Update On The Poorman's Autopilot By Donald E. Hewes (EAA 32101) Aero Research Engineer 12 Meadow Dr. Newport News, VA 23606

S,I NCE THE PRELIMINARY report on my homebuilt

autopilot appeared in the April issue of SPORT AVIATION, I have received letters from and had conversations with a large number of enthusiasts about installing a similar system in their own airplanes. At the Friday Forum that I conducted with the help of Doug Garner this year at Oshkosh, just about everyone of the

about 250 attendees indicated that they wanted to build a unit. (The majority of attendees were active builders of the VariEze, the KR-1 and KR-2, and the Thorp T-18.) Because of this very enthusiastic response, I have devoted several hundred hours in refining and testing the system so that it should be fairly easy to build, align and operate. There are literally hundreds of possible circuit arrangements and design details and I doubt that I have covered them all and found the very best combination. However, I think that the arrangement covered in this article represents a practical and effective system. In the June 1978 issue of SPORT AVIATION. Doug Garner presented recent refinements to his wing-leveler system which is the basis of my system. The rate sensor and pulse width modulator (PWM) circuits are essentially the same in both systems; however, other details of the circuits are different. The system can be used in (1) the manual-trim mode which provides roll trim while the autopilot is not being used, (2) the wing-leveler mode which holds the wings level and prevents the airplane from entering a spiral dive, and (3) the heading-hold mode which locks the airplane onto the desired compass heading. An additional feature is a left-right command switch which permits the pilot to command a steady turn rate or a small heading change depending on which mode is being used. In the flight tests with my BD-4, I have found that the wing-leveler mode would recover the airplane from a 45° banked attitude in a spiral to a wing level attitude in 8 to 10 seconds. Also I found that the heading hold mode would hold the desired heading within ±1° in fairly calm air and ±2 or 3 degrees in fairly turbulent conditions. When I wrote the original article about this autopilot, I had planned to provide complete construction and operating instructions in this present article. As I started preparing this material it was quickly apparent that there was a very large quantity of details which would be needed by some of those desiring to build the system but lacking a general knowledge of electrical circuits. Because this material would require a great deal more of my time to prepare, I have elected to limit the information covered in this article to those details which

describe the changes to the system and how the fin; system works in general. The information concerning circuit layout, component arrangement for easy assembly and compact packaging, and initial balancing of the circuit for optimum operation are not included in this article. Those readers who wish to obtain more information, especially these particular steps, should refer to the note at the end of this article. I believe that for someone who is capable of building their own airplane the system is not too difficult to build, install and operate. The use of only simple hand tools (drill press, disc sander, etc.) is required. If you intend to build the system, you should know how to solder and you should know a little bit about the fundamentals of electrical circuits. Such information is readily available in do-it-yourself type books at most electronic supply stores. You will need a 0-10 volt scale voltmeter and will find an oscilloscope very helpful. Inasmuch as I have had access only to my BD-4 airplane for installation and flight testing, I cannot guarantee that the system will work as well in some other airplane design. However, as an aeronautical engineer, I see no reason why the system cannot be made to perform equally as well in any other design with very little modification. The only modification required should be the sizing and installation details of the control tabs used to produce the aerodynamic rolling moment. System Design A block diagram of the improved system is given in Figure 1. This diagram is similar to that given in the

FIGURE 1

original article except that several features have been added, as noted by the asterisks. In addition several details of the circuit have been revised and a couple errors in the original circuit have been corrected. The following gives a review of these changes and additions. Heading Sensor — Figure 2 presents sketches and notes concerning the heading sensor which detects changes in heading of the airplane from the desired heading. This unit has been changed so that it now has two photo transistors rather than one. This change minimizes the heading errors caused by changes in ambient lighting conditions and greatly increases the range of heading errors over which the autopilot can "capture" the heading. Although the sensor will work using the letters at the cardinal points of the compass, as discussed in the original article, I have added a special "target" on the center of the compass card face which the sensor now tracks. The target is aligned with the cardinal points but the sensor does not actually "see" the letters. The target is much wider than the letters, is closer to the axis of rotation of the compass card, and has uniform contrast (solid white on the black background). Use of the target permitted the sensor to be moved closer to the center of the compass so that the view of the compass markings is less obscured. SPORT AVIATION 41

With these changes, it is now possible to simply turn the sensor to a new heading for changes up to about 30 degrees and the autopilot will automatically fly the airplane to a new heading without losing capture. This capability also minimizes the possibility of losing capture in turbulent flying conditions. With the original system, capture could be maintained only within about ±5° of the heading setting and was frequently lost when more than light turbulent conditions were encountered. It was found that the compass target could be added quite easily by carefully removing the compass front cover assembly and gluing the target in place on the rotating compass card.

Rate Sensor and Servoed Vac-

With this signal in the system, the speed of the motor changes automatically to compensate for changes in air jet velocity. A small potentiometer (pot) permits the speed of the motor to be adjusted to provide the proper velocity of the air jet thereby eliminating the need for a needle valve to adjust the flow as suggested in Reference 2. Command Switch — This optional feature was added as a convenience to provide the ability either to command a small heading change, such as required to compensate for wind drift, or to command a standard turn rate, such as might be used in an instrument holding pattern. The particular type of command provided is determined by the mode in which the autopilot is being operated. This feature was obtained merely by adding

uum Pump — Figure 3 presents some sketches and notes for the rate GLOSS race sensor. A small vacuum pump has been built directly into the body of the rate sensor. Originally I used the regulated vacuum system which I have in the BD-4 for the gyro compass, as a vacuum source for the jet. However, changes in power settings of the engine caused fairly significant zero shifts in the rate sensor output. Tests revealed that the normal flow velocity in the jet is generally in the range of about one-foot per second and that changes in this velocity of only a few percent can produce a full-scale shift in the output of the sensor. Earlier use of a small vacuum pump, similar to that described by Doug Garner in his article in the June 1978 issue of SPORT AVIATION, was unsuccessful because of poor open-loop speed regulation of the electric motor I was using. The FIGURE 2 motor that Doug uses apparently has very good speed regulation but is rather expensive and was not readily available to me. I used one • NOTE-I MNCCT CXHOU5T Tp INL^T OillTH % THICK -UKU-L. which was taken out of one of the RCMQVC COPPER PLQSTIC TUBING several radio-controlled model servos TuiN AjaU-TuBIN' Z g-EQ'D that I had available. This motor is very rugged and well built and costs about $9.00 when bought separately. After some discussion of this problem of jet velocity changes with Doug and also with Dale Walker of Eosmira, Houston, Texas (Dale is HOTEL SENSOR HOUSING CXUQU&T 1 •'4-4O I Mncn - METOL.OR PLASTjC also working on development of this system and has encountered the same difficulty), I decided to use a MOTORpiftMclosed-loop system employing two feedback signals. One comes directly from the input voltage to the pump motor and the other from two resistors that sum the output signals from the two amplifiers tied directly to the thermistors in the rate sensor. Dale has used this latter signal in another manner and suggested that TZflrc SENSOR £ VACUUM POMP DETQILS -— PRILL. CCNTER UOUC TO PKC£>5 FJT ON DR1VC &CLACJ ON it was not of much use for the pump DON W (2.5-76 because of the very low gain involved. However, I have found that the system appears to have a little FIGURE 3 better long-term stability and, consequently, have retained it.

l-t-

rvl —'

42 MARCH 1979

two adjustable resistors and a single-pole double-throw switch. Control Tab — I indicated in the first article that I intended to install a second identical control tab (Footnote 1) on the other wing tip so that the roll control power could be doubled. I have done this and the results showed a very marked improvement. With the control power increased to about 20% of that for the ailerons, the wing-leveler mode is now able to recover the airplane from a steep spiral in about 8 to 10 seconds after the autopilot is switched on. (The controls are left free throughout the test.) Furthermore, the heading-hold mode is able to maintain the desired heading in heavy turbulent conditions. I have performed numerous hard-over tests in which the control tabs were held at their maximum deflection while all normal flight maneuvers, including landings and take-offs, were performed. These tests showed that the handling characteristics and control responses of my BD-4 were essentially unaffected even though 20% of the aileron control travel was used to counter-balance the hard-over tabs. Based on these tests, I am recommending that the control tabs be sized to produce a total rolling moment from 15 to 25% of the maximum produced by the ailerons and that in no case should it exceed a 30% value. The relative magnitude of the rolling moment produced by the control tab is determined by measuring the amount of stick or wheel travel required to maintain straight and level flight with the control tab deflected to its maximum travel. (Be sure to maintain zero sideslip throughout by keeping the ball centered.) Compare this measurement with one-half of the total lateral travel of the stick. The following formula can be used to roughly estimate the size of the tab needed to produce the desired moments for any airplane equipped with standard ailerons:

ST = .20 SA x l Where Sj = total tab areas, SA = total area of both ailerons, YA = spanwise distance from centerline of the airplane to the mid-span of one aileron and Yf = the corresponding spanwise distance for one of the control tabs, as illustrated in Figure 4. The total tab area can be provided by using either one or two tabs.

7 FIGURE 4

Maximum control power can be adjusted, once the tabs have been sized and installed, by altering the arm length of the control-tab torque arms. These arm lengths should be adjusted so that maximum travel of the tabs is produced by a 90° rotation of the servo from its neutral or centered position. In no case should the tabs deflect more than ± 25° from their neutral position because of loss of control effectiveness and large increases in adverse effects. The method for determining the neutral position for each tab is discussed subsequently.

These comments apply primarily to the type of externally mounted tab I have used and I do not know how well they will apply to a tab of some other geometry. Tabs applied directly to the ailerons may be very much

smaller than those I have used and the above formula will not apply. I have had a number of discussions with various people who are building the VariEze and want to use the roll-trim tab on the right wing for the autopilot. I have discouraged this use because this tab is hinged at its leading edge and, therefore, has rather large hinge moments which will impose high loads on the servos. Furthermore, the single tab probably will not be sufficient to provide the necessary rolling moment to both trim the airplane and control it for the autopilot functions. Another factor was recently brought to my attention by Burt Rutan who pointed out that the single tab produces a pitching moment along with the rolling moment because of its placement aft of the center gravity. My recommendation for the VariEze, therefore, is to eliminate the original trim tab and install two auxiliary tabs, one on each wing, in a manner similar to that I have used on my airplane. I recommend, however, that the tabs be mounted below the trailing edge rather than above it. This will keep the tabs out of the low-energy wake which trails behind the wing. Use of the two tabs will eliminate the pitching problem and provide sufficient rolling moment to handle both trim and control functions. Voltage Regulators and Control Servo — A second servo was added to drive the added control tab, consequently a second separate 5-volt regulator was also added to handle the additional current load which would have over-loaded the single regulator. There was no need to add a second 8-volt regulator. The voltage regulator for the servos has been changed from an LM340T to a LM340K which is the same basic regulator but mounted in a TO-3 case that is larger and has greater heat dissipation capabilities than the other unit. It was found that the LM340T regulators ran quite hot when used with the Heathkit high-torque servos. It was desirable to make the change because the regulators had a tendency to cut off occasionally when they became overheated. This is a built-in feature of the regulators that prevents regulator failure but it also puts the autopilot out of commission until the units cool down.

Although the servos work almost constantly when the system is on, I have had no servo failure to date in many hours of operating (probably a hundred or more). Several months ago I was experiencing what I thought was a servo failure and I consulted with the Heath Company about the problem. However, subsequently I found that the problem was due to several other sources. In the course of my correspondence with Heath, they indicated that the high torque unit (GDA 1205-8) was not intended for continuous duty and recommended that the standard units, which are intended for continuous use (GDA 1205-4 or -5), be employed. In view of the fact that the high torque unit appears to have more than enough power to drive the control tab, I believe that either of these other two servos or similar units by other manufacturers will be adequate, provided the aerodynamically balanced tabs (Footnote 2) are used. In fact, I believe that if these units are used, it may be possible to use the original LM340T voltage regulators without overloading them, however I have not made any tests to verify this point. The "T" version is considerably smaller than the "K" version and probably is easier to obtain. (Radio Shack does not carry the "K" series but other regular electronic supply houses generally handle either type.) Circuit Details — A diagram of the revised system circuit is given in Figure 5. Aside from the features alSPORT AVIATION 43

sensed at the same time that the yawing rate is sensed, the airplane will be brought back to a zero turn rate but the airplane will be rolling at the same time. Consequently, the airplane will overshoot and go into a "dutch roll" type oscillation about the wing-level attitude. This problem is solved by tilting the rate sensor to about a 45° attitude so that it will sense both rolling and yawing rates (a rate to the right in either or both yaw and roll should produce a roll moment to the left). The wing leveler will now bring both rates to zero at or close to the same time. The wing-leveler system will tend to hold a steady heading in calm air for a few minutes so that you can safely pick up your charts and read them without holding the stick or wheel (assuming you have trimmed the plane first, of course). This system, however, is charac-

teristically incapable of holding a steady heading indefinitely and any amount of turbulence can cause the

FIGURE 5

ready discussed, several details have been changed. The manual trim has been moved to the input of the final rate amplifier and the switching has been changed so that outputs of the rate and heading sensors can be individually switched on or off. This change eliminated the need for two of the meters used originally. The gains for the rate and heading output amplifiers are now inputed separately to the sensing amplifier. The limiter now consists of two pots coupled to the input of a buffer amplifier. This eliminated the use of the light emitting diodes which were found to be the source of much electrical noise and frequent failures. Finally the single microammeter was connected in parallel with the pulse width modulator rather than in series so as to be less critical as to the type of meter used and to facilitate scaling of the meter. The control panel now incorporates the single manual trim pot (Tj^j), vacuum pump speed pot (Ty), final output meter, wing leveler switch (Sjj), heading-hold switch (Sjj), command switch (SQ), and servo power switch (Sg). All other pots and trimming or adjustable resistors require only occasional ground adjustments and do not need to be on the control panel. How The System Controls The Airplane The flight path of an airplane in level flight is composed basically of a series of straight and curved lines and an airplane should be made to fly these lines primarily by controlling the bank angle. Although some people think differently, this should be done primarily by use of the ailerons and not the rudder. For most airplanes, the rudder will do the job but it is not as effective as the ailerons. In some cases, however, the rudder will not work because of the lack of sufficient effective dihedral to roll the plane. (The plane must slip or skid to the right in order to bank to the left, and vice versa, when rudder-only is used.) Thus the ailerons or an equivalent roll control system, such as the tabs used for my airplane, are the logical methods of controlling the direction of flight with the autopilot. The wing-leveler system functions by sensing the presence of a turning or yawing rate and applying a rolling moment, through deflection of the auxiliary roll control tab, to oppose the turn. On the surface this sounds very simple and logical, but just sensing the turning rate is not sufficient. If a rolling rate is not

44 MARCH 1979

heading to change significantly in a short time. Furthermore, changes in lateral and directional trim of the airplane due to power changes, to fuel useage, or to inadvertent displacement of the ailerons and rudder will induce a constant turn rate in spite of the wing leveler. This turn rate can be overcome only by the pilot offsetting the rate sensor zero in the direction opposite to the turn or by retrimming the airplane. Thus it is seen that the wing leveler is not an "autopilot" in the sense that a true autopilot will hold a given course heading indefinitely. The heading-hold mode brings this system closer to being a true autopilot by sensing the change in heading and adding a corrective roll command as the heading error builds. If the airplane is in perfect trim, the system will hold the airplane on the desired heading and will return to it if a gust disturbs the airplane. However, if the airplane is not trimmed properly, the out-oftrim moments acting on the airplane will cause a small amount of heading error. If the out-of-trim moment is about the roll axis, the system will balance this moment directly without disturbing the airplane in yaw or sideslip. However, if the moment is about the yaw axis, the airplane will sideslip so that the directional stability of the airplane will develop the balancing moment required about the yaw axis. The roll control actually will be opposing the rolling moment caused by the sideslip and the effective dihedral of the airplane. Generally, this sideslipping will not be very large but it may be objectionable. Thus it can be seen that even the heading-hold mode does not provide a true autopilot capability and that some pilot attention may be required, at least occasionally, to maintain the airplane in near-trim conditions. Have heart, though, the heading-hold mode will do a very good job of substituting for one of those very sophisticated commercially made (and expensive) autopilots. Speaking of sophistication, I added the optional feature of the command switch SQ which merely adds a small out-of-trim signal to the system so that it is biased in one direction or the other. If the system is in the wingleveler mode, the switch will cause the system to perform a specific banked turn, depending on the setting of the adjustable resistor, as long as the switch is held in the desired direction. The turn is not necessarily coordinated and some sideslip will develop. As soon as the switch is neutralized, the system will return the airplane

to the wing level attitude. If the system is in the headinghold mode, the same switch will cause the system to change heading by a specific amount, and so on. One peculiarity of my airplane is that it has essentially no geometric dihedral, and as a result, the un-

coordinated turn produced by this system causes the fuel in the partially filled wing tanks to shift in the direction

of the turn. This out-of-trim shift tends to hold the

airplane in the turn after the turn command switch is returned to zero, especially if the turn is held for more than a few seconds. This is not a problem with the heading-hold mode because the turns involved are very small and are short in duration. An airplane with only fuselage tanks or with some geometric dihedral in the wings should not have this minor problem. Installation and Flight Adjustments

Following bench checkout of the system to adjust it

for proper alignment, the system should be installed in

the plane with the rate sensor mounted right side up with the appropriate end of the sensor facing aft and tilted

downward about 45° (not critical). The proper sensing

of the heading sensor also should be checked by rotating the compass card away from the neutral position in which the target is directly beneath the two photo transistors.

With the heading-hold switch (Sj-j) on, the meter nee-

dle should deflect in the direction of the bank that would be required to recenter the compass card if the airplane were flying. (Be careful, this takes a little bit of thought

doing it on the ground. It's easier doing it in the air, but

then you may have to land to reverse the heading output.) If the output is backwards, it can be reversed by

rate amplifier C should be slowly reduced until the oscillation gradually disappears. This point represents the maximum usable gain for the rate signal. If the oscillation is not evident, the gain should be increased until the oscillation is evident and then the gain reduced as before. Now switch to the manual-trim mode and let the

plane diverge in the direction of greater spiral tendency. As a moderate spiral develops, take hands off the controls, switch the wing leveler on and observe the automatic recovery which should take only a few seconds

to return the airplane to the wing level condition. If it doesn't actually return it to full level conditions, it should at least keep the spiral from diverging further. If it doesn't even do that, would you consider building a new airplane? Heading Hold — The next adjustment is the heading gain which can be adjusted after switching on the headinghold mode and setting the sensor index a few degrees off from the current position of the N or S letters on the compass card. The autopilot should automatically bank

the airplane in the direction of the setting, that is, in the direction to place the target directly under the sensor (the N or S should be aligned with the sensor index). If the airplane banks in the other direction, you goofed in the ground check (you see, I told you it was tricky);

so land and switch the connection at amplifier E.

Okay, if you didn't goof, the airplane will automatically

switching the output of D through the 50 KOHM resis-

turn out of the bank as it approaches the new heading, or it may overshoot and bank back again and then continue

flight in smooth air after short lengths of coarse thread

some more; just back off on the gain G j-j at amplifier D until the oscillation ceases. On repeated attempts, the plane should bank smoothly to the new heading and then level its wings as the heading is reached without an overshoot. If no oscillation is experienced at first, increase

tor to the + terminal of E. Roll Trim — The final adjustment steps are done in

or fine yarn have been attached to the trailing edge of each tab. Establish straight and level flight at cruise conditions and with the system operating in the

manual-trim mode, note the angle of the trailing threads relative to the angle of each tab. Use the flight

controls to maintain straight and level flight if needed. With the meter needle centered (consequently the servo centered), the threads should be aligned with the tabs. If they are not, land and re-adjust the pushrod lengths and

repeat the flight test until "eyeball" alignment is achieved. Be sure the servos stay in their neutral position throughout this operation. Next, release the controls with the airplane controls trimmed and the meter centered to check hands-off trim. Retrim the airplane if necessary but do not move the servo control tabs. If the airplane has ground adjustable

trim tabs and is not in trim at this point, land and readjust them until trim is achieved. If aileron trim is

lacking, the airplane will have to be trimmed in roll by using rudder trim alone and accepting the resultant sideslip. An alternate to this sideslip condition is to offset the system using the manual trim control of the autopilot system. However, this will make the system

unsymmetrical with possible loss of effectiveness in one direction. Of course, a compromise of a little bit of both the two trimming methods may be the best solution to this problem. Wing Leveler — The next step is to switch to the wingleveler mode and check first to see that the meter is centered as a straight flight path is maintained. Retrim with TJ^J, if necessary, until the airplane will maintain essentially straight flight with hands off the controls. If straight

and level flight is achieved but the needle is not centered, the airplane is out of trim in roll. Retrim the ailerons, if possible. If straight flight is achieved (no turn rate) but the wings are not level (ball not centered), the airplane is out of trim in yaw and perhaps roll as well. Retrim rudder and then ailerons, if needed. At this point, the airplane may be oscillating in roll

at a constant amplitude, and if so, the gain GR of the

in that good old dutch-roll fashion. No, don't tilt the gyro

the gain Gpj until the oscillation does show up and then

back off as before. If it doesn't show up by the time you

get to full gain, forget it and leave it there. You got your

money's worth already! Command Switch — The final adjustments have to do with the resistors attached to switch SQ. You have the choice of adjusting these for either a desired turn rate or a heading change. You can't have both, or at least not

very easily (you might if you want to go back and change the gains, but I don't think you really want to do that).

Anyway, simply go to the mode you want and throw SQ in one direction. Adjust the appropriate resistor until

the desired turn rate or heading change is obtained. Then

switch to the other direction and repeat with the other resistor.

You now have a single-axis roll-command electro-

fluidic autopilot. HAPPY FLYING!

PROBLEM AREAS

There are three basic problem areas that have not as yet been solved. The first of these has to do with zero drift in the rate sensor circuitry. This is believed to be due to thermal changes either within the sensor itself or the associated amplifiers. The amplifiers are operating at a large gain and therefore are very sensitive to minute changes in the circuitry due to thermal effects. The second problem also has to do with temperature and was brought to my attention by Dale Walker. He

found that the rate sensor would not operate for about

15 to 30 minutes at room temperature after leaving it in the freezer of a refrigerator until it was cold soaked. In discussing the problem subsequently with Doug Garner, it was decided that the problem had to do with the fact that the air and the interior of the sensor body have to SPORT AVIATION 45

be above a certain temperature for the thermistors to operate correctly. Otherwise the thermistors dissipate their heat too rapidly and saturate the amplifiers A and B.

A solution perhaps to both problems would appear to be the use of insulation around the sensor and a small regulated heater. However, I have not pursued either of these problems as yet because they have not seemed to be at all serious in actual operation. I have flown in some freezing as well as torrid weather without noting serious adverse effects of temperature, however, the cabin has usually been at fairly comfortable temperatures at these times so the rate sensor was not actually exposed to the extreme temperature. The third problem has to do with radio-frequency induced voltages whenever the communications transmitter is keyed. These voltages cause the amplifiers to saturate and drive the servo hardover as long as the mike key is pressed. The system resumes normal operations whenever the transmission is completed. I have been merely switching off the servo switch, Sg, whenever I use the transmitter for anything other than a very brief message. The solution to the problem obviously is to provide shielding which has not been done. However, I have not been bothered very much by this problem and have been too busy to correct it. POSSIBLE FUTURE DEVELOPMENTS

There are several intriguing avenues for future development of this general system. The first is the development of a simple magnetometer which will eliminate the need for a compass and the heading sensor. This magnetometer currently is under development by Doug Garner, who fathered the whole idea of the simple fluidic autopilot in the first place. Doug says that he has several approaches going and I expect you will be hearing from him directly sometime in the near future. Next, there is the use of a sensor to detect sideslip and apply a corrective yawing moment through a small auxiliary rudder tab. This will help overcome some of the out-of-trim conditions caused by changes in power settings and shift or usage of fuel in the wing tanks. It is very easy to see how this single-axis system can be applied almost directly to the pitch axis so as to provide a form of speed holding capabilities. My plane has stable operation in pitch and can fly indefinitely with hands off in non-turbulent air. However, it can be easily disturbed when turbulence is encountered. Surprisingly enough, having the single-axis autopilot operating makes me more conscious of pitch difficulties, even in calm air. The addition of a VOR-holding mode is a very likely development which I already have looked into to some extent. All that is needed is an effective way of coupling into the OBS drive signals. I believe that the VOR signal can merely be added to the heading-hold mode so as to provide a heading bias signal. Different forms of roll, yaw, and pitch control devices is one area which needs to be developed further. The use of the small auxiliary tab suited my airplane very well,

but it is not as applicable to other designs. Certainly, the direct coupling of the tabs onto the existing aileron, rudder and elevator needs to be pursued because it tends to be more universally applicable. However, the use of various types of spoilers seems to have some merit. One of the prime considerations in the selection of a control system is the requirement for low torque in order to

utilize the low cost model servos. Some types of spoilers

will operate with essentially no aerodynamic hinge moments. Finally, there are some alternate types of servos that should be explored. The auto industry uses a servo in 46 MARCH 1979

the speed control system that may be adaptable. There may be some types of small pneumatic or hydraulic servos which can be obtained inexpensively and adapted to this use. CONCLUSION

In closing I merely want to comment that this effort has been a hobby activity. Without receiving any monetary compensation, I have derived much enjoyment and satisfaction from it and have received a liberal self-education in electronics and opera.ion of autopilot systems. But most of all, I have experienced the great pleasure of meeting and conversing with hundreds of fine and interesting people whom I, otherwise, would never have encountered. This is one of the fine benefits of this country of ours and being a part of the EAA. NOTE

Additional information concerning the simple electro-fluidic autopilot can be obtained from the two articles mentioned in this article. Also, there are two known suppliers of kits for the autopilot: Eos mira, 1108 Hedwig Green, Houston, TX 77024 and Omnics Corporation, P. O. Box 371, Ansonville, NC 28007. As noted in the introduction of this article, I have collected a large amount of detailed information dealing with construction and operation of the system. I elected to leave out a large portion due to the sheer quantity of material and the size of the effort required to assemble it in a suitable form for publication. Furthermore, I had no way of knowing just what demand there would be for this information. I am prepared to expend the additional effort and money required to compile a complete booklet with detailed construction information with sketches and photos, covering the sensors, circuit board design and fabrication, component layout and assembly, wiring procedures and alignment or balancing procedures. Also included will be the information presented in this and my previous article so that the booklet will be a complete compilation. Before doing this, however, it is necessary for me to know that there will be a sufficient demand of at least 200 copies to warrant the effort. I estimate that the cost of this booklet, which would be printed on high quality paper and bound with a heavyduty cover, should be between $8 and $12, with cost of handling and postage included, depending on the final number of orders. If you are sincerely interested in this booklet, please write to me with an enclosed self-addressed and stamped envelope so that I can contact you subsequently. Please include a check for $5 as a deposit, as well as any comments on the information you would like to have included. I will try to answer any specific detailed questions that you may have at the time but would prefer to handle most of these in the booklet. I will return a letter indicating the final details of the booklet. I expect that it will take about a month to determine the demand and to return the letter, and about 3 to 4 months to prepare the booklet. The deposit will be returned if the booklet is not published. Footnote 1 — A more technically correct terminology is auxiliary roll control surface but I prefer to use "tab" for convenience. Footnote 2 — An aerodynamic balanced tab is one which has some of its area ahead of its hinge line. The amount of area, generally ranging from about 10% to

30"# of the total area, depends on the amount of balance desired and the specific geometry of the tab and its location on the wing.