INTRODUCING . . . AIRGUIDE I AND AIRGUIDE II A MODULAR ELECTRO-FLUIDIC AUTOPILOT SYSTEM FOR HOMEBUILT AIRPLANES By Don Hewes (EAA 32101) Research Engineer 12 Meadow Drive Newport News, VA 23606

IN THE MARCH '78 issue of SPORT AVIATION, I gave an update report on the continuing development of my version of the single-axis electro-fluidic autopilot referred as as the Poorman's Autopilot. This relatively low-cost autopilot system as based on the pioneering work of H. Douglas Garner, who is an Instrument Research Engineer at NASA's Langley Research Center and is also continuing with development efforts with his versions of the autopilot as part of his official duties at the Center. Doug has reported on his efforts in earlier issues of this magazine. The interest shown by homebuilders has been growing rapidly over the past couple years, and a number of enthusiasts have successfully built their own versions of the autopilot based on the information both of us have provided. I was greatly encouraged to continue with my own efforts by the many letters I received in response to the March '78 report in which I offered to prepare a booklet containing complete detailed information on the design, construction, and operation of an easily-built version of the autopilot. Up to that time, only sketches, photographs and notes on some key aspects of the system had been published, and I had felt that many enthusiasts were hesitating in constructing their own autopilots because of the many necessary details that were still lacking. The letters I received confirmed my assessment and, furthermore, the number of deposits for the booklet helped to cover the outlay of money involved in printing and distributing the booklet. The purpose of this article is to give a report on this latest project which has resulted in the development of a modular design approach. The new autopilot system

based on this approach has been named AIRGUIDE, which makes reference to the very small wisp of air in the rate sensor which is the heart of this simple but effective guidance system. AIRGUIDE I and II refers to two versions, the first of which provides the same single-axis roll control function as the original Poorman's Autopilot. The second version provides two-axes control in both roll and yaw. The purpose of the additional control axis is to provide an automatic sideslip trim capability that should reduce the errors caused by out-of-trim yawing moments. These moments result

from changes in engine power and lateral position of the center-of-gravity as well as inadvertent displacement of the primary flight controls caused by small amounts of friction in the control systems. 12 FEBRUARY 1980

In this report, I will discuss various details of the two autopilot versions as well as to recount just how and why some of these details came into being. Not long after completing the update report last year, I found that I had considerable time on my hands to contemplate the autopilot system. My flying activities had been halted as a result of the implantation of a pacemaker to regulate a somewhat erratic heart beat. While recuperating from the surgery, I began to collect my thoughts as to how to proceed with compiling all of my material. The big obstacles appeared to be 1) deciding how much material to present, and 2) designing the system components in a form such that a person unfamiliar with electronic principles and fabrication practices would not have too much difficulty in building and operating the system himself. I especially wanted the design to be such that the system could be built quite easily from scratch without the need for special tools or a kit of prefabricated parts. Just prior to my heart problems, I had been making some very interesting flight tests of some additional features that could result in expansion of the single-axis system into a two-axes, and ultimately, a three-axes system. Unfortunately, I was unable to finish the work and was left with only some encouraging but inconclusive indications that the schemes would be successful. Therefore, knowing that I might be successful in proving out the additional features, I wanted to provide a design that would permit expansion at a later date into the more advanced version. The problem was that of doing so without compromising the simplicity and ease of construction of the basic system. As it turned out, these problems were overcome by a single solution. The solution was the use of a readily available circuit board (via Radio Shack) which is intended for experimental breadboarding of circuits utilizing 14 and 16 pin integrated circuit chips or modules. Each ll/2 x 2 inch board provides mounting holes at the center for a single module and associated soldering contacts which are connected to a series of small copper pads around the periphery of the board. The pads serve as soldering contacts for wires leading to other parts of the circuit and as mounting terminals for small components directly associated with the module. By using a series of these standard circuit boards, it was possible to build up various electronic subassemblies as separate units and then to merely tie them together to form

whatever type of complete system was needed. This eliminated having to make up a single customized circuit board which would have to be discarded if the system was to be modified or altered. Although it was still necessary to use some special circuit boards, these turned out to be small and relatively easy to fabricate because of the modular design approach.

Sewsotz CIRCUIT So«wzo —

FIGURE 1

pump used with the rate sensor. Each board can be assembled and given a preliminary operational checkout as a separate circuit prior to assembling together into the final complete circuit. The other c i r c u i t components, which were not mounted on the standard circuit boards, were grouped together in functional assemblies mounted on the customized circuit boards. To complete the fabrication of the system, the individual assemblies merely had to be interconnected by wires bundled together into a neat wiring harness.

I do not want you to think that because I am presenting the AIRGUIDE I as a completely flight-proven system that AIRGUIDE II is likewise fully proven. To the contrary, the complete two-axes system has not been subjected to any flight tests whatsoever, although it has received extensive bench checkout and testing. Some of its components have been subjected to limited flight tests and the results were encouraging, as I have mentioned earlier. I have introduced AIRGUIDE II at this time primarily to demonstrate how the basic modular design for AIRGUIDE I permits the system to be expanded quite easily into a more advanced system with improved capabilities. Of course, because of the same modular approach, it is also quite possible to improve or expand the capabilities of AIRGUIDE I as a single-axis system by replacing or adding additional sensor signals and the related circuit modules. For instance, the optical

final design details for the single axis system and also to

developing (see June '79 SPORT AVIATION), merely by

The complete circuit, therefore, was broken into a series of separate circuits, each of which fit on one circuit board and performed some particular functions. For instance, the sketch in Figure 1 shows the rate sensor

circuit board which provides the rate sensor output signal and also provides the speed control for the vacuum

Once the general scheme for the modular system was developed, it was a simple matter to proceed with the incorporate the additional elements needed to expand

this basic system into the two-axes system. A design was developed which made the chassis quite easy to build and adapt to whichever system the builder decided to assemble. A sketch showing the general layout of the chassis and components is given in Figure 2 and a

photograph of the partially assembled prototype unit is given in Figure 3.

heading sensor circuitry could be replaced with the magnetometer circuitry that Doug Garner is currently adding the standard circuit board with the magnetome-

ter circuit components mounted to it. Addition of a VOR tracking signal also appears to be a very simple step.

(Although I have not accomplished this as yet, I have heard from at least one builder who has done so to his unit with very good results.)

For those who might be interested in experimenting with the additional features of AIRGUIDE II, let me SPORT AVIATION 13

The yaw-control function was conceived to eliminate the need for keeping your feet on the pedals or for mak-

ing periodic rudder trim inputs. To make the yaw sys-

c c

ds> L_«j._

Ir-

1 1

©

PVlrtR O

O PWMV

®[®]

h CMS

1 -3

O

ftdtl.

'•©

©

O

Goo -^-

c

(.AT

©I

©

ssa

«® "o"

O

LJ

O*IZv OC

1

"C-1

j

D€H C,/G/Y9

CM

—I— I

1——^——1

PARTS £. Mojoa I

FIGURE 2

explain what they are and how the extra control function is supposed to work. The purpose of the added yawcontrol function is basically to eliminate the stable band of steady-heading sideslip conditions that exits with the single-axis roll system. This band includes the desired wings-level condition for which the heading error will be zero; however, the band also includes a wide range of combined sideslip and wings-banked conditions in which the airplane will maintain a steady heading. Not only do these conditions create an uncomfortable and a high-drag attitude for the airplane, they also fool the autopilot into accepting a heading error of perhaps several degrees without applying a corrective roll input. The only way to eliminate this problem is to ensure that the yawing moment acting on the airplane is zero at the same t i m e the wings are level. This, of course, is achieved with AIRGUIDE I or any other single-axis roll-control autopilot by using the rudder to manually trim the airplane in yaw. Since the yawing moments

acting on the airplane change from time to time during

the flight, it is usually not possible to maintain trimmed yaw conditions throughout the period that you would be

using the autopilot. Changes in power, unsymmetrical consumption of fuel from the wing tanks and the effects of friction in the rudder control system are the primary factors which influence the yaw out-of-trim condition.

14 FEBRUARY 1980

tem operate correctly it was necessary to develop a method for sensing the sideslipping condition that results from the out-of-trim yawing moment. A very simple device used for indicating slip is the inclinometer, the ball, incorporated in the standard turn-slip panel meter. It would have been a simple task merely to attach an optical sensor to the face of this meter in much the same manner as was done for the heading sensor. However, this would not be a very satisfactory method since it would completely block the ball from view and also would result in another set of wires dangling from the instrument panel. An additional inclinometer mounted separate from the panel also could have been used. A more direct method for measuring the sideslip was found, however. It turned out to be a very interesting and simple adaptation of the electro-fluidic angular rate sensor used for the wing leveling function of AIRGUIDE I. During my original development work a couple years ago, I had found that, when the air jet was turned off, the output from the sensor was a direct function of the angle of tilt of the sensor body. However, this output was obtained only if the body was tilted about the axis of the air jet with the jet axis held in the horizontal position. Investigation of this effect revealed that the sensor was responding to the gravity component aligned with the two thermistors in the sensor. In other words, the thermistors were acting as a linear acceleration sensor. I made a sample acceleration sensor which consisted only of the two thermistors m o u n t e d in a hollow chamber at the center of a small block of plastic and found that it worked beautifully. I had no need for the newly-created sensor at that time but I did submit a patent application to protect this very simple but effective idea. However, as it turned out, this unit was the perfect solution to the problem of measuring sideslip for the yaw-control function in AIRGUIDE II. All that use of this sensor required was essentially duplication of the wing-leveler system with the acceleration sensor substituted for the rate sensor. Because of its very small size, the accelerometer could be mounted inside the autopilot chassis. The servo actuator for this new circuit was used to drive a yaw-control tab. If the rudder control system of your system has a very low friction level, the actuator can be hooked directly to a trim tab at the trailing edge of the rudder surface. However, if the friction level corresponds to a force greater than a couple ounces exerted at the trailing edge, the friction very likely will lead to unstable operation of the autopilot in the same way that aileron control system friction will lead to the same problem for AIRGUIDE I. Use of a small auxiliary yaw-control surface, separate from the rudder control system, is dictated for airplanes where the rudder friction is high. There are several features involved in both versions

of the system that I'd like to discuss briefly to illustrate the steps I took to simplify construction, checkout and operation. As mentioned before, the circuit was broken down into different sections which could be built and checked out separately, if desired. All circuit wires and connections were given letter, number and color codes to help identify all parts of the circuit. Practically all of the individual components of the complete circuit were connected either directly at the mounting pads of the

standard circuit boards or indirectly through the color

coded wiring to these pads. Consequently, the pads,

which were arranged in a neat pattern, provided very convenient test points for checking the operating voltages of any point in the circuit. The circuit boards were

arranged on the chassis so that these test points could be readily exposed by simply removing the chassis cover. All of the potentiometers that are used only to align and adjust the system initially, and at very infrequent intervals afterwards, were arranged along the sides of the chassis where they are accessible through holes in the cover so that it does not have to be removed. These a d j u s t m e n t s can be made even w h i l e the u n i t is mounted in the airplane as long as access to the two sides is available. Of course, all of the regular operating controls were located at the front panel of the unit. While pondering the problem of simplifying the fabrication process for the several special circuit boards, a rather u n i q u e idea for preparing the etching masks struck me. I was looking for a way to eliminate the process of drawing the circuit patterns on the copper-faced boards with the special pen which contains an ink resistant to the etching fluid. I found this pen to be very clumsy to use and the results obtained with it to be very unsatisfactory. I also wanted to avoid the use of the photo-resist technique which I found to be very involved and t i m e consuming, although the results were of somewhat better quality than the ink method. The simple solution to the problem was to use a few strips of Scotch-brand Magic Transparent Tape, the type that is very easy to write on with pen or pencil. The strips are laid side-by-side, with edges overlapping slightly, to form an adhesive sheet covering the pattern of the circuit board to be reproduced. A sheet of wax paper is laid over the pattern before applying the tape so that the mask sheet can be easily removed for subsequent application to the copper sheet after the mask is prepared. The circuit pattern is transferred to the mask by tracing the pattern with a normal pencil or pen. After removing the mask from the original pattern and then applying it to the copper-clad side of the circuit board, the pattern lines on the mask are cut using a sharp razor-blade knife and a straight edge. The pattern is laid out with simple straight lines to facilitate the process. Two parallel cuts spaced about 1/32 inch apart are made for each pattern line. The scotch tape cuts very easily and cleanly. The thin strips of tape between the two cuts, representing the copper to be removed, are then carefully removed. The board is now ready for immersion in the etching solution which takes about twenty minutes. Following the etching process, the boards merely have to be trimmed to final shape and drilled to accept the pin mounted components and the hook-up wires. The results from the very first attempt had a "professional" quality, even though a very small amount of effort was involved. This report would not be complete without mentioning a few of the details included in the book that should help anyone attempting to build the system. The book was written on the assumption that the reader would be starting from scratch and making all the fabricated parts himself. (Please note that I have made no plans for s e l l i n g i n d i v i d u a l parts, kits or completed subassemblies.) In addition to the complete circuit schematic diagrams, detailed drawings of all the subassemblies are included as an aid in locating the individual components and making the wiring harness connections. There are, of course, complete construction drawings for the chassis, special circuit board assemblies, and the sensors. The text provides step by step assembly checkout, troubleshooting, installation and operating instructions. The text also includes some theory of operation which should help in the checkout and troubleshooting process. A complete list of parts giving the codes for their locations in the physical layout is given in the accompanying table along with information as to where the parts can be obtained. Practically all commercially-available

FIGURE 3

parts can be bought at the local Radio Shack or equivalent electronics retail store. The book comes unbound with the pages punched for f a s t e n i n g in a 3-ring notebook folder so as to facilitate removal of some of the pages used repeatedly during the fabrication steps. This saves continual page-flipping back and forth between the desired drawings, tables and instructions. As a final help-hand gesture, I included the names and addresses of all the people who had paid a deposit on the book up to the time that it went to the printers. This list included almost ninety names with about a dozen of these for people living outside this country, as far away as France, Yemen Arab Republic, and Australia. I felt that the list would be helpful to the individual builders who wanted to share their experiences or get help from others nearby. I have received many very fine complimentary letters from those who have seen the book and have also received much encouragement to continue the development work. It has been a most emotionally rewarding experience for me to know that this hobby effort has been worthwhile to others. There have been numerous suggestions concerning various features and I plan to continue along several lines as time will permit. Since releasing the book for distribution, a few items have shown up that needed correction or amplification. There is one item that came to light which might be of interest to others who have started to build a unit based on the drawings given in my last report on the Poorman's Autopilot. I learned that the particular style of photo transistor used in the design of the heading sensor may not be readily available. This photo transistor has SPORT AVIATION 15

its sensitive axis at the side of the case. Radio Shack

originally carried this style but now provides an equivalent photo-transistor with its sensitive axis at one end of the case which is opposite the terminal leads. This particular case will not fit in the sensor housing shown in the drawings, consequently I have redesigned this sensor so that this style can be used. The drawings for this sensor are given in Figure 4. For those of you who might be interested in securing a copy of this book, I will refer you to the classified ad which appears in the November and December 1979 issues of SPORT AVIATION. If you do not have one of these available, you can order directly from me by enclosing a check or money order for $14.00 per copy along with your request. Write Don Hewes, 12 Meadow Drive, Newport News, VA 23606.

FIGURE 4 —SKETCH OF OPTICAL HEADING SENSOR

5 - CONDUCTOR CABLE

BODY DRILL #4-40 2 HOLES

PHOTO TRANSISTORS

BULB

CIRCUIT BOARD PATTERN

16 FEBRUARY 1980

DEH

11/8/79