Test of Servos for Free Flight Models

In comparison with RC models, there are higher requirements for some servo parameters in free flight class. Very important is temperature drift of servo position. In case of RC model, there is no problem to trim the model during the flight. In free flight, there are fixed servo positions, and even a small deflection can cause a poor flight result. Therefore, this test is focused mainly on servo temperature drift.

1   Servo Motor Principle

There are two kinds of servos for aircraft models on the market, analog and digital. But the general principle of both kinds is very similar. So, how it works inside? There is a DC motor, gears for servo lever and electronic circuit for feedback control.

Servo position is controlled by PWM (Pulse Width Modulation) signal. The period should be about 20 ms. The servo lever position depends on width of PWM pulses. For example, 1 ms corresponds to fully left deflection, 1.5 ms to center position and 2 ms corresponds to fully right deflection. But the sense of direction may vary by manufacturer.

PWM (Pulse Width Modulation) signal for servo control.

1.1   Analog Servos

There is a RC circuit that generates PWM pulses which corresponds to the real servo position. The RC circuit consist of a fixed capacitor and a potentiometer which is connected to the servo lever. So, the resistance of potentiometer is changing with the servo position. Then an analog circuit compares the input PWM signal (with desired position) and the generated signal (with real position). If there is a difference, it makes an action to the DC motor to eliminate that.

The resistance ratio of the potentiometer itself is temperature invariant by principle. The problem is with capacitor which depends on temperature. But there are some capacitors with very low temperature error, or two capacitors with complementary characteristics can be used to compensate the temperature error.

1.2   Digital Servos

Digital servos contains also potentiometer which measures servo position. But the potentiometer voltage is connected to the A/D converter of some MCU (microprocessor). Then the MCU compares the input signal and the real servo position and makes an action to DC motor.

It sounds very easy, but there may be some problem with reading the length of input PWM pulse. If a crystal oscillator is used for MCU timing, there is no problem. But some digital servos doesn't contain crystal, and clock is given by embedded RC oscillator which has strong temperature error. Then, the same input signal is read differently at various temperatures.

2   Measuring Equipment

I made a prototype board to control servos by PWM signal. It contains LCD, buttons and potentiometer. There is a temperature sensor that can be attached to the servo body. I used 180 mm long arm, which is attached to the servo lever, to measure servo deflection.

   

3   Servos in the Test

I chose both analog and digital servos that looks suitable for free flight models. It means minimal dimensions and fast motion.

4   Measured Parameters

I measured a pulse width for center position and an increment needed for 60 degrees right deflection. It is evident that some servos have an inverse sense of motion. The gear backlash is a deflection, you can see on the servo lever, when the servo is switched off. The centering represents the difference when the servo returns to the center position from the left and from the right side. These values are measured at the end of the 180 mm long arm and evaluated in degrees. The supply voltage for all measurements is 5.0 V.

5   Temperature Drift

I put the measured servo to a freezer for several minutes to achieve low temperature. For high temperature, I used a hot air gun with temperature regulation. The temperature value isn't very accurate, but it is sufficient to see temperature drift in the servo position.

The temperature drift of analog servos is caused by temperature error (volatile capacity) of used capacitor. Only two servos (Hitec HS-56 and HS-65) have no measurable drift. It is probably achieved by precise foil capacitor.

   

Disassembled analog servo Hitec HS-65HB.

In comparison with excellent Hitec servos, there is GWS NARO with strong temperature drift almost 9 deg. It is 7.5% error at 120 deg trajectory. So, this servo is totally not suitable for free flight models.

I was surprised by temperature drift of Saturn digital servos. So, I looked inside Saturn S-62D. On a very small board, there is a 8-bit microcontroller PIC12F675, but no crystal oscillator!!! So, I searched the datasheet for this PIC. I saw it has an internal oscillator with accuracy, at that temperature range, about 2%. It corresponds to my measurements.

   

Disassembled digital servo Saturn S-62D, no crystal inside!!!

6   Conclusion

This test shows that there is a huge difference between servos. There are some analog servos which have zero temperature drift (Hitec HS-56, HS-65). On the other side, there are some digital servos with significant temperature drift, because of missing crystal oscillator (Saturn S-62D).

Based on this test I recommend the following servos for free flight models.

Hitec HS-56HB, Hitec HS-65HB

Very precise analog servos. HS-56 is faster and HS-65 is stronger. Ideal for stabilizer or for wing wiggler.

Futaba S3154, S3157

Very fast and miniature digital servos. Contains crystal oscillator. Ideal for rudder. It may be used also for stabilizer and wing wiggler, but servo shaft is not so thick as Hitec servos have.

Servo S3157 is not in the test, but it is a newer version of S3154 with the same parameters. It looks better, because of thicker shaft and holders.