In this case it would seem that he fought it initially and the electric trim fought back, then consequently, once he allowed it to climb into cloud, all the control forces felt wrong and the autopilot stalled and span.
Peter wrote:
For that simple method to work you do need reasonably good vis though and be sure there are no obstacles.
I agree completely about the vis, but please nominate the approaches to UK runways where obstacles on a 3° slope would actually be a problem. I am sure that there are a few, but I can only think of 19 at Oban and 25 at Little Staughton as definites, and 36 at Wellesbourne and 28 at Sherburn which might be a bit iffy if you got below the slope.
Yes, there are, of course, considerably greater risks, but are they proportionate to the risks of scud-running?
Unfortunately many owners have no appreciation of the effect of fighting the autopilot in pitch and having the auto-trim running so far that when they finally disengage the autopilot, the aircraft is almost uncontrollable.
Timothy wrote:
and it is for sure that Philip Garveyand David Norris werenot attempting any kind of sensible straight in approach, but were attempting to scud run.
He crashed because he lost control in IMC. It just happened to commence at the beginning of a DIY let down.
Your analysis is amazing. Thank you.
It’s far more objective than an assertion that he was scud running and the accident (probably) wouldn’t have happened if there was an LPV in place. 
It appears clear that scud running deaths far far exceed DIY IAP deaths. One has to really trawl the records to find [clear examples of] the latter but the former is everywhere.
The problem is that most scud running is done by pilots who could not fly IFR legally. If you can fly IFR legally, and there is no other factor (e.g. nearly empty tanks due to earlier bad decisions / lack of a fuel totaliser) then you just divert to the nearest ILS. For Dunkeswell, just go to Exeter; 5 mins away.
Once you start analysing CFITs like that you get into a more complex debate. For example an IR guy did a CFIT in SE France (Tyson?) which could have been trivially avoided by flying at the altitude where Nice Approach want everybody, FL150, but that needs oxygen, which like many IR holders he was not carrying. A number of IR holders have done CFITs primarily due to not carrying oxygen. N2195B was another. There is a big culture of not carrying oxygen and just flying a “de iced” plane through whatever IMC.
It isn’t really an argument for LPV.
The NTSB each month records IR pilots crashing in IMC, often on an approach. Lack of currency in real IMC?
(I let my IMC rating lapse for 2 reasons – I didn’t want to do enough instrument flying to keep current, and a realisation that I didn’t have reliable enough instruments.)
On the SkyDemon/virtual GS debate, I have tried it this morning again on a commercial flight. It was as always (when I tried it) showing the plane at least 1.5 dots high. I think that is because an ILS takes you to the middle of the TDZ whereas SkyDemon uses the runway threshold as a reference? Also, there was no lag judging from lateral position shown which corresponded well to the ground features. Using internal GPS of an original iPad mini (the very old one).
Rwy20 wrote:
On the SkyDemon/virtual GS debate, I have tried it this morning again on a commercial flight. It was as always (when I tried it) showing the plane at least 1.5 dots high.
This is my experience as well, when using Skydemon on commercial flights.
Yes; we did that in the GPS-ILS thread. The localiser is shown correctly but the glideslope on a real ILS is some way away from the glideslope depicted on these apps.
I don’t know if the data on an ILS glideslope is published. If it was, it would make obvious sense for these apps to use that data. The LPV data is definitely published, as a “hex block”, in the AIPs, and any such app ought to be using that, instead of doing a straight line from the FAF to the MAP which I think is all they do.
