using this configuration instead of individual prismatic cells because we need to draw a large current for a short period.

We have a an isolated DC to DC converter (a SynQor PQ48033QNA25NKS )that takes in a 48 volt nominal (35 to 75 volts) input from a charging buss. This board when modified allow adjustible output of 3.4 to 3.6 volts and up to 25 amps. These are mounted to a circuit board that allows us to adjust the output of the DC to DC unit to voltages with with 0.xxx digit accuracy.

Depending on the SOC of the cell(s) we can pass up to 25 amps of current into each battery stack until the output voltage reaches reaches the desired cell voltage, in our case 3.550 volts, at that point the individual charger can no longer push current into the battery so it stops charging that battery stack and floats.

Meanwhile each individual charger board is working to reach the same status. By monitoring the cell voltage of each battery and the input current of the 48 volt power supply buss we can see when all batteries have reached fully charged status.

This is a prototype system that is built with off the shelf components and requires manual monitoring. A purpose built system could have a lot of bells and whistles added that would eliminate need for manual monitoring.

One thing we will check during the next summer is, if as we beleive, we can just leave it plugged in because when the battery cells and charger boards reach equalibrium nothing will happen. The chargers should just float.

We chose 3.550 volts to start out because that is just below the knee of the charge curve. We may go lower. Whatever end voltage we choose the final result will be all cells charged equally.

So we have a system that will safely charge a LiPo battery without worry of over volting and damaging a cell and each charge ends with all cells charged to the same voltage i.e. a balanced pack.

Our 50 cell system can be probably be duplicated for under a $1000 dollars and does include in that price cell voltage monitoring and LVC that is part of the system (Cellog 8Ms with the low voltage alarm for each monitored cell set to trigger a throttle cut back). So we have a safe LiPo battery charger with a Battery Monitoring and Management system for around a $1000.00.

Jim

 

[DIYguy]

It really is a nice system Jim. If I had not already purchased a charger for almost $2,000.... I would have likely gone this way. I think you can get these bricks that pass a lot more than 25 amps for a fairly low price also... yes/no?

 

[Jimdear2]

It really is a nice system Jim. If I had not already purchased a charger for almost $2,000.... I would have likely gone this way. I think you can get these bricks that pass a lot more than 25 amps for a fairly low price also... yes/no?

I guess it is an idea that has kind of been around for a while but no one, at least no one willing to spread the word, has woked with it. It does seem to make sense to limit the voltage right at the battery with individual monitored chargers that cannot damage a cell then with a very expensive charger and and after the fact shunts or disconnects.

Two areas that seems to need further investigation by someone with more knowledge then me.

One is ths systems ability to deliver large current right at the battery, current produced by the DC to DC brick from a higher voltage relitivly low amprage feed buss. Seems like a good way to deliver 1 hour recharge without a dangerous voltage/current, produced by a very complicated and expensive charger in a cable handled by the end user.

The second is monitoring using coded information packets piggybacked on the charging buss or main current cableing. If each board has a unique ID, that information could be processed in a central unit and a simple user friendly interface developed. There is also the RF that is assigned to systems like tire pressure monitoring.

I still think it's the battery cell OEMs responsibility to design an intigrated modular system that at least can be purchased with the battery for a few dollars more per cell.

When I first though of trying this type of recharge set up and posted our thoughts. RWaudio was the only person who respondid had been activlly persuing it. He was very helpful as well.

The bricks can be a bit pricey, shopping helps. The types we are using have been on eBay priced at from $5.00 to $20.00. I believe they are obsolete but there seem to be replacement units with the same specs available I got lucky and found someone who had 63 new ones for just under $5.00 each shipped.

Oh Heck, sorry I've gone and done it again . . . Run on and on and on.
Jim