WHITE PAPER

Improved RC Aircraft Transmitter and Control System Design

Jef Raskin jef@jefraskin.com
650 359 8588
AMA L88
KE6IGI
October 2003 -- January 2004
Version 4

PART 1: Scope

This white paper discusses how transmitters for Radio Control (RC) aircraft and other devices could be designed to

• make model airplanes easier and safer to operate,
• make better use of the radio spectrum,
• be less expensive to manufacture, and
• make programming and setup easier to learn and use.

PART 2: Electronics

2.1 High Frequency Operation

The RC industry should be striving harder to move to higher frequencies, such as to the 800 MHz or higher bands, which would make receiving and transmitting antennas smaller and more convenient.

High frequencies work best on a line-of-sight basis, which is precisely the condition under which models are usually operated. The popularity of consumer high-frequency devices has brought the price of receiver and transmitter chips and other components down to practical levels. The use of high frequencies also minimizes the size, weight, and expense of passive components.

2.2 Digital coding

Digital coding can eliminate most frequency conflicts. A current example are the low-priced Radio Shack "Zip Zap SE" cars, which allow 6 cars to be run on the same frequency. In spite of the low price, the receivers are automatically set to match the channel set on the transmitter used to charge it. A discussion of some of the technical means that can achieve these goals without increasing unit costs is in Part 5.

With a digital code-adaptive receiver and digitally coded transmitters, fliers at the same field could not interfere with each other. Even flyers unknowingly setting up at sites near other fliers would obtain some protection from interference from those already flying.

Such methods become more practical at higher frequencies because higher data rates can be used. Part 5 also discusses a technique where even non-listening transmitters could be devised so as to not interfere.

At present, RC transmitters such as the Polk Tracker II, which will not turn on if there is another flier on its frequency, are an intermediate step in the right direction.

2.3 Lower Power Via Digital

As digital transmitters need only transmit when sending a frame instead of continually as at present, power consumption can be lower, reducing size, weight, and expense.

2.4 Another Power Saving Technique

Conventional (PPM) transmitters could be designed to operate at much lower power (e.g. 0.5 mW) than normal when the user is setting up an aircraft. This would make it less likely to interfere with another plane accidentally; even at a controlled field (without impound) a person sometimes uses their transmitter to set up without getting a frequency pin.

PART 3: Ergonomics

3.1 Smaller Size

Lithium batteries and today's smaller components give us the opportunity to build lighter and thinner transmitters, making them easier to hold and carry. Small size, as with cell phones, is currently more popular than it has been: At one time large and heavy was equated with quality, this is no longer as true as it once was. There are considerable cost-of-material, warehousing, shelf-space, and cost-of-shipping advantages in smaller equipment.

As anybody who has hand-launched an RC model knows, insufficient attention has been paid to one-handed operation. Future designs should take this need into consideration, but small size alone is a major help in this regard.

3.2 Antenna Human Factors

Helical ("rubber") antennas offer not only greater durability and lowered probability of striking things (or getting into someone's eye), but most important, you do not have to remember to extend them. Many modelers have lost control of a model because they forgot to extend a telescoping antenna. While there may be a slight loss of range compared to a telescoping antenna, the overall safety benefits of providing helical antennas are substantial enough so that they should be standard equipment.

3.3 Patch Antennas

If there is a move to the GHz range, patch antennas built into the case, exhibiting no protruding parts at all (as with most GPS units) may be feasible. At 800 MHz, antennas are small.

3.4 Stick Design and Placement

Potential size and cost reductions in the transmitter case seem to be limited by the need for two gimbals in four or more channel transmitters. The solution, which has many advantages besides the size and cost reduction, is to use a three axis (3D) stick for roll, pitch, and yaw. This is probably the one most important ergonomic factor in making transmitters not only smaller, but -- which is important in our market -- easier to learn and easier to use.

Many are leery of this approach, saying that it has been tried and rejected by the marketplace. However, both the design and the times have changed. In particular:

• 3D sticks in the past had large knobs and had a different look and feel than standard sticks. The designs discussed here do not. They look and feel familiar
• 3D stick-equipped transmitters were only sold at the high end of the market, so very few modelers got to experience their advantages
• The computer revolution has popularized 3D sticks, and many potential and existing customers are already comfortable with them.

If you hold the transmitter horizontally, as is always recommended, 3D stick motion is analogous to the airplane's motion. Picture a tiny airplane glued to the top of the stick. Push the stick forward and the plane on the stick tilts its nose down. Ditto the model in the air. Lean the stick left or right and the plane on the stick banks. So does the model in the air. Now yaw the tiny plane on the transmitter (rotate the tail left or right). If you have a 3D setup, the model in the air follows what the stick does. With the conventional two-stick transmitters most of us fly, you bank the left stick in order to get yaw. It doesn't make much sense and is much harder to learn.

There's a good reason why, when 3D sticks were available, a majority of the top fliers used them in competition.

3.4.1 Experiments Prove that 3D Sticks are Better

I recently did a number of tests using a RC flight simulator and a customized transmitter that can be switched to be a standard Mode II transmitter or a 3D stick transmitter.

Detail of a custom transmitter for testing 3D vs. dual-stick learning. A switch allows it to be used as a 3D stick transmitter or as a conventional Mode II transmitter.

Working both with novices and experienced fixed-wing airplane fliers, I had them fly fixed-wing aircraft on a simulator and compared their ability to track the runway on landing and takeoff with standard (Mode II) sticks or a 3D stick. It soon became clear (and nearly all subjects said so) that it was easier to work with the 3D stick. It was also easier to fly knife-edge and to do hammerheads. There were fewer flying errors except initially with experts who had habituated to mode II, though they quickly learned 3D control. This tells us that a 3D transmitter will be popular with the many newcomers to our hobby.

An especially interesting test was with the helicopter simulator. All of the subjects learned to fly more quickly with the 3D stick. Also, they all preferred it. Better still, the nose-in hover was far less of a problem because when you rotate the stick, the chopper rotates the same way. There is no apparent yaw control reversal as there is with a standard rudder stick.

Research by others back up my observations (See, for example, "Twisting a device is a direct technique for adjusting rotation." Rosson & Carroll, Usability Engineering, Morgan Kaufmann 2002).

3.4.2 3D Sticks Are Widely Used

Currently, most popular computer game controllers have 3D sticks, so many users are familiar with them. Even Microsoft makes them. Because of this, it is now a good time to move to the 3D stick. We need to introduce one difference from the old 3D sticks to assure consumer acceptance. The new sticks would have the yaw pot in the gimbals assembly and not in a lump at the top of the stick. The stick would then look like any other stick in current use, but you could rotate it on its long axis, operating a pot or optical sensor in its base. This would keep the familiar look, allow a seamless transition from aileron-elevator piloting to aileron-elevator-rudder piloting, and decrease pilot learning time and error rate.

For the manufacturer, the three-channel and four (and higher) channel radios can share a case design. The early adopters of this variety of stick will have a marketing advantage due to clear product differentiation and the marketing cachet of having an easier-to-use product.

3.5 Control Design and Location

Present transmitters are often a minefield of protruding levers. It is all too easy to move them accidentally. There are a number of solutions.

3.5.1 Locking Switches

Switches that are used infrequently in flight and not at all during maneuvers should be of the locking type. These have levers that must be pulled to allow the switch to change position.

A custom transmitter with pull-to-operate switches. Note also the four-point harness and the yaw trim control (upper right), which moves in the same way as the yaw control.

3.5.2 Protected Switches

A transparent guard that allows operation of switches that are not normally used in flight yet prevents their accidental operation can be used in place of the pull-to-operate switches

A home-made switch protector, clear so that you can see switch position, with finger holes so that the switches are easily operated.

3.5.3 Avoid Momentary-Contact Toggles

Any button that toggles a flying feature on and off can also cause errors because the flyer cannot tell which way it is set or if it has been activated. One transmitter that has a throttle lock/unlock button causes problems because the flyer cannot tell whether it is in the locked or unlocked position. This, and all similar controls, should be a switch.

3.5.4 Keep All Controls Visible

Controls (for example, the throttle lever) on the back of the transmitter is poor design. Such a position may be easy to reach, but its current setting cannot be seen. On one brand of transmitter not only is the throttle on the back, but it moves right and left, so that the pilot cannot guess which way it has to be moved for high or low throttle. I have seen more than one crash caused by this design mistake. A slide control or lever for throttle should move in the same general directions as the pitch control and be on the front of the transmitter. It should move forward (away from the flyer) for high throttle.

Quasimodal controls buttons do not have to be visible. There is not room here to explain quasimodes or their advantages. It is covered in Raskin, J. "The Humane Interface". Addison-Wesley 2000.

3.6 Gated Throttle

At present, most non-electric models use the trim lever to stop the motor. Much better would be to have a gate (a strong detent) on the throttle lever motion. With the present system, to kill the motor you have to move the throttle down, then move your finger to the trim lever and move it down. In case of an emergency, this is too slow. With a gate, as is found on full-size aircraft throttles (for example, for going to "war emergency power") you just pull or push harder on the throttle lever to get through the gate. With electrics, moving the throttle lever below the gate would also help keep the throttle from being pushed up accidentally.

A doubly-gated throttle might allow, in conjunction with a reversing controller (as are available for cars and boats) a reverse thrust position for braking upon landing, taxiing backward (as some airliners can do), and extreme aerobatics. You would not go into reverse thrust accidentally because you would have to pass through two gates.

3.7 Knob Design

Some transmitters have round knobs for positionable functions such as flap. You cannot tell by feel where such knobs are positioned. Any function on a transmitter that forces pilots to take their eyes away from the aircraft in flight is potentially dangerous. The easiest fix is to use knobs with readily feelable pointers on them. Programmable "beep" points could also be valuable, such as one beep for 15 degrees of flap, two beeps for 30 degrees, and three beeps for 45 degrees.

3.8 Auto Shut-down

For most applications, a transmitter that turned itself off after a certain number of minutes with no stick or switch operation might prevent leaving a transmitter on accidentally where it can interfere with another model or run down its batteries. By sounding a sequence of beeps and putting a message on the display (or blinking a LED) when the auto shut-down is about to occur, the extremely rare case where a plane is being flown without transmitter input for long periods will not be impacted.

3.9 Build in a Stand

Allow a transmitter to be safely put down on the ground or on a table in a vertical or near-vertical position. This is especially useful if the transmitter has a telescoping antenna.

3.10 Include a three-point or four-point harness

A very solid four-point harness is illustrated in the Kraft-built transmitter that I designed shown above. The one-point method usually provided is less secure and the strap can obscure the panel. A proper harness is as effective as a tray, and is both lighter and less expensive. Very small and light transmitters do not need a harness.

3.11 Use Distinctive Switches

Too often all the switches on a transmitter are similar. Errors would be fewer if each had a distinctive look and feel. For example, on human-carrying aircraft, the landing gear lever often has the appearance of a wheel and tire. The same could be done on an RC transmitter.

PART 4: Cognetics

4.1 Trimming A Model in Flight (InstaTrim)

A model that is out of trim is difficult to fly, and it is often hard to operate the trims while dealing with a model that is badly out of trim. A few transmitters have offered "auto-trim" features. The kind on the Ace Micropro transmitter, however, was too slow. Testing has shown that a different method of automatic trimming, which I call "InstaTrim(TM)" (IT) works well.

As the plane is flying, the pilot holds it in trim with the stick. The InstaTrim button is tapped. This transfers the held position to the center position of the stick so that the plane is now trimmed with the stick released. A pilot can use this to trim any one, any two, or all three flight axes at once. One possible position for the button is on the top of the stick, because a 3D stick is held with two fingers to allow yaw control. Otherwise the button can be in a convenient position to be operated by the hand not on the stick.

A transmitter with four channels and automatic trim. The case and gimbals design can be cleaner with no trims and a 3D stick. Note the helical antenna. A transmitter that is small and light does not need a tray or harness. The control in the thumbwell is the throttle.

4.1.1 Flying With InstaTrim

The pilot flies the airplane holding the stick or sticks so that the plane flies straight and level. The pilot presses and holds the IT button and then immediately releases the stick(s) to its (their) center position(s), and then releases the IT button.

The plane is now trimmed. This is far faster and easier than reaching for trim levers. You can trim one axis at a time if you wish, or all at once.

I built this into a computer transmitter I designed a few years ago (1986), and it was well liked by everybody who tested it. After using it, I eliminated the trim levers on the design as they were unnecessary (it is illustrated below).

It is important for automatic trim to work in a certain way and not to cause accidents if it is operated incorrectly. The exact details are given in PART 5.

An experimental transmitter designed by the author. It used automatic trimming only and did not have standard trims and used a 16-key pad so that programming did not require many repeated button presses.

4.2 Dual rates

Using a conventional switch for dual rates causes crashes (I've heard many stories, but I most vividly remember a friend snap rolling a new and expensive model into the ground during a landing because he had forgotten that he had the high rate set and over controlled). Usually, a pilot would like to engage dual rates while performing a maneuver and then disengage it between maneuvers.

For most uses, dual rate selection should not be by a standard two-position switch, but by a pushbutton, but it would NOT be a toggle. The pilot would hold the button during the maneuver and immediately release it to fly smoothly between maneuvers. Because it is a non-toggling pushbutton and not a switch, there is little workload on the pilot, and it cannot be accidentally left in an unexpected position.

4.3 Programming

The topic of programming requires a separate document, because this white paper is already rather long and programming is a subject with considerable detail. But a few general points will be mentioned.

4.3.1 You Can't Use the Display While Flying

The display is one of the most costly and delicate parts of a computer-based transmitter. It adds weight and size as well as cost. Yet it is totally useless when you are flying a model plane because you usually have no time to look at it. In fact, good piloting demands that you keep your eyes on your model at all times when flying it.

Here are a few approaches to solving this problem:

• Design the programming so that a large display is not needed,
• Put the display and programming controls on back,
• Have a separate display and programming panel that attaches to the transmitter, or
• Connect to a PDA, cell phone, or personal computer for programming.

The first method is possible, but difficult to design. In addition, it can make a product look "old fashioned."

The third choice is what was used with the experimental transmitter shown above. A separate box held the display, and was plugged in when the aircraft was being programmed. The transmitter could also be attached to a computer.

The idea of putting the display and controls on back has the advantage of giving the designer a lot more space to work with. Because you do not program as you fly, there is no loss

The large number of pixels in the display and the availability of color and a keyboard make using a PDA or computer for programming a viable option. Better interfaces, not constrained by the limitations of an on-board display and computer, can be provided. There would be room for self-teaching setups and manuals.

4.3.2 Never ask for Confirmation

This is usually done with an "Are you sure?" query or having to use a Save screen. The first doesn't work because answering "yes" (usually by pressing a certain button) becomes habitual. The second is annoying because we seldom make a change by accident but we often forget to use the Save feature. In practice, having a Save feature causes more errors than it prevents and an "Are you sure?" feature does not prevent errors!

Simple works best: whenever users make a change, assume that that they were doing it on purpose, and just store the change. An "Undo" button can be provided to correct errors. This is a controversial issue to those who have not studied interface design, but experienced practitioners know that confirmations do not work. A detailed treatment of why is in Raskin, J. "The Humane Interface" Addison-Wesley, 2000

4.3.4 Use a Keypad

A serious design flaw in many of today's transmitters is trying to do too much with too few buttons. It is ridiculous to have to enter a number by tapping a key dozens of times. If I want to set channel 58, why can't I just enter "58" on a keypad with just two clicks? If I want to set the throw to 75% of normal, why can't I just enter "75" at the proper time?

In general, we try to do too much with too few buttons on most computer transmitters. More buttons does not make a product harder to learn or harder to use, and with conductive plastic dome buttons, the cost for additional buttons is almost nothing and the buttons have good tactile response and are at least as weatherproof as the rest of the transmitter.

On one experimental transmitter I designed, a 16-key pad was used, and many common functions were assigned to well-labeled buttons which could be accessed immediately. For example, you could choose "Ailerons" with one button press, choose "Reverse" with another button press and not have to traverse any menus at all. A safety button had to be held while setups were being made so that pressing buttons accidentally had no effect.

4.3.5 Avoid Menus

If there's one thing we all hate it's having to traverse menus to find a function we wish to use. One partial solution is to have more buttons as described above. It is almost always possible to eliminate all or most menus from a transmitter without losing functionality, but just how it is done has to be determined on a case-by-case basis.

4.3.6 Store many planes

With memory now being very inexpensive, and it taking only a few hundred bytes (at most) to store the setup for a particular model, there is no reason to have fewer than 10 models in storage. From a practical standpoint, there is also probably little reason to have more than 100. But a transmitter that stores only a half-dozen models seems silly.

4.3.7 Organize by Servo Rather than by Function

It is much easier to work with one servo and control surface at a time. That is, if you want to set the end-point adjustment for the elevator, you should navigate to the elevator setup section, not the end-point adjustment section.

4.3.8 Use Longer Labels

There is a custom in industrial design to use short labels (usually one word) or no labels even where there is room for -- and need for -- a longer and more descriptive label. This custom should be resisted.

4.4 Servo Reversing

It is probably impossible to make this kind of change given the millions of servos already sold, but servo reversing should be a function settable at the servo (possibly by reversing a plug or jumper) and not on the transmitter. That way, once a plane was set up correctly, changes to the transmitter program could not reverse servo direction. This rule should certainly be followed in RPVs. Very rarely can a pilot recover if ailerons or elevator functions are reversed. Note that most pilots carefully check control surface direction the first time they fly a plane, but sometimes omit this check thereafter.

PART 5: Technical Details

5.1 Multi-plane Operation On a Single Frequency

There are a number of ways of achieving this. For example, there is the Ethernet strategy of using random delays between short burst transmissions, each plane responding only to signals prefixed with its code. Frame rates would be sufficiently high to make control tracking at least as good as with present equipment. This kind of technology would not require a large number of frequencies in a new band.

Code assignment could be automatic, a monitor receiver in the transmitter would note those codes in use and choose an unused one. If there would be too many transmitters in use, a new participant's system would simply not go on the air, and its display would tell the flyer that there were no available slots. As soon as any other transmitter was shut down, the new flier would be able to fly. Note that the monitor receivers needed in this method can be electronically simple, single-frequency devices.

Receivers would self-set to the code of an adjacent transmitter when the receiver was turned on, but not to transmitters that were more distant. Once set, it would retain the code until turned off. The present "Seeker II" receiver from Polk sets its receive frequency in this way. Setting to a code is electronically easier and far less complex.

Manufacturers and dealers would not have to stock transmitters, receivers, modules, or crystals in different frequencies.

5.1.1 Another Method

Every transmitter can be given a unique code during manufacturing. A 4 byte serial number would suffice to identify over 4 billion RC transmitters. Agreements between manufacturers would assign blocks of numbers to each. These serial numbers could also be used to track stock, would allow identification of stolen or lost property, and can be used to identify a transmitter from its signal, which would answer certain security concerns.

This method would not require a monitor in the transmitter, lowering cost. Receivers would be self-setting to the nearest transmitter as already described.

5.2 Further Details Available

It is easy to show that with reasonable data rates at least 50 models would be able to be flown simultaneously. This information can be obtained from the author.

5.3 InstaTrim Algorithm

This is the method needed to make InstaTrim work properly. I assume that this code is in the main polling loop. It can be readily modified for interrupt-driven code.

Start

If InstaTrim (IT) button is not down, Exit. Otherwise:

Store current stick position for aileron, elevator, and rudder

Start timer

If IT button held for more than 5 seconds:

restore normal operation even though button is still
down (pilot may not know that button is down and you
want to restore normal control; this is also a fail-safe
the button sticks closed through mechanical or electronic failure). Exit.

If IT button is released before 5 second limit:

check if aileron, elevator, or rudder controls
are centered.

If none are centered, Exit (Pilot has made an error or
decided not to trim so ignore the button press).

Otherwise:

Adjust internal values so that center position
of the stick corresponds to the stored positions of
the servos for all channels for which the stick was
centered when the IT button was released.

PART 6: Mechanics

Gimbals can be manufactured with far fewer parts than is customary. With InstaTrim, trim levers and other trim components are unnecessary. Here is a very simple way to make RC gimbals, pioneered by Kraft. The X and Y pivots are the pot shafts. Digital position sensors could be used instead of pots. Centering springs are not shown.

Acknowledgments

I am grateful to the many modelers, designers, and manufacturers who have read and commented on the earlier drafts. Fred Marks has been especially helpful. Cover illustration by David Baker through the courtesy of Maplegate Media. Photographs by the author; all transmitters shown were built, modified, or designed by the author.