> (but I found out that at least I can balance the cells at 3.6V; later...)
this is the pack on tricle charge:
View attachment trickle.png
from sample 0 to 15 the cells are being brought down by the VMSes and the charge FET is off; this is the result of an earlier OVP force and release pair of commands (not shown). at sample 15 the FET is turned on and trickle begins.
"bunching" shunt action keeps each cell bunch together from 3.5V to 3.65V as expected. above 3.65V the shunts are fully turned on and overcome the trickle current, preventing further charging.
past sample 70 my script detects no progress and shuts down the FET.
the surprise here is the change of slope at 3.6V. obviously the shunt current is regulated in 3 stages: maximum above 3.65V, medium from 3.65V and 3.6V, and low from 3.6V to 3.5V.
EDIT: this is not true! see my post below...
clearly the system expects a charging voltage of N*3.6V (not N*3.65V as I previously thought) and at that voltage it will balance all cells in a multi VMS pack correctly during trickle (no problem B).
why wasn't this behavior observed earlier? for now I'm guessing it emerges only after an overvoltage event. this can be a true 4.1V overvoltage or one simulated via the force OVP command. so apparently the 4.1V peak is always expected, and the cells won't balance unless the peak happens. (or maybe ANE wanted to balance the cells only when sufficiently unbalanced? I don't think so...)
the VMS docs mistakenly say the system expects a N*3.65V charger. but it's worth noting that the charger sent by PSI is spec'ed at N*3.6V and was precisely set there before I touched it!
the complete picture of ANE's system is beginning to take shape:
-must be used with a compatible charger that switches to trickle when the charge FET is momentarily turned off.
-trickle current must be less than max shunt current for the board version in use.
-must be used with a charge FET.
-problem A, the 4.1V peak, is always expected (otherwise cells won't be balanced).
-the charger must be set precisely at N*3.6V (otherwise problem B shows up on multi VMS systems).
under these conditions, and if you accept the 4.1V peak, the system will work.
furthermore, I have a hunch that the 3.6V balance level can be set via the infamous "average command". continuously feeding back the cell average to the VMS chain would solve problem B for a charger set anywhere in the N*3.5V to N*3.65V range.
the stand-alone charging can be enhanced in 3 ways by an external controller (such a PC running my scripts):
-the 4.1V peak can be avoided by briefly forcing the OVP signal at a lower preset level, such as 3.67V. [reason for needing an external controller: unknown!]
-the charge FET can be shut off by forcing OVP when end of charge is detected to avoid keeping the cells at 3.6V for extended periods. after this the VMS will quickly bring the cells down to 3.5V and then let them be. [reason for needing an external controller: the VMSes in a chain are isolated and don't know the state of the cells managed by other VMSes]
-the cell average can be continuously fed back to the VMS chain so that problem B is avoided for chargers not set precisely at N*3.6V. (it allows using chargers in the N*3.5V to N*3.65V range. it allows you to choose your preferred cell target voltage.) [reason for needing an external controller: the VMSes in a chain are isolated and don't know the state of the cells managed by other VMSes]
of these things, the first two are implemented in the scripts posted above (ane-0.85.zip); the average command haven't worked so far... I need documentation!