You are right, to some extent. But FETs have a positive temp coeff so if one gets hotter, it's on resistance rises so it takes less current which provides a pretty good self-balancing capability. IGBTs on the other hand, may have a +,- or 0 temp coeff which means they may not current share in the same way. That temp coeff may even vary in the same batch, so they tell me. The smaller current ones tend to have a + temp coeff and they are more uniform which seems to be why its better to use 20 x 50A ones than 10 x 100A ones. I'm only going on what I've read in a few App Notes over the last week or two as I have no experience in using the things at all. Tritium-James appears to have been working on that ac controller they have just released and would, I expect, know a fair bit about this. But FETs are really limited to a few hundred volts whereas IGBTs go to, what, 1000 or more? so the choice is fairly clearcut in that way. It's the grey area where the overlap that is hard.
My measurements of the gate drive on the LS controller indicate to me that the trace length is not so critical with relatively slow switching times generated by somewhat large gate resistors and these slow times mean that the devices are turning on slow enough so that one will not "hog" the current before others start to share the load.
It certainly is something I'd like to do a lot more study on.
192 cells is still a lot! 24 slave boards.
My original idea was to use 3R3 balancing resistors which give a current of 3.6/3.3 = 1.09A for SE cells and to have the charger switch down to that current when one cell got to 3.6V while the master turns on the shunt for that cell. That means that 1.09 amps continues to flow through the battery, but not through any cell that has reached 3.6V. Because I "know the man" who is designing the charger, I can get him to make that lower current limit programmable to suit whatever the BMS is doing. 1A or so balancing current seemed to me to be a reasonable figure. But after reading a bit more about balancing etc, I decided that the heat generated was going to be a bit excessive so I decided to increase the balancing resistors to 18R and drop the charger current down to .2A. It seems that this is still a quite adequate charging current to do the final balancing act and it should be if the cells are fairly close anyway. It wouldn't be much good if you had 100 AHr cells with one 100% charged and one 100% discharged!
The beauty of this scheme is that you can really make the balancing resistors as large or small as you like within your capacity to keep them cool and you can program the reduced charger current to match.
In your case, with 3.7V per cell and 18R shunts, the charger would be set to .206A. Or you could use 3R3 resistors, program in 1.21A and stick a fan on each slave board to keep them all cool.
The charger can deliver up to 5400 watts if you have 3-phase available, but it is limited to 180V which means it could deliver as much as 30A. In my case with 162V needed and only single phase power, it will be set at around 20A.
If you were to use a different charger, you could build a programmable current limiter and control that with the BMS output that says "reduce to lower current".
All this is just a long-winded way to say "yes, everyone but you DID know that"
I have further considered that if one were to use two FET switches for each cell, one in series and one in parallel, you could turn off the series one and turn on the parallel one simultaneously. The parallel one would then carry the whole charge current bypassing the cell and series switch and the power dissipation would only be i x i x Ron so the balancing would be done at the full charger current and therefore very much faster. You'd want FETs that can carry 20 A or so with an Ron down in the milliohms region, but that should be able to be a lot less power wastage than the resistors. More expensive of course.
). If load sag becomes an issue, that's how we intend to handle the current measurement with our digital BMS.