After 18 months of work, we've got a fully functioning coefficient of pressure AOA system that generates the old F-4 style performance cues! Lots left to do, but it's nice to have our baseline capability validated. With science in progress, you're never quite sure what's going to work and what's not...thank god the primary stick monkey is a history major--I'm not over-burdened with over-thinking.
Although Lenny built the mounts for the test boom last year and completed initial flight test, we are finally at the point where we can bolt it to the RV-4. Besides physically attaching the boom to the airplane, it was necessary to install power and data wiring to allow the boom data to be written to the same file as pressure, accelerometer and EFIS data on the micro SD card installed in the V3 box. But Lenny has done the serious ditch-digging: developing software that allows all of these data streams to flow seamlessly and simultaneously. The data can be easily downloaded via the micro-USB port on the box or wirelessly post-flight. A single file contains all of the data, synched to the GPS time hack provided by the navigation system via the EFIS data stream.
The boom is powered by a small lithium battery. It can be mounted on the right or left wing tip using the same nut plates that hold the wing tips in place. The mounts are CNC machined and can be mounted on any RV with nominal #6 nut plate spacing (per Van's drawings). The boom transmits wirelessly to a small receiver in the cockpit that is wired into the V3 box. LEDs on the boom and receiver let the pilot know that the system is operational and transmitting. The boom was built by Andrew and the folks at Spin Garage out in Mojave. It weighs only 250 grams. In addition to alpha and beta transducers, the boom is equipped with a sensitive Kiel probe that measures pitot pressure +/- .004 PSI and static pressure +/- .06 PSI. Accuracy of the alpha (angle of attack) and beta (sideslip angle) is within .1 degree. The boom should allow us to determine the accuracy of our coefficient of pressure angle of attack solution as well as measuring lag and response rate. It also provides an independent calibrated air source for analysis. We gotta' get the physics right!
Our heavy-duty CNC mounts use a pin for alignment--I managed to miss the mark by .1 degree low relative to the FRL. We'll use a correction factor during analysis to correct for this installation error.
As Many Curves as it Takes
We've had a few debates over beer on how many "curves" are going to be required to accurately capture the performance-based AOA cues needed to accurately map the power required/drag curve of the airplane. Like most things in aviation, the answer is "it depends..." apparently, on the type of flaps installed. We use GPS speed runs currently to map the aircraft curve in each flap configuration: flaps up, half flaps and full flaps. For the mighty RV-4 with it's NACA 2315.5 constant thickness airfoil and plain flaps, that is Flaps 0, Flaps 20 and Flaps 40. Turns out if we plot those curves on a graph relative to absolute alpha, they are pretty close to over-lapping.
As a matter of fact, these curves are so overlapping, it's practical to compress them into a single curve:
The equation shown on the plot is the algorithm the software uses to compute angle of attack. Because the relationship between the pitot pressure and AOA pressure from the Dynon probe is parabolic, the equation is a second order polynomial. Unlike history majors, computers don't mind math; so no big deal for the processor. The "R2" value on the chart is a statistical test applied to determine how well the data correlate with the curve. 1.0 is a perfect fit, so we'll settle for 99% as being good enough for experimental flight test. Looks like just a single curve is what we'll need to calibrate the system in an RV-3/4/6/7 and -8. We will be testing in Cecil's Citabria to verify the plain flap theory--if it holds, that airplane will only require a single curve as well; but we'll see. Science in progress.
On the other hand, if the airplane has split flaps, the data produce three distinct curves. Here is the plot of Lenny's RV-10 with it's proprietary airfoil and split flaps for Flaps 0, Flaps 15 and Flaps 30:
It's not practical to compress all three curves into a single algorithm without inducing some pretty good error in the solution. Since we want the best cues in a normal landing configuration, it may be practical to compress the data into two curves, however, with the most accurate reflecting normal landing condition. In the plot below, I've compressed the Flaps 0 and 15 curves which results in a usable calibration for those flap settings, and still provides a very accurate curve for Flaps 30 (normal landing). Since the RV-10 is non-aerobatic, we can accept some minor error in a flaps 0 configuration. In fact, this is essentially what's happening with the Dynon EFIS % Lift signal that is currently driving our first generation system. The curve derived by the Dynon calibration is a compromise. Unfortunately, we don't know the particulars of the Dynon algorithm; so it's not practical to perform a direct analytic comparison other than to measure both systems relative to boom data--which is on the list of things to do.
Lenny will be flight testing a two-curve solution; and as we go into beta test, we'll have plenty of opportunity with different types of airplanes to develop a good understanding of this.
As always, we are always open for critique, collaboration or donations to support the work and thank every one that helped this year. We are always happy to interface through this blog or our forums page. We hope to make substantial progress in 2020 to be ready for Oshkosh next summer with a throughly tested system. Best wishes to everyone for a great new year!