Thinking through selecting a servo for the rudder control system

Ahhh, the wonderful world of hobby electronics – Where opinions and data do not correlate.

I’ve just finished my motor monitoring routine, and i’m pretty happy with it. The function will determine if LOCARB has been powered on for the first time, and if so, will wait for a GPS fix before initializing the motor. After that, it will continue to monitor the motor speed every x seconds in the event of a motor stall (due to whatever reason). If the motor has stalled more than three times, the Arduino power cycles the ESC and re-initializes the motor again. If this still doesn’t fix the motor, the Arduino will send an SOS message (I haven’t written this portion as I haven’t received my satellite modem yet). I’m pretty happy with it and have tested the inter-operability of the Arduino and ESC in all sorts of strange power on conditions. I think i’m not even going to throttle the motor speed as it’ll just run off 12-13.6v all the time.

Get on with it…

Anyways, this post isn’t about motor monitoring but about what came after, which introduced a few more kinks in the build process. If you’ve been following any of my recent posts, I haven’t actually mentioned much about the rudder servo used to steer the boat. Its a fairly simple item, but its extremely important to think through as its equally important as motor selection and power supply design. Since I’ve read lots about autonomous boats since I started this project, I’ve noticed a few boats which have just veered off course or had a rudder servo burn out (ig. Seascout). This nudges me to be diligent in selecting a servo, as I don’t really want a fully working boat to just be adrift at sea because of a bad rudder servo.

Why is it so hard?

Have you ever tried to find the life expectancy of any type of RC hobby electronics? It’s very difficult to say the least, it could be because most usage cases do not require such a level of dependability. I’m guessing many items are just manufactured using low quality parts to make for a lower cost item so the manufacturers don’t specify the expected lifespan. After searching the web for “servo life expectancy” and reading through many opinions, the consensus is that it depends on how its made, being used, and who makes it…and also, they last about 200 flights. :/ So back to square one.

I actually purchased a Turnigy waterproof servo a while back as it met my requirements of being low cost and waterproof, and if it didn’t make the final build it would still provide useful for testing. But thinking about its qualities and after finding nothing of consequence when taking it apart (it really did look like a decent quality metal gear servo), I decided to write Turnigy and check what kind of life expectancy I could expect from the components. Here is what I encountered:

Just a bunch of bad code, and a non-functioning submit form. Considering that google actually cached the page in the state I found it said to me Turnigy was in no hurry to fix it. Anyways, I was back to square one…again.

Motor Life Cycle

I decided if I couldn’t get a straight answer from Turnigy, I would have to figure things out myself…A servo is made up of a little motor, some electronics chips, and a potentiometer. It’s unlikely the MCU will die given stable voltage and signal inputs, with the higher probability the parts exposed to mechanical actuations would fail instead. This would mean either the motor or the potentiometer fails. Doing a quick search on the specs of the servo, I see its a coreless brushed motor, which are typically rated from 1000-3000 hours of expected life. Doing some simple math, the life expectancy comes out to about 41-125 days of use. I don’t believe the rating is for 24/7 continuous use, but this isn’t actually all that bad. If I program LOCARB to sail to Hawaii, it should take about 40 days to get there depending on how fast the boat goes. If I can eek out 100 days of life, I may be able to bring LOCARB back to SF in time for a servo swap.

Potentiometer Life Cycle

This brings us to the next item which is likely to fail, the servo potentiometer. From this article, we see that:

  • Precision Potentiometers are generally capable of mechanical cycle life in the range of 300,000 to 25,000,000 shaft revolutions.
  • Panel Control Potentiometers are generally capable of mechanical cycle life in the range of 25,000 to 100,000 full cycles.
  • Commercial Potentiometers are generally capable of mechanical cycle life in the range of 10,000 to 25,000 full cycles.

According to the data above, the lifecycle of potentiometers vary greatly from one type to the next. I assume the servo manufacturer just uses whatever cheap commercial potentiometer they can find to put in their servo (as the potentiometer is not specified as a precision servo), so Ill assume some number from 20,000-100,000 which I think is pretty generous. If we assume that on the ocean, there is 1 full cycle (adding up all the little course corrections) done by the servo every 5 seconds, there will be 12 cycles per minute (conservatively), 17,280 per day, and roughly 518,400 cycles per 30 days. Since I estimate it will take ~40 days to get to Hawaii, I need the potentiometer to last up to between 691,200 cycles to arrive in Hawaii and 1,382,400 cycles to return home. I think it might be asking too much for a potentiometer of questionable specifications to withstand the abuse.

Where does this leave us?

In my search for higher end servos, I found there are robotics servos which DO use precision potentiometers, and even better yet…contact-less encoders (no more potentiometer wear)! These specialized servos vary in build quality and price but offer durability, precision, and position feedback. The issue with these special servos is that they sometimes use serial instead of PWM, so it was “iffy” if they would work with the Pixhawk navigation unit I have. As luck would have it however, it turns out there are a few servos which will work, which are not too expensive!

Two contenders: Price range and features

Relatively inexpensive and using quality components, I was thrilled to see higher end servos which were meant for robotic (Dynamixel servo) and Heli (KST servo) applications. The Dynamixel AX series use precision potentiometers, but the Mx series use contact-less position sensors. They also run at 12Vdc which is what my battery will be providing so I wont need to use any voltage regulation to run the servo. The only issue i’m having at the moment is deciding if I need more torque than the Dynamixel MX-12W provides (its intended use is probably for robotic wheel applications needing higher rpm with less torque). The MX-12W also only comes with plastic gears, a cored motor, and costs about $70. The next jump up is the MX-28T, costs about $230 and comes with a Maxon motor and metal gears… I just looked up Maxon coreless motor life and its still rated at 1000-3000 hours so the only real benefit to using the MX-28T are the metal gears (it also uses more amperage)…

The second contender (which I found after about an hour of research into the Dynamixel servos) are the KST MS series servos (models MS805 through MS3012). These high performance models provide an RC hobby type servo which can operate from 6-8.4v and provide quite a bit of torque. There are no current draw specs that I could find, but the MS servo line offers a nice aluminum case for heat dissipation, metal gears, a brush-less motor (brush-less motors are rated for tens of thousands of hours), and a magnetic position sensor (contact-less) which means if everything was designed correctly, this servo will handle the trip to and from Hawaii without any issue. It also runs like a standard hobby RC servo using PWM signals. The only thing it doesn’t do is operate on 12v but that isn’t too big of an issue. At a $60-$150 price point, they definitely sound like a good buy.

What now?

At this point of the build, its really a toss up between the Dynamixel and a KST servo. At its price point, the MX-12W would be a great option for replacement after each journey and still not break the bank. The only thing holding it back in my opinion are the plastic gears and brushed motor. On the other hand, for $80-$150, the KST MS series provides sturdier metal gears AND a brush-less motor with higher torque (and longer life expectancy). I am seriously considering the KST MS series because of the brush-less motor and metal gears, but I hope the average power draw is around 1amp and lower.

NOTE: There are other producers of high quality robotic servos such as the Volz brand, but they are way out of the price range of LOCARB ($400+).