MOTM

>It's not the steady-state load, it's charging up all the caps on the 
>boards. The
>current in the first 100ms is probably 1A or so.
>
>No real easy solution.....power sequencing is the best.

An overkill solution might be some kind of relay circuit with 
pseudo-random delay times (using different resistor values to charge 
up the caps) on each of the MTA connectors. Like on a Yammy TX816 
when power is applied and then a second later you hear 8 relays 
click, but not all at the same time.