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

[tomw]

Could someone please direct me to the location in this thread that discusses how the charger current is controlled? Does this require a certain charger type with the proper input? I don't see the mechanism for interfacing between the charger and bms using HVC and PWM.

Thanks,
Tom

 

[GGoodrum]

Could someone please direct me to the location in this thread that discusses how the charger current is controlled? Does this require a certain charger type with the proper input? I don't see the mechanism for interfacing between the charger and bms using HVC and PWM.

Thanks,
Tom

All you need is either a standard CC/CV charger, or a power supply that has a constant current limit. Many "lab"-type supplies have adjustable voltages and adjustable current, but can be expensive if bought new. You can usually find surplus units on ebay, though, that can be reasonably priced. Most of these can be pretty bulky.

One option many of us are using these days are inexpensive Meanwell power supplies sold by places like Sure-Electronics, in Hong Kong. The one I've been using is the S-350 model, which comes in a variety of voltage and current combos, but are rated for 350W. For my 24s LiPo setup I use four S-350-24 units in series. Each of these are 24V and 14A. There is a pot on the front that lets you adjust the voltage from about 18-27V. I have all of them set to 24.9V, so the total output voltage is 99.6V, or about 4.15V per cell for my 24s3p LiPo pack. The only problem with these Meanwell supplies is that although they do have constant current limiting, they have it set to 130% of the rated output. For the S-350-24, the rated current output is 14.8A, but the CC limit is set to about 20A. That's fine for applications where these might be used in stereo equipment that might have surge currents that spike up to 20A, but for use a charger, the limit needs to be lower. It turns out this is very easy to do, though, by simply tacking on an extra resistor inside, in the current limit section. There's a whole thread on this here: http://endless-sphere.com/forums/viewtopic.php?f=14&t=4125.

In any case, whether you use a supply like this, or "regular" charger with a CC/CV profile, what you need is to have the voltage set to the number of cells x the max charge voltage per cell for the type of cells you are using. For LiFePO4 cells, 3.65V is the recommended "full" charge voltage and for LiPos it is 4.20V. The thinking these days is that it is better to undercharge the cells slightly, to increase longevity, so I use 3.60V for LiFePO4 and 4.15V for LiPo. For a "typical" 16-cell LiFePO4 48V pack, 16 x 3.60V is 57.6V. For a 12s LiPo-based pack, 12 x 4.15V is 49.8V.

In the BMS, the shunts are set to turn on at these same set points (3.60V for LiFePO4, or 4.15V for LiPo...). The HVC signal is set for just slightly above that (3.605V/4.155V...). When the cell's voltage hits the HVC point, the opto output for that channel is turned on. The control section looks at all of the opto outputs, which are all logically "OR'd" together. It uses this signal to control the duty cycle of a PWM circuit controlling a FET that interrupts the charge current. When the FET is off, the cell voltages will drop slightly, which turns off the optos, The cycle repeats at a fast rate and the net result is the current is limited to whatever it needs to be to keep all the cell voltages from going over the HVC set point. At some point, the high cells are going to reach the point that they just can't accept any more current with the voltage being allowed to go higher, so if the shunts were not there, the current would drop to zero. Since all the current has to go through all the cells, the cells that aren't full yet, would not get anymore current. What the shunts do is bypass current so that the cells that aren't quite full yet can continue to charge until they are full as well. The amount of current that the shunts can bypass is limited by the shunt resistor used. Typically, we've been using 6.8 ohm 5W shunt resistors, which limit the current to about 1/2A. With LiPos, I've been using 6.2 ohm 5W shunt resistors, which allow about 700mA of shunt current, but it does generate a fair amount of heat. With this version of the BMS, shunt currents can be higher, up to about 2A, which is why we've added a control for a fan.

The PWM-type control technique for limiting the current is basically what the charger/supply uses to limit current as well. Controllers do the same thing to limit battery current.

-- Gary

 

[tomw]

"The number of connections between the 6s sections and the control board have been reduced to just two, by having the LVC and HVC detection functions share the same opto outputs. This makes total sense as these functions are never used at the same time."
Except in the case of regenerative braking. This system cannot prevent overcharge of a high voltage cell by regenerative braking. The motor controller can be setup to limit regen current based on pack voltage, which should avoid overdischarge in most scenarios. A rare exception might be if a cell is at much higher voltage than other cells. I suppose even this can be avoided by simply setting the controller to limit regen above 80% charge.

 

[GGoodrum]

Except in the case of regenerative braking. This system cannot prevent overcharge of a high voltage cell by regenerative braking. The motor controller can be setup to limit regen current based on pack voltage, which should avoid overdischarge in most scenarios. A rare exception might be if a cell is at much higher voltage than other cells. I suppose even this can be avoided by simply setting the controller to limit regen above 80% charge.

Well, the shunts will come on, which will help a bit, but I feel the same way about this as I do about controlling discharge current. I think both should be done, and are in fact are already being done, in the controller, not in the BMS.

 

[fechter]

Since the LVC and HVC are shared, I don't see any reason that the HVC signal couldn't be used to inhibit regen. It would depend on the controller how you would interface this. If you're using the LVC alarm function, the alarm would also go off if you hit HVC during regen.

 

[Lock]

This msg just came across the EVDL. Curious whether this issue(?) has come up already in this thread but unsure how to search for this...
tks
Lock

Re: [EVDL] tech info or shopping news?
Wednesday, December 23, 2009 2:16 PM
From: "Rick Beebe" <rick@beebe.org>
To: "Electric Vehicle Discussion List" <ev@lists.sjsu.edu>

Bill Dube wrote:
> You are going to ruin the cells pretty fast in a high-voltage pack
> without a BMS.
>
> You also must have a reliable BMS. An unreliable BMS is almost worse
> than no BMS.
>
> With most Li-Ion cells, you are going to probably set the pack on
> fire at some point if you don't have a BMS.

I said it was a lively discussion

His "discovery" after destroying a few cells was that most BMS's--especially those derived from the Lead/Acid world--focus on equilizing charge at the top. For example, shunting current around full cells so you can pack more juice in the not-so-full one. At that end of the cycle you're pushing just a few amps into the cells.

But they ignore the bottom of the cycle. Since the cells will have slightly differing capacity, a cell may become depleted while others still have charge left. The problem is that it's _very_ difficult to tell when a cell has become depleted and during the drive cycle you may be pulling hundreds of amps from the cell. Instant death.

End-of-charge in lead/acid is pretty easy to detect because voltage drops fairly smoothly to "done." In LifePO the voltage remains flat and drops suddenly at the end. If you're pulling 150amps from a cell, by the time you detect the voltage has dropped to "dead" the cell may already have gone so far it's ruined.

As Willie said, "There are really only two things the BMS must do:
1) stop the charger when any cell reaches the upper voltage limit
2) inform the driver or stop or slow the car when any cell reaches the lower voltage limit."

Detecting that lower voltage is what's problematic (the cells may touch that voltage briefly under load which should be "ok"). No one has come up with a really good solution yet, though there have been a lot of interesting proposals put forth.

--Rick

 

[GGoodrum]

The new BMS now has an option to either stop the charge when the first cell hits the cutoff, or continue on to balance the cells. There is also a new timer that will limit the balance time to a jumper-selectable 15 minutes, 1 hr, 2 hr or 4 fours. While I agree, balancing doesn't have to be done with every charge, it still needs to be done periodically, to get the cells to a common state of charge. Otherwise cells of roughly equal capacities, but differing states of charge, will cause a reduction in the "useful" capacity/range of the pack.

In either case, you simply have to do cell level low voltage protection, whether it is simply to turn on an audible alarm, or actively limit the throttle, or a combination. Without that level of protection, you will kill cells eventually. Guaranteed. As for picking the trigger point, it just needs to be the lowest point that the cells will hit under load, just before the resting cell voltage drops off the cliff. For LiFePO4, it is pretty much anything above 2.0V, and for LiPos, about 2.5V.

The new BMS now has connections for directly driving a buzzer/alarm whenever the LVC line trips. This is in addition to our "standard" implementation of tying directly into the controller brake or throttle inputs.

-- Gary

 

[Lapwing]

...............

The new BMS now has connections for directly driving a buzzer/alarm whenever the LVC line trips. This is in addition to our "standard" implementation of tying directly into the controller brake or throttle inputs.

-- Gary

Probably not a issue for most users, but I there any way to temperature compensate the LVC in the control section, using some kind of thermistor. Making the LVC less at very low temps? and higher at much higher temps. I would like to see something like 1.8V at -5F and more like 3.1V at 95F. The thermal probe would need to be in the pack glued to a cell.

Certainly with the LiFePO like Thundersky, in very cold weather the voltage can dip a long way, when around -5F below 2.0V even with full batteries under acceleration. The full batteries are not in danger of damage, just have a higher IR, cause of temp effects. After a few miles they warm up, but in the meantime, a nice safe 2.9V cutoff would drive a man nuts.

 

[fechter]

Probably not a issue for most users, but I there any way to temperature compensate the LVC in the control section, using some kind of thermistor. Making the LVC less at very low temps? and higher at much higher temps. I would like to see something like 1.8V at -5F and more like 3.1V at 95F. The thermal probe would need to be in the pack glued to a cell.

That would be pretty hard with the way the LVC sensors are used. There might be a way to put a thermistor on each cell circuit to do that, but I think it would be hard to make it simple and still work. I'll look at it when I get a chance.

Plan B is to just use a lower LVC voltage all the time. A momentary dip to 2v even on a warm pack proabably won't damage the cells.

A microprocessor based system could probably do this with a single temperature sensor.

 

[CroDriver]

....we are finally going to be able to do a surface mount version that can be automatically assembled. These we will make available sometime in the 1st quarter of 2010.

I'm very interested in large quantities (100 pieces for the beginning)

Will this units be smaller? (I have used V2.6, seems like the new one is larger)

Btw. I can't see the figures for maximal discharge currents. What could be a continuous and a peak rating?

 

[fechter]

....we are finally going to be able to do a surface mount version that can be automatically assembled. These we will make available sometime in the 1st quarter of 2010.

I'm very interested in large quantities (100 pieces for the beginning)

Will this units be smaller? (I have used V2.6, seems like the new one is larger)

Btw. I can't see the figures for maximal discharge currents. What could be a continuous and a peak rating?

The through hole version is slightly larger than the ver.2.6. If we go surface mount, we can shrink it a little.

There is no limit on discharge current. The batteries go straight to the load. It's up to the motor controller to do the discharge current limiting.

 

[ELETTRONICABERTINI]

Ok, everything worked , I am very happy and thank the designers and all of you
I wish everyone a happy 2010

http://www.youtube.com/watch?v=8fOnTta1tdc

 

[Doctorbass]

Ok, everything worked , I am very happy and thank the designers and all of you
I wish everyone a happy 2010

http://www.youtube.com/watch?v=8fOnTta1tdc

nice Work ELETTRONICABERTINI !

It appear that the celllog8 are the great complement for those 24s BMS !!

Here is mine assembled for 40s: and modded using two FDP2532 mosfet in parallel for handling 150V at 20A continuous of charging capability:

http://www.youtube.com/watch?v=-yMgWl2xF2Y

 

[ELETTRONICABERTINI]

Doctorbass congratulations for the excellent assembly.

In the BMS that I assembled, I installed a heatsink on the power resistors to be channeled (3.9-5W) obtaining a maximum temperature of 50°C with 20°C tempertura environment, the fans are activated by the insertion of the plug recharging and are powered by a DC/DC converter outside.
Ignition of monitors is with the activation of charging or starting the scooter or a test button.

Maurizio
(Sorry for the horrible english I'm using a translator automatic)

 

[ryan_lirui]

Ok sorry to crash the thread here guys, but I'm just itching to buy a Goodrum/Fetcher BMS and TPPacks.com is all sold out. What can I do to get my hands on one of these?
Also, is there a chance I can order a pre-assembled one?

Cheers.

 

[fechter]

You might try checking with andyh. He was building them. How many cells do you need it for?

I'm afraid there may not be any stock available for a while as we are trying to get the next generation boards going.

 

[ryan_lirui]

need it for 24 cells

 

[wookey]

Oh bum, that's bad news. I just ordered my headway cells, and having read all the 'fancy new BMS' threads decided an assembled G/F analogue one was going to be the best plan for the immediate future (and save a whole load of messing about). Planning to order tonight/tomorrow. Hmm, might have to fall back on some more exotic plan...

 

[AndyH]

Ok sorry to crash the thread here guys, but I'm just itching to buy a Goodrum/Fetcher BMS and TPPacks.com is all sold out. What can I do to get my hands on one of these?
Also, is there a chance I can order a pre-assembled one?

Cheers.

Hi Ryan,

I still have V2.6c boards in stock. I can supply a bare board or a completed BMS. I plan to keep this version 'alive' until the V4 boards are ready.

Andy

Rechargeablie Lithium Power

 

[fechter]

I have one 24s bare board in the garage I could sell.

 

[GGoodrum]

Update...

I think we are finally close to finishing up the testing. We had a bunch of layout changes, for the control section, and a couple of missing trace connections on the 6s sections, and then I did a new run of boards. Here is what the latest one looks like:

View attachment 12s BMS-06.jpg
View attachment 12s BMS-07.jpg


Again this is for a 12-channel version, but the additional 6s sections can be added. As I said, most of the changes were to the control section, mainly to move things around and closer together. The big resistor is for the fan. I'm using a single 12V 40mm x 40mm fan, which consumes about 130mA. To run this right off the 50V, a resistor is needed, in this case 300 ohms. If two 12V fans are used in series, the resistor needs to be around 200 ohms, but in this case a 3W would be sufficient. I needed to test how well a single fan works, to move enough air over the 1A shunt circuits to keep the heat in check.

Anyway, if all goes well, I'm going to make these available starting early next week.

-- Gary

 

[number1cruncher]

This is awesome! I'm assuming that this setup will be used in conjuction with the LVC boards already available, right?

Not that I care much, because I'll be using it for LiPo, but will this thing work for LiFe? Or will there be two different models.

Bryan

 

[GGoodrum]

This is awesome! I'm assuming that this setup will be used in conjuction with the LVC boards already available, right?

This is a full BMS, so it includes the LVC function, but it can also be configured without the LVC function, and essentially used as an on-board charge controller/balancer. The onboard LVC function, however, has an opto-coupled output, so it can be used with throttle controls on some of the "big boy" EV controllers. I now it seems redundant to have an optocoupler on an optocoupled output, but this removes the charger/pack ground reference.

that I care much, because I'll be using it for LiPo, but will this thing work for LiFe? Or will there be two different models.

Bryan

Yes, this will also work with LiFePO4-based setups. There is a resistor change, and one other part number change, per channel, and zero changes in the control section.

-- Gary

 

[alangsam1@me.com]

When will this be available?
I have a 4 x 6S 5800 lipo packs configured 2S 2P that i have waiting for a proper BMS that i can use to manage the pack. My goal is to have the battery and BMS in the electric motorcycle and not need to take it out to charge and balance just use my 48v charger and have the BMS manage the cell by cell balance, LVC, stop at 4.2/cell limit current flow etc.
will this new offering work for my needs?

 

[GGoodrum]

They will be available one week after I finish testing, and get a quantity of boards ordered. I hope to finish testing in the next two days. Right now I'm doing thermal tests with and without fans. With 1A shunts, there needs to be some sort of active cooling. Yesterday I tried a single fan, mounted on the left, as shown below. The fan actually moves quite a bit of air, but in this "open" configuration, the air is going to find the path of least resistance, so most of the air is flowing towards the bottom right, in front of the shunt resistors.

View attachment 12s BMS-06.jpg


This is causing the shunt resistors on the far right to get pretty hot, pretty quick, like around 250F. The first shunt resistors, closest to the fan, ran cool, like around 103-104F. The ones in between got progressively warmer. I simulated an air dam, of sorts, using my fingers, to try and vector the airflow over the shunt resistors, and that did help. The temps dropped on the end resistors. So, if a single fan is used, some sort of deflectors are going to be required in order to make sure the air is forced over the shunt resistors.

What I'm trying today is two fans. I mounted a second one on the right side, also oriented to blow air inwards, and wired it in series with the first fan. I also changed the resistor to 200 ohms.

We are also thinking adding 1 or 2 thermostatic switches to the input of the 12V regulator, which can be mounted at the hot spots, so that the control section will shut down if there's a fan failure.

-- Gary