GGoodrum's excellent new LIPO balancer development

[mikehains]

Instead of high-jacking two other threads (Turnigy, and 1010B+ series charging) ... I've quoted and replied here ...

I've got my balancer working pretty well, but even with "only" three packs in parallel (12s3p 45V/15Ah...), with the shunts set at 350mA, it takes quite awhile to get them completely balanced. If you don't charge all the way to 4.2V per cell, and don't discharge down below about 20%, the cells will stay pretty well balanced, but over time, this will change. Anyway, on the next one I do I might go back to at least 1/2A.

I'm also going to make a few other changes. The cell circuit can be simplified by removing the logic that keeps the LED off below 3.80V. I originally was doing this as a full BMS, and so the circuit would have to be connected to the cells all the time. For the balancer, this doesn't matter, so I was able to eliminate four parts per channel. I also played around with the divider values, in order to reduce the full-on shut voltage to about 4.15V. It initially was too low, but then I picked a set that was a bit high, at 4.19V. the lower value is better on the cells, but the main reason I did it was to make sure the high cell don't go too high, before the charger/supply's CV mode kicks in. For now, I'm not using the throttling circuit, to cutoff the charge current when a cell goes to high. It seems that with these high-C LiPo cells, it takes longer for the voltage to drop back under the cutoff voltage, so the FET controlling the charge current stays off longer. As soon as the first cell hit the cutoff, it caused the rest to not get the full shunt current. So, what I did was lower the cutoff voltage, and let the high cells overrun a bit, but they are still under 4.20V. This lets the low cells catch up faster.

Fechter is looking at a replacement for the throttling logic, for the regular BMS board, based on using a PWM chip to control the current, so as soon as he sorts this out, I'll do a new version of the balancer that includes the new charger control section.

-- Gary

Rc-groups has a circuit to balance LIPO's that doesn't involve shunting. The thread is here: http://www.rcgroups.com/forums/showthread.php?t=270580

It is very similar to fetcher's and Goodrum's circuit, but has a much lower resistor in the shunt, and a high power Darlington to run it.

I've built this circuit on a breadboard, and it holds cell voltage very well, up to 2A shunting. Over about 2.5A it gets unstable. There is no throttling, but instead a "protective" cut-off if shunting current goes over 2A.

According to reports (I haven't yet tried it on my pack - but will) high shunting currents are unlikely, unless the pack is seriously out of balance. If it is seriously unbalanced, then the reports indicate simply charging with low current that first time, bringing the pack back into balance.

My plan is to charge my 10S4P pack very, very quickly (2C) and have it balanced. The balancer has to be foolproof, meaning redundant safety systems (and no smoke, ever).

I'm thinking the TL431's to shunt current, through a Darlington and 0.5ohm resistor ... with the protection circuit to cut power in the event shunting exceeds 2A ... and then a seperate ("redundant") circuit to trip power off immediately if any cell receives in excess of the set voltage.

- Mike

 

[GGoodrum]

Hi Mike --

We actually use the same 2.5V regulator chip in our BMS/balancer circuit. It comes in many flavors (FAN431, LM431, etc...). In the first BMS design I worked on with Bob Mcree, we also used Darlington power resistors, but one of the big problems was trying to deal with the heat. To run this circuit at even 1.5A, that is over 6W, and that is only one. Multiply that by 12 and you've got a heck of a lot of heat to deal with, unless you go to active cooling. I don't remember the exact math, but I'm pretty sure we needed like a 1/2 pound of aluminum to dissipate that much heat. We were doing 16 channels, but for LiFePO4, so this has to be similar. On my current balancer, I'm only using 10 ohm/3W shunt resistors, which sets the shunt current at about 350mA. With 12 channels, that's about 20W, but the box get hot enough, when all the shunts are on, that you can't keep your hand on it.

Assuming you solved the heat problem, you still have to do something about limiting the current. If your charger puts out more than the 2A you want to run this at, the high cells will go over the shunt cutoff voltage. Your "protective cutoff" that cuts the current off is pretty much what the throttling logic in our current design does. If any shunt is fully on, the voltage will rise above the cutoff, and the charge current is turned off. When the voltage drops back down under the cutoff, the charge current is turned back on. This oscillation/"throttling" will keep the cell voltage fixed so that it can't go above the cutoff. The only problem is if there is a significant imbalance. In that case, the high cell will get full first, and the fuller the cell gets, the longer it takes for the voltage to fall back down below the cutoff, so the charge current stays mostly off. This means the low cells don't get the full shunt current, so it increases the amount of time it takes to balance the cells. What we are working on now is a new scheme that replaces the throttling with a true PWM controller to cut the charge current down to the shunt current, as soon as the first channel hits the cutoff. That will ensure that each channel gets the full shunt current to catch up.

-- Gary

 

[mikehains]

What we are working on now is a new scheme that replaces the throttling with a true PWM controller to cut the charge current down to the shunt current, as soon as the first channel hits the cutoff. That will ensure that each channel gets the full shunt current to catch up.

I get it. Great idea !

This will solve one of the problems with using shunts ... as I see it, shunts effectively reduce the the resistance in the circuit as they come on ... which pulls more current from the power supply just when less is needed !

I can't wait to here more.

MJH