Menu Sign In Contact FAQ
Banner
Welcome to our forums

KAP 140: can be dangerous when it starts a turn while climbing

Hello there,

Just returning from an IFR flight, where I had the occasion to confirm a poor behavior of my KAP 140 autopilot, which I don’t think is faulty, rather not perfectly engineered.

As you may know, it’s a rate-of-turn based autopilot, which all in all performs nicely.
When climbing or descending, you can select a “rate”, which is pretty constant (so your speed will move if there is any up or downdraught).
I also learned that the true rate of climb was 50 to 100ft/min higher than selected, that’s why I prefer to set a rate a little lower than what I actually want, but that’s not big deal.

Now, imagine you are cruise climbing at 700ft/min, 95 IAS, 7° pitch up. At one moment, you select a new heading (or the AP is following a new track after passing a waypoint).
What I observed is that, when initiating the turn, the AP is going to pitch up the aircraft to maybe 10° to reach a RoC of 1000ft/min, and the speed bleeding obviously. If you let it do, after a few seconds, while still turning, the rate of climb (and pitch) will normalize down to what you had asked initially. And the speed rising again.

So the point is that, this AP (or maybe my AP), is not able to accurately keep the RoC during the turn. I didn’t observe the same thing during the descent.

I wondered if you have had such experiences, ideas about what is going on, and (why not !) solutions to improve this behavior.

Happy flying !

Last Edited by PetitCessnaVoyageur at 26 Feb 14:50

PetitCessnaVoyageur wrote:

I wondered if you had such experience, …

Not yet. But I can’t recall to have come across this type of autopilot yet.

PetitCessnaVoyageur wrote:

… ideas about what is going,

My idea would be this: The autopilot – like the human pilot – is programmed to pull on the yoke during a turn. In your case, the amount of “pull” required to compensate for the reduced lift and g-force in a turn seems to be set too high. Whilst in the turn, the normal control law for maintaining constant rate of climb takes over and stabilises things again. The KAP 140 being a digital autopilot, it should be possible to adjust those settings in some way through a digital interface.

EDDS - Stuttgart

PetitCessnaVoyageur wrote:

What I observed is that, when initiating the turn, the AP is going to pitch up the aircraft to maybe 10° to reach a RoC of 1000ft/min, and the speed bleeding obviously. If you let it do, after a few seconds, while still turning, the rate of climb (and pitch) will normalize down to what you had asked initially. And the speed rising again.

I’ve used the KAP140 a lot and have not observed this behaviour. But maybe I haven’t been paying attention if it only lasts a few seconds. I’ll try it next time…

ESKC (Uppsala/Sundbro), Sweden

In my experience the KAP140 can be dangerous in other ways too. Here is what I wrote 4 years ago:

Kamikaze initiated by KAP140
This happened to me on a flight with a KAP140-equipped DA40. As a flight instructor, I had a student converting to the DA40, and he was new to the KAP140. When trying to select lower altitude from FL90 on the KAP140, he must have hit the BARO and turned the QNH setting way off. He then managed to set correct target altitude, toggled the VS key, which made the A/P enter VS mode, and selected 500 fpm down. The A/P obeyed, IAS was around 125 K and all was well.

Then my student checked the BARO setting. It was around 1070, and QNH was around 995, a huge difference. He turned the knob (a lot), setting 995 as the proper BARO setting for the descent. Then I noticed the trim wheel moving steadily forward, the nose lowered and in very few seconds the airspeed approached Vne. I pressed the copilot’s A/P disconnect botton, steadied the aeroplane and retrimmed, then re-engaged the autopilot, which now worked fine again. We were IMC, and perhaps due to c/s propeller, light turbulence and a generally low noise level the student had not noticed the dive in time to react.

On the next flight with the same aeroplane I managed to re-create the upsetting KAP140 behaviour. During descent on A/P in VS mode, I changed the BARO setting. Moderate (realistic) changes produced no reaction, but after turning to an absurdly high setting like 1070 and then turning down to the actual setting of around 1000, the same thing happened – a dramatic dive towards redline.

Apparently there is no warning in the manual against changing BARO setting during descent – which would indeed be a nuisance, especially IFR, since descend clearences to a low flight level are often extended, while descending, to an altitude below transition altitude, requiring the pilot to set QNH.

The autopilot works flawlessly in every other respect as far as I can tell. I would like to hear from someone with similar experience or perhaps an explanation.

huv
EKRK, Denmark

huv wrote:

Apparently there is no warning in the manual against changing BARO setting during descent

Actually, changing the baro setting during descent is the only way to do it on the KAP140 and that is the way I have done it too. If you do it with ALT mode engaged (as opposed to altitude capture) it will indeed try to maintain the set altitude.

I never had the problem you described, but then I never entered such big baro changes.

LFPT, LFPN

what_next wrote:

In your case, the amount of “pull” required to compensate for the reduced lift and g-force in a turn seems to be set too high. Whilst in the turn, the normal control law for maintaining constant rate of climb takes over and stabilises things again

W_N: I do understand your interpretation, which seems to correlate pretty well with the phenomenon.

What I could add is that in ALT mode (level), turn are perfectly done, and the altitude won’t move from a foot. But, you will answer that ALT and VS mode are not the same :-)

I don’t know exacty how the KAP 140 works. For what I know, it has its own static pressure port. I don’t think it is the origin of the problem, as rate (in VS mode) are accurate (except climb), and altitude capture exactly match PFD and back-up altimeter.
Does the KAP 140 uses a G-load sensor ? In that case, this one could have a problem ?

Anyway, I need to perform tests with different climb rates (and thus speed), and different aircraft loading, to provide better informations. I keep you informed :-)

I wonder how behave other King AP, including attitude based, such as @Peter KFC225, when climbing (VS mode) and turning ?

(I wish I had a GFC700 set up with its wonderful capability to maintain airspeed, in a clean way for what I’ve seen.)

@huv
I never experienced the kamikaze mode

I am not actually sure whether these autopilots have any coupling between pitch and roll behaviour. FWIW, I have never seen any evidence of it.

The KFC225 has a VS mode and along with the other vertical pitch modes that involve baro data the feedback on this loop is is implemented with a combination of the baro input (which is measured with a 16-bit pressure sensor), an internal accelerometer, the pitch data (analog LVDT) from a KI256 vacuum AI, and the gray code altitude data from the KEA130A encoding altimeter. Apart from a stupid bug which fairly randomly adjusts the VS by 100fpm up or down, it works pretty well.

The roll feedback is implemented using the roll data from a KI256 (analog LVDT) only.

The KAP140 (IM is here) I know almost nothing about. I would need to dig out the MM and read it up. But the IM says it was designed in 1996, which is about the same era as the KFC225, and curiously it is specced with the same servos as the KFC225 (KS270C, KS271C, KS272C). In fact it seems really similar in many ways. It is probably digital.

Administrator
Shoreham EGKA, United Kingdom

Between 3:15 and 4:00, you can see a T182T (equipped the same), initiating a turn (with the AP), during the climb.



You can see that the climb rate is rising from 650 to 750 ft/min, then down.
I feel the same, but maybe a bit worse.

I can’t see it clearly (not familiar with the tapes) but a variation in VS, in the VS mode, from 650 to 750fpm, due to turning at say 25 degrees, is fairly normal.

VS is measured barometrically – no other way to do it – and that is quite a poor feedback input, with a lag which makes it hard to control accurately.

Administrator
Shoreham EGKA, United Kingdom

Yep, and the speed is decaying from 105 to 95 approximately.

I need to take a video of what happens with mine, and we’ll see the thing more clearly. Not sure it is only a static lag problem, it could however.
Of course, if you have the speed, that’s not a problem, but if you’re short…. Hmmmmm !

One thing we could hold as a lesson here: with VS mode, and limited power (that’s NOT a JetProp !!), limit climb rate significantly.

36 Posts
Sign in to add your message

Back to Top