TLDR: The question is about retrofitting a balancer to the Ni-MH battery of an existing appliance (that doesn't trickle charge the battery), preferably in the way that will not require frequent access to the battery. Changing the charge/discharge logic of the appliance doesn't look feasible - this way it would be really better to rework the circuit for Li battery.
While there's a lot of Lithium balancers for every common Li chemistry, Nickel (Ni-MH or Ni-Cd) balancers seems to be nearly non-existent.
I know it's told that Nickel cells would balance if charged with a standard C/10 charge for 16 hours. And in pre-Li era, enthusiasts would make the same C/10 "forming charge" of the pack after assembly, then use it as is, without caring too much about disbalance.
But recently I've run into a situation where it's desirable to have a Ni-MH balancer.
The device (robotic vacuum cleaner) charges 12S Ni-MH 2200mAH battery with ~C/2 current, with dV/dt and dT/dt end-of-charge controls seemingly in place. (And discharges them at ~C/2 when running.)
After few hundred cycles (a year or so of nearly everyday use) the usable capacity of the battery significantly drops.
Cell-by-cell test clearly shows some cells where chronically undercharged, that resulted in polarity inversion during discharge and rapid deterioration of affected cells.
I think that proper balancing could significantly (2 times or so) prolong overall usable battery life (as the healthy cells in the battery still have 80%+ of their initial capacity).
Searching the net resulted in few proposals for Nickel balancer:
1) Very practical and simple voltage limiter: https://www.electroschematics.com/balancing-ni-mh-battery-packs/
A pair of rectifier diodes per cell that will partially shunt the cell once it's nearly full, thus limiting it's charge, and giving runners-up a chance to catch up.
Requires all diodes to be on a common heatsink for temperature equalization, will drain batteries fast if not disconnected, will limit usable capacity of the battery.
While cruder, it looks to be OK for occasional balancing, but it's impossible to integrate such a "balancer" into the battery.
2) Microcontroller-based inductive balancer: http://cache.freescale.com/files/32bit/doc/app_note/AN4428.pdf
Good all-around solution, but it looks like an overkill for the case. The feasibility criteria is the cost of the balancer that should not significantly exceed the cost of cells for a new battery.
3) Simple charge shuttle balancer: https://easyeda.com/Popov_Alex_r/Battery_Balancer_Ni_Cd_Active1-840522a2a4e44fe89bf70d560e4607f9
Looks great for integration with standby current of 80uA (that's 60mAh/month, comparable with self-discharge of 2200mAh Ni-MH cells), but charge shuttles are not particularly effective, and I'm in doubt if it would be able to reach balanced state on a used battery with maximum balancing current of just 10mA.
Looks like none of purpose-built balancing ICs are able to work with Nickel cell voltages, and most are kind of specialized on Li cells with undervoltage protection on ~2.75V.
Any other ideas on Nickel battery balancer that could be integrated into the battery of an existing appliance?