How not to play nice with a GPS

I had an interesting experience earlier today. I thought after all this
time I had the CNX-80 figured out, but I found a new way to screw things up.

I was on the POU GPS-24 (http://204.108.4.16/d-tpp/0508/00286VDG24.PDF),
full procedure, Kingston IAF. We were IFR, in and out of IMC. We had been
cleared for the appoach, which was loaded and executed in the GPS. We were
on a heading of roughly 180 to IGN, and about a mile from the VOR, when the
controller said something like, "I need to you to stay at 4000 and keep
tracking outbound for a while for traffic below you". I agreed to do so
and turned to a 060 heading.

At this point, things started going downhill. I suppose, technically the
controller committed an error. He had already cleared me for the approach,
but then gave me instructions contrary to the procedure. I was in a sort
of half-way place between flying the full procedure pilot nav and being on
vectors. Better to have just canceled my approach clearance and given me
vectors to avoid the traffic then back around for a fresh start.

I was equally at fault for accepting this. But, now that I had made my
pact with the devil, I needed to convince the GPS to play along with our
game. The box hadn't yet sequenced past IGN, when I started my turn
outbound. This seemed to have confused the computer. We were heading
outbound on the PT, but the GPS was still navigating to IGN. In an attempt
to convince it otherwise, I pulled up the flight plan page and selected
direct to CFGUY, which got me what looked like positive (and correct)
course guidance on the outbound leg.

Eventually the controller told me I could turn back inbound and descend to
2900. I did so and found myself tracking inbound, but with the CDI giving
me indications opposite of what it should have been, apparently the box
thinking I was still trying to track outbound. I could see (from the
moving map) that I was south of the final approach course, but the CDI was
giving me "fly left" indication. At some point, while we were puzzling
over this, the CDI suddenly flipped to "fly right"; I think this might have
been as we passed CFGUY inbound. Everything proceeded normally from that
point.

I think the moral of the story here is that the computer is not good at
ad-libbing procedures. Accepting an early turn outbound and an extended PT
would have been a no-brainer with a real VOR-DME setup. With the GPS,
since that wasn't what it was programmed to expect, things got ugly. On
the other hand, with the moving map, it's hard to go too far wrong. As
long as I had a picture of where I was relative to IGN, I could pretty much
fly the desired track by eye as I sorted things out.

You can get reverse sensing from GPS units, for much the same reason
that you get them from VOR units.

First and foremost, you are correct that ATC incorrectly amended your approach
clearance. Trouble is, Frick plays with Frack too much, so this stuff gets sort
of "codified."

Had you said, "Telling me to stay at 4,000 is an altitude assignment, which
cancels my approach clearance" could have sent you to pergatory. ;-)

There is no easy answer. It's humans dealing with humans in a very imperfect
system.

Where there are a lot of repeats (read "traffic") in similiar circumstances,
particularly heavy iron traffic) the kinks usually get worked out. Not so for
light G/A ops..sadly.

Roy Smith wrote:

I had an interesting experience earlier today. I thought after all this
time I had the CNX-80 figured out, but I found a new way to screw things up.

I was on the POU GPS-24 (http://204.108.4.16/d-tpp/0508/00286VDG24.PDF),
full procedure, Kingston IAF. We were IFR, in and out of IMC. We had been
cleared for the appoach, which was loaded and executed in the GPS. We were
on a heading of roughly 180 to IGN, and about a mile from the VOR, when the
controller said something like, "I need to you to stay at 4000 and keep
tracking outbound for a while for traffic below you". I agreed to do so
and turned to a 060 heading.

At this point, things started going downhill. I suppose, technically the
controller committed an error. He had already cleared me for the approach,
but then gave me instructions contrary to the procedure. I was in a sort
of half-way place between flying the full procedure pilot nav and being on
vectors. Better to have just canceled my approach clearance and given me
vectors to avoid the traffic then back around for a fresh start.

I was equally at fault for accepting this. But, now that I had made my
pact with the devil, I needed to convince the GPS to play along with our
game. The box hadn't yet sequenced past IGN, when I started my turn
outbound. This seemed to have confused the computer. We were heading
outbound on the PT, but the GPS was still navigating to IGN. In an attempt
to convince it otherwise, I pulled up the flight plan page and selected
direct to CFGUY, which got me what looked like positive (and correct)
course guidance on the outbound leg.

Eventually the controller told me I could turn back inbound and descend to
2900. I did so and found myself tracking inbound, but with the CDI giving
me indications opposite of what it should have been, apparently the box
thinking I was still trying to track outbound. I could see (from the
moving map) that I was south of the final approach course, but the CDI was
giving me "fly left" indication. At some point, while we were puzzling
over this, the CDI suddenly flipped to "fly right"; I think this might have
been as we passed CFGUY inbound. Everything proceeded normally from that
point.

I think the moral of the story here is that the computer is not good at
ad-libbing procedures. Accepting an early turn outbound and an extended PT
would have been a no-brainer with a real VOR-DME setup. With the GPS,
since that wasn't what it was programmed to expect, things got ugly. On
the other hand, with the moving map, it's hard to go too far wrong. As
long as I had a picture of where I was relative to IGN, I could pretty much
fly the desired track by eye as I sorted things out.

Where there are a lot of repeats (read "traffic") in similiar circumstances,
particularly heavy iron traffic) the kinks usually get worked out. Not so for
light G/A ops..sadly.

I have a feeling there may be something else at play here. The GPS
should have a mode similar to "suspend", which would allow you to do
"whatever you want" in the middle of an approach, and then gracefully
come back to it. But approaches aren't supposed to work that way, so
getting the FAA to approve a device that lets the user do stuff he's not
supposed to do in the first place is not likely to happen. (I recall
the flap about the FAA requiring that sferics devices NOT link to the
moving map display).

So, the devices are designed to only work when rules are followed, so
the devices enforce the rules on us. But the controllers aren't using
those devices and are free of the tyranny of [that particular] machine.

It is happening in many walks of life, aviation being only one of them.

Jose
--
Quantum Mechanics is like this: God =does= play dice with the universe,
except there's no God, and there's no dice. And maybe there's no universe.
for Email, make the obvious change in the address.

Jose wrote:

Where there are a lot of repeats (read "traffic") in similiar circumstances,
particularly heavy iron traffic) the kinks usually get worked out. Not so for
light G/A ops..sadly.

I have a feeling there may be something else at play here. The GPS
should have a mode similar to "suspend", which would allow you to do
"whatever you want" in the middle of an approach, and then gracefully
come back to it. But approaches aren't supposed to work that way, so
getting the FAA to approve a device that lets the user do stuff he's not
supposed to do in the first place is not likely to happen. (I recall
the flap about the FAA requiring that sferics devices NOT link to the
moving map display).

I don't know anything about the Garmin 480, which I believe Roy was using. But,
with a Garmin 530 I could select OBS mode while IGN was the active waypoint, then
selected a bearing of 062 to have a correct outbound reference. Once I turned
inbound I could select 242 for proper sensing, then as I established track cancel
the OBS mode and the active leg should be correct.

, wrote:
I don't know anything about the Garmin 480, which I believe Roy was
using.

Yes, I was using a GNS-480 (nee CNX-80).

But, with a Garmin 530 I could select OBS mode while IGN was the
active waypoint, then selected a bearing of 062 to have a correct
outbound reference. Once I turned inbound I could select 242 for proper
sensing, then as I established track cancel the OBS mode and the active
leg should be correct.

I can certainly do "Course To", "Course From", and "OBS" on the 480. These
are all minor variations on what you're describing, with OBS being the most
exact match. I'm not sure, however, if the software will let me do those
things on an approach.

Like any other instrument failure, the hard part is recognizing that there
is a problem. Flying partial panel isn't really that hard; the trick is
noticing that the primary instrument has gone TU and not following it into
oblivion in the first place. Once I figured out something was not right
with the GPS (because the CDI and map displays disagreed), there were lots
of ways to fix it. The trick is noticing in the first place.

I can certainly do "Course To", "Course From", and "OBS" on the 480. These
are all minor variations on what you're describing, with OBS being the most
exact match. I'm not sure, however, if the software will let me do those
things on an approach.

Like any other instrument failure, the hard part is recognizing that there
is a problem. Flying partial panel isn't really that hard; the trick is
noticing that the primary instrument has gone TU and not following it into
oblivion in the first place. Once I figured out something was not right
with the GPS (because the CDI and map displays disagreed), there were lots
of ways to fix it. The trick is noticing in the first place.

This is why those involved in the human-factors aspects of RNAV have concluded
that it is next to impossible to manage "curve balls" without a good moving
map. I wouldn't characterize your circumstances as an instrument failure, but
it certainly was a setup for a route discontinuity.

I further checked it with the Garmin 530 using the 530 in conjunction with a
Sandel EHSI. The correct (or best) solution in that case was to go to OBS mode
and set a course of 242 as you worked through the delay "vector." That
resulted in both a magenta line on the correct side of the VOR station and, at
least with an EHSI, correct CDI sensing. Then, when I turned inbound (at about
12 miles) and intercepted the 242 track I cancelled OBS mode. Then, I had to
line select the "PT" to GFGUY" flight plan leg (direct, direct, enter) and I
was all set, just as if I had rolled out of the procedure turn.

It works okay with the OBS set to 062 except the active track is white, not
magenta because the magenta line extends from the VOR station to the southwest
along 242. But, that still works, but is a bit confusing.

It could be somewhat different with a fixed CDI. I don't know.

Bottom line: unlike VOR or ILS, being qualified on RNAV equipment "A" doesn't
give anyone a franchise to operated RNAV equipment "B" or for that matter "B
through Z."

And, think of the issues ahead with the advanced procedures like the one I
posted for DCA.

Roy Smith wrote:
I had an interesting experience earlier today. I thought after all this
time I had the CNX-80 figured out, but I found a new way to screw things up.

I was on the POU GPS-24 (http://204.108.4.16/d-tpp/0508/00286VDG24.PDF),
full procedure, Kingston IAF. We were IFR, in and out of IMC. We had been
cleared for the appoach, which was loaded and executed in the GPS. We were
on a heading of roughly 180 to IGN, and about a mile from the VOR, when the
controller said something like, "I need to you to stay at 4000 and keep
tracking outbound for a while for traffic below you". I agreed to do so
and turned to a 060 heading.

At this point, things started going downhill. I suppose, technically the
controller committed an error. He had already cleared me for the approach,
but then gave me instructions contrary to the procedure. I was in a sort
of half-way place between flying the full procedure pilot nav and being on
vectors. Better to have just canceled my approach clearance and given me
vectors to avoid the traffic then back around for a fresh start.

I was equally at fault for accepting this. But, now that I had made my
pact with the devil, I needed to convince the GPS to play along with our
game. The box hadn't yet sequenced past IGN, when I started my turn
outbound. This seemed to have confused the computer. We were heading
outbound on the PT, but the GPS was still navigating to IGN. In an attempt
to convince it otherwise, I pulled up the flight plan page and selected
direct to CFGUY, which got me what looked like positive (and correct)
course guidance on the outbound leg.

Eventually the controller told me I could turn back inbound and descend to
2900. I did so and found myself tracking inbound, but with the CDI giving
me indications opposite of what it should have been, apparently the box
thinking I was still trying to track outbound. I could see (from the
moving map) that I was south of the final approach course, but the CDI was
giving me "fly left" indication. At some point, while we were puzzling
over this, the CDI suddenly flipped to "fly right"; I think this might have
been as we passed CFGUY inbound. Everything proceeded normally from that
point.

I think the moral of the story here is that the computer is not good at
ad-libbing procedures. Accepting an early turn outbound and an extended PT
would have been a no-brainer with a real VOR-DME setup. With the GPS,
since that wasn't what it was programmed to expect, things got ugly. On
the other hand, with the moving map, it's hard to go too far wrong. As
long as I had a picture of where I was relative to IGN, I could pretty much
fly the desired track by eye as I sorted things out.

You sound like a real knowledgable guy with your GPS, but what you are
describing is the effect of leaving an approach in effect on your GPS, but
not actually flying it. There are two immediate ways to deal with that.
If your GPS has a "suspend" function, that should be activated. If not,
the approach should be canceled and/or a new waypoint entered.

Roy Smith wrote:

Eventually the controller told me I could turn back inbound and
descend to 2900. I did so and found myself tracking inbound, but with
the CDI giving me indications opposite of what it should have been,
apparently the box thinking I was still trying to track outbound. I
could see (from the moving map) that I was south of the final approach
course, but the CDI was giving me "fly left" indication. At some
point, while we were puzzling over this, the CDI suddenly flipped to
"fly right"; I think this might have been as we passed CFGUY inbound.
Everything proceeded normally from that point.

Another option at this point would have been to select VTF (Vectors to Final). When it became clear the
box was still stuck on the outbound leg or PT, this would have cleaned up the flight plan and made the
inbound leg to the FAF active. The CDI would then indicate normal deviation to the inbound course also.

Mike

Mike Adams wrote:

Roy Smith wrote:

Eventually the controller told me I could turn back inbound and
descend to 2900. I did so and found myself tracking inbound, but with
the CDI giving me indications opposite of what it should have been,
apparently the box thinking I was still trying to track outbound. I
could see (from the moving map) that I was south of the final approach
course, but the CDI was giving me "fly left" indication. At some
point, while we were puzzling over this, the CDI suddenly flipped to
"fly right"; I think this might have been as we passed CFGUY inbound.
Everything proceeded normally from that point.

Another option at this point would have been to select VTF (Vectors to
Final). When it became clear the box was still stuck on the outbound leg
or PT, this would have cleaned up the flight plan and made the inbound
leg to the FAF active. The CDI would then indicate normal deviation to
the inbound course also.

In retrospect, I think going into VTF mode would have made the most sense,
at least on this particular approach, where the PT-inbound course was the
same as the FAC. On an approach with a dog-leg, that wouldn't have worked.