Hi Mick,
Slow reply because I couldn't think of anything to add other than "mine doesn't do that". And I couldn't suss it out from just the drawings in the manual. But I've got the top off my saw at the moment for some new bearings, so thought I'd see if I could help you out while I'm at it.
I have no such issue with my saw, and have been able to do some nice square mitres recently so I know my saw remains true even when tilted. Looking at the assembly, I can see 4 possible causes for this movement:
1. The Table is not securely fastened to the Apron. Sounds stupid, but this looks to be the place where I could get most play on mine if the bolts weren't tight. Looks like dowel 41 helps with this too (although I had not refitted these after my refurb job and everything seems fine even without them).
2. Shoes 37 & 51 are not tight in Segment 39, or there is bad wear in these parts.
3. Pivot Pin 34 (or the corresponding holes in Arms 36 & 82) is worn - I find it hard to imagine this could cause the 4-5mm deflection you are seeing.
4. Your damaged trunnion nut. At first, I thought this was unlikely to cause this problem, thinking it would only cause sloppiness in the tilt adjustment. My trunnion nut does not have a slot in it, and it is fixed to the arm with a hex head bolt and washer, just as shown in the manual. (So it looks like yours has been butchered in the past.) I notice that when I pull on the Saw Arm to simulate your deflection, the tilt Feed Screw 31 is securely anchored to the Tilt Arm by the Trunnion nut, and at the other end to the Apron by Tilt Shaft Bracket 6. So maybe this has some bracing effect and helps to keep the whole assembly rigid.
I'll have the top off for a couple of days whilst waiting for new bearings, so if you'd like any more details or photos of a working unit just shout.
Regards,
Andrew (yes there are 2 Andrews on this thread!)
P.S. On re-reading this thread I noticed some questions about manuals. I found the exact manual for a 275 model on ALT Saws website (very nice people), and the component references I quoted are from there.