New 16-cell Battery Management System (BMS)

Hi Guys

Great work!! just brilliant!! I am so pleased that you guys have got this off the ground, exiting times indeed!! some of the Chinese BMS systems that I have seen have been poorly implemented and badly made, its great to see this on here, does the BMS offer any short circuit protection? or are you just using fuses?

Series connection was indeed a problem on the stock Point-1 packs and Bob you are quite right those BMS boards used 55V fets and if they disconnected under series load they would often fry, both me and Jozzer replaced the stock fets for 100V fets on twin packs and it isn't a problem any more, the BMS does warm up a little under load.

I think that you are on to a winner with these packs fellas, most of the Chinese batteries that I have stripped have been very poorly made and they either lasted for a reasonable amount of time or failed very quickly, I cant vouch for your cells and time will tell how they preform, however I think poor construction and bad BMS design was the Achilles heel for a lot of the Chinese packs and not the cells themselves.

Looking forward to seeing some board photos and pack pictures, these things should fly off the shelves guys! I wish you all the best with them.

Knoxie
 
thanks for the kudos, knoxie. it means a lot coming from someone who has seen what is available. i think our shunt based voltage equalization system will get the best performance possible out of the cells by simply making sure they all get fully charged and making sure none are overcharged/overdischarged. With the cells easily capable of delivering over 100A we figured fuses were a reasonable compromise between complexity and protection. The cutoff circuit uses 2 100V .004 ohm fets, so it should barely get warm. If enough people want overcurrent protection we can add it to the next version.
 
Hi Bob

Yes I have seen quite a few BMS boards up close! ha ha and its hard to trust some of them, I dont think that short circuit protection should be as much of an Issue, this of course was essential with the Lipo packs to prevent the end user from fuse nailing a pack, it would add a little more cost and complexity but may be a feature people would want as a double back up.

Again real nice work and nice design from Gary on the board layout as well, a well designed BMS should help protect the packs, the biggest problem with a lot of high capacity packs is cell replacement, its very well having the BMS trip the pack but it needs to be nice and easy for the end user to change out a bad cell ,this was of course the problem with the point-1 packs as they were wax sealed in a metal can! of course the BMS may well save the battery venting and making a mess! which I guess is more relevant, however its still hard to take that the 750 bucks you paid for a battery is out of action just because one cell is bad.

Looking at your pack designs it doesnt look hard to swap out a bad cell, this is good news!! most people I am sure would undertake it else they would likely have it done under warranty or paid service replacement?

Great News Bob!! well done again!

Paul
 
Bravo, Bob and Gary, well done! :D

I am building a 15s4p (57V 11.6Ah) emoli cell pack; this week I got the last Milwaukee drill packs I needed. I've been worried about how to provide low-voltage protection, and how to recharge it.

For LV protection I was going to use the 10-cell LVC schematic Gary provided and build a 15-cell version. For charging I was going to use 5V wall-wart power supplies with a high-current diodes in series to drop the voltage to 4.3V.

It makes me real happy to see a complete solution appear that I can purchase.

One of my questions has been answered already; I can jumper this 16-cell board to work with 15 cells. But I have another question...

Can I substitute different voltage detectors, and/or change resistor values, to change the lower- and upper-voltage limits to 2.5V and 4.25V?

The lithium-maganese emoli cells have different values than the LiFePO4 cells.

disadvantage
 
disadvantage said:
Bravo, Bob and Gary, well done! :D


For LV protection I was going to use the 10-cell LVC schematic Gary provided and build a 15-cell version. For charging I was going to use 5V wall-wart power supplies with a high-current diodes in series to drop the voltage to 4.3V.

Can I substitute different voltage detectors, and/or change resistor values, to change the lower- and upper-voltage limits to 2.5V and 4.25V?

The lithium-maganese emoli cells have different values than the LiFePO4 cells.

disadvantage
Click to expand...

if you are planning on using a bunch of wall cube supplies to charge a serial string you may run into the problem that they are not isolated from each other and cannot be used that way. if you want to build a single cell type charger you need to use isolated modules which are generally DC powered. If you are using regulated voltage charging of each cell our shunt regulators would not gain you anything, but if you did find some wall warts that are isolated it would keep their voltage limited to the set level.

the lvc circuit can work at any voltage above 2.1v by using a voltage divider on the input. this by necessity increases the current drain from a microamp to 100x that, if 1% accuracy is desired. there are also devices available at different voltages; 2.7,2.9,3.0 V if one of those works you can keep the drain at a microamp.

the shunt operating voltage is adjustable with a pot, nominally from about 3.5-4.0v but this can be changed to any voltage above 2.5v by changing resistor values.

i have given credit to gary previously for doing a great job on the pc layout, but he was also instrumental in developing the design. he has killed a lot of these cells and knows how to do it, so it stands to reason he should know how NOT to do it as well?
 
Talking about battery overcurrent protection during the ride, I have a 30 A automotive fuse on my battery.

The diode discussion got me interested - are you saying that 2 packs in series with BMS, when the LVC on one pack trips it presents an open circuit? Can't figure out how you could hook a normal diode in parallel to each pack. Seems like attaching it the same + - as battery would short it, reverse -+ would block current when the LVC opened the pack. is there any leakage current from the diode while storing, or is diode only connected during the ride? Quick sketch ?

I have a 36V 12 ah pack now - the rectangular cells that YPedal has. What would be the affect or issues with connecting a 18 or 36 V 40138 pack 10 ah in series with it? Each pack using your LVC circuit.. Does 10ah in series with 12 ah cause any problems?

DK
 
:idea: Oooo, that's a good idea.

If you put a diode across each pack in the non conducting orientation, when one pack BMS opens, the diode will conduct and keep the voltage across the BMS switch from seeing more than one pack voltage (preventing it from blowing up). The diode would need to handle the full current.

If you had two 36v packs for example, when one opened, the controller would suddenly see only one pack and drop to half speed.

Edit:
I don't know why I didn't think of this before. Not enough kimchi I guess.

If you have multiple battery packs with a built in BMS, like duct tape, ping, or factory DeWalt drill packs, If you put a backward diode across the output you could stack as many as you want in series without risk of blowing the BMS when one trips.

In the normal operating mode, the diode would have essentially zero drain on the battery.

There's something to think about Gary...

If you had smaller chunks of BMS in series, you could make a "fail soft" kind of feature like the new Valence packs that would allow limp-home capability if one part of the pack failed.
 
knoxie said:
Hi Bob

Yes I have seen quite a few BMS boards up close! ha ha and its hard to trust some of them, I dont think that short circuit protection should be as much of an Issue, this of course was essential with the Lipo packs to prevent the end user from fuse nailing a pack, it would add a little more cost and complexity but may be a feature people would want as a double back up.

Again real nice work and nice design from Gary on the board layout as well, a well designed BMS should help protect the packs, the biggest problem with a lot of high capacity packs is cell replacement, its very well having the BMS trip the pack but it needs to be nice and easy for the end user to change out a bad cell ,this was of course the problem with the point-1 packs as they were wax sealed in a metal can! of course the BMS may well save the battery venting and making a mess! which I guess is more relevant, however its still hard to take that the 750 bucks you paid for a battery is out of action just because one cell is bad.

Looking at your pack designs it doesnt look hard to swap out a bad cell, this is good news!! most people I am sure would undertake it else they would likely have it done under warranty or paid service replacement?

Great News Bob!! well done again!

Paul
Click to expand...

With these cells having 6mm studs on each end, changing out a cell is not all that difficult. Just unbolt one side, swap the bad cell out, and bolt everything back up again. I'm not guessing this will be something that will need to be done much, or LiFeBatt wouldn't have felt confident enough to give such a long warranty, but nonetheless, it won't require major surgery. :)

-- Gary
 
Ypedal said:
Quick question ( preparing ahead of time so i don't have to cut and re-weld later ) ..

Dimentions of the BMS board ( or VMS or CMS.. i don't know anymore lol )

aprox 3 x 6 inches.. but how thick/deep once all the components are mounted.. ?
Click to expand...

I'll post some pictures of the first one a bit later today. It is indeed 3" x 6", and with the heatsink, it is just about 1" high.

I'm assembling/testing the first three boards today. They will go on one 16-cell and two 12-cell LiFeBatt packs that I am building up as well. I sent Bob some boards as well. Hopefully he will get them today, so he can start buiding/testing as well. I got all the parts together already.

-- Gary
 
it is also possible to use diodes to combine parallel strings so that each string has its own bms board. you use a dual diode at the + end of the cells with the common cathode as the output. now if one cell drops out the bms will open the fet at the minus end of that string and the parallel string will contrinue to support the load at the same voltage. this is the way to get protection on each cell, where if you combine cells in parallel first you can only protect each group. (which may be fine in many cases)
 
As Bob says, the easiest way to raise the LVC cutoff voltage would be to use the higher voltage parts. They don't have 2.5V parts, but 2.7V is available. That's what I was using for the a123-based LVC circuits. They should work fine for eMolis as well. The reason I went with 2.7V for a123s and 2.1V for LiFeBatts is because of the very high C-rating for the a123 cells. What I found is that with two or more a123 cells in parallel, even 50A+ loads would not pull the cell voltage, for healthy cells, to below about 3V per cell. At least not until the end of the duration, when it drop very quickly. Using 2.7V allows the LVC to detect when this happens as soon as possible. If the LVC ever started tripping in the middle of the pack's duration, that would be an indication that there is a bad cell/block of cells.

With the LiFeBatt cells, I've seen 85A loads pull down the voltage to the 2.7V-2.9V per cell range, towards the middle-to-end of the duration, which would trip the 2.7V parts. Using the 2.1V versions seems to be just right. I've done several tests now, where I've run the packs down to where the cutoff starts happening. At first you just feel it "hit" once or twice, under basically full throttle. You can back off the load and get another mile, or two, and then even about half throttle will cause it to start oscillating at about a 2Hz rate. At that point the Ah readout on the WattsUp units were showing 9.6Ah-9.8Ah as being used.

It's been awhile since I've played with eMoli cells (we used to rip out the 7 cells in Milwaukee tool packs, and use them in our RC helicopters, before the a123 packs came out...), but I think the optimum CC/CV cutoff voltage is 4.1V, instead of the 4.2V used by Lithium-Cobalt-based LiPoly packs. As Bob points out, the crossover point for the shunts can be adjusted via an on-board pot, but the adjustable range can be modified by simply changing one resistor.

There are single cell 4V chargers that are isolated, but these ae typically limited to about 1.5A-2A. With this BMS you can use a single higher power charger or even a regulated supply, like the Mastech units, or like the the 0-60V/0-15A HP supply that Bob found for me on ebay (it was $125...). The shunts in the BMS act like individual cell CV chargers. Once the cell voltage gets to the crossover point (3.65V for LiFePO4 cells, or 4.1V for emolis...), the shunt clamps it there. The current the cell can accept then tapers off, but what ever it doesn't need/use gets bypassed so that the next cell in series can have all of the current available, if it isn't at the cutoff yet.

-- Gary
 
The answer got too complicated for me.

Fechter - you might be wrong. For 2 packs in series, seems like the diode would have to be in the same polarity as the battery it's across. In the conducting mode.

Otherwise if the BMS opens that particular pack, the reverse conducting diode wouldn't let any current pass through the series circuit as a whole.

Bob-Gary That's why I asked for a sketch (with the little diode arrows, +s and -s, current flow arrows, etc) so I wouldn't misunderstand. :wink:

I'm hoping different type packs could be used in series.

The chemistry or format should not matter, 30A flowing through both packs in series should not care (if both packs have a high enough "C" rating.

With this diode thingy used with a BMS, would surmise I could use a 12 ah battery in series with a 10 ah battery.
Suck the first bottle dry, open and bypass, then suck the 2nd bottle dry of it's remaining 2ah.

First the controller would see 72V, then 36V with 2ah left.

?? :?:
 
Plus you wouldn't use the standard 1A diode...

Maybe one of these?
http://ixdev.ixys.com/DataSheet/L273.pdf
 
Here's a shot of the almost finished board:

BMS-02a.jpg


It is minus a few parts and the heatsinks/top plate, ready to test and adjust each channel. After that, I'll install the rest of the parts, and do the final checkout. I need to do one more iteration on the board layout, to pick up a couple of minor issues. I should get these back by Wednesday or Thursday.

Because of the fair amount of labor involved, I'm probably only going to offer this version separately, in kit form. We are going to do an SMT version soon, and outsource the fabrication/assembly. Until then, the only ones I'm going to do myself will be for use with LiFeBatt packs.

I'll post some more pics when I get the boards completely finished, and mounted on a 16-cell LiFeBatt pack with the heatsinks/plate installed.

-- Gary
 
Gary, you said it could be downsized. Can it be upsized to 18 Cells?
 
drewjet said:
Gary, you said it could be downsized. Can it be upsized to 18 Cells?
Click to expand...

Think about it ... you downsize by not populating the unused part of the board and soldering in jumper wires.

"Upsizing" a board would be like trying to stretch a plank that is too short. Cutting off material is easy, adding it is kinda hard.

I think you might need two boards.
 
"Upsizing" a board would be like trying to stretch a plank that is too short. Cutting off material is easy, adding it is kinda hard.
Click to expand...

Yes, but a 20-cell board for those of us with 60-volt systems may be a useful addition in the future.
 
PJD said:
"Upsizing" a board would be like trying to stretch a plank that is too short. Cutting off material is easy, adding it is kinda hard.
Click to expand...

Yes, but a 20-cell board for those of us with 60-volt systems may be a useful addition in the future.
Click to expand...

I get that, but the question was about modifying the existing 16 cell board.
 
Maybe when we go to the SMT version, I'll make the layout so that it can accomodate 20-24 channels, but to fit into the 3" x 6" form factor that would still fit on the front of a LiFeBatt pack, 16 was all that could fit. The design will accomodate pretty much any number of channels.

To do a 20s setup, you could use two boards, but only populate one of thm with the LVC and charge current FETs, and associated logic. The easiest way to do this would be to populate the first five of the eight channels, on the top and bottom rows on the main board, and then do the last five channels on the secondary board. Two jumper wires would be needed to connect the channels between the boards, and then 4 small gauge wires to connect the opto logic lines.

-- Gary
 
Thanks Gary. I am sorry for my bad wording, I meant, well you know what I meant. 2 boards of 9 or even 3 boards of 6 would work in my current application.

I guess I should of added that these are 4.2 volt lithiums as well. What is the max voltage? Or would it not matter since it is only looking at each individual cell. Just thinking of the future when I get enough funds to do a car or truck.
 
Or just use all 16 on two boards and make a 106V battery.

The perfect bicycle battery. Go Speed Racer, go ...
 
drewjet said:
Thanks Gary. I am sorry for my bad wording, I meant, well you know what I meant. 2 boards of 9 or even 3 boards of 6 would work in my current application.

I guess I should of added that these are 4.2 volt lithiums as well. What is the max voltage? Or would it not matter since it is only looking at each individual cell. Just thinking of the future when I get enough funds to do a car or truck.
Click to expand...

I think we talked about other chemistries a page, or so, back. The cutoff voltage used by the shunts is adjustable, so you could simply adjust these to 4.2V instead of the 3.65V we use with LiFePO4 cells. The LVC cutoff is controlled mainly by which version of the TC54 detector chip you use. We are using the 2.1V versions here, but I have used 2.7V for a123 cells. I think they have 2.9V and 3.0V versions as well, so this could be a simple substitution. The lVC cutoff and charger current control logic doesn't know, or care, what the voltage for each cell is, as the only connections to the channels are through the opto couplers.

-- Gary
 
You must log in or register to reply here.

Similar threads

D
Bms help needed
Replies
5
Views
353
amberwolf
amberwolf
L
BMS or active battery balancer for 20s pack
Replies
3
Views
719
amberwolf
amberwolf
warrah
(UK) 3010wh, 288 cell battery pack with BMS and charger £150
Replies
2
Views
266
warrah
warrah
W
BMS compatibility with new controllers and battery build 16s9p
Replies
14
Views
259
WonderDrift
W
pistolpete1776
  • Question
BMS bad... or something else?
Replies
19
Views
991
amberwolf
amberwolf
Back
Top