Ignition Timing

Make sure the degree scale in the primary cover is accurately set, and make sure the Trispark rotor is with tolerance regarding how much offset from the ignition plate mounting surface it should be. For it to be working ok and then not sounds like something has slipped, broken, blocked or otherwise gone duff, wiring all ok? 

You can't determine TDC accurately by feeling with a pencil. Its better to use the 'split the difference' method with a plug stop and degree disk and forget about linear measurement.

I have always found that timing doesn't have to be spot on to at least start, if not to run well, so your method may be good enough to at least get you going.  The stroke is 89mm, then ignoring the con rod effect, which will be insignificant at a small angle like 28 degrees, the distance is 89/2 * (1-cos 28) = 5.2mm

When clueless (a lot of the time )  with a bike ,i just set the static (retarded) timing a few degrees /mm before TDC and adjust it after starting a few times so that i cant provoke  much of a kickback .(cant afford that anyway with old arthritic legs) . Probably takes the edge off the high performance but not by much as i still get told to slow down,and ,   its not a race !. 

Are you sure you are on the firing stroke when setting your timing?
A common slip up when your fixing rather than fettling.

My money is on the following............
The Firing Point is close to max advance at around 28*.
This will allow the Ignition to fire and power a stroke. But this stroke then over-advances the ignition until the system fights itself to a stop.
A similar effect will occur if a Dominator electronic ignition system is bolted on a Commando engine with timing cover points. The ignition will advance itself into a retarded mode.