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

Gary and Richard, if the changes are not too involved, one option you may want to consider is to offer the earlier version boards at a slight discount to people with more advanced electronics skills. I for one have the tools and skills to make small changes to make it work (Eg, bending the transistor leads to correct the layout error, cutting tracks and bridging with short lengths of Kynar wire, drilling 1mm holes in the PCB to mount extra capacitors etc, ). This may help you to defray some of the development cost and also allow people to get the BMS quicker. Others may be interested in this option also.

Keep up the excellent work,

JohnB
 
Ditto - I'm more than happy to play with a prototype board, and would happily accept the risk of it needing some tweaking. I have most of the components needed already, as I was going to build my own system anyway.

You guys deserve to get some return on the cash you've pumped in so far.

Jeremy
 
OK. I showed this post to Keywin and he was blown away. He thinks it is a great circuit design.
But he has no additional insight. No words of wisdom yet. There are many a123 BMS systems available in China.
Keywin can get them. But I guess they don't do what we are looking for.

So ... This means we must do it ourselves. Rich and Gary, this Bud's for you!

PS. Once we are happy, the Chinese will glady make 10,000 clones for us! :shock:

Ha Ha ... We so crazy!
World so crazy! 8)
 
Jeremy Harris said:
Ditto - I'm more than happy to play with a prototype board, and would happily accept the risk of it needing some tweaking. I have most of the components needed already, as I was going to build my own system anyway.

You guys deserve to get some return on the cash you've pumped in so far.

Jeremy

Me too.

Nick
 
I'd agree with JB_EB's idea, but the pricing thing is up to Gary.

I've been so busy working the bugs out of the design I haven't had much time to post lately.
Gary sent me a "kit" last week with the board and a set of parts. The small parts come sealed in plastic bag with separate compartments for each value. Some of the IC's and transistors come on machine gun belts.
View attachment 5

I started the build by placing all the low profile parts like the little resistors and the optocouplers first.
Board kit assembly 2.jpg

Here's what the board looks like with the big 500ma shunt resistors.


We may need to wait until the next version of the board to offer a 500ma version due to the pin configuration problem with the transistors. With 300ma or lower shunt resistors, I think we can find transistors that will fit the holes without having to twist the leads around.

In testing a fully built board, I discovered that under some conditions the gate driver chip could go into a very high frequency (like 100mhz!) oscillation that caused all sorts of problems. Using Randomly's suggestion of providing some negative feedback through a capacitor completely solves this problem and lowers the switching frequency to a nice range between 1k and 10khz.
When I tried this before, it messed up the current regulation, but I was able to correct that by changing the value of the input resistor. I also decided a voltage regulator is a good idea and the places for it are already on the board, but we'll feed the regulator from the #4 or #5 cell tap to keep the heat dissipation down. Here's what the updated schematic looks like:
LiFePO4 BMS-v1 4d.jpg

Here's what the FET gate signal looks like when the circuit first starts oscillating:
Scope trace 1.jpg
As you can see, the duty cycle is still very high and the circuit is just barely throttling the charge current. The frequency here is just over 1khz.

As the charging voltage is increased, the duty cycle will start dropping to keep the shunts at their set point. This would also happen if one cell got fully charged before the others and the circuit was actively balancing. The higher the charger voltage goes, the lower the duty cycle gets to keep the shunts happy. Here's what the gate drive looks like with about 10v over the string voltage. You can see the duty cycle is down to about 50% and the frequency is up to about 10khz. The operating frequency will probably be lower with cells attached. I was just using a capacitor to simulate cells. Next is to test with actual cells and verify the behavior remains correct.
View attachment 2
 
Tiberius said:
Jeremy Harris said:
Ditto - I'm more than happy to play with a prototype board, and would happily accept the risk of it needing some tweaking. I have most of the components needed already, as I was going to build my own system anyway.

You guys deserve to get some return on the cash you've pumped in so far.

Jeremy

Me too.

Nick

Me too! I also have access to a pretty broad supply of analog components and test equipment. I can lend a hand with testing or data acq.
 
I'm modifying my board to do the changes that Richard has come up with, and will test this today. The good news is that since al the changes are in the FET control area, the existing boards will still work. Basically, all we did was go back to using the regulator transistor, but still tapping off the 4th cell. There is still one trace that needs to be cut and jumpered, and we need to add the cap across two pins of the gate driver chip, but these are easy mods.

Using the big 6.8 ohm 5W shunt resistors definitely generates a fair amount of heat, and the KSA931 PNP drive transistors get pretty hot as well, so using these is going to require some sort of fan, or something. On the next board, which will have all the above changes, I will also redo the pinout for the KSA931s, so we can still use these, but I think it might be better to go up in resistance, just a bit, like to 9-10 ohms, for the higher-powered version, just to get the heat down a bit more. I think the "standard" kit will come with the 15 ohm/2W resistors. This will still allow 300mA of shunt current, which is going to be more than enough for most setups (i.e. -- 10-20Ah...).

Anyway, we are dangerously close... :wink:

-- Gary
 
Maybe "shunt(s) on" line can turn fan on? That way the fan does not run all the time?
Jeff K.

Thanks for this BMS again in advance!
 
jeffkay said:
Maybe "shunt(s) on" line can turn fan on? That way the fan does not run all the time?
Jeff K.

Thanks for this BMS again in advance!

Actually, the fan can be tied in with the jumper for the input to the voltage regulator, as it only needs to on when the charger is present, and charging. Even then, it only gets hot when the charge process is almost done. For the first 80%, or so, of the charge, the shunts are not on at all.
 
Even then, it only gets hot when the charge process is almost done. For the first 80%, or so, of the charge, the shunts are not on at all.[/quote]
Yeah, well it would be nice to not have fan(s) running 80% of charge time. That's a mod we can make on our own though EARTHQUAKE HAPPENED NOW...
 
jeffkay said:
EARTHQUAKE HAPPENED NOW...

Yeah, I felt it here too. I'm in the south part of Orange County and the news is saying it was about 5.6-5.8, and centered in Chino Hills, which is about 30 miles NE of here, as the crow flies. This one had a pretty good jolt, and then quite a bit of rolling for about another 15 seconds, or so.

I lived 2 miles from the epicenter of the Northridge earthquake, and had a ton of damage. The only reason we didn't have more is because five years before, a canyon fire leveled our house, and we built a new one on the same lot that was built to much tougher earthquake standards. Some of our neighbors that didn't have their houses burnt down, had a lot more earthquake damage. As it was, it took a week to dig through most of the crap in each room. I remember speniding about four hours cleaning out a small office we had upstairs when we got an aftershock that was about 5.9. All the books and computer stuff I had just finished putting back up, ended up in a pile in the middle of the room again. I then waited a few days, until the aftershocks settled down to something manageable. We had more than 9000, over the first month, or so.

I also lived in an area that was only about 5 miles from the epicenter of the Sylmar earthquake in 1971. In that one, the shock waves were more horizontal than the Northridge quake which was very much up and down. In the Sylmar quake, our street rolled, like someone was shaking a rug, but we had a lot less damage. I watched as all the water in our pool went up and over our garage, but the pool itself was fine and the only structural damage we had was that our brick chimney fell down. The Northridge quake scared the crap out of me, and was the main reason we ended up moving south, to Orange County.

-- Gary
 
Gary,
I work in the CNN building in Hollywood. I was remixing an Elvis movie when the shaking came on. I could not get any cell phone service or land line to go through for a half hour. Ironically, I was able to get ebay on the iPhone's internet connection...

My son e-mailed me a pic from my house. Our peach tree snapped off a huge branch which was laden with peaches and the quake was the last straw.
Jeff K.
 
Gary and Richard,

So, to confirm, I understand that the regulator section "throttles back" the charger current to whatever lower value is needed to keep the high cell from exceeds the set point. So with large cells packs, a high cell still can't "outrun" the shunt and go above 3.68 volts - but the final part of charging will take a bit longer. Is this correct?

In my case it I could simply put a cutout and 12V fan on the cover that I will be mounting. That will draw an extra couple hundred mA from those first 4 cells though - maybe I should consider a 48-12V DC-DC converter - or tap into the one on the scooter for the accessories so I draw off the whole pack for the fan?
 
PJD said:
Gary and Richard,

So, to confirm, I understand that the regulator section "throttles back" the charger current to whatever lower value is needed to keep the high cell from exceeds the set point. So with large cells packs, a high cell still can't "outrun" the shunt and go above 3.68 volts - but the final part of charging will take a bit longer. Is this correct?

In my case it I could simply put a cutout and 12V fan on the cover that I will be mounting. That will draw an extra couple hundred mA from those first 4 cells though - maybe I should consider a 48-12V DC-DC converter - or tap into the one on the scooter for the accessories so I draw off the whole pack for the fan?

I think you got it. As soon as any shunt comes fully on, it starts to throttle the charge current to keep the highest cell(s) from going over the set point.

Since the shunt resistors are supposed to be wasting energy anyway, it would be nice if some of the energy was used to power a fan. Anybody ever see a 3v fan? A 5v fan might work OK on 3v. Just wire a fan across one of the shunt resistors. This could be a problem if the one cell that doesn't come on has the fan.

I don't think running it off the tap would be so good. The fan current is almost as much as the shunt current. You could get a pair of small 24v fans in series and run it off the full pack. If you have a really small 12v fan, you could just use a big resistor to drop the voltage from full pack voltage.
 
I'm building my own dedicated charger, using 3 x 24 volt 6 amp switchmode power supplies in series. I'm feeding a 12 volt regulator off the 24 volt tap, and a third wire from the charger to the BMS. The 12 volt input to the BMS is also diode-OR'd with the 12 volt feed from the scooter (EVT-4000) when in operation, because I need a switched 12 volt signal to drive the e-brake input on the controller.

These power supplies have an adjustment pot, with a range of 19-27 volts each. Initially, I'm fitting 16 cells, but building the BMS as a 20 channel unit, so I can increase the voltage later if I wish to.

Amanda
 
More tests...

I matched Richard's latest configuration, and mine seems to be working quite well. The good news is that even with the 6.8 ohm 5W resistors, the heat is completely manageable, I don't think a fan is going to be needed. All you need to do is make sure there is a little bit of open room around the board, when charging. Of course, a fan won't hurt, for sure. :) For those that want to duct tape this board up with the pack, you just need to use the 15 ohm 2W resistors. I will include both with the kits.

I did my tests today with a couple of 4s4p a123 packs in series. My plan now is to use this board with the 16s6p setup on my Cyclone-equipped 20-inch folding bike. The first thing I will do is run the pack down to cutoff, just to make sure this works. Then, I'll connect up my 48V Zivan NG1, and see how the board works.
 
commanda said:
I'm building my own dedicated charger, using 3 x 24 volt 6 amp switchmode power supplies in series. I'm feeding a 12 volt regulator off the 24 volt tap, and a third wire from the charger to the BMS. The 12 volt input to the BMS is also diode-OR'd with the 12 volt feed from the scooter (EVT-4000) when in operation, because I need a switched 12 volt signal to drive the e-brake input on the controller.

These power supplies have an adjustment pot, with a range of 19-27 volts each. Initially, I'm fitting 16 cells, but building the BMS as a 20 channel unit, so I can increase the voltage later if I wish to.

Amanda

To charge 16 cells, you need just under 60v. Sounds like you can adjust the supplies down almost all the way and be right in the ballpark. Those should work great.

Where did you get the power supplies?
 
I did some testing today with a bunch of beat up 11p cell blocks. At first, the charging current was limited by the power supply and was cruising along at about 7 amps. No noticeable heat from the IRFB4110 at all.

One cell came up before the others and the LED for that cell gradually got brighter. Before the current started throttling, a second cell came on. As soon as the first cell reached set point, I could see the charge current start to taper down and it eventually settled at the shunt current, about 500ma. Shortly afterward, all but one cell was lit. The low cell was measuring 3.5v and climbing slowly. I'm sure it would have eventually caught up, but I had to dismantle the test setup before I could finish. More testing tomorrow.

There are a few component values that still need to get optimized, but I think the general layout will work OK.

The monster 500ma shunt resistors and their drivers do get pretty toasty after a while, but nothing ever got hot enough to sizzle spit. Since I screwed up on the transistor pinout, it would be better to build these first run boards for a lower current and use transistors that fit the holes.
In the next version, we can get the holes switched around so we can use the larger transistors.
 
GGoodrum said:
More tests...

I matched Richard's latest configuration, and mine seems to be working quite well. The good news is that even with the 6.8 ohm 5W resistors, the heat is completely ....

Gary, What current was this running at?
 
Since I screwed up on the transistor pinout, it would be better to build these first run boards for a lower current and use transistors that fit the holes. In the next version, we can get the holes switched around so we can use the larger transistors.

Any time frame on the next version?

I sure hope the lower current version will work for my two setups. Otherwise, I'm dead in the water with two, $1400 40AH packs going to waste. Help!
 
fechter said:
commanda said:
I'm building my own dedicated charger, using 3 x 24 volt 6 amp switchmode power supplies in series. I'm feeding a 12 volt regulator off the 24 volt tap, and a third wire from the charger to the BMS. The 12 volt input to the BMS is also diode-OR'd with the 12 volt feed from the scooter (EVT-4000) when in operation, because I need a switched 12 volt signal to drive the e-brake input on the controller.

These power supplies have an adjustment pot, with a range of 19-27 volts each. Initially, I'm fitting 16 cells, but building the BMS as a 20 channel unit, so I can increase the voltage later if I wish to.

Amanda

To charge 16 cells, you need just under 60v. Sounds like you can adjust the supplies down almost all the way and be right in the ballpark. Those should work great.

Where did you get the power supplies?

Australian OEM supplier, soanarplus.
https://www.soanarplus.com/

Manufactured by Meanwell. AU$36.12 each plus 10% gst. Catalog number is MP-P225.

Amanda
 
commanda,
The specs quote a "Hiccup" overcurrent protection mode. How does this affect charging?
I was also looking at this site for Meanwell supplies for my charger. I need 29.7 volts (8s5p) and was looking at the larger model (27VDC, 150W MPP238 - $65.94 +GST) as it could be adjusted up to ~30V and had Current Foldback rather than Hiccup mode. But it is twice the price. Have you adjusted the voltage output? I'd really like to know if it can get to >29.7V and how Hiccup affects the BMS and the charging process as this model is much cheaper.
JohnB
 
jeffkay said:
GGoodrum said:
More tests...

I matched Richard's latest configuration, and mine seems to be working quite well. The good news is that even with the 6.8 ohm 5W resistors, the heat is completely ....

Gary, What current was this running at?

I was using my HP 0-15A/0-60V supply, with the current set at 6A and the voltage at 29.6V.
 
Okay, more tests... :)

I hooked my new board into a 16s6p 48V/13.8Ah a123 setup that I have in a rack bag. Here's what it looks like:

16-Cell%20LiFePO4%20BMS-v1.4c-04.jpg


It looks a bit busy in there, but it is actually quite roomy. This bag originally had two 12s3p packs that I used with my Townie. I rewired them so that all the cell junctions are paralleled, into a 12s6p configuration. Then, I added a 4s4p pack and a 4s2p pack, also paralleled at the cell junction points, yielding the final 16s6p configuration, all wired to a Molex-type connector that plugs into the BMS board. The bag fits on a rack, on the back of my 20" folding bike. Here's what it looks like on the bike:

eMariner-18.jpg


My intent was to try and drain the pack down to cutoff, but I didn't have time. I did try shorting the LVC test pads, to make sure it would cut the throttle on the Cyclone 1000W controller, which it did. I then did a short 5 mile ride, which used up 3.117Ah. When I got back, I hooked up the NG1 which started charging at about 6A. The max rate for this charger is about 18A, but it only does this high a rate when the pack is pretty empty. I had only used about 22% of the capacity. Anyway, It took about 40 minutes to point when the first orange LED came on. Within about 5-6 minutes more, about half were lit. At the 55 minute mark, or 15 minues since the first one came on, all were on but three. At the 65 minute mark, all but one were fully lit, and the last one finally came on five minutes later.

Since having all the LEDs going is the worst case, for heat, I left the charger running for another 20 minutes, or so. The WattsUp showed the pack voltage was steady at 59.08V, and the current was bouncing around .40-.49A. I didn't have my temp gun, but I used my "calibrated" thumb, and touched all the shunt resistors. I could leave it on each one for about a second, or so, which tells me the temps were under 200F. Definitely no fan required. :)

All-in-all, I'm very happy with how this is working. Richard still is playing with the values of one cap and one resistor, but even they way it is, I'm a pretty happy camper. :mrgreen:

I'm going to look at "pre-tweaking" the leads on the KSA931 parts, for the kits, so that we can use these with the existing boards. Eventually, once Richard gets done futzing around with the "final" design, I'll do a new layout with the right pinout for this part.

So, the kits will come with the big 6.8 ohm/5W resistors, like those shown above, and with 15 ohm/2W versions as well. If your application is something like a Ping 15-20Ah pack, where it is desired to seal up the BMS with the cells, you would use the 15 ohm shunt resistors. These setups usually have charge currents in the 2-4A range, which is a godd match for the 15 ohm shunt resistors, which will allow a bypass current in the 250-300mA range, which is 5-6 times higher than what most of the typical Chinese BMS boards will manage. For configurations which have higher capacities, and/or higher charge currents, the 6.8 ohm "fatboys" can be used, so long as there is some room around the board.

More tomorrow...

-- Gary
 
Back
Top