I promise, I will provide a "buying guide" soon, one that explains all the configuration options between various setups. We just need to finish the testing first. Richard's board seems to be pretty stable now, and I just need to get mine in the same configuration, and then do another run of boards with all the changes. It didn't help that I spent the last week in San Diego, on a mini-"stay-cation", but I'm home now. I'll try and take/post some pics later.
Andy found a trace problem with the full BMS boards, that also is a problem on the balancer boards. I need to do another run of these, with the fix, before we can release the balancers. I'll do the fix to my existing test boards, and then I'll post some more picks of these, in use with the CellLogs/Control Unit.
For those that are totally confused, here's a brief explanation of the main differences between the various options: The so-called "full BMS", which is described more in the 1000-page long "sticky" BMS thread (
), contains all three main BMS functions on one board/group of boards, low-voltage detection/cutoff (LVC), overcharge protection (HVC) during charging, and cell balancing at the end of a charge. This sort of "all-in-one" BMS setup is really better suited to configurations that use LiFePO4-based cells. The reason is that many of these LiFePO4 cells tend to go out-of-balance quicker, especially the larger capacity variants like ThunderSkys and Sky Energys. Another reason is that these are usually custom setups which have enough room for the full BMS board(s).
LiPo-based packs are much more compact, usually, and generally use 6 or 8-cell 5Ah "sub-packs", combined together in parallel and in series, to create a larger ebike pack. These 6s/8s packs typically come pre-wired with balance plug taps, which makes it easier to make connections to whatever is used for the various BMS "elements". By simply connecting these balance taps in parallel, you can create higher capacity configurations, paralleled at the cell level, which is best, IMO, because the cells will self-equalize immediately, and can be treated as one larger capacity cell at that point.
Now you could certainly use the full BMS for a LiPo based configuration, but it creates some installation challenges. The balancer portion of the BMS board contains the heat-generating shunt circuits, so you can't just bury them in with the packs. Besides that, the other main advantage of using LiPo-based setups is that unless you are draining the packs down to LVC all the time, healthy cells just don't go out-of-balance very often. You just don't need to balance the cells but once every 5-10 cycles. So, what this means for the BMS is that you can take out the balancing part of the system, and make it a separate "external" unit. What that leaves, in terms of required functionality are the parallel adapters, the LVC function, the high voltage detection/protection function and the charger control function. If you then separate out all the functions/circuits related to charging, what you are left with that
needs to be on the bike itself are the LVC circuits and the parallel adapters. So, in this "bare bones" configuration, you can use a combination LVC/parallel adapter, like the one above, mounted with the packs, and bring out the combined parallel plug connections, one plug per 6/8-channel board. Then, these output pigtails would plug into connectors on the "external' unit that does high-voltage detection/protection, and charger control, when the pack is charged. Balancing, when required, is done separately. Call this configuration "Option 1".
A variant of "Option 1" is to move the HVC detection function to the pack-mounted LVC/parallel adapter board, and then just have a charge controller board as a separate unit. Balancing can still be done separately. This "Option 2" is essentially the first configuration that I started offering, awhile back. I added the Battery Medic Booster, to add a higher-power balancing option. The only problem with the latter option was it required hacking in some small wires to the back of the BM unit. Many have found this to be in the "too hard" category. The other problem is that if you wanted to balance the whole pack at the same time you charge, you needed multiple BMs and multiple boosters. One other minor issue was that the readout function on the BMs was not all that accurate. Oddly, the balancer portion was pretty close, but not what was displayed.
To address both issues, I did the standalone balancer units, which are basically just like the BM boosters, but with a bit extra circuitry. I then discovered that the CellLogs units were great to use in conjunction with these balancers, because they have 3-decimal cell voltage readouts, and are very accurate. I liked these so well that I started using them whenever I charged the pack as well. It was at this point I decided that it would be good to do one combination unit that combined the charge control logic, with connections and mounts for the CellLogs.
The CellLog units actually have a lot of smarts in them, and contain handy programmable HVC and LVC functions that can drive an alarm output. This opens up additional configuration options. A possible "Option 3" could be just having parallel adapters mounted with the packs, and then the balance connections bought out to box that has the CellLogs mounted on top. This box could be mounted on the top bar, for instance, so that the CellLogs can be made visible, and then the LVC and HVC function would be provided by the CellLogs, and the added support circuits. This unit would also contain the charge control logic so that there is just a single two-wire connection back to the charger/supply. There would still be extra connections needed somewhere, so that you could plug in the external balancer(s), when balancing is required. This "Option 3" configuration is what we are currently testing.
A variant of Option 3, similar to Option 1, is to not have the CellLogs/Control Unit do the LVC function, and not be permanently mounted on the bike. In this case, like for Option 1, there would still be combo LVC/parallel adapter boards mounted with the packs, and the CellLogs/Control Unit would be connected just when it is time to charge. This "Option 4" is actually my preferred configuration for my own setups, as all my packs already have LVC/parallel adapter boards buried in the packs.
Anyway, the new boards we are testing will support both Option 3 and Option 4 configurations. I have some parallel adapter-only boards that will be included with Option 3 configurations. The LVC/parallel adapter boards that are on the site already will be used for Option 4 setups. Part of what is causing a lot of confusion is that I went ahead and make these available, as they were finished, but the rest of the stuff is not quite ready. Since I haven't sold a BM Booster in a couple of months, I've just gone ahead and taken it off the site. In its place will be the new balancers. These initially will come in 6-channel and 12-channel variants, but soon after, I'll also do 8-channel and 16-channel versions. Finally, as soon as we finish the testing on the new control unit, I will make them available as well.
The balancers will be able to be used standalone, one 6s (or 8s...) section at a time, or multiple balancer units can be used at the same time as when the pack is charged. In this configuration, with the balancers and the CellLogs/Control Unit all connected, it is functionally identical to a "full BMS" setup. This could actually be an "Option 5" configuration, which is a CellLog-based, full BMS.
Like I said earlier, I will eventually do an illustrated "buyer's guide", but hopefully the above will help answer some questions. I get lots each day, so I know it is confusing. To minimize the confusion, I'm trying to keep all the CellLog/control unit/balancer info in this thread, and the full BMS-related stuff in the sticky BMS thread. It is becoming harder to do this, because there is a lot of cross-pollination lately. This is because for both the balancers and for the BMS shunt circuits, we've made a pretty significant change in how these circuits are used. In all the previous BMS versions we've done, and in virtually every shunt-based BMS design I'm aware of, the charge voltage is set at a point a bit above where the shunts are fully engaged, which in our case means they are passing about 1A of current when fully on. The HVC trip point is set right at the point just before the shunts get swamped/overloaded. This HVC signal is then used by the charge control logic to "throttle the charge current so that the net effect is that no cell voltage can go over the HVC set point.
This works quite well, because there's always at least 1A of current available for the "slow" cells to catch up to the cells that get full first. Without this, the full cells will block current from going to the less full cells down the line. The way electricity works is that all the current has to go through all the cells, so if one is full, and not taking in any more current, it will block the others from that are not yet full, from getting any more current. What the shunt circuits do is let current be "bypassed" around the full cells, so that the ones not yet full can catch up. Eventually, all the cells are full, and all the shunt circuits are in full bypass. Again, this works quite well, but the problem is that it generates a ton of heat. With LiPos, there's about 4W of heat generated, per channel, so for a 24s setup, there's about 100W of heat that needs to be dealt with in some fashion.
Our new scheme is quite different. Now what we do is set the charge voltage right at the desired balance point, so in the case of LiPo, 4.15V. This is also the point that the shunts first start to conduct (as opposed to where they are fully on, "bypassing" 1A...). The idea is that if the cells are perfectly balanced, they would all reach 4.15V at the same time, and the charger's CV mode would take over, keeping the voltage at 4.15V per cell while the current slowly drops. In this scenario, the shunts never come on at all. If a cell is out-of-balance, it might get full first, and reach the 4.15V point first. What will happen then is the voltage will try and go higher, but then the shunt circuit will start to conduct. the greater the imbalance the quicker the cell will hit this point, and the harder the shunt will work to keep it at 4.15V. The net result is that the high cell shunts will keep them in check, and eventually the rest will catch up. When they do, and the current drops, the shunts will all eventually go out completely. The cells are just as balanced as they would be with the "regular" BMS shunt scheme, but without all the shunts cooking away at 1A at the end. This simplifies the control logic significantly. Now what we can do is set the HVC trip point higher, and use it as a safety cutoff function, instead of having it drive a PWM-based throttle circuit. All that is left in the charge control logic is an adjustable low current detection function that is used to shut off the charge process when the current drops below some point, like 50 or 100mA. This is the new control circuit we are testing now. The LED on the front of the CellLog box will be red during a charge, and will be green when normal shutdown occurs. If the charge process is terminated due to a cell hitting the HVC trip point, the LED will be orange.
Anyway, this has gotten longer than I wanted, and I need to get back to testing.
-- Gary
Because the pinout was different than all the other dual opamp chips we've been using, I ended up using a bunch of different colored 22-gauge wires to map the pins into the correct positions on the PCB. It was pretty ugly, for sure. I told Richard that it looked like a dead gay spider. 
