Finding and testing a RELIABLE low power rudder servo for Snoopy's boat 11
Robin Lovelock
RELIABILITY: The rudder servo is a critical element in Snoopy's robot boats. For years we used the Acomms AS-17 because tests showed high reliability of typically 8 to 12 weeks MTF in continuous* use. * typically 10 rudder movements per minute. When the AS-17 became obsolete and no longer available, I have used the Carson "Modelsport part No 50 050 2015.
POWER CONSUMPTION: The rudder servos chosen draws an average of less than 7 mA, with small spikes during movement. The total electronics including Autopilot computer, Globalsat GPS, SPOT Tracking, is less than 40 or 50mA. This enables use of a small boat with limited deck space for solar panels.
Thursday 26th March 2015: Don't feel bad, Cesar in Romania, but your suggestion of trying the HS5086 servo did us a big favour: it reinforced the importance of testing a product before changing the design, based on what is seen on paper. It will also make us look more seriously, in the long term, into a vane-rudder clutch based autopilot: but that may take years before someone designs, tests, AND produces the right product. See when we put this product under 2.5x accelerated 24/7 test above, on Wednesday 4th March. At about 8pm this evening the house stank of something burning: it still stinks now, after having the patio door open 2 hours. I calculate this failure, after 22 days, equates to about 2 months of typical Snoopy autopilot use. i.e. not reliable enough for a trans-Atlantic crossing, where each critical element should have nearer 12 months or more as a MTBF (Mean Time Between Failures). I will be happy to connect any servo supplied to me into the same test rig. The trusty Acoms servo is still running: we will see when that fails. The earlier failure, in the Blog above, gave the same symptoms. See those scorch marks on the paper it was sat on, and is now stuck on. It's a good job it didn't start a house fire !
Today I also got a small mains 5v supply from Maplins, so we could put a complete autopilot here under 24/7 test, looking for a "frozen autopilot failure". The Acoms servo will probably be moved across, to join the autopilot Picaxe 08M2 computer, BR355 GPS, and Text-To-Speech, that tells us that it is working correcting, giving guidance to Bray Lake, as in our "Pub Tests".
Friday 3rd April 2015: The Spot4 tracker is STILL running off it's internal batteries. The second Acomms AS-17 servo failed on it's accelerated 2.5x reliability test. This new one lasted 17 days, equivalent to only 42 days, compared with the earlier, used one lasting about 30. Another one was started on test. This seems as if the AS-17 product only has a MTBF in the ballpark of 1 or 2 months: not enough for an Atlantic crossing. We will continue with the AS17 until we find a better product - with low standimg current. If anyone finds one, with low current ~10mA or less, they are welcome to put it in our test rig.
Sunday 14th June: The AS-17 servo stopped today, having been running since 4th May (according to desk diary). i.e. approx 6 weeks x 2.5 = 15 weeks / 3.75 months MTBF. New As-17 would have been put on same 24/7 test, but test rig mains PSU failed. Battery tests indicate that the servo had failed, drawing more current (and getting hot). This had probably overloaded the PSU.
Friday 24th July: AS-17 servo reliability test started again, at 1430, after pouring some Halfords Cycle oil into the lower part of the casing, in the hope that some might reach the motor brushes. If so, this might make things worse, causing the servo to fail earlier. However, there is a chance that the oil might have some beneficial effect, such as lubrication or dissipation of heat. This was after replacing the failed mains USB 5v supply. The new supply caused the Picaxe/Servo to "hunt", so a 5v NiMH battery was added in parallel. We will see how long this servo test lasts, until something fails !
Saturday 25th July: The servo stopped in less than 24 hours, so it seems the oil made it worse. It was hot, like the others.
Thursday 3rd September 2015: GOOD NEWS ON SERVO RELIABILITY ? I'm grateful to Peter, who recently suggested that our failures of servos, mostly due to overheating, might be due to our 24/7 testing of them at a higher rate of use (2.5x?) than how the servo is normally used in the boat.
Having looked again at our test program SERVOTST.BAS, I see that we are testing at 5x, not 2.5x. i.e. our guesstimate of the rudder servo lasting 1 or 2 months, should be doubled. Also, this accelerated use maybe, as Peter suggests, be the reason for overheating, and the reliability under normal use might be better. For those interested in the detail: our autopilot logic typically operates the rudder about 10 times per minute, of which about 4 or 5 seconds it is central, then moved to left OR right for 1 or 2 seconds. Our SERVOTST software has been operating the servo as fast as it can: about 24x/minute, with each operation being full left AND full right. Hence my realisation that this is 5x rather than 2.5x normal use. ALSO, under typical use, the rudder will only be moved a small ammount, especially if the boat is going roughly in the correct direction. So this full movement in SERVOTST is a more demanding test.
A new test program SERVOT2.BAS was made that repeats a more realistic pattern of 4 secs centre, 2 secs full left, 4 secs centre, 2 secs full right, etc. Note that the new test still uses full throw left or right. The test rig was started at 1000 today, with two servos connected in parallel: our trusty Acoms AS-17, and the Hitec HS5086. The bad news is that this more realistic testing will take longer to show results: i.e. it will not be until we are well into 2016, that we will see if the reliability is better than we have been calculating ! :-)
27 December 2015:SERVO FAILS ! The Acoms AS1-17 servo failed, in a hot condition yesterday. We can do the arithmetic in slower time :-)
Meanwhile Snoopy's boat10 remains ready for the next Atlantic Attempt, probably in early 2016. It has a new AS-17 servo.