I carefully borescoped my pack through the pressure vent. Then it was simple deduction after studying the bottom module PCB.
The bottom module (and I assume all the others) has qty126, 18650 style cells. 96 of the cells are "hardwired" into groups of 6 parallel cells 16 strings (series) long producing a typical (6p x 16s) configuration. The remaining 30 cells are also configured into groups of 6 parallel groupings but are NOT hardwired. The 21 sets of 6 parallel cells make a module of essentially 21 individual cells (instead of 126) and as far as the system is concerned are treated as such when it comes to typical BMS functions (charging, discharging, temperature probing, etc). I'll call these "subcells". Therefore, if I'm correct, the "Pack" would require 84 conventional BMS control circuits (4 pack modules x 21 subcells each). Most of the better industrial BMS controllers can handle up to 20 individual "subcells" so it makes sense that each pack module would need its own controller with all four pack controllers being coordinated by a main "remote" BMS processor which I believe is integrated into the black box directly behind the steering head. It communicates with the subcells via the onboard CAN system and probably a few analog channels via the telemetry connector on the pack.
This is where it gets interesting. The 30 non-hardwired cells (5 groups of 6 cells) all connect to high current terminations that go deep into the multilayered PCB. This means the final connections can be manipulated in a myriad of ways. Looking again specifically for "high power" switching components I noted what appeared to be 5 switching transistors. Hummm . . . . . , 5 high power switching transistors and 5 "non" wired subcells . . . . . . . . . coincident? Maybe, maybe not. Then it struck me that the switching circuits could be used to cut those 5 sets of subcells into or out of the overall pack to change total bus voltage instantly! Maybe not a novel idea but one that never occurred to me before. I'll coin the 5 suspect subcells "active subcells"
The more I thought about it the more it made sense. I'm pretty sure that Alta uses a BMS system where during discharge lower capacity/voltage "subcells" are pumped up to actively balance the pack. Also, the other electronics on the bike require low voltage. both require a separate source of power so I'm assigning 1 of the 5 active subcells for this purpose leaving the remaining 4 active subcells for the purpose of variable battery configuration. Hummmm again . . . . . . ., four power modes and 4 active subcells . . . . . . coincident? Again, maybe, maybe not.
Of course, this is the most simplistic application. I estimate the 16 total active subcells in the pack could be electronically switched in infinite ways. Hummm . . . . . . . 16 active subcells per pack and 16 hardwired subcells per module . . . . . . . coincident? Maybe, maybe not. Gotta think about that one. The most basic schemes would be series and/or parallel variations.
Electrically, the series connections offer the least complexity as the active subcells would discharge in relative concert with the other subcells in the module, the total bus voltage (speed) being the main and most noticeable performance variable. Parallel variations are more complex and probably used for different reasons. For example, when connected in parallel the active subcells would be assisting other subcells essentially sharing the electrical current (heat producing) demands. This would cause the "shared" subcell to get out of balance with the others so there must also be the ability to electronically select and switch sharing duties to any of the hardwired subcells in the module to keep things relatively balanced. The "hardwired" subcells do have terminations that COULD go into the PCH just like the active subcells, but the terminations do not look like the high current carrying terminations of the latter. But wait! Since only a fraction of the total bus current needs to be "shared" between active and passive subcells when paired in parallel for this purpose the existing "cell balancing" bridge would be quite sufficient. Therefore, in this case manipulation of the active subcells could be another method of active balancing effectively extending the range of the pack.
Finally, when shared in parallel the thermal load between shared subcells is also reduced. Could this be used as an electronically controlled thermal control system? Sounds viable to me! It might be possible to generate switching algorithms that can move heat from the hotter cells to cooler ones, or from the rearmost subcells to the front where cooling from outside air movement is more effective.
Speculation:
Mode 1 - 3 active cells in parallel and 1 series. Lowest bus voltage but greatest range potential and best thermal control.
Mode 2- 2 in par and 2 in series.
Mode 3 - 3 in par and 1 in series
Mode 4 - All in series. Highest bus voltage but least range potential and little thermal control.
1 module - 6p x 16s = 66 volts, 288,000mah + 4.1 to 20.5 boost volts and another 90,000mah. Total 70 to 87 volts and 378,000mah.
4 modules per pack - 280 to 350 volts (depending on configuration) and 1,512,000mah (approx. 5.8kwh)
I haven't thought this completely through yet but it's intriguing. There's probably a dozen other logical explanations as well. All this, or it could simply be what was physically needed to connect the 21 subcells electrically as the truncated corners at the front of the pack forced the designers to reconfigure the cell arraignments producing the 5 oddball subcells. That's very possible as well but I like my reasoning better! LOL.
Other comments: IF we had access to the onboard data loggers we'd probably be looking at electrical data for each of the 84 subcells and physical data (temps, etc) for the same (assuming indirect measurements). Note: I did not see any connections for direct reading external Thermal Couples but suspect there must be several integrated into the cell support block and wired to the backside of the PCB. That's a lot of data!
