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

[gismo]

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?

I know the reverse has happened out east where ATC was bugging pilots using GPS to fly victor airways and not following the wildly scalloped path the controllers were used to seeing when pilots accurately followed the curved VOR radial.

As to the freaking out part, I suspect that when a new generation controller handles your VOR only airplane the first thing he'll freak out about is your refusal to fly direct to some distant fix you can't track to legally. But when he sees you waiting to start your turn after crossing a VOR he'll probably assume (incorrectly) that you're just a lousy and/or inattentive pilot.

 

[peter-h]

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 don't see any evidence of that in my video (the URL is in my last post above).

The EHSI course pointer just flips around when it sees the new RS232 data.

However, an interesting question is what happens with a HSI without auto-slew, like a mechanical KI-525 type, or the Sandel with auto-slew disabled. The KLN94 owners guide would have been written for the KI-525. I have been flying with the KI-525 since 2002 and I am pretty damn sure the CDI doesn't move until far too late, so no turning takes place. You have to rotate the CP to the new track.

 

[SCCutler]

I don't see any evidence of that in my video (the URL is in my last post above).

The EHSI course pointer just flips around when it sees the new RS232 data.

However, an interesting question is what happens with a HSI without auto-slew, like a mechanical KI-525 type, or the Sandel with auto-slew disabled. The KLN94 owners guide would have been written for the KI-525. I have been flying with the KI-525 since 2002 and I am pretty damn sure the CDI doesn't move until far too late, so no turning takes place. You have to rotate the CP to the new track.

I was definitely referencing a non-EHSI setup (HSI or regular Nav indicator) in my hypothet.

If the CDI deviation starts very late, obviously, my idea would be wronger than wrong.

Carry on.

 

[peter-h]

My video shows no movement of the CDI bar prior to the EHSI flipping around (the last video was done at 1nm FS on the GPS). I will try it sometime with auto-slew disabled to see what the deviation is, but I am tell you that the KFC225 won't turn anywhere near the new track until the CP is turned around. The adjustment is just too small.

I suppose it is possible that my autopilot lateral sensitivity has been set up incorrectly (too low) but it tracks very accurately when in flight.

 

[John Collins]

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.

I viewed the video many times and offer the following observations:

The turn anticipation warning as noted by the flashing WPT on the KLN94 is at :14. The KLN94 DTK changes at :33 and the CP responds at the same time by moving to the new DTK value. At the same time, the KLN94 WPT goes solid indicating that turn anticipation should commence. The aircraft rolls into a left turn of approximately 25 degrees of bank. At :55, the new waypoint becomes the active waypoint, from what I could see on the moving map, this appears to be when the aircraft is abeam the original waypoint.

From what I observe, the KLN94 is doing its duty as intended. What you are witnessing is the fact that auto slew is not roll steering. In roll steering the angle of bank is controlled and should roughly produce a standard rate turn. Assuming an airspeed of 140 Kts, I would estimate this to be 19 degrees of bank. But the autopilot does not fly a standard rate turn, but instead uses 25 degrees of bank as its only bank limit is from the FD and appears to be a maximum of 25 degrees. A single bank angle is fixed into the autopilot and doesn't get adjusted for differences in winds or cross-track error in the turn. Roll steering would make these subtle changes in the bank angle to compensate. As you note, flying the route with a tailwind gets you closer to the course at the completion of the turn. This is due to the higher groundspeed requiring a higher bank angle to complete the turn so that the aircraft rolls out on the new course. But as stated before, a single bank angle is used and there is no compensation for these differences. In control theory it is an open loop system with no feedback verses with roll steering the bank angle is continuously computed based on the actual track and groundspeed to achieve the desired flight path. Based on the groundspeed, there will be one that will work out just perfectly, but it is going to require a good tailwind.

 

[peter-h]

Yes, I agree, John.

With about 100kt of tailwind it would be spot on

But the key thing in that video is there is no deviation on the deviation bar prior to the CP turning around.

One small detail... I don't think the AP-limited bank angle is quite 25 degrees, or at least should not be. It is about 21/22 degrees. This is very close to Rate 1 at 140kt, and the TC agrees. I have spent quite a bit of time on this; there is an offset null adjustment on the KC225 front panel with which one can balance this up, and currently it is doing 21/22 degrees when settled in the turn.

The TB20 KFC225 STC requires the bank angle to be set at 22 degrees, IIRC.

 

[peter-h]

I've just done some long flights with the Sandel and there is more to this... The wp-35s behaviour is not consistent.

For small angle changes, say under 10 degrees, it is something like wp-5s. And such turns happen precisely.

For larger angles, the time advance rapidly increases and the system then always undershoots.

I had some replies from Honeywell:

The RS 232 is not intended for Flight Control. It is mainly intended for Moving Map Displays and Fuel Flow. The KLN 94 should transition to the next active waypoint halfway through the Turn Anticipation Arc.

This begs the question: what would a KI-825 do? It must do the same as the Sandel SN3500, because it is fed from the same RS232 data.

Presumably the intention is that a separate roll steering channel is installed when an EHSI is installed.

 

[SCCutler]

What you describe seems inconsistent with BK's own description from the user manual.

---

I always enjoy your threads, Peter - excellent analytical approach to problem solving. Makes me miss my toolbag and oscilloscope days.

 

[John Collins]

I've just done some long flights with the Sandel and there is more to this... The wp-35s behaviour is not consistent.

For small angle changes, say under 10 degrees, it is something like wp-5s. And such turns happen precisely.

For larger angles, the time advance rapidly increases and the system then always undershoots.

I had some replies from Honeywell:

The RS 232 is not intended for Flight Control. It is mainly intended for Moving Map Displays and Fuel Flow. The KLN 94 should transition to the next active waypoint halfway through the Turn Anticipation Arc.

This begs the question: what would a KI-825 do? It must do the same as the Sandel SN3500, because it is fed from the same RS232 data.

Presumably the intention is that a separate roll steering channel is installed when an EHSI is installed.

Auto Slew is there to eliminate the task of manually selecting the next course. It is not intended to provide a roll steering function. Auto slew is synchronized with the KLN94 changing the DTK output which is coincident with the turn anticipation display to begin the turn and is based nominally on a standard rate turn. The time that this occurs should not be fixed, as you have noted. A small turn angle should have less time required to make the turn and therefore turn anticipation should be closer in time to the mid point of the turn when the aircraft is abeam the fly by waypoint. Since the actual turn does not have any feedback and the rate of turn is dependent on the autopilot and the heading or course error, smaller turns will give better results. If you want roll steering performance in the turn, hook it up.

 

[peter-h]

Makes me miss my toolbag and oscilloscope days.

I miss those days too. Occassionally I get to do some fun analog scope work (got a 400MHz Tek one) and I am trying to get back into software (asm+C) after a break of about 7 years. Avionics is good fun; there is a fair bit to learn about the conventions but at the same time most of it is banally trivial as "electronics" goes.

What I find especially funny is looking at schematics of e.g. a KFC225, seeing the designer used say a 1.1k resistor when a 1k would obviously work just as well. Nobody working for himself would do that

Auto Slew is there to eliminate the task of manually selecting the next course. It is not intended to provide a roll steering function.

I don't think one can categorically state that, John, unless you know the KLN94 designer personally and he told you that.

I don't know how long the KI-825 has been around but probably it was around when the KLN94 came out in 2000 or so. This issue would have been obvious to anybody flying with the KI-825. They would quickly notice that all big turns happen much too early. And the KLN94 analog roll steer output works only with the KFC225 which was a huge failure in the market, due to silly pricing and an appalling failure rate.

If I was designing the KLN94 I would have arranged the flight plan data change to come out "right" for larger turns. It is clearly not rocket science.

On small turns it doesn't matter. I've just done 6 hrs behind this kit and all enroute turns are just perfect, but they would be perfect even without any turn anticipation.

 

[John Collins]

I don't think one can categorically state that, John, unless you know the KLN94 designer personally and he told you that.

I don't know how long the KI-825 has been around but probably it was around when the KLN94 came out in 2000 or so. This issue would have been obvious to anybody flying with the KI-825. They would quickly notice that all big turns happen much too early. And the KLN94 analog roll steer output works only with the KFC225 which was a huge failure in the market, due to silly pricing and an appalling failure rate.

If I was designing the KLN94 I would have arranged the flight plan data change to come out "right" for larger turns. It is clearly not rocket science.

On small turns it doesn't matter. I've just done 6 hrs behind this kit and all enroute turns are just perfect, but they would be perfect even without any turn anticipation.

I don't know the KLN94 designer, nor have I talked to him. The KI825 is a 2002 design, the KLN94 is a 2000 design. The KLN94 Pilot Guide is silent on DTK being used for anything other than to note that it changes in a synchronized fashion with the commencement of turn anticipation. The KI825 supports automatically setting the course pointer to the DTK when GPS is the source. I don't know why you assume that the timing of DTK output will match the timing required to complete a course change when there are so many variables, there is no system control or feedback, the KLN94 has no configuration that describes the autopilot or HSI or EHSI, etc. It simply changes the DTK based on the assumption that the autopilot or person will turn the aircraft using a standard rate turn and has no knowledge of the actual turn rate of the autopilot. Once the DTK is changed, there is no more output from the KLN94 to the EHSI that has any effect on the turn rate.

Auto slew is not roll steering and never will be. It will at best work under some conditions to look somewhat like the effect of a roll steering. Roll steering actually controls the autopilot bank angle and through it the turn rate. In the case of the KFC225 and most Bendix King autopilots the bank angle is fixed at the maximum bank angle, somewhere between 20 and 30 degrees. Without roll steering, the bank angle is fixed in the established turn. This will result in a turn rate above standard rate for lower airspeeds and lower for turn rates at higher airspeeds.

I think it is rocket science in that you can't describe a control system that will work for all situations with a fixed bank angle that the GPS doesn't know, a single output that triggers the turn (DTK changes), the effects of wind can not be factored into the bank angle, and deviation from the desired course can't affect the bank angle. The answer is if you want this capability hook up the 3 or 4 wires.

 

[peter-h]

All I was getting at is that Honeywell's own KI-825 would have quickly revealed that the KLN94 undershoots turns by nearly a factor of two.

It's not a case of being unable to guess the wind etc, which is totally true but which does not make such a huge difference.

The KLN94 with the analog roll steer output is the 069-01034-0102 version and came out, according to Honeywell, late 2002. Also the roll steer output is not usable with any autopilot except the KFC225, so would not have been an option in many/most cases.

This is all moot to me anyway since both KLN94 units I have are -0101 and probably not upgradeable without spending way too much money.

There is incidentally something which confused me and, looking at google, other people: the "boot firmware" (whatever that means; the latest internal software is "ORS3") can be 01/01, 01/02, and one unit I have is 01/05. But even the last unit has a P/N of 069-01034-0101 The similarity between the last 4 digits of the P/N and the "boot software" are apparently coincidental. I have a description of the 01/05 "boot software" and it has some map improvements and little else.

So the roll steering will need to wait till I rip this stuff out and fit a GNS430W, but that will be a big job so I am not in a hurry.

 

[gismo]

All I was getting at is that Honeywell's own KI-825 would have quickly revealed that the KLN94 undershoots turns by nearly a factor of two.

I think those two products (EHSI and GPS) were developed for entirely different markets (i.e. turboprops vs the kind of planes you and I fly) and the concept of "autoslew" wasn't on the minds of KLN94 development team. AFaIK, the KLN94 was among the first GA products providing announced turn anticipation so the implementation probably didn't have much chance of being done to perfection.

It's not a case of being unable to guess the wind etc, which is totally true but which does not make such a huge difference.

Without an external ADC, the GNS430W can't anticipate the effects of wind either. It can and does compensate somewhat for groundspeed by estimating a greater turn radius for higher speeds but AFaIK it assumes that GS==TAS so it's not going to start the turn at the correct point if there's much wind.

The KLN94 with the analog roll steer output is the 069-01034-0102 version and came out, according to Honeywell, late 2002. Also the roll steer output is not usable with any autopilot except the KFC225, so would not have been an option in many/most cases.

This is all moot to me anyway since both KLN94 units I have are -0101 and probably not upgradeable without spending way too much money.

I suspect that the ORS upgrade isn't all that expensive unless hardware changes are required but as you say it's a moot point without a KFC225.

There is incidentally something which confused me and, looking at google, other people: the "boot firmware" (whatever that means; the latest internal software is "ORS3") can be 01/01, 01/02, and one unit I have is 01/05. But even the last unit has a P/N of 069-01034-0101 The similarity between the last 4 digits of the P/N and the "boot software" are apparently coincidental. I have a description of the 01/05 "boot software" and it has some map improvements and little else.

It's common to separate the small bit of software used to load and execute the "operating software" (e.g. ORS2) from the operating software which actually performs a device's intended functionality and that small software piece would be the "boot software". Often the "boot software" also performs the startup diagnostic tests (power on self test), facilitates the updating of the operating software, and provides some of the underlying hardware support (e.g. image drawing routines) in addition to starting the operating software. At one time it was also typical to use a different storage medium (e.g. ROM vs Flash) for the boot software and today the boot software is likely to reside in s separate and protected block of the flash memory. Updating the boot memory often carries a risk of completely disabling a device to the point where development tools are required to resurrect it.

 

[peter-h]

I've put the 01/05 firmware changes list here.

The PDF blocks copy/paste so I couldn't quote selectively, without hacks.

Oddly enough, the KLN94 is one of the very few bits of Honeywell avionics for which I have never seen a maintenance manual (schematics etc) but I suspect the -0102 variant has extra D-A channels so would be a PCB swap. I have the KLN90B MM but that seems a very different product to draw conclusions.

The Sandel SN3500 (LED version) seems to have loads of A-D and D-A pins, which also double up as RS232/ARINC serial I/O, and can be "woken" up and do things, just by software upgrades. They implemented the external display brightness control in that way, very recently.

 

[peter-h]

Just found out:

1) The KLN94 can be upgraded from -0101 to -0102 with just software.

2) Production stopped Jan 2011. No plans for a replacement.