Why would “VS hold” be a good idea if the wind (and therefore the GS) is changing?
I wouldn’t fly a NPA in VS mode either. It kind of defeats the purpose because you want to vary VS to meet the desired glide. I would fly it with the trim wheel only. Keep watching your VS and your check altitudes and adjust power and trim.
Flying a NPA to many GA airports in the US and using a constant angle descent is not going to result in an approach being completed with a landing when the visibility is low. What often is needed is time at the MDA to locate the runway out of the mist. The more time available at the MDA, the greater the chances of success. MITRE did a study comparing the safety of CANPA (Constant Angle Non Precision Approach) with turbo jet operations and found clear evidence that it was safer. They then repeated the study for piston aircraft, and guess which method was safer although not statistically significant. Dive and Drive! The point is that one technique does not fit all and that there were no measurable differences in safety for piston GA aircraft. Here in the US, there is no TERPS provision for ensuring that on an NPA that the visual segment is clear of obstacles. Most GA airports do not have a control tower, many have narrow and short runways, tree lined, with obstacles and challenging terrain, they typically do not employ approach lighting systems, are not always aligned with the final approach course, can be poorly marked. It takes time to figure out if what you are actually looking at is a runway or a Walmart parking lot.
We are not permitted to use GPS in lieu of ADF or VOR or localizer or LDA for the lateral guidance once inside the FAF. Only recently has there even been a clarification that permits using GPS coupled to the autopilot when flying an NDB or VOR approach as long as the underlying equipment is on board the aircraft and being monitored on the approach. The facility must also be in service.
Sure, and I agree with you that dive and drive (at least in light aircraft) is better in terms of finding the airfield.
However, like it or not, here in Europe, dive and drive is totally on the way out and even frowned upon by many people.
Even commercial flightschols tend to teach only the CDFA, because they use the PA28 to "simulate"’ flying a bigger aircraft…
NCYankee makes his usual very well argued point in favour of descending down to the MDA after the last SDF, which is fine because your obstacle clearance is assured after each SDF.
I don’t disagree with that at all, but my original Q was (let me rephrase it, and I was tempted to modify the threat title):
How to best fly an NPA using an autopilot with lateral (LNAV) guidance, when the autopilot has VS hold (but obviously there is no synthetic glideslope etc etc etc otherwise I would not be asking the Q)?
You have two options
The former method is possibly trained… to the negligible extent that anything to do with that mysterious box called an autopilot is trained. It is also what do you, more or less, if flying by hand.
The latter method is arguably making better use of the available automation because you are using it for what it is good at, and leaving you the pilot to do what you are good at (operating the throttle). It is also helped by the fact that if you start the “glideslope” at a high-ish speed, say 100-130kt (often desired by ATC), you end up with a natural IAS decrease on the way down, which is exactly what you want to go to full flap and land.
This topic cannot be disposed of with one-liners 
I don’t see what the value of using VS should be. What you want is a constant IAS in the approach. That’s the main criteria of a stablized approach. And of course you control the descent rate with power. As always: Speed with elevator/trim – descent rate with power. Very simple. Setting a constant VS would mean you have to adjust the IAS a couple of times and that is not what you want in a stabilized approach. I try to set the trim once, and I do not try to capture the GS with trim or elevator changes.
I find the “dive and drive” method much easier to learn, and safer too. But that is a different topic.
EASA now preaches CANPA/CDA* for all non precision approaches. Many of them will have artificially generated glideslopes in the 430W/G1000 or your FMC of choice. If so, you can follow that quite happily. If not, the technique taught everywhere, on every type, since I have started doing them is exactly as described by what_next
You will fly a more stable and accurate approach using your first method Peter – I do not know anybody who teaches varying approach IAS to maintain any form of glide path. Your typical VS adjustments will very rarely be more than 100fpm at a time. Does your autopilot allow you to adjust VS in finer increments than 100fpm? Some I have come across can do 50fpm increments if the VS is below +/-1000fpm which is very helpful. I don’t see what is wrong with requiring hand-flown approaches as part of skills tests or training – you need to demonstrate the ability to do it properly and can then have the autopilot to make your life easier.
On basic automatics like I had in the AA5, I would fly non-precision approaches manually and I would adjust the VS the old fashioned way, but it was VS I was scanning to maintain an accurate vertical profile at constant IAS.
Leaving you to find your own methods and operating procedures as long as you pass your revalidation is ultimately one of the freedoms private flying gives you.
*Constant Angle Non Precision Approach/Continuous Descent Approach
I think the what Peter is trying to say is that the “constant VS, constant GS” method allows, in theory, for a more optimal approach where you have continuous feedback as to whether you are on target, and if you maintain that target throughout (correct constant VS and correct constant GS), you should fly perfectly down the glideslope the whole way.
Whereas if you use the normal “constant IAS” method, you only get feedback that you can act on at each altitude reference point, and at a given point (including at the MAP) may be higher than you would really like to be. You have no information to act on between each reference height unless you interpolate the whole way down.
I can’t decide if this is an “innovative” new way of doing things, or just a bad idea in general :). Certainly never heard of it before myself.
Either way I think it is fairly irrelevant once you have LNAV+V or better, which gives reference to a glideslope fixed in space that you can use instead.
Shoreham is an interesting example though, as the approach is very steep and there is terrain which means you can’t afford to be too low, and both of these factors create a tendency to be too high at the MAP.
I’m a bit confused about what the whole problem is here.
I definitely don’t agree with adjusting your speed to maintain a glidepath though. I’m not saying it wont work because it will (most of the time) but it definitely seems like the wrong way to do it to me.
At work if we don’t have a coded path to follow then we use vertical speed mode (with a selectable vs) and fly it like that. As an example if we are doing a vor/dme approach and have to use vertical speed, we will configure before descent and then start our descent at the FAF. For a 3 degree slope we typically need a VS of about 750, so normally we select a higher one first because unless you start descending before the FAF you are going to ‘overshoot’ slightly in the transition into the descent. thereafter we cross reference the distance/altitude and adjust VS as required.
This is how I would fly it without an autopilot too, except instead of using a thumbwheel to change VS I would use the control wheel, and I’d then adjust speed with the thrust levers/throttle.
On my recent initial IR checkride the examiner asked “tell me the speed on the approach”. You then give them a speed and this is a fixed value that you have to follow within a certain allowable tolerance.
The initial VS in calculated based on GS and adjusted according to the checkpoints along the decent.
No autopilot allowed.
