A different approach to battery management...

[dogman]

Been a few cells ruined by leaving a bms connected all winter on the shelf. Or leaving a controller turned on, I never do that!, much.

[mwkeefer]

Yes I was - there is an updated app note on the Atmel site but really this ASIC (application specific integrated circuit) just as any designed for this purpose could simply be adapted with a cap or inductor and an addressable bus for discharge/charge direction.

Their appnote is ok but it would be better to design from scratch after selecting the MCU and drawing up the stacking interface connectors (6s per board x 3 boards for 18S - each with it's own ADA chip tied back to the master MCU.

I am still awaiting confirmation (which doesn't usually arrive before the sample units do) on shipment and qty of samples shipped, I asked my atmel rep for a small number of samples for testing in pump balance/BMS mode for a 24S pack but he usually sends me 2x what I ask for...

We shall see... in the meantime, I am excited about the HK balance mod GGoodrum came up with on the external discharge board for higher current balancing of our packs - I have several of the units here and plan to arrange them in a similar fashion until I have somthing better to use on my big packs (hint hint gary).

In the long run, the charge pump style balance system would be best for efficiency and lowest heat generation... I believe this system could be allowed to run even durring discharge - this may seem trivial to many people, why would you want to balance while discharging? In the event of a weaker cell, the power stored in a stronger cell would transfer durring discharge at up to a 1A rate which should extend the cutout of the pack as a whole and increase the retrievable pack level energy. This could be very useful in smaller packs too (a single series string of LiPo gets more oob at end of cycle than several strings in parallel configuration) to increase their lifespan and usefulness.

PS: I just lost another iCHarger... no more for me, 135.00 each is too much to keep toasting them... I'll stick with my homebrew charger solutions from now on and hopefully a pair (up to 18S each) of GGoodrum/Fetcher BMS HVC/LVC systems when they are completed.

-Mike

[Doctorbass]

I heard alot of these problem overbleeding cells to Voltage under lvc using the MKS or any other BMS that balance to the lowest.

The best way i found to avoid that with every of my setup is to link every balancer line thru a relay contact.

Yes.. it exist some very small 6pst or 4pst relay that you can use to switch the balancer ONLY when you use or charge the ebike battery!

When i precharge the contactor on my setup and that the contactor activate.. it also activate these relay... and when the ebike is OFF and left for prolonged period or just not be used, There is absolutely no link between the BMS and the cells..( relay contacts are open) so no danger to drain all your cells !

It cost around 15$ to avoid that problem for a 24s battery.

Doc

[CamLight]

Nice setup Doc, I like it!

[AndyH]

The best way i found to avoid that with every of my setup is to link every balancer line thru a relay contact.

Brilliant! I guess that's why you're the Doc, Doc!

[AussieJester]

FANTASTIC to see progress Gary! When do you envisage
they will be available for purchase mate?

KiM

[richerson]

HI Doc

That would work well for my pack, but I can only find 24v relays, I need 75vdc. I have trashed a whole lipo pack leaving hk cell monitor conected.

I have just bought some zippy packs so i want my setup to be full proof this time.

[GGoodrum]

What Doc is doing is mainly for permanent installations of these RC-type monitoring/balancing widgets. The LVC/HVC-only boards draw microamps, and can be left connected. They draw less than the typical self-discharge rate for most cells. The charge control boards also draw nothing, when the charger/supply is not hooked up.

Using Battery Medics, boosted or not boosted, and/or other units, as external balancers is still a problem, if left connected to a pack, but at least if you use the Battery Medic in the discharge mode, all the ones I have do seem to stop when they hit the programmable set point, so at least they won''t quickly drain the pack dead.

I have the first of the new charge controller cards built (shown below...), which I will be testing today. This is the first one with the new adjustable current limiter. Later, I'm going to try this with the CellLog unit, which I've added an isolated opto output to so that the programmed HVC "alarm" output might be used with the throttling logic. I tried this before, but there was hysteresis between when this signal comes on and when it goes off, which was playing havoc with the PWM throttling logic. What I will try now is using a higher value filter capacitor on the opto signal, to smooth out the swings.

Kim -- We're getting close, buddy. My hope is to get you something by the time you have everything back together on your blinged-out cruiser.

-- Gary

[AussieJester]

Kim -- We're getting close, buddy. My hope is to get you something by the time you have everything back together on your blinged-out cruiser.

Won't be a problem there Gary I'm waiting on you now LoL..i can't go any further at this stage until i have the BMS, i can't have the battery housing painted until i know what holes i need to placed in the enclosure, its being professionally painted all openings etc must be done to it before it sent for paint alternatively if you can give
me exactly what size holes are needed for the charging/balancing plugs, LEDs or display window to see status of charge etc? etc? So i can do it before the BMS is ready? lemme know when you have one ready and ill paypal the money for one immediately, they look and sound EXCELLENT anywayz Gary fantastic effort mate hopefully i can keep my lipos 'alive' a lil longer this time round hahaha ...

cheers mate much appreciate your efforts

KiM

[GGoodrum]

The hysteresis in the CellLog alarm output is just too great to be able to use these as HVC outputs for the PWM/throttling logic. Even with a very large cap across the input, the current was jumping all over the place.

The new controller works great,however, with the LVC/HVC boards. the CC limiter also works quite well. You can set the current pretty much anywhere you want. I also tested another new feature, which is the auto-shutoff based on the current dropping down below a minimum value. Right now, it is set for about 190-200mA. As soon as the current drops down under this point, everything shuts off.

Simplified Charge Controller-v.4.0.6-02.jpg

(186.96 KiB) Downloaded 284 times

[methods]

Someone needs to measure the current at each cell to see how they are tapping the pack. Have you taken that measurement yet or reverse engineered the circuit?

Even a few mA is enough to trash a pack in no time flat - even if it does not drain a subset of cells down to the damage point it will create the worst possible balancing situation where only one cell is down and the balancer has to bring all the rest down to meet it - or worse yet if the user does not give the balancer enough time - that one cell falls off the cliff early.

*I suggest cutting the power to the LCD screen to minimize leakage current* It may look cool but I am sure that it is power hungry.

If you are lucky you might be able to find the power traces for the chips, cut them, and fit a low cost DC-DC that would power the entire circuit from the full stack - that would minimize the current draw and imbalance.

Say there is a 5mA leakage (which sounds reasonable)

5mA/hour
120mAh per day
840mAh per week
3.36Ah per month
10Ah in three months

Here is a product with 8mA leakage - granted this has a power hungry blue LCD
http://www.hobbycity.com/hobbycity/store/uh_viewItem.asp?idProduct=10952&Product_Name=Cell-Log_8M_Cell_Voltage_Monitor_2-8S_Lipo


If it turns out that they draw power equally from all cells then you are sitting pretty - but I suspect they did it the lazy way.

-methods

[GGoodrum]

I agree, you wouldn't want to leave a Battery Medic connected all the time, especially if you have it in the balance, mode, which brings down the level of all higher cells to the level of the lowest. Sometimes, a channel will overshoot, so that cell now becomes the lowest. This can keep happening, which would drain the cells down. I use the discharge mode, which brings all the cells down a programmable set point, and then it stops. I still wouldn't leave one connected for an extended period, but I've yet to see the discharge mode drain any cells past the set point.

The point I'm trying to make is that with LiPos, balancing just isn't required with every charge, so whatever balancer-type that is used, doesn't have to be left connected all the time, like with a regular BMS. For basic charging, the LVC/HVC part of the BMS, and the charger controller card can be used by themselves. An example of this is shown below. This is an 18s2p 66V/10Ah pack made from six Zippy 6s-5000 packs.

18s2p LiPo Pack-01.jpg

(90.89 KiB) Downloaded 193 times

Buried inside the pack are three 6s LVC/HVC boards, like the one shown earlier in this thread. The opto outputs from all three boards are ganged together, and brought out as a single two-wire JST-BEC plug, which plugs into a matching connector coming from the charger controller. I also brought out a separate set of 12-gauge leads for charging. These have standard 45A Andersons, which also plug into mates, coming from the charge controller. The picture below shows the charge controller plugged into the pack. Ahead of the charge controller unit is a Turnigy Watt Meter.

18s2p LiPo Pack-02.jpg

(111.87 KiB) Downloaded 193 times

For balancing, each LVC/HVC board has a set of 18-gauge wires, brought out of the pack, into a Molex Microfit 3.0mm 7-pin connector. As I said, most of the time, these are simply not connected to anything, but if balancing is required, they can be used to connect external balancers, such as the ones shown below. These are basically working prototypes that I put together, using modified shunt boards and end plates, from earlier versions of the 4.0 BMS unit. Shown are both a 6s and a 12s version. For my 18s pack, both will be used.

12s LiPo Balancer-01.jpg

(167.64 KiB) Downloaded 193 times

12s LiPo Balancer-02.jpg

(180.38 KiB) Downloaded 193 times

6s LiPo Balancer-02.jpg

(165.92 KiB) Downloaded 193 times

6s LiPo Balancer-03.jpg

(222.6 KiB) Downloaded 193 times

The balancer units have fans, and are completely self contained. They simply plug into the MicroFit connectors coming from the pack, whenever balancing is needed. When used while charging, the combination acts just like the full BMS, but they will also work standalone, after a charge. In this case, cells above about 4.14V will be brought down to that level.

More later...

-- Gary

[methods]

I like the idea of keeping the sensitive electronics off-board.
* Less bulk on the bike
* Less wear and tear on the BMS
* Fewer failure points at ride time

We need to find a good source for "Full Charging Solution" connectors.
I am picturing a mill-spec connector with 4 large gauge contacts and 20 - 50 smaller gauge contacts.
Something we could use to allow us to have the balancer off board but can still connect with one motion.

I hate connecting the charge leads AND a bunch of balance taps. (though I suppose having all the taps in one connector is not that bad - I am thinking of 6 JST-XH connectors for 36S)

I have seen a lot of connectors with SMA's built in but I have yet to see any with different gauge wires
(but I have not really been looking either)



Obviously it could be built up with Anderson connectors - or two different types of connectors could be "glued" together.

-methods

[GGoodrum]

Yes, connectors are always a pain, especially the smallish JST-XH variety. I also don't like the fact that they only use 22-gauge wires on these, This is fine for hooking sub packs into a parallel adapter, but for balancing, it is far better to use heavier gauge wires between the pack and whatever is used for balancing, especially with 1A+ shunt currents. There's too much voltage drop in the wires otherwise. The problem, as you know, is finding a suitable multi-pin connector that can handle larger wires sizes, and a decent amount of current. I've found two that work well, and come in various sizes, the AMP/Tyco VAL-U-LOK 4.2mm PE Series, which have pins rated up to 9A, and the Molex MicroFit 3.0mm, with a current rating of 3-5A. The problem with these is that there's the labor of putting pins on the wires. I have an expensive special crimping tool, but it is still a pain. I am able to use this same crimper on the smaller MicroFit pins, but it is even more time consuming.

There are board-mountable versions of the connectors, which eliminates having to do pins for that end, at least. What I'm doing now is having Keywin get a bunch 18-gauge multi-colored wires made up with MicroFit pins pre-crimped on each end. He couldn't source the housings over there, but all that is required is snapping in the pins. Anyway, by putting the board-mountable connectors on each end, no more crimping.

-- Gary

[liveforphysics]

I would much rather not have the balance and charge connectors all in the same plug.
I want to be able to just throw some charge in a pack with nothing more than a meanwell when I'm opertunity charging somewhere. Hooking up a balance plug and charge plug doesn't really seem like a big inconvience IMO.

[amberwolf]

Except for the cost of the connector shell/pins/etc for the opp.chg meanwell, there's nothing wrong with just using the charging pins on the meanwell connector, leaving the balancing pins open.

Unless you mean to use the same meanwell for balancing and charging when at home, in which case you have a definite point.

[GGoodrum]

Actually, I was just talking about finding good connectors for high-current balancers. You get too much voltage drop in the 22-gauge/JST-XH pigtails, even with just 1/2A of shunt current. I still run separate 12-gauge leads for the main charge current, and use Andersons for the connectors. My 18-gauge MicroFit balance plugs won't get used all that often, but when they do, I know they can handle the current.

18s2p LiPo Pack-03.jpg

(130.6 KiB) Downloaded 296 times

18s2p LiPo Pack-04.jpg

(69.2 KiB) Downloaded 296 times

18s2p LiPo Pack-05.jpg

(102.11 KiB) Downloaded 296 times

The MicroFit connectors have the same locking tabs as the larger VAL-U-LOK plugs, so they have the same positive connection, but are much lower profile and they come in single row variants.

-- Gary

[recumbent]

Opportunity charging is something i do often also, shopping and commuting is mostly done on my ebike in the summer.

Charging without balance leads is perfectly fine (in the field) with a $60 bulk charger in the pannier using some sort of monitor like the Cycle Analyst or CellLog.
The 18awg Microfit connector for balancing is a good idea but carrying the expensive balancing charger or other means is not practical (in the field), yet.

Although I like the black Microfit connectors, and with 18awg wires, we could charge and balance through just the balance leads with modest current, i hope.

[methods]

we could charge and balance through just the balance leads with modest current, i hope.

I actually drew up a 12S charger that is nothing but an array of 4.2V isolated DC-DC converters - input is 12V to 38V, output goes directly to the balance taps. Similar to what people are doing with these "cell phone chargers" but cheaper, on a PCB, and with a DC input. You could leave it attached to your pack (serving as a parallel connector board) and then opportunity charge with whatever you can find - a car battery, a Lead charger, a laptop supply, a rectum fryer, etc.

Talk about opportunity charging. . .

-methods

[recumbent]

... You could leave it attached to your pack (serving as a parallel connector board) and then opportunity charge with whatever you can find - a car battery, a Lead charger, a laptop supply, a rectum fryer, etc.

Talk about opportunity charging. . .

That would be so economical charging with a laptop charger if possible, while we park our ebikes. Great idea's you put together Methods.

Now all I have to is assemble a bunch of 8 series Microfit, or HX connectors with my new crimper from Westmountainradio.
Hope it's not too difficult if i use the correct die attachment i've also ordered.

[Hyena]

I actually drew up a 12S charger that is nothing but an array of 4.2V isolated DC-DC converters - input is 12V to 38V, output goes directly to the balance taps

Cool, I was looking at making something similar in 6S guise to plug into the balance tap of each pack for a short period of time to balance them if needs be. Got an online source for these ?

[Ykick]

we could charge and balance through just the balance leads with modest current, i hope.

I actually drew up a 12S charger that is nothing but an array of 4.2V isolated DC-DC converters - input is 12V to 38V, output goes directly to the balance taps. Similar to what people are doing with these "cell phone chargers" but cheaper, on a PCB, and with a DC input. You could leave it attached to your pack (serving as a parallel connector board) and then opportunity charge with whatever you can find - a car battery, a Lead charger, a laptop supply, a rectum fryer, etc.

Talk about opportunity charging. . .

-methods

How would these things work?

http://www.callpod.com/products/chargepod?tab=demo

I wonder if they could be readily adapted?

[liveforphysics]

These will do what Methy described. 4 isolated 4.2v CC/CV power supplies in a case. Buy 3 of them to perfectly balance charge a 12s pack, but 6 to perfectly balance charge a 24s pack. These things are tiny.

Or, even cheaper:

http://www.hobbycity.com/hobbyking/stor ... rger_2S-3S

Those things are $4.25 each, and contain 3 isolated DC/DC converter CC/CV power supplies inside a little box scarcely larger than a matchbox.
For $20 (for a 12s pack) or $40 (for a 24s pack), and some time wiring up the required JST connectors, you can have 800mA balance charging to perfectly balance charge a pack no matter how off the voltages were when you started. You could power them all from a laptop charger.

It would be useless for opportunity charging on large packs due to the slow rate, but for something like putting a pack in balance, just plug them in and leave it overnight, and the next daay, all cells would be sitting right at 4.2v.

[Hyena]

Those things are $4.25 each, and contain 3 isolated DC/DC converter CC/CV power supplies inside a little box scarcely larger than a matchbox

Sweet, I didn't realise they were isolated I take it they can charge through the balance taps only then without the main discharge leads being connected ?

[liveforphysics]

Those things are $4.25 each, and contain 3 isolated DC/DC converter CC/CV power supplies inside a little box scarcely larger than a matchbox

Sweet, I didn't realise they were isolated I take it they can charge through the balance taps only then without the main discharge leads being connected ?

That's right. You don't use main charge leads. You charge entirely through the balance taps. This method is no replacement for the type of charging you can do with a setup like a meanwell and Garys balancer, because its limited to the current that can pass throught the balance taps, so 20Ah packs take like 24hrs to charge.

What it is good for though is taking a pack that is out of wack, and over the course of a day or so, you can return all cells to perfect balance.
A BMS struggles to find balance if it has a few very low cells. A setup like this doesn't matter if you have some cells at 4.25v, and some cells at 2.8v, it charges each one seperately, and doesn't know or care what any cells happen to be at, it just slowly raises each one to 4.2v through the balance taps.