How long before a waypoint does a GPS emit the new flight plan?

[peter-h]

The KLN94 starts emitting the next leg's data 35 seconds (evidently computed at the current GS) before the waypoint.

What does a Garmin GNSx30/W do?

I have been doing tests with a SN3500 EHSI. The KFC225 is actually driven from the KLN94's lateral deviation outputs directly; these also drive the EHSI so the EHSI is not relevant to the autopilot turn performance.

I find the EHSI course pointer moves to the new bearing at the t-35s point and the turn commences immediately then so evidently the GPS emits the new lateral deviations at the same time as the new flight plan data.

The problem with a t-35s advance is that it is far too much. From obvious geometry, if you want an intercept of the next track, with a 90-degree turn done mostly at Rate 1, you need something like t-20s.

Or, of course, roll steering

This stuff is mostly undocumented.

 

[Geico266]

The Garmin 496 is about the same. It gives a pop up with the new heading about 30 seconds prior with no voice prompt that I could hear. I think I had all of the warnings and voice prompts on.

The Cross Bow GPS in my Cheltons is lasar accurate and starts a turn based on your speed & cross wind drift to intercept the next course line. It's like being on a string.

 

[John Collins]

Peter,

I am not quite sure what you mean, but the GNS430W flashes the waypoint annunciator 10 seconds prior to commencing the turn to a new course. At the same time, a countdown message is displayed in the status area of the GNS430W screen. When the 10 seconds expires, the waypoint annunciator stays on and on the GNS430W displays a message to turn now to the new course. Roll steering starts the bank coincident with the turn now message. When past the waypoint (abeam it), the waypoint annunciator extinguishes. Also about a minute before the turn, a dashed curved path is depicted on the GNS430W map showing where the turn will occur

 

[poadeleted20]

To add to what John said, the 400/500-series GPS navigators normally use "turn anticipation" in which the point to start the turn is computed so the aircraft will roll out on the outbound courseline. This point is based on the number of degrees of turn, the ground speed, and the turn radius based on that ground speed and an IFR-standard 3-deg/sec turn rate. Obviously, unless there is no wind, this will not produce an exact result for a "blind" standard rate turn, but it will provide steering commands and course deviation data to the CDI to keep you on the curved path it computed from the computed roll-in point before the turnpoint itself and the roll-out point it computed on the outbound course line on the other side of the turnpoint as the wind changes relative to your heading around the turn.

 

[peter-h]

It appears that the Garmin does some kind of variably-computed turn anticipation.

The KLN94 doesn't - just a straight 35 sec advance before the WP.

I know a GNSx30/W can provide ARINC (serial data) roll steering direct to the autopilot but I did not realise this is tied in with what it displays on the front panel.

The KLN94 does have an analog form of roll steering (apparently designed for the KFC225 specially) but the behaviour of this is undocumented.

I have simply been doing some flight tests, to see how well the aircraft turns now that the EHSI course pointer auto-slews.

All turns finish well short of the next track segment.

One can work out the likely thinking behind the fixed 35-sec advance: it is about right for a 90-degree turn at Rate 1, with a bit of a lead-in and lead-out at a smaller turn rate, and smaller turns are much less critical anyway. Unfortunately, somebody messed up because it's obvious that the geometry is plain wrong and it should be about 20 secs if the exit track of the turn is to be aligned with the new track. The 35-sec advance is not configurable and is not dependent on the full-scale sensitivity of the KLN94 which can be set manually to 5.0, 1.0 or 0.3nm, or which varies automatically through 1.0 to 0.3nm when flying a GPS/RNAV approach loaded from the database. The pics show that increasing the GPS sensitivity does not yield a more "intelligent" turn, although of course the general tracking before and after the turn is most accurate at 0.3nm FS.

 

[John Collins]

Peter,

There is a difference in concept between auto slew and roll steering. Auto slew occurs when a new desired track to the next waypoint is provided by the GPS. When auto slew occurs, the course pointer changes to the new course. I don't know precisely when the new track information is provided to the RS232 interface, but where ever this occurs, is when the course pointer moves to the next course. The effect of an adjusted course pointer on the autopilot is unique to the specific autopilot. Some will turn to the new course when in a Nav mode, others will not. An autopilot using the CDI deviation to determine deviation from a straight line course as selected by the course pointer will determine course drift due to wind and will adjust the heading to re-intercept the course and apply a wind error correction to attempt to compensate for wind drift. This is iterative and will eventually track the course after a few iterations as long as the wind correction remains constant. If anything changes, such as the wind speed or direction or the aircraft speed, the autopilot will drift off and go thru the correction process again.

With roll steering, the course pointer is not relevant, neither is the CDI indication. Roll steering is a bank angle to fly to intercept and track a GPS path. Normally this path is displayed on the GPS as the magenta line. In the case of flightplan course changes, a curved path is calculated to turn to the new course using a rate of turn, typically standard rate. The bank angle for the turn is calculated and sent as the roll steering command. It is continuously updated as the turn is in progress and the turn completes when the aircraft is aligned with the new course and the track equals the desired track (in other words it is wind corrected). There is no equivalent CDI indication for the curved turn used in turn anticipation. In a similar way, if the course is not a series of straight segments, but involves turns such as a DME arc, a procedure turn, a hold entry, a holding pattern, etc, there is no equivalent CDI deviation that describes these turns, but the GPS calculates and transmits the bank angle necessary to turn and follow the intended GPS calculated path.

 

[poadeleted20]

I keep forgetting whether Rate 1 is standard rate and Rate 2 is half-standard, or vice versa.

 

[BillTIZ]

To add to what John said, the 400/500-series GPS navigators normally use "turn anticipation" in which the point to start the turn is computed so the aircraft will roll out on the outbound courseline. This point is based on the number of degrees of turn, the ground speed, and the turn radius based on that ground speed and an IFR-standard 3-deg/sec turn rate. Obviously, unless there is no wind, this will not produce an exact result for a "blind" standard rate turn, but it will provide steering commands and course deviation data to the CDI to keep you on the curved path it computed from the computed roll-in point before the turnpoint itself and the roll-out point it computed on the outbound course line on the other side of the turnpoint as the wind changes relative to your heading around the turn.

Turn anticipation, reminds me of OAS (Offensive Avioncs System) navigation waypoint loading. Either OFD overfly destination, then turn and accept the track line to the next waypoint. Or TSD, turn short destination, turn early to make the track line between waypoints good. Depends on what you needed, going down an approach I would want TSD.

 

[peter-h]

I know what turn anticipation is and I know what roll steering (or GPSS) is.

Occassionally I have used the term "roll steering" for what is just automatic autopilot steering which is implicit with an EHSI which has course pointer auto-slew enabled.

My Q was really what does a Garmin x30/W do when doing turn anticipation, via (or with) an EHSI. What it comes down to is how long before the next waypoint does the GPS change its flight plan data output. The KLN94 seems to change it a fixed 35 secs before the next WP.

It looks like this stuff is essentially undocumented.

I've installed a Sandel 3500 EHSI which obviously gives you implicit "automatic turning" because the HSI course pointer (CP) is rotated from the GPS flight plan data, and the autopilot follows the CP.

This all works fine but flight tests show that the flight plan data coming out of the KLN94 changes at a time which bears no relation to the kind of turn one is doing.

The EHSI just follows the flight plan data and flips the CP to the next track immediately, and the aircraft of course turns immediately because the autopilot uses the CP as the initial heading.

Clearly the KLN94 does it in a totally "dumb" manner. I wonder if a GNS430W does it "right" (or nearly so).

It is evident from obvious turn geometry that, for still air, commencing a R1 turn about 20 secs before the waypoint will get you a reasonable intercept.

If you have a tailwind you will overshoot the turn and vice versa, but not by all that much.

 

[John Collins]

Peter,

You are assuming too much and asking too much of the equipment. The GPS switching to the next DTK was not designed to be used for turn anticipation.

 

[John Collins]

Peter,

I just ran a test using the GNS530W simulator. DTK changes to the new course within a second or two of when the simulated aircraft starts its turn to the new course. The simulated aircraft is following a roll steering course. Since the amount of time prior to the turn commencing is a variable, depending on cross track, ground speed, track, track change required, etc., this can't be a fixed amount of time such as 35 seconds, but has to vary with the specific situation. So my best guess is that with a GNS430W/530W, the auto slew and autopilot operation will give approximately the same turn performance as roll steering on these kind of course changes.

 

[poadeleted20]

My Q was really what does a Garmin x30/W do when doing turn anticipation, via (or with) an EHSI.

Exactly what it does without an EHSI. The CDI type has nothing to do with what the unit does internally regarding turn anticipation.

What it comes down to is how long before the next waypoint does the GPS change its flight plan data output.

The flight plan cycles to the next waypoint as the current waypoint goes through the perpendicular to the track ("passes through the wingtip") during the turn. At that point, to give you a visual cue, the to/from flag is commanded to go briefly "from" and then "to" as the flight plan cycles.

However, the CDI positioning output remains independent of all that and is based on lateral deviation of the aircraft from the plotted course. In the turn, that means how far laterally you are off the computed arc from the computed roll-in point to the computed roll-out point. And it doesn't matter what sort of course deviation display you have, be it a traditional CDI, mechanical HSI, EHSI, whatever.

Now, if you have a bearing pointer display (like an RMI or BDHI or similar on your HSI), the bearing pointer will remain pointed at the old waypoint, swinging towards the outside of the turn, until the waypoint cycles to the next point as the pointer reaches the wingtip (approximately, depending on wind drift). At that point, the bearing pointer will swing into the turn, pointing at the new turnpoint, and will continue to swing towards the nose (again, approximately, depending on wind drift) as the turn continues until roll-out. You can see this in the movement of the green bearing chevron on the Nav1 (so-called "HSI") page of the 530.

Finally, if you have an autoslew function on your EHSI, the CDI presentation on the EHSI should rotate to the new DTK when the DTK cycles to the next course about the time the GPS commands the start of the turn. However, the CDI needle will, if you stay on the computed path during the turn, stay in the center through the whole turn.

 

[peter-h]

the auto slew and autopilot operation will give approximately the same turn performance as roll steering on these kind of course changes.

That would be a good result.

IMHO, coding any other behaviour is just pointless.

The flight plan cycles to the next waypoint as the current waypoint goes through the perpendicular to the track ("passes through the wingtip") during the turn

This is odd because what if you were flying in say the HDG mode and thus approached the WP without doing a turn at all? At which point would the GNSx30W RS232 flight plan data change? It has already changed to the next segment, at some computed point before the WP, and it is only after it changes that the airplane can commence the turn.

the CDI presentation on the EHSI should rotate to the new DTK when the DTK cycles to the next course about the time the GPS commands the start of the turn.

Indeed, but that cannot be when the WP is abeam, because you need the flight plan data to flip to the new section to commence the turn in the first place.

 

[poadeleted20]

This is odd because what if you were flying in say the HDG mode and thus approached the WP without doing a turn at all? At which point would the GNSx30W RS232 flight plan data change?

No different -- it will change when the waypoint goes past your wingtip.

It has already changed to the next segment, at some computed point before the WP, and it is only after it changes that the airplane can commence the turn.

Not exactly. As I said, it computes a flight path to fly, and that's what drives the roll steering and CDI deflection. This is irrelevant of what nav data (time/distance to go, etc) is being displayed. The nav data presentation is a separate issue, and it cycles to the next point as you pass abeam the current point even though the airplane has started to turn to follow the computed track over ground.

Indeed, but that cannot be when the WP is abeam, because you need the flight plan data to flip to the new section to commence the turn in the first place.

No, you don't. The computed flight path and the nav data are separate issues. If you do what you said, and keep it in HDG mode straight towards the current waypoint, as you pass the point where you should have started the turn, you'll see the CDI start to deflect in the direction of the turn as you deviate off the computed flight path, but the nav data will remain on the current point until you pass it.

 

[gismo]

Exactly what it does without an EHSI. The CDI type has nothing to do with what the unit does internally regarding turn anticipation.

The flight plan cycles to the next waypoint as the current waypoint goes through the perpendicular to the track ("passes through the wingtip") during the turn. At that point, to give you a visual cue, the to/from flag is commanded to go briefly "from" and then "to" as the flight plan cycles.

However, the CDI positioning output remains independent of all that and is based on lateral deviation of the aircraft from the plotted course. In the turn, that means how far laterally you are off the computed arc from the computed roll-in point to the computed roll-out point. And it doesn't matter what sort of course deviation display you have, be it a traditional CDI, mechanical HSI, EHSI, whatever.

Now, if you have a bearing pointer display (like an RMI or BDHI or similar on your HSI), the bearing pointer will remain pointed at the old waypoint, swinging towards the outside of the turn, until the waypoint cycles to the next point as the pointer reaches the wingtip (approximately, depending on wind drift). At that point, the bearing pointer will swing into the turn, pointing at the new turnpoint, and will continue to swing towards the nose (again, approximately, depending on wind drift) as the turn continues until roll-out. You can see this in the movement of the green bearing chevron on the Nav1 (so-called "HSI") page of the 530.

Finally, if you have an autoslew function on your EHSI, the CDI presentation on the EHSI should rotate to the new DTK when the DTK cycles to the next course about the time the GPS commands the start of the turn. However, the CDI needle will, if you stay on the computed path during the turn, stay in the center through the whole turn.

Ron, that's not even close to what I've observed with a GNS430W or GNS480 and IIRC the KLN94 coupled to an autopilot via a mechanical or electronic HSI.

First of all I believe all three GPSs do provide some form of "turn anticipation" well prior to reaching the "perpendicuar to the track" you mentioned, at least to the pilot. And with most autopilots coupled to HSIs, the position of the course pointer as well as the CDI deflection influences the autopilot's actions so the behavior may indeed be affected by the type of heading and course display. IME, if you adjust the HSI course pointer to the next track when a 480 or 430 says to do so, your turn will begin as soon as you move that pointer. Depending on the autopilot and wind this turn may end up short or long for the new course but it will likely always put you a lot closer to the desired track than you'd be if you didn't start turning until station passage on the old course.

I know the 480's anticipatory course change will occur at a variable distance from the fix depending on both groundspeed and the amount of turn required. I believe that it's calculation assumes a standard rate turn which may or may not be feasible depending on the plane's TAS and autopilot. The 480 also claims to be able to compensate for wind if the appropriate information is supplied to it (My 480 does have wind data but still tends to overshoot turns in a tailwind so I'm not convinced they got that part right although this could be due to my autopilot's limited banking ability).

And what Peter's really concerned about is the point in time (relative to crossing/passing the fix) when the GPS sends the information about the new course to his EHSI because that's when the EHSI "autoslews" the course pointer causing the autopilot to initiate the turn. I don't have an EHSI so I can't say for certain but I do have a SAM which announces/displays new waypoints and can show the desired track as sent by the GPS serially. I haven't yet verified this but IIRC the new waypoint and track does show up on the SAM at approximately the same time as the GPS recommends beginning the turn with a text prompt.

That said, I've flown in airplanes with GNS430s tied to Sandel EHSIs with autoslew enabled and IME that combination does a noticeably inferior job on 90° corners than my GNS480 (and likely a GNS430) does when coupled via GPSS. For one thing the turn entries are much smoother with GPSS. I also believe the resulting over/undershoot is less and I've definitely observed that the acquisition of the new course and appropriate wind correction is achieve much more quickly with GPSS than via autoslew.

 

[John Collins]

I have confirmed everything Ron has stated on the the GNS530W simulator, great analysis!

The first indication does not generate any output other than to the status line that a turn is upcoming in 10 seconds, 9 seconds, ...

Next, the DTK changes about a second or two after message to turn is displayed.

As the turn progresses, until abeam the waypoint, the bearing pointer points to the up coming waypoint. As the midpoint, abeam, the waypoint sequences and so does the bearing pointer.

While in the turn, it will provide a CDI left/right indication if not on the depicted turn on the map.


With respect to the serial interface using aviation data, there are two message types repeated on a once per second basis. The DTK is part of message type 1 and should switch at the same time as DTK changes on the GNS530W screen. This message should be used for turn anticipation.

Message type 2 contains the flightplan waypoints and the active waypoint is a flag associated with each of the waypoints, only one of which should have the flag on at any time. This should change when the next waypoint is the Navigation fix which changes at the mid point of the turn, when abeam.

 

[poadeleted20]

Ron, that's not even close to what I've observed with a GNS430W or GNS480 and IIRC the KLN94 coupled to an autopilot via a mechanical or electronic HSI.

First of all I believe all three GPSs do provide some form of "turn anticipation" well prior to reaching the "perpendicuar to the track" you mentioned, at least to the pilot.

Yes, they do, and that's exactly what I said -- the GPS starts commanding the turn at the point at which a standard rate turn will result in a roll-out on the line from the current waypoint to the next waypoint. However, even though it's turning and knows the path to fly, it will not show waypoint passage or display the nav data to the next point until the waypoint goes past your wingtip in the turn.

And with most autopilots coupled to HSIs, the position of the course pointer as well as the CDI deflection influences the autopilot's actions so the behavior may indeed be affected by the type of heading and course display.

That depends on what sort of steering you're using. If you're using roll steering, none of that affects the input from the GPS to the autopilot. It will start the turn at the computed roll-in point, and make roll inputs to keep it on the computed arc path throughout the turn.

However, with an autoslewed EHSI, if you're using non-GPSS nav steering (which is driven primarily by the needle position and biased by the selected course), then it's a different situation. The desired course data in the HSI will change to the next course at the roll-in point and the CDI will rotate to that new course, and the autopilot will start the turn. However, the CDI needle will still reflect deviation off the computed flight path arc from the roll-in point to the roll-out point, so the steering will not be as accurate as it will be with GPSS roll steering. Most likely, it will try to make the turn too fast initially, and then slowly ease itself onto the outbound courseline.

IME, if you adjust the HSI course pointer to the next track when a 480 or 430 says to do so, your turn will begin as soon as you move that pointer.

If you have a purely manual HSI, and you're in non-GPSS nav tracking, you are correct. But again, with the CDI needle deviation being part of the steering command structure, and the needle being driven by deviation from the arcing path, it will not track the arc as accurately as GPSS roll steering. OTOH, if you fail to adjust the HSI course pointer in non-GPSS nav mode, the autopilot will not see much deviation on the CDI needle until you're significantly past the desired roll-in point, and it will be playing catchup ball thereafter -- usually with a damped s-turning across the path until a couple of miles past the turnpoint.

And what Peter's really concerned about is the point in time (relative to crossing/passing the fix) when the GPS sends the information about the new course to his EHSI because that's when the EHSI "autoslews" the course pointer causing the autopilot to initiate the turn.

That will depend on whether he's using GPSS roll steering or needle-based nav steering. If the former, the turn will start at the roll-in point, and it won't matter what he sees on the EHSI. If the latter, the autopilot will start the turn as soon as the CDI course pointer slews, just as if you had turned it manually, and that will happen when the DTK changes at the roll-in point. But, as always when it's using CDI needle deviation as an input, it won't track turns very well, and the greater the turn, the worse it gets.

That's why I teach folks without GPSS roll steering to go from NAV to HDG when the "turn to xxx in yy seconds" comes up, select the new course (DTK) on the CDI, and when the display says "turn now," use the heading bug to make the turn to the new heading. After the turn is complete, and the wings are level, then re-select the NAV mode -- the course should match the DTK, and the needle should be close enough to centered that it should pick up the track and damp out very quickly. With an HSI and an autopilot that senses selected course (i.e., not a Century I or S-Tec 20/30), you can bypass the HDG mode and simply rotate the HSI course selector to the new DTK when the GPS says "turn now." Again, it won't be as smooth as with GPSS roll steering, but it will work reasonably well.

That said, I've flown in airplanes with GNS430s tied to Sandel EHSIs with autoslew enabled and IME that combination does a noticeably inferior job on 90° corners than my GNS480 (and likely a GNS430) does when coupled via GPSS. For one thing the turn entries are much smoother with GPSS. I also believe the resulting over/undershoot is less and I've definitely observed that the acquisition of the new course and appropriate wind correction is achieve much more quickly with GPSS than via autoslew.

...and now you know why.

 

[peter-h]

Interesting discussion

I have some real auto-slew turn performance data on the KLN94 and the SN3500 here

For reasons of centre stack access I can't presently implement the only available KLN94-KFC225 roll steering which is the special analog one.

That's why I teach folks without GPSS roll steering to go from NAV to HDG when the "turn to xxx in yy seconds" comes up, select the new course (DTK) on the CDI, and when the display says "turn now," use the heading bug to make the turn to the new heading. After the turn is complete, and the wings are level, then re-select the NAV mode

Flying with a mechanical HSI, one soon discovers that one can get accurate turns by turning the CP earlier or later than the annunciation by the GPS. With the KLN94, the right time is invariably later. Or one can turn it very slowly...

What I didn't realise till I got the EHSI was just how early the KLN94 emits the new flight plan and thus requests the EHSI CP rotation. I never paid much attention to when the "waypoint coming up" annunciater popped-up, except that it was way too soon.

I think Sandel missed a business opportunity here. The EHSI knows, the track, the heading, the GS, etc and could easily act as a roll steering converter, controlling the turn using its CP rotation intelligently, instead of just flipping the CP around in about 1 second.

 

[gismo]

Yes, they do, and that's exactly what I said -- the GPS starts commanding the turn at the point at which a standard rate turn will result in a roll-out on the line from the current waypoint to the next waypoint. However, even though it's turning and knows the path to fly, it will not show waypoint passage or display the nav data to the next point until the waypoint goes past your wingtip in the turn.

OK I must have misinterpreted what you wrote, starting with:

Exactly what it does without an EHSI. The CDI type has nothing to do with what the unit does internally regarding turn anticipation.

I took that to mean that having an EHSI with autoslew would behave the same as any other course deviation display / heading indicator type.
And I took your last statements to mean you thought that the centered needle on the EHSI would keep the plane from turning when the autoslew switched the course pointer. In my defense I don't see where you described the autopilot's reaction to the change in course pointer heading:

The flight plan cycles to the next waypoint as the current waypoint goes through the perpendicular to the track ("passes through the wingtip") during the turn. At that point, to give you a visual cue, the to/from flag is commanded to go briefly "from" and then "to" as the flight plan cycles.

However, the CDI positioning output remains independent of all that and is based on lateral deviation of the aircraft from the plotted course. In the turn, that means how far laterally you are off the computed arc from the computed roll-in point to the computed roll-out point. And it doesn't matter what sort of course deviation display you have, be it a traditional CDI, mechanical HSI, EHSI, whatever.

Finally, if you have an autoslew function on your EHSI, the CDI presentation on the EHSI should rotate to the new DTK when the DTK cycles to the next course about the time the GPS commands the start of the turn. However, the CDI needle will, if you stay on the computed path during the turn, stay in the center through the whole turn.

What you're saying in this post is much clearer and consistent with what I've experienced and "know":

That depends on what sort of steering you're using. If you're using roll steering, none of that affects the input from the GPS to the autopilot. It will start the turn at the computed roll-in point, and make roll inputs to keep it on the computed arc path throughout the turn.

However, with an autoslewed EHSI, if you're using non-GPSS nav steering (which is driven primarily by the needle position and biased by the selected course), then it's a different situation. The desired course data in the HSI will change to the next course at the roll-in point and the CDI will rotate to that new course, and the autopilot will start the turn. However, the CDI needle will still reflect deviation off the computed flight path arc from the roll-in point to the roll-out point, so the steering will not be as accurate as it will be with GPSS roll steering. Most likely, it will try to make the turn too fast initially, and then slowly ease itself onto the outbound courseline.

If you have a purely manual HSI, and you're in non-GPSS nav tracking, you are correct. But again, with the CDI needle deviation being part of the steering command structure, and the needle being driven by deviation from the arcing path, it will not track the arc as accurately as GPSS roll steering. OTOH, if you fail to adjust the HSI course pointer in non-GPSS nav mode, the autopilot will not see much deviation on the CDI needle until you're significantly past the desired roll-in point, and it will be playing catchup ball thereafter -- usually with a damped s-turning across the path until a couple of miles past the turnpoint.

That will depend on whether he's using GPSS roll steering or needle-based nav steering. If the former, the turn will start at the roll-in point, and it won't matter what he sees on the EHSI. If the latter, the autopilot will start the turn as soon as the CDI course pointer slews, just as if you had turned it manually, and that will happen when the DTK changes at the roll-in point. But, as always when it's using CDI needle deviation as an input, it won't track turns very well, and the greater the turn, the worse it gets.

What you teach may result in a better corner but I'll bet this varies among autopilots as well. With the Century III I have a course error of 90° completely the CDI error (or lack therof) so at least initially turning the course pointer will have exactly the same effect as switching to heading mode with the heading bug set 90° off the current course, at least initially. But either way the cross track error at the end of the turn will vary considerably depending on wind and the rate at which the autopilot chooses to turn the plane. Also for some autopilots (including mine IIRC) switching to heading mode zeros out the wind correction integrator. This is probably a good thing on a sharp turn but less helpful when the turn is shallow enough that the wind remains on the same side of the plane. YRMV.

That's why I teach folks without GPSS roll steering to go from NAV to HDG when the "turn to xxx in yy seconds" comes up, select the new course (DTK) on the CDI, and when the display says "turn now," use the heading bug to make the turn to the new heading. After the turn is complete, and the wings are level, then re-select the NAV mode -- the course should match the DTK, and the needle should be close enough to centered that it should pick up the track and damp out very quickly. With an HSI and an autopilot that senses selected course (i.e., not a Century I or S-Tec 20/30), you can bypass the HDG mode and simply rotate the HSI course selector to the new DTK when the GPS says "turn now." Again, it won't be as smooth as with GPSS roll steering, but it will work reasonably well.

...and now you know why.

I think I already knew but thanks anyway.

 

[gismo]

I think Sandel missed a business opportunity here. The EHSI knows, the track, the heading, the GS, etc and could easily act as a roll steering converter, controlling the turn using its CP rotation intelligently, instead of just flipping the CP around in about 1 second.

Perhaps, perhaps not. DAC attempted to generate a roll steering signal from the 94's serial output (track, gs, cross track error, track angle error) but found that the 1Hz update rate was insufficient to produce acceptable results. OTOH, DAC didn't have aircraft heading as an input and that could help smooth things out.

 

[peter-h]

I wonder if anybody has looked at the PDF link I posted.

but found that the 1Hz update rate was insufficient

They could have easily solved that by fitting a curve to the points. 3 points is good for a 2nd order polynomial. 1Hz update rate is loads good enough.

It's all been done before because the same issues rise with e.g FLARM.

Actually I thought it was 0.5Hz on the KLN94. The old GNS boxes were 1Hz. The W boxes are 5Hz.

Back to the EHSI, I don't think the lateral deviation (as indicated by the deviation bar of the EHSI) has any real effect on the turn performance because nearly the whole turn is done at a rate limited by the autopilot. The CP flips around in about 1 second (for a 90-deg turn) and the airplane banks to the AP-limited roll angle, and sits there. The lateral deviation comes into it only when the turn is more or less finished and the system is trying to minimise the off-track error.

So it is all down to timing and how well one determines the turn commencement point. What the system does after that, for another 30 secs or so (for a 90-deg turn), hardly matters.

In the KLN94 installer config there is a curious config for the autopilot roll angle when using the analog roll steering feature. I wonder why they need that?

I just wish the box had a config for reducing the silly flight plan change advance from 35 secs to 20 secs. That would give you good turns most of the time and probably as good as roll steering will give you unless it takes actual wind into account, which none of the roll steering converters will do that.

If the KLN94 was a major product today, with EHSI products, there would be an obvious market for a converter box, RS232 to RS232, which did turn optimisation and worked wholly on the RS232 flight plan data. It would be easy to do because the new leg's data comes out at WP-35 secs which is always going to be early enough (unless you have 150kt of tailwind ).

 

[poadeleted20]

I think Sandel missed a business opportunity here. The EHSI knows, the track, the heading, the GS, etc and could easily act as a roll steering converter, controlling the turn using its CP rotation intelligently, instead of just flipping the CP around in about 1 second.

This may be why Aspen included roll steering in their unit -- and Aspen PFD's seem to be selling a lot better than Sandel EHSI's right now.

In any event, this whole discussion points out the problems inherent in training folks on the various GPS/display/autopilot combinations -- there is a myriad of combinations, and no two combinations work alike.

 

[John Collins]

In the KLN94 installer config there is a curious config for the autopilot roll angle when using the analog roll steering feature. I wonder why they need that?

I assume it is to adjust the sensitivity to match the autopilot for a particular installation.

I just wish the box had a config for reducing the silly flight plan change advance from 35 secs to 20 secs. That would give you good turns most of the time and probably as good as roll steering will give you unless it takes actual wind into account, which none of the roll steering converters will do that.

I have attached a page from the KLN94 Pilot Guide on turn anticipation. According to it, turn anticipation is preceded by Waypoint Alert (flashing of Waypoint annunciator) by approximately 20 seconds. When the flashing stops is when the DTK changes and turn anticipation begins. I talked to Sandel technical support and the fellow said that the CP moves when DTK changes. DTK changes coincident with the beginning of turn anticipation and the flightplan doesn't advance to the next waypoint until later when the aircraft is well into the turn and abeam the current waypoint. In other words the "silly flight plan change" occurs later than turn anticipation.

I question whether your software is at a current revision. Let me know the specific software version and I will contact HBK technical support on Monday and inquire if there are any known problems. The application software configuration number is available on the top line of the Auxiliary (AUX) 14 page. It reads “SW REVISION” followed by a four digit configuration number. Also the ORS level is displayed on startup.

The roll steering converters don't take actual wind into account, as they are mere converters of the bank angle that the GPS outputs. However, if the GNS530W has air data and heading input, it will adjust the bank angle to account for the wind. This can be seen particularly on holding patterns that will change their depicted shape from an oval.

If the KLN94 was a major product today, with EHSI products, there would be an obvious market for a converter box, RS232 to RS232, which did turn optimisation and worked wholly on the RS232 flight plan data. It would be easy to do because the new leg's data comes out at WP-35 secs which is always going to be early enough (unless you have 150kt of tailwind ).

You are making an assumption that this is not a software issue. It is a simple matter to tap into the RS232 data stream and record it using a PC. I don't think you will find that the new active waypoint data will change until well after the DTK changes. You can also watch the data change sequence on the KLN94. Note when the DTK field changes and note when the active waypoint changes. Do you have a video showing the KLN94 screen as the turn anticipation happens?

 

[peter-h]

My software is ORS3. This was the last one they did as far as the pilot guide is concerned.

I think the user guide description is what actually happens. At something like wp-50s the wp warning appears. Then a bit later (wp-35s) the new rs232 data is emitted, the 3500 cp flips around and the aircraft of course immediately turns at rate 1. With the old HSI, this is also the point where the kln94 reminds you to turn the course pointer to the new track.

Let me do a movie of it... should be done on Tuesday.

The rs232 data is indeed easy to capture especially as I have it available on a connector, and it may contribute to knowledge (though I doubt it since it is obvious when the new data reaches the 3500) but it won't deliver a solution because nothing in this department is configurable.

However I did find something curious. The kln94 I fly with shows "application software mod status 01/02" and this I believe is what initially did the roll steering. A while ago I bought another kln94 (from a US pilot who installed a G500) and this shows mod status 01/05.

The Jan 2003 IM (rev 4) describes 01/01 and 01/02 and says 01/02 does the roll steering.

Google hits on 069-01034-0102, not on -0105 but it does one hit on -0104.

I don't see Honeywell having documented any of these changes.

If any of them relate to the rs232 data timing then I will try to get these GPSs upgraded, but it would be great to see some confirmation of it.

I can of course try the -0105 unit and see what it does...

 

[SCCutler]

I am probably way over my head here, but when I read the posted manual page, it appears to me that the BK folks were providing a measure of home-boy GPSS here - as I read it, if you have your AP set to NAV tracking, the CDI or HSI deviation provided by the KLN94 will direct the autopilot to turn accordingly; this, as opposed to utilizing the 3500's autoslew with the autopilot in heading mode.

Please, shoot me down here, just trying to learn.

 

[peter-h]

That doesn't work.

If you have a mechanical HSI, and you don't turn the course pointer, what happens is that the deviation bar moves out and the aircraft will gradually turn a little bit but it never really does it because the autopilot is still flying a heading determined by the CP.

 

[SCCutler]

That doesn't work.

If you have a mechanical HSI, and you don't turn the course pointer, what happens is that the deviation bar moves out and the aircraft will gradually turn a little bit but it never really does it because the autopilot is still flying a heading determined by the CP.

Hmmm.

I guess this is behavior endemic to the BK autopilot? My S-Tec will follow the CDI without regard to the course pointer, if I have it in Nav tracker mode.

Sorry to digress - carry on.

 

[gismo]

Hmmm.

I guess this is behavior endemic to the BK autopilot? My S-Tec will follow the CDI without regard to the course pointer, if I have it in Nav tracker mode.

Sorry to digress - carry on.

Assuming you're talking about a 20/30/40/50, your S-Tec can't do intercepts and AFaIK any autopilot that can will use a combination of heading error and nav deviation to control bank.

 

[denverpilot]

This conversation taught me more about how GPS in modern cockpits fly autopilots than the giant number of repeated FAA Instrument test pool questions that state "VOR passage is when the first full-reversal of the TO/FROM flag takes place."

It also shows how utterly behind the times the FAA question pool is.

Anyone unlucky enough to be flying VOR to VOR these days (me) without an IFR certified GPS (me) and no DME (not me) out here in the Wide Open West with only one VOR receiver (not me) or low enough to not have a suitable crossing Navaid to the route (not often me) is also laughing their butts off at the concept of leading a turn prior to the fix.

Wonder if over time a controller seeing an aircraft actually cross directly over a fix and THEN turn will start freaking the new generation of ATC folks out who are used to watching over you guys with your fancy computer-derived arcs that turn prior to the fixes?

 

[poadeleted20]

Assuming you're talking about a 20/30/40/50, your S-Tec can't do intercepts and AFaIK any autopilot that can will use a combination of heading error and nav deviation to control bank.

But not the ones you mentioned. They are entirely rate-based with no input from the heading indicator other than the bug when in HDG mode. In NAV, it's pure needle position and movement rate which is why they don't do intercepts and are lousy at tracking turns at navaids (other than with GPSS, when the CDI isn't even in the loop).

 

[poadeleted20]

Anyone unlucky enough to be flying VOR to VOR these days (me) without an IFR certified GPS (me) and no DME (not me) out here in the Wide Open West with only one VOR receiver (not me) or low enough to not have a suitable crossing Navaid to the route (not often me) is also laughing their butts off at the concept of leading a turn prior to the fix.

Since you seem to have DME, I thought you might want to know that the Navy taught us how to use TAS, DME, and altitude (AGL) to compute DME distance to start turns so we'd roll out on course on the other side of the TACAN station. I haven't used that in decades, so I'd have to work the formula again, but it wasn't that hard.

 

[John Collins]

Assuming you're talking about a 20/30/40/50, your S-Tec can't do intercepts and AFaIK any autopilot that can will use a combination of heading error and nav deviation to control bank.

As you noted, the Stec 20/30/40/50 autopilots can't intercept a course, but can track the course. I have used a Stec 50 to fly a DME arc with reasonable success, as it continues to try to catch the CDI deviation that the GNS430 puts out. I am not sure how much of a radius of curve it has the capability of following, but it will follow it to some degree, even if it was not designed to do so.

 

[Everskyward]

Since you seem to have DME, I thought you might want to know that the Navy taught us how to use TAS, DME, and altitude (AGL) to compute DME distance to start turns so we'd roll out on course on the other side of the TACAN station. I haven't used that in decades, so I'd have to work the formula again, but it wasn't that hard.

I know that there is a more complicated formula but I have used the rule of thumb that for a 90 degree turn it's GS/200. For example, if your GS is 100 kts. you start the turn .5 mile early.

 

[poadeleted20]

I know that there is a more complicated formula but I have used the rule of thumb that for a 90 degree turn it's GS/200. For example, if your GS is 100 kts. you start the turn .5 mile early.

That may work for horizontal distance off a GPS, but since DME is a slant-range device, with DME you have to account for altitude, too, e.g., at 6000 feet, you never see less than 1nm, and then only as you pass overhead.

 

[Everskyward]

That may work for horizontal distance, DME being a straight-line distance has to account for altitude, too, e.g., at 6000 feet, you never see less than 1nm, and then only as you pass overhead.

That's true, I was speaking about horizontal distance.

 

[peter-h]

The KLN94 with 01/02 firmware was replaced with a freshly updated unit with 01/05 firmware (actually 01/05 is dated 2001 so it is 10 years old) and the turn behaviour is exactly the same as before.

I did a video which shows the turn. My filespace does not allow files above 50MB so I could not do an HD one.

You can see the KLN94 'waypoint' annunciator flashing at 0:14. The EHSI CP flips around at 0:33.

This time there was headwind and the ground speed was about 120kt so the turn ended up about 0.5nm short of the next leg. Previously there was tailwind and the GS was about 160kt which reduced the shortfall to about 0.25nm.

The DME shows the correct DME distance to the waypoint (SFD) and so does the GPS although the GPS is much harder to read. The KI-204 CDI and the KI-229 RMI don't show anything relevant.

Clearly this stuff was coded in the KLN94 without much thinking.

 

[John Collins]

Peter,

I was unable to view the video. I am using IE 8 and Media Player 11, latest versions.

 

[SCCutler]

Assuming you're talking about a 20/30/40/50, your S-Tec can't do intercepts and AFaIK any autopilot that can will use a combination of heading error and nav deviation to control bank.

I know the referenced S-Tec models do not do intercepts; but...

The discussion of CDI behavior in the KLN94 manual page posted above appeared to be consistent with synthesizing an intercept. The documentation describes a gradual transition of CDI *prior* to crossing the intersection of the old course and the new course. I believe that, based upon this description, if I had my S-Tec 30 engaged in Nav-tracking mode, high-sensitivity, and the CDI did that, the airplane would tend to turn in that direction, and would approximate the turn I wanted.

 

[peter-h]

The best multi-format media player is VLC.

However I uploaded a WMV file here. It is 45MB. This shows it better than the previous one.

 

[gismo]

But not the ones you mentioned. They are entirely rate-based with no input from the heading indicator other than the bug when in HDG mode. In NAV, it's pure needle position and movement rate which is why they don't do intercepts and are lousy at tracking turns at navaids (other than with GPSS, when the CDI isn't even in the loop).

I think we're on the same page here. I was responding to Spike's "My S-Tec will follow the CDI without regard to the course pointer" and pointing out that while those autopilots don't use course or heading information from the HSI or DG when tracking a nav signal, most other autopilots do use the additional info and as you say therefore are affected by the setting of the heading or course pointer. And further that the reason that the more sophisticated autopilots use heading/course input when tracking is (among other things) so they can perform intercepts without overshooting then wandering back and forth across the desired course for a while.