Some conclusions about servos

In a recent job I got to test 50 servos from several manufacturers (Hitec, JR, Futaba, GWS, Volz, FMA) for a UAV manufacturer. Perhaps this summary will help. I set up a 100 hour test with the servos cycling +/- 45 degrees once per two seconds lifting a 1 kg load on a 10 mm moment arm. In normal flight conditions there would be many small deflections rather than the continuous sweeping I used in this test. The 1 kg.cm torques are fairly representative of flight loads on a small UAV or RC model.

All tests were done at 5.00 volts with each servo having its own 1 amp power supply and a 1000 uF cap to supply a short time peak current a bit over an amp. Note that digital servos can pull considerably more than one amp when stalled but running in this test they drew less than one amp.

• There is NO true standard in the hobby RC servo industry except that 1500 uSecs was the centre position on ALL servos tested.
• 1000 to 2000 uSecs works with EVERY servo tested. 900 to 2100 or 800 to 2200 worked with SOME analog servos but caused almost instantaneous destruction with newer digital servos. Digital servos draw 4 or 5 times more current in their quest to get to the set point faster. If the internal gearboxes, or a binding linkage, prevents this, they go into very rapid meltdown internally.
• Just because your new fancy transmitter lets you dial in 120% throws, do NOT assume your servos will comply. Digital ones may die.
• Rotation angle is not standard, even within one maker's range. 1000 to 2000 uS can be +/- 45 degrees, +/- 60 degrees or even +/- 90 degrees.
• All nylon gearboxes typically have lowest backlash and remain tight over 100+ hours of cycling. Metal gearboxes wear substantially over 100 hours.
• Metal shaft/metal gear servos like the Futaba micro servo are very robust but the metal shaft conducts interference into the pot and the servos chatter with nearby RF interference - bad news on a UAV with on-board transmitter until you fit ferrites & bypass caps.
• Hitec servos, in general, have motors far too powerful for their gearboxes and instantly strip gears if linkages bind. Other servos may do this but Hitec stood out as poor in this regard. Volz failed in the shortest time under load tests from electronic failures, not gearbox troubles.
• Ball bearing servos performed no better than servos with the output shaft just rubbing on the plastic case. Observed case wear on the no bearing Futaba and GWS servos was negligible.
• Driving the servos with 50 Hz refresh rate gave 100% of makers specs for response time and torque. Driving faster (only went to 60 Hz) did not improve response times. Going down to 25 Hz refresh rate worked for all servos tested but holding torque and response rates suffered.
• Lowest power with highest speed was to drive the servos at 50 Hz rate until into position then drop the refresh rate back to 10 Hz. Only works for lightly loaded servos however.
• Price was absolutely unrelated to lifetime. The most expensive (Volz) failed first (all three of a sample of 3 at 5, 22 and 35 hours).
• Cheaper servos have more backlash when new and tended to have highest backlash at end of test. Backlash was very small in every servo tested and your linkages are guaranteed to have more slop than the servos.
• Digital servos have a genuine 1000+ steps between 1000 and 2000 uSecs. Analog servos gave 500+ steps from cheapest to most expensive.
• How long do servos last??
  Unless you physically stress them by manually moving the output arms, you can be almost certain to get 75 continuous hours. That is probably plenty for normal RC hobby flying but for UAV use I would suggest replacement at 50 hours maximum. Your mileage will vary depending on loads, vibration (the wiper on the feedback pot can gouge a pit into the track in high vibration), power supply voltage and current limits, temperature extremes, moisture ingress, etc, etc. Lifetime could be as low as 10 hours if you insist in pulling the full rated torque and loads out of the servo with every movement.

Artikeln ovan finns refererad på Internet under diverse diskussionsgrupper. Har dock inte återfunnit originalet varför det inte är någon garanti att artikeln är komplett eller manipulerad. Innehållets korrekthet får författaren stå för (under förutsättning att texten är oförändrad). När originalet egentligen skrevs är också höljt i dunkel. /Webmastern