DB, I'll try to keep this simple (not that I'm accusing you of being simple).
AS EN48 says, the TDZ is the first 3000ft of an instrument runway and the notional touchdown point is found within this area. An ILS GP is set up such that an aircraft is guided down to a point in space that we would commonly understand as coincident with the TCH (in other words, a point about 50ft above the runway threshold). Beyond this point, the ILS GP signal is nonsense (it's actually pretty useless below about 120ft) but the notional GP is extrapolated to a point on the runway which is about one quarter distance into the TDZ (let's say 800ft). The ILS DME is then normally adjusted to give a zero reading at this point (
try it next time you fly an approach - check the DME reading over the threshold and you will note is is normally a smidgen over 0.1nm)
The system minima is designed around the highest centreline elevation of the runway within the TDZ but the elevation of this point may not be coincident with the DME zeroed notional touchdown point. Indeed, in the most extreme cases, the highest TDZ elevation may be 2000ft further down the runway than the notional touchdown point; you could be short changed!!
What does this all mean?
It means that an ILS GP 'origin' is assumed to be a point some 800ft within the TDZ but the system minima is based upon the TDZE which will often be different to the elevation of the GP origin - the runway threshold plays no part. However, if we keep things simple and assume that TDZ, threshold and GP 'origin' are all the same elevation, we will find that a 3 degree GP crosses the threshold at about 50ft and will be 200ft at 0.65nm from the GP origin (about 0.52nm from the runway threshold). The two examples I gave, albeit 3.5 deg GP, demonstrate a higher ground elevation as an aircraft passes system minima and your radalt will, at this point, read less than 200ft (there are also buildings at these locations further reducing the radalt indication). I've personally flown one example (not in Europe) where the radalt was less than height above TDZE by nearly 100ft. London City also gives some interesting indications.
In simple terms, DAs/DHs aren't normally increased due to obstacles that occur on final approach prior to system minima as any such obstacle would probably demand an increased GP angle. Increased DAs/DHs are far more common due to problems in the missed approach segment which for a precision approach commences at system minima, i.e. the 200ft/0.65nm point.
I think I've rambled on; sorry.
