What happens with an ILS (or LPV) glideslope below the DH?

Timothy wrote:

But the good news is that we all seem now to be on the same page. It flies super accurately to ~10’ above the GPIP, then levels off and flies along the centreline to the end of the runway.

That is the sim behavior. The vertical guidance from the GTN ceases past the GPIP according to RTCA DO 229, so what the sim or the trainer does will not represent the actual outputs of the GTN or how the aircraft autopilot would react.in an actual aircraft. My expectation is that there is no vertical guidance beyond the GPIP, so if you are still 30 feet in the air, what happens is going to be autopilot dependent.

What does “no vertical guidance” mean?

In the GPS-autopilot connection, we are talking about either the analog signals (which are accompanied by flags) or about an ARINC429 connection (which I don’t know the protocol for).

At the GPIP, are the LOC and GS signals flagged as invalid, or are they just held at the previous levels but held constant? An analog signal doesn’t just disappear; a voltage is always a voltage But an ARINC429 stream could be simply terminated.

I am sure some thought went into this stuff, because you don’t want to kill everybody who left the autopilot engaged, by gratuitously inverting their aircraft

I would expect, on the loss of analog GS guidance i.e. the GS gets flagged as invalid, for the autopilot to switch to PIT mode. That is the only safe status. One could make a case for the ALT mode, perhaps.

Peter wrote:

I would expect, on the loss of analog GS guidance i.e. the GS gets flagged as invalid, for the autopilot to switch to PIT mode. That is the only safe status. One could make a case for the ALT mode, perhaps.

In the sim, the CDI remains centred, the CDI Flag is FROM, the A/P remains in NAV & GS modes.

Yes, agree very close to the edge of acceptable pzrameters and things will start to depends on code compilers and platforms not just avionics providers, a good lesson is here:

https://en.m.wikipedia.org/wiki/Cluster_(spacecraft)

Btw, no doubt a lot of smart guys were involved on that project

This could be probably the reason why we dont have auto-taxi as an extension to auto-land on Cat IIIc runways (current, requirement of ILS IIIC is land/stop)?

A fascinating link Apollo used mostly integer maths which needs a lot more understanding of what is actually going on. In embedded systems, floats, especially 64 bit floats, tend to show the opposite…

Poorly implemented integer maths is most likely responsible for the self destructing KFC225 servos.

the A/P remains in NAV & GS modes.

Which AP is that?

Very interesting Timothy and others.

I guess what is not a bad idea for anyone having a regular plane to play with is to simply try this stuff out in VMC and to a safe height. Down to 50 ft in VMC and with the finger on the AP Disconnect button it should well be possible to try out what to expect of the setup in fog. So in the case of cases you know what to expect.

One bit to remember is also that fog conditions actually favour autopilots as there is very few if any wind. My S-TEC is not very accurate in holding an ILS as soon as there is wind but it is pretty much ok without wind. I am told that it is much better in LPV.

But it has to be clear that these things are real emergency “outs” which will in almost every case be the result of lacking preflight planning and poor decisions, it should in no way be a “freebee” to use to get home in a normal case. In many such cases you would also have to expect the ATCO and quite possibly the airport authority to ask questions and possibly file a report and there will be quite a lot of explaining needed. So the foremost thing must be to avoid the need to employ such antics in the first place.

As for the older Autoland systems: One of the first was certified for Aeropostale’s Caravelles. Incidently, we had the hardware in ours but never used it (it was illegal to use anyway for us). I guess this whole thing was a bit of a “French” solution and their idea that cargo is less valuable than passenger lifes. It was a single channel system and therefore not the most accurate. We used to fly it down to CAT I minima and it was resonably ok in fog conditions (which as a precondition has very few wind) but lost it rather fast with crosswind.

Peter wrote:

Which AP is that?

It’s not marked, but it’s a legacy Bendix/King / Century / Altimatic. The closest picture I can find on the interweb is the Bendix/King KC290, but the switches are a little different.

I would bet you a year’s supply of TKS fluid that whoever wrote that sim didn’t know what it does when the inputs disappear (or get flagged, or whatever). A KC290, on a quick google, pre-dates the “bendix” in bendix-king and looks totally analog, and while the schematic is obviously “obtainable” I would be amazed if the sim programmer had re-drawn it into an analog sim (spice, electronic workbench, etc) and incorporated that into the sim.

@wigglyamp might know but I think that when the GS signal disappears, the GS outputs will get flagged as invalid, and the autopilot will then do something supposedly benign, like switching to PIT mode.

Just for completeness, I did the same on the ILS, just to see if it’s an autopilot artefact

It did a perfect autoland!