uAvionix AV-30-C AoA calibration

[dtuuri]

I've been flying a Cessna 172 with one of these blenders that does everything but chop carrots. On takeoff the AoA red bars annunciate a stall warning even though airspeed is accelerating while in or while climbing out of ground effect at Vx. The Pilot's Guide raises more questions than answers, so I'm wondering if somebody knows whether this is an accurate display of angle of attack, considering these red bars are supposed to light up with the aural stall warning, but it is silent. The system "derives" AoA by measuring the difference between pitch attitude (via AHRS) and a calculated flight path slope deduced from indicated airspeed and vertical speed — which makes me skeptical. The VSI is known for having a lag, so is the AoA system affected likewise? Or do the mysterious accelerometers mentioned in the Pilot's Guide supply the vertical input, i.e., no lag.

Another thing, the installation is supposed to be calibrated with 20° of flaps and partial power, as in the "base leg to final" configuration. Takeoffs with full power during transition out of ground effect with zero flaps are a different and a dynamic animal, so how can the red bars set with 20° flaps be valid? Looking for some thoughts about this...

 

[Morgan3820]

Just another tool. Auto pilots can get you into a lot of trouble also and yet they seem to be very popular.

 

[nauga]

The system "derives" AoA by measuring the difference between pitch attitude (via AHRS) and a calculated flight path slope deduced from indicated airspeed and vertical speed — which makes me skeptical.

It should. If AOA is truly estimated using only that method (which is also what's described in their documentation) I would disable it, as it's only valid in wings-level flight. It'd rather have no indication that an indication that varies wildly in its accuracy.

ETA: I firmly believe 'accurate' AOA in the cockpit is very useful and I have an AOA indicator in my airplane, but it's driven by differential pressure and augmented by an AHRS, GPS, and Kalman filter.

Nauga,
who turns

 

[Capt. Geoffrey Thorpe]

The 172 has a NACA 2412 airfoil, but the data in Abbot and VonDoenhoff does not include a flap, so looking at the 2410 (slightly thinner):

Looks like the angle of attack for a stall changes by about 4 degrees with a split flap - not the same as the Cessna flap, but it does show the stall angle changing with flaps.
So, is the question about the accuracy of the AOA indication or when to turn on the red bars?

 

[nauga]

So, is the question about the accuracy of the AOA indication or when to turn on the red bars?

Seems to me the first one should be addressed before the second one.

Nauga,
and carts and horses

 

[dtuuri]

The 172 has a NACA 2412 airfoil, but the data in Abbot and VonDoenhoff does not include a flap, so looking at the 2410 (slightly thinner):
View attachment 111689
Looks like the angle of attack for a stall changes by about 4 degrees with a split flap - not the same as the Cessna flap, but it does show the stall angle changing with flaps.
So, is the question about the accuracy of the AOA indication or when to turn on the red bars?

Looking at that chart, wouldn't the 60° split flap act kind of like a spoiler to reduce the stalling angle? A slotted flap like the C-172, on the other hand, might allow a higher AoA before it stalls? Or, is there a thumb rule that any flap amount of any kind would stall at a lower AoA? If I knew the answers I could try to go further, logically, trying to explain the behavior I see, but I agree with @nauga that if the system design is a poor one, it isn't worth the mental anguish (at least I think he'd agree with that).

 

[nauga]

...but I agree with @nauga that if the system design is a poor one, it isn't worth the mental anguish (at least I think he'd agree with that).

I feel maybe just a little stronger than that about it but will hold my tongue for now. I'd also be careful about extrapolating 2d airfoil (section) angle of attack and stall data into real-world 3d wing and aircraft AOA and stall data even with an accurate AOA system. It might give you a good starting point for a real calibration, but it's not means to an end.

Nauga,
with the needle at 3 o'clock on the gauge.

 

[Capt. Geoffrey Thorpe]

Looking at that chart, wouldn't the 60° split flap act kind of like a spoiler to reduce the stalling angle? A slotted flap like the C-172, on the other hand, might allow a higher AoA before it stalls? Or, is there a thumb rule that any flap amount of any kind would stall at a lower AoA? If I knew the answers I could try to go further, logically, trying to explain the behavior I see, but I agree with @nauga that if the system design is a poor one, it isn't worth the mental anguish (at least I think he'd agree with that).

Assuming we measure angle of attack from the fixed geometry of the airfoil, I'm reasonably confident that the airfoil will stall at a lower angle of attack with flaps deployed because the flaps effectively increase the AOA relative to the unflapped airfoil geometry by dropping the trailing edge.
I would suspect that the AV-30 red bar can be calibrated with or without flaps but would not expect a perfect match for both conditions.

 

[Capt. Geoffrey Thorpe]

It might give you a good starting point for a real calibration, but it's not means to an end.

Correct. Just an qualitative illustration of which directions the curves move.

 

[dtuuri]

Assuming we measure angle of attack from the fixed geometry of the airfoil, I'm reasonably confident that the airfoil will stall at a lower angle of attack with flaps deployed because the flaps effectively increase the AOA relative to the unflapped airfoil geometry by dropping the trailing edge.
I would suspect that the AV-30 red bar can be calibrated with or without flaps but would not expect a perfect match for both conditions.

Ok, I think I can see that. In other words (?) for a given (safe) speed in level flight, extending flaps would cause the nose to pitch down due to dropping the trailing edge and result in a lower AoA of the fixed airfoil, right? Then it reasons (my reasoning ) that retracted flaps would raise the nose resulting in increased AoA w/respect to the fixed part of the airfoil. Since the fixed part of the airfoil is married to the AHRS pitch angle by rivets, this would result in red bars at the same IAS where, with flaps, there weren't any, right?

 

[Capt. Geoffrey Thorpe]

First, we need to define angle of attack. Given an airfoil with the flaps up, the stated angle of attack is the angle of a straight line (chord) from the tip of the leading edge to the trailing edge.
When we put the flaps down (assume no pitch change to keep things simple). We typically keep using the original line and just say the AOA has not changed (the simple, sensible thing to do - particularly since flaps are not full span.) and you get a graph like I had previously shown. But in reality, we have changed the wing into a totally different airfoil (even though we say we have just lowered the flaps) with a lot more camber and if you were to draw a new chord line you would claim that it is at a higher angle of attack. But that would end up being a complicated mess, so that isn't done.
And, this is where it gets messy. The flaps change the airflow over, under, behind, and ahead of the wing.
In front of the wing the air comes up to the wing at an angle - you can see it in this picture at the left where the general airflow is horizontal, but you get up flowing stuff out ahead of the wing. Using flaps will increase that upwash at the leading edge (at least until things start to stall). So, depending how and where you are measuring AOA, you will get different results as you change flaps. (there are better pictures, but I don't have one handy at the moment - sorry)



In the end, I would expect the red bars vs stall relation to change with flaps just because there is a lot of stuff going on. The stall horn/ thingie mounted at the leading edge is going to be reasonably close even with flaps because of the change in upwash - but I suspect (don''t really know) that the exact margin between the horn and the stall may change somewhat. I think.

An experiment in the airplane stalling with and without flaps would probably be your best indicator of what is really going on and what the red bars really mean.

 

[AlphaMike]

I have the AV20 in my 182. I got it as a replacement for my clock but also for the AOA (I expected it to be more of a novelty). After running the calibration on the AV20 it's surprisingly accurate in all configurations, even steep turns.

 

[nauga]

After running the calibration on the AV20 it's surprisingly accurate in all configurations, even steep turns.

If that is the case then they are using something more to estimate AOA than the method they describe in their manuals. That's entirely possible, and probably smart - misleading descriptions aside.

Nauga,
with attitude

 

[AlphaMike]

If that is the case then they are using something more to estimate AOA than the method they describe in their manuals. That's entirely possible, and probably smart - misleading descriptions aside.

Nauga,
with attitude

Yeah I agree. When I bought it I figured the AOA function would be nothing more then a novelty. I got it because my clock was flaking out and it has some cool clock / timer functions. But, to my surprise its actually pretty functional as an AOA.

 

[AlphaMike]

If anyone is interested, this video is pretty accurate to what I see with the AV20s. I have flown chandelles, lazy 8, power on and off stalls in all configurations and this is pretty much what I see on the AV20s.

 

[Capt. Geoffrey Thorpe]

If that is the case then they are using something more to estimate AOA than the method they describe in their manuals. That's entirely possible, and probably smart - misleading descriptions aside.

Nauga,
with attitude

They have pitch, G loading, airspeed, rate of climb (should have a lot less delay than with the analog device). Missing is weight and flaps.
Effect of weight you may be able to sort from pitch/ climb / airspeed / G data / pressure altitude - level flight seems reasonably easy to solve for the fudge factor from (if I did the math correctly)
L = 1/2 *rho *V^2 *S *Cl and Cl = 2 *pi *AOA *Correction for aspect ratio

AOA = FudgeFactor* density * Velocity^2, AOA approx = pitch in level flight
FudgeFactor = density*Velocity^2/pitch.

The question becomes "How long does it take to sort out the fudge factor if you are climbing off the runway?"

I think.

 

[nauga]

The question becomes "How long does it take to sort out the fudge factor if you are climbing off the runway?"

I think.

I admittedly did not dive into the math to check, but I've worked on stuff that used what is probably a similar method but accounted for weight and normal acceleration...with mixed results. If you're going through those gyrations to get an approximation and have (a) air data, (b) accelerations, and (c) an AHRS, all of which are ostensibly available in this unit, why not just do the Kalman filter architecture the rest of the industry uses and get better accuracy, assuming it's implemented correctly? I can't help but wonder if they do, but have just dumbed down the description a little for the masses.

Nauga,
one of the masses

 

[Capt. Geoffrey Thorpe]

"Reference Sperry Patent #3,948,096 for additional implementation details." 1976 - appears to be done in circuits. No idea how uAvonix did it in software.

 

[nauga]

"Reference Sperry Patent #3,948,096 for additional implementation details."

Oof. That's a reminder that I've had this conversation with someone here at least once before.

Nauga,
whose head is sometimes like a sieve

 

[dtuuri]

"Reference Sperry Patent #3,948,096 for additional implementation details." 1976 - appears to be done in circuits. No idea how uAvonix did it in software.

I looked at that yesterday, but it put me to sleep. Can you tell me how they measured the vertical velocity and whether or not it would suffer from a lag?

 

[Capt. Geoffrey Thorpe]

I looked at that yesterday, but it put me to sleep.

Cheaper than Sominex...

Can you tell me how they measured the vertical velocity and whether or not it would suffer from a lag?

They being uAvionix?
The AV-30 can read the changes in static pretty darn quickly - and they have control over any filtering without relying on the "calibrated leak" technology used by the VSI. The accelerometers read quickly and the gyros are most likely rate gyros to begin with so they will pick up the changes very quickly. They almost surly have some averaging / filtering algorithm in the software - but as for how fast/slow it is the only way to really find out would be to test it in the airplane.

 

[asicer]

After running the calibration on the AV20 it's surprisingly accurate in all configurations, even steep turns.

What was the procedure that you used? I think I did it once a while ago and it wasn't very close.

 

[Ed Haywood]

So what does it indicate when you spin it?

 

[Tools]

Pretty sure it’s gotta be calibrated for each wing configuration to be accurate for each wing configuration.

we were taught our AOA systems in the navy were such, and therefore accurate in any configuration. And they were ACCURATE.

 

[nauga]

Pretty sure it’s gotta be calibrated for each wing configuration to be accurate for each wing configuration.

we were taught our AOA systems in the navy were such, and therefore accurate in any configuration. And they were ACCURATE.

Yes and no...You don't have to change the calibration for each wing configuration but you do need to know the effect, e.g. at stall is 20 units with flaps and slats full but 15 flaps/slats up. That's the beauty of 'units' instead of degrees, they have an operational meaning that is clear but they don't necessarily correspond to any physical measurement like degrees. Nowadays with air data computers as critical equipment you can have real degrees in the cockpit calibrated for different configurations, but flying at 22 units or 8.1 deg on approach doesn't really tell you anything different in terms of performance or how to fly it, as long as both are at 3 o'clock on the gauge, on-speed indexers, and a centered E-bracket.

Nauga,
unitized

 

[Tools]

So my choices are:

a. Memorize different units for different flap settings

2. Make engineers apply a bias so all I gotta remember is 15 units (3 o’clock… er… 3 bells) on speed.

Is that really a choice?

 

[nauga]

So my choices are:

a. Memorize different units for different flap settings

2. Make engineers apply a bias so all I gotta remember is 15 units (3 o’clock… er… 3 bells) on speed.

Is that really a choice?

As the trend towards degrees in the cockpit continues it's looking more like (b) but you still have different angles of attack to remember for different situations, like (paraphrased) 10 deg max AOA with flaps full, OEI; but 12 deg max AOA flaps half, OEI; and I daresay a Vmc departure is somewhat worse than FAW.

Nauga,
from when dinosaurs flew over the earth

 

[dtuuri]

I'm stumbling over this "filtering". If the AV-30 stall warning is calibrated with 20° flaps, my take on it is that such filtering could not predict a stall accurately (other than random chance) for 0° flaps, am I right? I.e., the filtering is useful for smoothing out the recorded datapoints when approaching a stall with 20° flaps only. Correct?

 

[AlphaMike]

What was the procedure that you used? I think I did it once a while ago and it wasn't very close.

It's in the manual. It is very simple but you need to be on level ground and its best to do it in a hangar or on a 0 wind day.

 

[Capt. Geoffrey Thorpe]

I'm stumbling over this "filtering".

Data can be noisy - just for a random example, you look at an accelerometer just sitting on your desk it will read (in gs) 1.013 .9999 1.002 1.010 blah blah blah (I just made those numbers up)- and of course, it will be a lot worse in the airplane with the engine running and vibrating the sensor. So, either in hardware or software (or both) you do some smoothing / averaging / filtering of the data to get rid of some of the extraneous variations in the data. This does introduce some lag, but it should be nothing like the analog VSI. Filtering changes the speed of the response to a change in input - for example, you get hit by a big downdraft - filtering will change how fast the display changes.

Mr. @nauga mentioned a Kalman Filter - this is an algorithm where you take a bunch of inputs, combine with some knowledge of how much the readings bounce around for no good reason and try to converge on the actual state of the system in an optimal manner - it's a bunch of math ****.

You want some filtering so you get a smooth display - you don't want it bouncing up and down with every bump - I'm sure some engineer spent more than a few hours testing the stupid thing to come up with a number that gives a fast but stable number of bars. Way more than a few hours. And probably a bunch of meetings. With "help" from management. Engineering would be a lot easier if management didn't "help".

The issue with flaps and stalls is that the actual stall AOA will vary some with flaps - perhaps with no flaps you stall at 2 red bars and 40 degrees of flaps you stall at 1 bar (I just made that up). That has nothing to do with filtering. I suspect that uAvionix suggests setting the stall AOA at 20 degrees of flaps to get you within one red bar or so of the actual stall at 0 and 40.

The only real way to know how sensitive this is to flaps would be to stall it at 0 and 40 while counting bars. And, to evaluate the filtering look at how fast the display changes when you yank on the yoke or you hit a bump and how stable it is when things are rocking and rolling.

 

[dtuuri]

It's in the manual. It is very simple but you need to be on level ground and its best to do it in a hangar or on a 0 wind day.

Not in the manual I'm reading: Microsoft Word - D-0032-0, AV-30-C, Pilots Guide UAV-1003946-001, Rev B.docx (aircraftspruce.com)

 

[Capt. Geoffrey Thorpe]

It's in the manual. It is very simple but you need to be on level ground and its best to do it in a hangar or on a 0 wind day.

Not in the manual I'm reading: Microsoft Word - D-0032-0, AV-30-C, Pilots Guide UAV-1003946-001, Rev B.docx (aircraftspruce.com)

There be Two kinds of calibration.

One is calibrating the gyros (and possibly the accelerometers) to compensate for the fact that they don't quite read "0" (or 1) at rest. That's done in the hanger. Don't wiggle around i in the seat. Make sure the airplane is level. This is done at the factory which may be good enough.
The other is setting the angle of attack where the airplane stalls. That is done in the air. Well, you collect the data in the air. I wouldn't be futzing around with it in the air unless you have someone else flying and paying attention while you are heads down. But that's me.

 

[dtuuri]

One is calibrating the gyros (and possibly the accelerometers) ... Make sure the airplane is level. This is done at the factory which may be good enough.

Who's factory, uAvionix? The plane's owner said there's no need to calibrate the system when it's new, so is he referring to just the gyro calibration, then?

 

[Capt. Geoffrey Thorpe]

Who's factory, uAvionix? The plane's owner said there's no need to calibrate the system when it's new, so is he referring to just the gyro calibration, then?

I would assume that is what was referred to.

In the manual for my AV-20 (surprisingly easy to see for how small it is) they say to only re-do the gyro calibration if something seems wrong. But the stall angle has to (should be) done. One of these days, I might even actually do it.

 

[Capt. Geoffrey Thorpe]

Gyro calibration is in the installation manual

https://uavionix.com/downloads/AV-30-E/AV-30-E Installation Manual UAV-1004234-001 Rev F.pdf
"13.2.2 Gyro Calibration
As units age or experience unusual conditions, the precision gyroscopic
sensors may need to be re-calibrated. This can be accomplished in the
field using the procedure noted here. At least 10 minutes of warmup is
required before performing the AV-30 gyro calibration.
...
3. It is critical to ensure the aircraft is completely motionless and on the
ground during the next several steps. The aircraft should be in a
hangar and must not be affected by wind or other sources of aircraft
movement. Do not leave the aircraft during the countdown"

Looks like there are procedures to calibrate the magnometer (and other stuff if necessary).

 

[AlphaMike]

My AV20 defiantly needed the gyro calibration after install. It was way off until I ran the calibration. I also had to set the AOA limits after doing the gyro cal.

 

[Ed Haywood]

Semi OT, but I have an RCA 2600 AHI and fly acro frequently. Is amazing how well it holds attitude and recovers. I can sometimes get through a full 10 maneuver Sportsman sequence without losing reference. If it does tumble (red X), it will usually recover to accurate indication in about 10 seconds of semi-stable flight, even in a steep bank. Not sure how the little accelerometers know, but they are awfully smart. And BTW there is no GPS or pitot-static input.

Pitch can be calibrated in flight using up/down buttons at the top of the gauge. Roll is calibrated by leveling the aircraft on the ground and loosening the mounting screws.

 

[dtuuri]

Many thanks to @Capt. Geoffrey Thorpe and @nauga for taking time to help me have a better understanding of the AV-30. It looked pretty alarming during climb-out to the airplane owner while I hadn't even noticed the indication because my attention was split between outside references and the ASI, mostly outside on the nose attitude, though. My practice is to raise the nose just barely clear of the runway and let the plane run on the mains until it flies off naturally while holding that attitude and no higher. I find that technique results in a comfortably short takeoff without the monkey-motions some pilots go through with multiple pitch adjustments (not to mention sudden flap extensions) all while trying to remain in ground effect. In short, I've concluded it isn't worth the bother because (Alaskan bush pilots, divert your eyes please) the advantage isn't significant enough and if you need to remain in ground effect then you should probably just trailer the plane out of there. Of course, I could be wrong. So, does anybody know of a serious study that empirically shows the actual difference between my method and the theoretically better way that's in all the textbooks for light aircraft (not transport/military)? In other words, how many feet sooner would the theoretically more optimum method clear the obstacles.

 

[Ed Haywood]

Many thanks to @Capt. Geoffrey Thorpe and @nauga for taking time to help me have a better understanding of the AV-30. It looked pretty alarming during climb-out to the airplane owner while I hadn't even noticed the indication because my attention was split between outside references and the ASI, mostly outside on the nose attitude, though. My practice is to raise the nose just barely clear of the runway and let the plane run on the mains until it flies off naturally while holding that attitude and no higher. I find that technique results in a comfortably short takeoff without the monkey-motions some pilots go through with multiple pitch adjustments (not to mention sudden flap extensions) all while trying to remain in ground effect. In short, I've concluded it isn't worth the bother because (Alaskan bush pilots, divert your eyes please) the advantage isn't significant enough and if you need to remain in ground effect then you should probably just trailer the plane out of there. Of course, I could be wrong. So, does anybody know of a serious study that empirically shows the actual difference between my method and the theoretically better way that's in all the textbooks for light aircraft (not transport/military)?

Remaining in ground effect is only necessary for soft field takeoff. For everything else, you described the textbook method. So not sure what difference you are looking for. You are doing it right.

 

[dtuuri]

Remaining in ground effect is only necessary for soft field takeoff. For everything else, you described the textbook method. So not sure what difference you are looking for. You are doing it right.

I think paragraph 13.1 is pretty close to my method (I don't "rotate", per se, nor start with full aft stick), but the author shows other (more better) techniques. I just question whether they are worth the hassle and might even result in worse performance if not executed perfectly: 13 Takeoff (av8n.com)