For those interested, I'm doing a new LVC-only series of boards that are really geared towards LiPos, but can also be used for LiFePO4-based packs as well. This one can handle up to 24s, broken down into four 6s "tearoff" sections. Each of these sections has pads for up to three 2.5mm (.100") 7-pin JST-XH board-mounted connectors, for mounting on top, or on the bottom of the board. These come with either vertical or 90-degree mounting angles. The idea is that you can connect the balance connectors from, say, two parallel LiPo packs and still have one more that can be used for a balancer. This way, once the packs are properly connected, they don't have to be disconnected to balance/charge the packs. Less chance for F-Ups.
There's also an onboard active cutoff "tearoff" section, in addition to the normal controller ebrake opto outputs. This active cutoff section has provisions for up to four high-power, low RDS on resistance, like the IRFB4110s we all use in controllers. This will allow for some pretty high power setups, for those that prefer not to use the opto-controller brake connections. With some thickening of the traces, and some big-ass wires, 200A+ is certainly doable, I think. For my stuff, I only use two FETs, as most of my setups rarely go over 100A.
Here's a couple cell-phone pics of the first of these boards, which are 6.5" x 2":
View attachment 6-24 Channel LVC v3.3-02.jpg
View attachment 6-24 Channel LVC v3.3-03.jpg
Actually, I think the PCB drawing file shows a bit more:
View attachment 24-Cell LVC-v3.3-PCB.png
I'm also doing a new widget that I'm calling a "charge balancer". Basically, it is in between a standalone balancer, and a BMS-like shunt regulator. It will work without the charger connected, but only down to a certain point, just below the cell's "full" voltage level, after the surface charge burnoff. For LiPos this is around 3.75V-3.80V and for LiFePO4, its around 3.36V-3.38V. With the charger connected, it acts pretty much like a standard BMS shunt circuit, but the shunt is turned on a bit earlier, and it comes on gradually, instead of all at once. From the pretty extensive tests I've been doing with the prototypes, it tends to keep the shunts from getting swamped all at once. My first prototype unit, shown in the pics below, didn't have any sort of throttling logic, so the overall circuit is much simpler. What I found is that with my a123-based packs, and a 4.5A charger, the typical "overshoot" was about 3.75-3.80V, for healthy cells, that stay fairly well balanced. The worst case I saw was with cells that I purposely had way out-of-balance (from empty to completely full...) and the full ones hit about 3.92V, before settling back down to about 3.70-3.75V, when the low cells finally caught up. Anyway, with a123s, I don't worry about overcharging them a bit, even frequently. Most of my cells are 2-3 years old, and still going strong, and I abused the hell out of them. With LiPos, however, overcharging is definitely not something to tolerate, so what I did was added the basic throttling back in, but left out the automatic cutoff function, which adds a helleva lot of circuit complexity. By keeping this part simpler, I was able to add a bit more functionality to the channel circuitry. Each channel has its own red-green LED, which comes on red as soon as the cell voltage rises above the normal "nominal" voltage (about 3.38V for LiFePO4s and about 3.80V for LiPos...), which only happens when the charger is connected. Once the shunts start to conduct (LiFePO4: 3.58V; LiPo: 4.05V...), the LED color will transition from red to green, becoming fully green when the shunt is fully on (LiFePO4: 3.70V; LiPo: 4.19V...). Here's some pics of my initial prototype unit, which is 16-channels, and mounted in a small extruded aluminum box:
View attachment CMS Lite-03.jpg
The shunt resistors mount from the bottom of the board, and make thermal contact with the bottom of the case, turning the whole box into a heat sink. I used 10 ohm/2W shunt resistors, which ends up supplying about 350mA of shunt current. 16 channels of this generates a fair amont of heat in a box this size, not enough to melt plastic, but hot enough you wouldn't want to hold on to it very long.
View attachment CMS Lite-04.jpg
Here's a shot of it connected and working with a 15s4p a123 pack:
View attachment CMS Lite-05.jpg
This was the first full charge for this pack, which was made from 15 4p blocks that were of varying SOC levels, from about 1/2 full, to already full. It took several hours, but eventually it balanced out just fine, and now, after 4-5 cycles, the cells stay within about .05V of each othre, if I don't run the pack down to LVC cutoff.
I've got a new version I'm going to start testing in the next week or so, that includes the throttling, and has 12-channels. I'll do one version for LiFePO4 and one for LiPos. Here's what the layout looks like:
This one is sized for the regular 6.8 ohm shunt resistors, but it will go in the next size up of extruded aluminum box. I'm still trying to sort out the connector issue, because what I realy wan is to have one multipin plug, that then plugs into the pack/LVC board/harness. Using two 7-pin plugs leaves open the possibility of swapping the two, which will fry some wires.
Anyway, I will probably start a new thread for this effort, once I get some more data to report, but I've had quite a few unrelated inquiries about this type of setup for LiPos.
-- Gary
