Fullscreen HTML5Med Res (??MB)HTML4

Electric RPAs... with wings! by Bob Young* In August, we looked at the burgeoning field of multi-rotor RPAs and SGMAs. But long before multi-rotor aircraft had enough computer grunt to actually keep them in the air, conventional (ie, fixed-wing) electric-powered model aircraft were being flown by radio control. I n this article we will be examining a small fixed wing Remotely Piloted Aircraft, designated as a Self-Guided Model Aircraft (SGMA) by the Model Aeronautical Association of Australia (MAAA) – and the technology incorporated into these little mini marvels. Electric-powered RPAs The choice of electric power as against internal combustion (IC) motors is a difficult one due to the severely limited energy density of batteries, even modern Lithium Polymer (LiPo) batteries. Because of the severe power and endurance limitations currently imposed on electric powered RPAs, great care must be taken in the design and in-flight tuning. However limited capacity not withstanding, the advent of the LiPO 12  Silicon Chip battery with its light weight and 3.7V terminal voltage has revolutionised electric flight for miniature aircraft. But they are still only suitable for short endurance flights at the moment, typically 15 to 120 minutes. As we mentioned last month, electric powered aerobatic model aircraft are becoming a dominant force in international model aerobatic competitions, with over 50% of competitors now using electric power. This event only requires 15 minutes endurance and is thus ideal for electric power. However, LiPos come with certain drawbacks including higher cost, they are easily damaged if not handled correctly and there is a higher risk of fire, especially in a crash. In addition, charging is not a simple process, taking much longer than refilling a fuel tank and it usually involves A transmitter not suitably equipped for LiPos in which the battery caught fire. Fortunately the fire burned itself out before any real damage was caused, probably due to a lack of oxygen in the battery box and carrying case. siliconchip.com.au Fig.1: this screen grab of the “Happy Killmore” Ground Control Station program shows an autonomous flight, plotted on Google Earth. Note the large variety of instruments and flight data available on screen – and there is much more data available under the various tabs and Google Earth settings. Happy Killmore GCS is a very powerful piece of software and it’s free multiple batteries to keep the flying session going. Charging can occasionally be fraught with risk, especially when fast-charging. It is always a good idea to charge them on a fire-proof metal tray that can be easily carried outside in the event of a fire. And it is best not to leave them on their own when charging. Fire is not a frequent occurrence but it does happen, particularly if the battery has been damaged in a crash. Models can be completely destroyed by the intense heat generated by burning LiPos. When using LiPo batteries which can be damaged if the cell voltage falls below 3V so a low voltage alarm or cut-off is a must. One word of warning here: there is a trend towards using LiPOs in transmitters which mostly (certainly older models) do not have low voltage cutoff. If that TX is left on inadvertently, then it is good-bye LiPo. So be very careful with this one. However, this situation is rapidly changing, with faster-charging batteries and improvements in battery siliconchip.com.au construction coming thick and fast. Even so, a twofold increase in energy density or even more is required to lift the electric RPA into the really useful endurance category enabling it to begin to compete successfully with the IC engine. Rumour has it that this improvement is not far away. Despite the foregoing, there are numerous advantages to electric power, including an almost complete lack of motor vibration (a boon for aerial photography), increased reliability over IC engines, ease of starting, the possibility of stopping and starting the motor in flight, (a great aid to increased flight times and further reducing vibration) and finally, an almost complete lack of noise. In view of these advantages, the Author would use electric power exclusively were it not for the limited endurance. Before we move on to an analysis of the electric motor and electronic speed controller (ESC) in the Cub, perhaps a few words on electric power are in order. The table below is a widely understood, rough guide to the power required for different model types. The “watts per kilogram” rating is calculated by dividing the wattage available to the motor by the gross take-off weight of the model in kilograms (kg). 20-30 W/kg: Minimum level of power for decent performance, good for lightly-loaded slow flyer and park flyer models 30-40 W/kg: Trainer and slow flying scale models 40-50 W/kg: Sport aerobatic and fast flying scale models 50-60 W/kg: Advanced aerobatic and high-speed models 60-70 W/kg: Lightly loaded 3D models and ducted fans 70-90+ W/kg: Unlimited performance 3D models The wattage available is one thing but that wattage must be transformed into thrust – and that is accomplished via the propeller. Broadly speaking, as with all prop-driven aircraft, the October 2012  13 Not all SGMAs (self-guided model aircraft) are ten pound weaklings! This Flamingo, designed and built by the Author, is twice as long as he is tall and is powered by a “pusher” Moki 135 glow-plug motor. Actually this one is designated as an RPA because it is intended for commercial and even (hush hush!) military use. bigger the prop, the more thrust it will deliver but with a consequent increase in required input power. However electric motor theory tells us that the lowest current draw will occur with the motor unloaded, thus again broadly speaking, the smaller the prop, the lower the current consumption albeit with reduced thrust. Aerodynamic theory tells us that in level flight thrust will equal drag, with the drag increasing with the square of the airspeed. Double the airspeed, four times the drag, so for the highest speed combined with the lowest drag (thus lowest current consumption) a high efficiency, low drag aircraft is called for. Therefore the challenge for electric powered RPA designers is to get the correct mission-oriented balance between endurance and airspeed, by choosing the correct aircraft design, motor, battery and prop combination. Thus we can now begin to see some of the problems for electric RPA designers. To get to the target quickly requires high speed but speed calls for a serious increase in current. Loitering over a target calls for a sailplane type Fig.2: screen grab of the Electronic Speed Controller (ESC) data file for Flight 6. Note cursor (red line center) and data at the current cursor location (box bottom left). This flight is discussed in detail in the article. 14  Silicon Chip aircraft that can virtually soar with the motor off. As a matter of fact it is here that electric powered RPAs shine, as the motor can be easily stopped and started in flight and by using thermal soaring, endurance can be extended dramatically, certainly by at least two to three times the motor-run endurance. So you see, the design and operation of an electric powered RPA is a very involved and delicate balancing act. Piper Cub SGMA The Piper Cub is obviously not the sort of aircraft discussed above. It is intended to be a pleasant to look at, easy-to-fly and boxy aircraft able to accommodate a wide range of test equipment, fit into the MAAA SGMA specs, teach people the fundamentals of RPA flight and serve as an example for articles such as this – all tasks it fulfils admirably. This particular Cub is 1.9m (69”) in wingspan, with over 1m2 (670in2) of wing area and a wing loading of 7900g/ m2 (26oz per square foot). Therefore it’s a very lightly loaded and quite safe model as needed for training. It weighs 3.3kg (7.25lb), and is powered by a 780W Scorpion 3020/890 out-runner electric motor controlled by a data logging Electronic Speed Controller (ESC). siliconchip.com.au Inset below: the Scorpion motor and ESC. The 3-phase leads to the motor are clearly visible in the fore-ground. Lurking in the background is a lead balance weigh. Note the toroid on the servo lead (just visible at bottom left) to prevent RFI. The Piper Cub self-guided model aircraft we’re looking at in this feature. It has a 1.9m wingspan and weighs just 3.3kg. The 890 is an interesting figure commonly used in out-runner motor specifications. This figure is a crude expression of rev/volt in an unloaded condition. It is expressed as Kv – not to be confused with kV (Kilovolt). Originally the Cub was fitted with a 2.4GHz manual control R/C system, an ATTOPilot autopilot V2 Thermopile autopilot and a 900MHz 9Xtend data link feeding data to a Happy Killmore Ground Control Station (GCS) on a laptop. Power is provided by one or two (parallel) 3S (11.1V) 5,500mAh LiPo batteries. The endurance of the Cub is typically 10 – 20 minutes with one battery, depending upon the prop fitted. Airspeed is measured using a Pitot tube connected to the autopilot. The Cub is not fitted with a camera. However, it could be fitted with one if required for the mission. Under the bonnet We will begin by examining the Fitting out the body of an RPA or SGMA like this twin boom Flamingo is a matter of Finding space for everything. Along with the radio control receiver, you need to find room for the motor (of course!) plus autopilot, attitude sensing, servos . . . and don’t forget the batteries! This particular plane is powered by an internal combustion engine so a fuel tank is also required. Photo: Notre Dame University, Indiana, USA. siliconchip.com.au October 2012  15 The ATTOPilot V3. From left to right: GPS module, 6 DOF IMU (6 Degrees of Freedom Inertial Management Unit) and ATTOPilot control board. The twisted pair is the cable for the LED which indicates the state of the Autopilot and GPS. model from front to back. The 780W electric motor is a brushless outrunner driving props of various sizes, depending upon the mission requirements. The motor is controlled by a data logging ESC. One of the nice features of modern processor-controlled electronic devices, in addition to their programmability, is their ability to record and graph almost everything that goes on inside that unit – and the units in a small RPA are no exception. Good electronic speed controllers (ESC) used to control electric motors come with a built-in data logger which includes such valuable data as battery voltage, current consumption, RPM, ESC temperature and throttle setting, all plotted against time. This kind of data is invaluable when deciding upon motor types, prop sizes, battery capacity etc. Modern ESCs are also fully programmable and feature a wide range of options, including: • programmable low voltage cutoff • programmable cutoff types (soft cutoff/hard cutoff) • programmable brake type (disable/soft brake/hard brake) • programmable time advance (low/standard/high) • some are even programmable to brushed or brushless mode. In Fig.2 we see a data graph for Flight 6, an early 20-minute test flight for this Cub, with the Y-axis calibrated for current. The Y-axis calibration can be changed simply by ticking the box at the middle right. This screen grab was chosen because it shows data which will be used in a later comparison with data graphs taken from the autopilot log for Flight 6. Along the top of the graph are peak readings recorded during the flight. In this particular screen grab the mouse pointer (vertical red line at left) shows the voltage at cruise with the throttle at 46.6% as 11.2V, current 11.8A, thus power being 132W and RPM as 4321. (See Mouse pointing data box bottom left). The diary note bottom right notes that for this flight the prop was a 13 x 10 and there was strong thermal activity. Fig.3, however, taken from the ATTOPilot log file, shows a similar graphic pattern but with a much lower current figure of approximately 8A. Calibration of the current draw was previously carried out with a 0 – 100A meter showing the A/P figure was correct. So the moral is? Trust nothing and always calibrate where possible! Thus we now have a take-off power of about 540W with a power loading of 33.86W/kg (74.5W/lb) but a cruise power of say 11.2V x 10A = 112W for a power loading of 7W/kg (15.5W/ lb) for an average speed of 60km/h, a figure well below what is suggested in the power loading tables. From the foregoing we can begin to see the enormous advantages that data logging provides for people interested in trying to get the best performance from any aircraft. Being able to compare motor power to airspeed now opens the way for some serious mathematical analysis of aerodynamic characteristics of the aircraft under examination. For this reason alone, fitting this sort of equipment to an aircraft is a worthwhile exercise for any pilot serious about improving aerodynamic performance and endurance and the electric model in particular lends itself well to this sort of analysis. 2.4GHz radio control The Digital Spread Spectrum (DSS) radio control system used in the Cub is a 2.4GHz 8-channel receiver running from a separate 6V battery driving The Thermopile sensors on the Cub. Here the horizontal sensor set (Top of wing) is arranged in the “X” format. Note the calibration sensor (vertical) on the lower side. Fig.3: current graph for Flight 6 taken from the ATTOPilot log file shows a lower current reading than shown in Fig.2. 16  Silicon Chip siliconchip.com.au four servos (elevator, rudder and two aileron servos) and the ESC. For a full discussion on 2.4GHz DSS radios see SILICON CHIP February 2009. It is advisable to use a separate RX battery rather than the ESC regulator for a variety of reasons. Amongst these are servo motor noise being induced into the receiver and to prevent overloading the ESC regulator when using more than three servos in the model. Also, the motor can take the main drive batteries to quite a low voltage under some conditions and one does not want to lose control when the receiver “browns out”. One of the nice things about 2.4GHz receivers is that they are largely immune to all of the little horrors such as interference from electric motor noise, servo noise, spark ignition noise and processor noise; all problems that sometimes caused the pilot serious grief when operating receivers working on 29 and 36MHz. They are also immune from interference from other flyers operating on the same flying field. Thus frequency control is no longer a major issue. The receiver used in the Cub features a fail-safe activated in the event of the TX being switched off in flight or an inadvertent loss of control signal. There are two fail-safe conditions, one in which the servos hold the last known position. The second fail-safe type sends the servos to pre-set positions. When embarking on a long range flight (out of TX range) the TX is usually switched off and the last known servo position fail-safe is used. This keeps the RPA in trim while handing over to the autopilot which then takes control of the aircraft. Autopilot Here we arrive at the heart of the SGMA or RPA. An autopilot (A/P) is essentially a feedback system aimed at keeping the aircraft on a pre-plotted course, at a set airspeed and flying in straight and level flight unless changing course as directed by the A/P. There is a wide variety of autopilots available, ranging in price from $500 to $50,000 or more. The difference in performance between the little low cost A/Ps and the high end models is staggering. The low-cost units usually control only the rudder for GPS steering while the high end A/Ps coordinate turns using rudder and ailerons, feature excellent cross-track correction and give the appearance in flight of a piloted aircraft. Plotted on a map, the high-end A/P flying a square or rectangular circuit will present sharp, right-angled corners with the sides absolutely straight and parallel and completely free of bowing due to an excellent cross-track correction system, eliminating sideways drift caused by wind. One of the most popular A/Ps with the SGMA pilots and in the lower cost range is the little American ATTOPilot. The ATTO comes in two versions; the V2 is fitted with a Thermopile sensor for attitude control and the V3 comes with an inertial measurement unit (IMU). For the full story see www.attopilotinternational.com The control board measures just 30 x 35mm and weighs 9 grams. The ATTO is a tiny package fitted with a staggering array of features and programming options. Here are just a few: • stabilisation with automatic PID gain scheduling. • gains for roll, pitch and yaw, adjusted continuously depending on airspeed. • stabilisation gains tuned at any airspeed and they automatically adjust at other airspeeds • proprietary navigation method automatically corrects for wind, flight speed and attitude. Fig.4: GCS showing some of the SET file parameters for Flight 40 just prior to uploading to the Cub. The Google Earth screen shows the Dalby (Qld) model field where these flights took place. It’s one of the best model flying fields in Australia. siliconchip.com.au October 2012  17 • airspeed can be controlled via pitch, throttle or a blended combination of both. • likewise, altitude is controlled by throttle, pitch or a blended combination of both. Proportional blending of the two methods is possible over user-defined altitude bands and mix ratios. • the processor is multi-core (8) and 32 bits, with 160 million instructions per second. • on-board SD card data-logging provides a high bandwidth “Black Box” data record of all flights as commadelimited text files with descriptive column headers. • filename is based on flight date. Unprecedented flexibility in setup The user may define lists of missionselectable loiter, radii and duration, as well as camera trigger repeat intervals based on either time or distance between trigger events. Flight plans can then be accessed via index number. In addition, ATTO gives pilots over 120 configurable parameters that can be used to tailor the A/P for use in a wide variety of aircraft, from conventional monoplanes through to flying wings with elevons (combined ailerons/elevators). These parameters are accessed via the GCS under the configuration tab and then uploaded to the aircraft from the GCS. These little RPAs may look and feel like toys but when combined with satellite-based GPS, long range data links and video downlinks, they represent a staggering achievement is terms of human endeavour. So much so, that Governments the world over worry about their ability to deliver lethal payloads and impose strict limits on their use including the mandatory RTL (return to launch) if the 300kmfrom-home limit is exceeded. Attitude sensing The ATTO V2 fitted to the Cub features a thermopile attitude sensor. These sensors keep the aircraft level by looking at the horizon and comparing the temperatures on the left and right hand sides and the front and rear of the aircraft. They can be arranged in an “x” or “+” configuration and the autopilot calls for the correct arrangement to be programmed into the “SET” file. The ground temperature is always higher than the sky temperature and the small 2-element thermopile sen- sor mounted vertically on the side of the Cub compares the sky and ground temperature and provides the calibration for the horizontal sensors. There is no elaborate pre-flight fiddling with the ATTO V2 in regards to sensor calibration. Thus in flight if the aircraft enters a dive the rear sensors looking at the sky record a lower temperature than the front sensors looking at the ground and it applies up-elevator correction. In a climb, the action is reversed ,with a down-elevator correction as a result. Likewise, left or right rolling deviation from level flight will result in aileron corrections being applied to correct the roll and restore the aircraft to level flight. In this way the aircraft is held in straight and level flight at all times. However there are limitations to the thermopile system. Fog, snow and glare from large bodies of water can reduce the system effectiveness. Nevertheless, the thermopile sensors work very well under most Australian conditions. In one of the less pleasant affairs during a recent Dalby (Qld) trip the wind blew the aircraft off the table and it landed upside down on the sensor head, smashing one of the thermopiles. A screen grab from the Happy Killmore GCS showing the track-plot of Flight Six painted on a Google Earth display. It is impossible to count the 16 orbits as they are all on top of one another. The Alarm sounds when the aircraft is recovered and switched off or if the data link is lost in flight. The vertical lines are called extrusions and are plotted upon receipt of each data packet. Uneven spacing indicated poor data reception. 18  Silicon Chip siliconchip.com.au RPA PIPER CUB’S FLIGHT 39 The 39th flight of the Piper Cub was a very early tuning flight for the Cub after being fitted with the V3 ATTOPilot autopilot. The ATTO resides between the radio control receiver and the servos, and performs the stabilisation and navigation functions when the R/C transmitter is switched from manual mode to autonomous mode. The graphs below are taken from the LOG file which records 49 data items, at a rate of four times per second. The LOG file is in comma-separated format and can easily be imported into Excel for data analysis. It is invaluable when fine tuning the ATTO to the aircraft. “Happy Killmore” Ground Control Station (GCS) The GCS program used during this flight is the Happy Killmore GCS, version 1.3.34. It’s a free download (with an option to donate) and you’ll find it at http://code.google. com/p/happykillmore-gcs/downloads/list This is excellent software and well worth a donation. It allows programming of the waypoints directly onto a Google Earth map, as shown elsewhere in this feature. The programmed course consisted of five waypoints aligned North/South and designed to force the Cub to perform left and right turns with two cross-wind, 700m long straight parallel runs. The flight was undertaken with a crosswind from the east gusting at 10-40km/h. Altitude was set at 120m AGL and air speed at 60km/h. One final point on the HK GCS is the provision for a tracking antenna which will deliver optimum range for the data link. This automatically aims the antenna directly at the aircraft during flight. Flight data analysis Looking at the graphs, Fig.i shows the autonomous section of the flight. We can see that during Flight 39 the transmitter was switched from manual control into autonomous mode about 20s into the flight with the Cub well below the target altitude. The 20s was a minor mistake on the pilot’s part as the autopilot prefers at least 30s of well-trimmed, stable, straight and level flight below the target altitude before switching into autonomous mode. The climb to target altitude took approximately another 35s at which point the Cub levels off exactly on altitude target with zero overshoot. Fig.ii shows from that point on there are small variations in altitude of ±9m or less. While this is a less-than-ideal result, from the ground this level of deviation is not noticeable. A well-tuned ATTO will stay within ±3m from the target altitude when installed in an airframe designed to track well and respond rapidly to small control inputs. The Cub is not that sort of airframe and the results illustrate this point. By far the weakest point in the ATTOPilot tuning at this point were the turns at the waypoints. Fig.iv shows the distance from each waypoint, with the long runs approximately 700m apart and the short runs approximately 200m. Fig.v shows the distance from the planned flight path between waypoints. Note that Fig.v shows that the Cub hit each waypoint exactly on target. Fig.vi however shows that the ATTO is a bit soft on coming back on track after the turns but finally settling down almost exactly on track. Thus the tuning needs to be more aggressive in regards to returning the Cub onto the track after turns. Better waypoint planning would also help in this regard. Longer straight runs even with this level of tuning would show excellent crosswind tracking accuracy and that is despite quite a strong crosswind component. Finally, Fig.vi is a plot of airspeed against GPS groundspeed, showing a variation of 80km/h in the groundspeed indicating a headwind/tailwind component of 40km/h at various times during the flight. Once again we see indications of the airspeed tuning being insufficiently aggressive enough to hold the airspeed to the 60km/h target in these conditions. All in all, the Cub and ATTOPilot handled well considering the weather conditions and the lack of fine tuning (on the ATTOPilot). The overall result was quite successful and would have resulted, even at this very early stage, in a very successful photographic aerial analysis of the area, had that been the planned outcome of the mission. siliconchip.com.au Fig.i: TX mode showing the autonomous period. Fig.ii: The distance from the target altitude. Fig.iii: GPS altitude. Fig.iv: the distance to each waypoint. Fig.v: the distance from the planned flight path. Fig.vi: air and ground speed O October ctober 2012  19 Fig.5: the graph of GPS ground speed taken during Flight six; a 20 minute, 16- orbit autonomous flight on a day with winds gusting up to 20km/h or more. Fig.6: the graph showing airspeed for the Flight 6 flight Fig.7: Flight 6 altitude graph. Note the small altitude variations tend to follow the upwind/downwind pattern. The thermopile sensor was replaced with the inertial managent unit (IMU) and the V2 firmware updated to V3.5 firmware and flying continued despite the wind. The IMU is located inside the fuselage and is thus not exposed to this sort of danger. The ATTO V3 uses the same control board as the V2 but with a 6-DOF (degrees of freedom) IMU instead of the thermopile sensors and carries more advanced software (V3.5). The IMU is a solid state device and works more precisely and responds more quickly than the thermopiles, giving the aircraft a crisper response to attitude changes. While the thermopiles can in theory operate at night, something the Author has never tested personally, the IMU certainly can. The IMU also eliminates the above-mentioned thermopile limitations, thus the ATTOPilot V3 is a 20  Silicon Chip very sophisticated little unit and works extremely well in action. Operational techniques As mentioned previously one of the nice features of modern electronics devices is their recording ability and the ATTO is no exception. The ATTO features two major file sets: the “SET” file in which the pilot sets the parameters for his particular aircraft and the “LOG” file which is the actual recording of the flight data. It is the RPA operator’s task to fine tune the values inserted into the SET file by test flying and examining the LOG file data and adjusting each parameter accordingly. ATTOPilot offer a good back-up service in this respect and will offer hints on tuning to the tyro Remote Pilot. The LOG file begins by recording the data in the SET file so that the actual parameters used during that flight are available for future comparisons and then goes on to record the flight data. The LOG file can be quite large in a long flight with 49 data columns recorded, data being updated four times a second in a comma-separated variable format. Thus the data can be inserted directly into an Excel spread sheet and graphed accordingly. The actual flight under examination was a 20-minute flight in which the aircraft orbited a single waypoint 16 times in autonomous flight on a day with winds gusting up to 20km/h or more. Thus we see the above graph indicating upwind/downwind ground speed variations of up to 40km/h during each orbit. Fig.6 shows the airspeed on that flight and in theory that should remain constant throughout the flight as there is no upwind or downwind as far as the aircraft is concerned. A glance at Fig.6 is all you need to confirm this was indeed the case. The slight variations in airspeed indicate a small degree of adjustment to the throttle gain value in the A/P SET file is required to overcome the small upwind/downwind variations in speed. These graphs provide an invaluable service in fine tuning the A/P for best performance. The aircraft. when switched into autonomous mode was at a height in excess of the target cruise altitude of 90m above ground level (AGL) set in the A/P SET file. Hence the aircraft dived to return to the target altitude thus increasing the airspeed temporarily until the system stabilised and entered the correct cruising airspeed envelope. Once again referring to Fig.5, we can see that the elevator parameters are not set correctly with the altitude deviations while being close, are in excess of the ideal. And again we can just see the repetitive pattern of upwind/ downwind variations. Altitude hold should be within ±3m in a well-tuned aircraft and ATTO. Even so, the final result is quite good SC for an early test flight. * Bob Young is the principal of Silvertone Electronics, a company at the forefront of design and building radio controls (especially advanced digital) and remotely piloted aircraft such as the Silvertone Flamingo shown in this feature. Contact Bob on 0423 098 418 siliconchip.com.au