The only thing that changes when going from dual to triple conversion is that the front side of the SW-2F wafer shorts the upper 455KC winding of T1, and enables the lower 3035KC winding, while the rear side of the wafer turns the 2580KC oscillator portion of the V3 circuitry on or off by shorting L8.
Everything else is "in circuit" regardless of selected band. If I were to hazard a guess, my first thought would be that the upper winding of T1 isn't being shorted when in triple conversion mode, or something in the 455kc coupling network is amiss, maybe the caps (mica?) in one of the cans. I've heard these can sometimes be problematic.
Of course the oscillator components are suspect as well.
I'm no HQ expert by any means, I just happen to have finished
refurbishing my "new" 170 a couple days ago, so this stuff is fresh in my mind(?).