Yesterday I was milling along on an IFR flight at 5000 ft altitude (on local QNH 1028) and ATC gave me a ‘climb flight level 60’.
I armed 6000 in the KAS297B, entered a climb rate of 500 feet/minute, pressed engage and the Bonnie started to climb.
I then remembered to change the altimeter to standard QNH 1013 and did just that. The difference in feet was (1028-1013) = 15 × 30ft = 450 feet.
Upon which the autopilot suddenly started to increase the pitch. Which, in IMC, doesn’t seem to be good idea.
I disconnected the AP which was now out of trim, retrimmed, and reengaged the AP.
My question here is: has anyone ever encountered that behaviour ?
The system is: Bendix King KFC150 autopilot, KAS 297B altitude preselect, KEA 130A encoding altimeter.
The guy who ought to know is @wigglyamp.
What I know about a similar system, KFC225 fed from a KEA130A, is this:
The autopilot gets the gray code pressure altitude from the KEA130A.
During the climb or descent, the target altitude from the preset box is compared with the value from the KEA130A, and upon a match (slightly before a match actually) the autopilot switches to an internal 16-bit-encoded analog barometer for holding the altitude. They use this internal baro because it has a high resolution of approx 1ft, which is what you want for the altitude hold control loop, whereas the KEA130A has high accuracy and stability but lousy resolution (25ft; useless for altitude hold). There remains a very slow background comparison against the KEA130A, to take out any drift in the internal barometer, but picks up only when any drift exceeds the 25ft**.
So that’s you fly a flight level.
How does flying an altitude work?
The KEA130A’s subscale knob is attached to a multiturn pot whose output is connected to the autopilot, and it tells it how much to shift the target preselect value. So if e.g. QNH=1023 and you have set a 5000ft target, the autopilot will go for a match at a pressure altitude of 4700ft or 5300ft instead (don’t ask me which one; this is not the Met exam and I am writing up the Elba trip ) and the plane will level off at 5000ft on the QNH.
Now, what happens if you are flying on 1028 and change the subscale to 1013?
What happens in detail in the software I don’t know but very clearly the autopilot notices that the preselect target has changed, most likely via the slow background process **above. It will thus order a climb or a descent – exactly as if you entered a downdraught in which case it has to pitch up to hold altitude.
So unless I misunderstood you, it did exactly what it should do.
Well, except it pitched the airplane up to a climb rate of more than 1000fpm, airspeed dropped rapidly. That was more than the re-engaging of the original climb rate I had dialed into the KAS297B, which was 500 fpm.
OK; that’s not right. The vertical speed resulting from just turning the KEA130A subscale is something like 50fpm.
I struggle to think what failure mode could cause this.
Does the autopilot work fine in all the other modes?
yes, everything ok. I just had the KAS297B installed…..
Changing the altimeter setting made the autopilot suddenly see a lower altitude, it thus thought that the VS was too low and raised the nose to increase the VS back to the intended value?
Yes that seems to be the case. It seems to be normal.
I consider it dangerous. Given the reciprocal situation – I descend through transition level, dial in from standard to local QNH, which is higher. KAS 297B thinks I‘m higher than I am, thus thinks it needs to increase descent rate, and goes into a nose dive. Nice….
It is not normal to do it so aggressively.
As If found out, there actually is an Allied Signal Service Bulletin, which at the time was customer option only, which resolves the issue by replacing an EPROM.
Can I upload it here somehow ?