Updated: May 24
I've been way behind the power curve posting progress reports over the past several months, no excuse! So, in no particular order:
VectorNav VN-300 Gyro Integration
This winter, we modified the RV-4 by adding a Vector Nav VN-300 reference gyro. The system combines GPS derived data with INS derived data. It is referred to as a “GNSS/INS” and is the gold standard for flight test work. By combining the two systems, it provides earth frame and body frame references, i.e., information relative to the earth (GPS) and information relative to the air mass (INS). The gyro itself is ridiculously small—about the size of a large postage stamp and not much thicker. It contains a full set of temperature and pressure compensated MEMS gyros, 2 sensitive GNSS receivers and advanced Kalman filtering algorithms to provide optimal estimates of position, velocity and orientation. The system also incorporates a GNSS compass into the INS filter and accurately estimates heading in both static and dynamic conditions without reliance on a mag compass. The system has two basic modes: static where it functions as a basic AHRS and dynamic where it functions as a precision INS. Since the V2 box is located on the aircraft centerline very near the center of gravity, we simply bolted the VN-300 to the top of the V2 box (Figures 1 and 2).
To provide accurate GPS information (including GPS derived heading), the system uses two antennas. The forward antenna is located on the upper firewall underneath the fiberglass cowling, and the rear antenna is located on the turtledeck underneath the rear canopy. The antennas are compatible with the US GPS, Russian GLONASS and European Galileo satellite navigation systems; so there are always plenty of satellites in view for accurate measurement. The system automatically transitions from AHRS mode to GNSS/INS mode when the airplane begins to move and a “dynamic” alignment is performed. The antennas are separated by 2 meters and mounted on the longitudinal axis of the airplane. This provides a heading solution accurate to within .15 degrees.
The system is fully automatic and powers up when the V2 box is turned on. When initially powered up, the system is operating in AHRS mode with basic pitch, roll and yaw measurements accurate within ½ degree and an on-board magnetometer provides coarse heading estimation. Satellite lock-on occurs within 30-40 seconds and the system achieves an accurate GNSS position for INS alignment. Full alignment occurs in 3-5 minutes after GPS lock-on and is refined after the aircraft begins to move. Full dynamic alignment is complete during the initial stage of the takeoff run.
The gyro measures parameters relative to the body frame and earth frame and records data to the V2 box at 50Hz for post flight analysis. Gyro data is combined with indigenous sensor data (pressure and IMU) from the V2 and air data boom for post-flight analysis. The gyro is calibrated to the three aircraft axis with the airplane up on jacks and leveled, thus recorded pitch and roll are aligned with appropriate aircraft references and no post flight “installation error” correction is required (Figure 3). The gyro data replace the EFIS data that we previously used as a baseline reference. The gyro is not only more accurate, but uses better filtering and records data near real-time, a big improvement over the EFIS.
Designing the V4
We’ve gotten many inquiries about whether we’ll develop either a kit or AUR (USAF speak for "all up round”), i.e., a fully built box ready to bolt into the airplane. While we still hope one or several manufacturers adapt the technology and incorporate the logic into existing production systems; the phone hasn’t exactly been ringing off the hook. We haven’t had any inquiries from manufacturers. Apparently, they are content with the performance and mechanization of their angle of attack systems.
We realize that building a small electronic device isn’t everyone’s cup of tea; so, we are currently coordinating with MakerPlane to develop a production unit. We are tentatively calling this derivative the “V4” since it is the fourth iteration of the Gen 2 system hardware. The big difference between the V3 and V4 will be a transition to “surface mount” components. The V3 design is referred to as “through hole,” which means it’s optimized for ease of assembly using COTS components by hobbyists with basic soldering and computer programming skills. One of the neat things about the EAB community is extent that folks are willing to experiment and push their limits as they learn new skills. We’ve had multiple V3s built in the field from scratch. Some of these folks also assisted with developing the assembly instructions and QC’ing the parts list currently maintained on our GitHub site. What a cool "crowd sourcing" project!
The V4 will use surface mount components to the extent practical to ensure reliable performance. We will still source components deemed critical for reliability and accuracy. It will have 2 IMU’s vs the V3's 1. Use of surface mount components will reduce parts count and facilitate ease of assembly. Since we are an all-volunteer program, we don’t have a target date set for V4—it will happen organically. Incidentally, the MakerPlane team shares a similar mission to the FlyONSPEED team: open-source, non-profit. The heart of their work is a common bus structure that allows all the various electronics to talk to each other. They also publish all their work online and at GitHub as well. We are looking forward to partnering and getting more hardware to the field.
Never as fast as You’d Like…
Seems hard to believe that we started this project 5 years ago. Turns out it’s a lot like building an airplane—everything takes longer than you think and there are lots of “do over” opportunities. As an all-volunteer group, folks contribute as they have the time and inclination to do so. We are non-timeline driven. Having a strict schedule doesn’t work well when two of the primary objectives are to learn and have fun. We don't want to turn our hobby into work, and we’ve all got day jobs (even if that means enjoying retirement :). Our biggest time consumer is the learning curve. We are constantly learning new things, improving software and hardware and revisiting previous tests to validate performance. To reduce data reduction time, we initially incorporated quite a few algorithms into the code so that the data stream contained usable information. I now regret this approach. Unfortunately, we didn’t always get the aero correct; or, more likely, fat-fingered something in the software. This leads to lots of “do over” flying for yours truly. Had we recorded raw data, we can always go back ex post facto and apply any algorithm we wish for analysis. We’ll be fixing this in our next software iteration. Another good lesson learned.
One feature we still lack is automatic calibration. Until we get that working to our satisfaction, we have only shared hardware with folks that have the wherewithal to install systems and “hard tune” them, which can require some extensive flight test and data reduction, not to mention the ability to wire components into an airplane. We do get to cheat a bit, since the RV-3/4/6/7 and 8 have a nearly identical wing and we can largely apply those curves interchangeably. We’ll distribute the rest of our “low-rate production” V3 hardware to beta test volunteers once we have a reliable easy to use calibration wizard that accommodates a private pilot aircraft handling skill set.
Woohoo, summer vacation is on. We are really looking forward to Oshkosh this summer! Us and a few thousand of our friends, no doubt. It should be a rage. We are planning a couple of presentations and will announce times and dates when the Forum schedule is published. We also entered the final round of the Innovation Prize competition. If we are selected to present, we’ll be notified by mid-July. For those interested, I posted our submission on the Technical Resources page. Click on the header above, and then scroll down to the bottom and you’ll find our white paper and a ten-slide power point briefing. The briefing is a large file (216MB) as it has a couple of imbedded videos. If I was more IT savy, I would have figured out a better way to do that, but you get what you pay for ;).
We could use some help with what we call “green eggs and ham syndrome.” That is, convincing folks that an audio cue is about the easiest way to do energy management in an airplane. We’ve succumbed to social pressure and have developed an optional visual energy display; so that addresses that concern; but the heart of the system is the cave man simple ONSPEED, slow, fast tone logic—do I push, pull or hold what I have when I’m coming into land or overshooting final approach because I failed to assess the strength of the crosswind?
We are looking for suggestions on how to best demonstrate the technology. We know that if you try it, you’ll like it Sam I Am; but how do we show folks the utility and get them to try it? I initially thought if we simply built a prototype and made a video folks would immediately grasp the idea. When we used the system in the USAF, we didn’t even provide training per se. It was simple enough after hearing the tone, there wasn’t any need to do anything other than demonstrate it once or just simply go fly with it. The thing I find interesting is that when folks watch a video, they immediately focus on the tone itself to the exclusion of everything else. In the airplane, the tone is natural background noise that only comes on at low speed/higher alpha. There’s a volume knob, after all. We also know that our Gen 1 demonstration had a very “tinny” tone that was harsh if you cranked the volume (which I always did for demo videos since hearing the tone was the whole purpose). Unfortunately, many of our demo videos use the Gen 1 system. The quality of the tone is much improved in Gen 2; but the idea of remaking lots of video isn’t too appealing to me as it takes lots of flying and editing to make that happen. So, we are looking for inputs: what sort of video, audio, printed, graphic, etc. information can we provide to folks to demonstrate the utility of this type of system? Post ideas, inputs, BS flags, etc. on the Forum or drop an email to team.flyonspeed.org.