360 degree servo position

[rickalty 0]

OK, all you electronics Guru's..... I have an idea and want to know if it's feasible.I've been playing with a cheap way to use RC servo's to drive analogue gauges - kinda like the "realcockpits / SimKits" fancy ones. (Still can't get that servo controller to work, but that's a hardware problem)The instruments that don't go all the way round aren't a problem, but those that need to be able to go round and round and round are. A servo can be modified to do that easily, but the problem is figuring out exactly where it's pointing, so as to know which way to move it.My idea is this. I will connect a 360' (non-stop) potentiometer directly to the shaft of the needle. This will tell me where it's pointing, but not very accurately - I've seen 128 steps as being the most accurate I can reasonably expect. Let's say 120, and that I have a plus / minus one step possible error. This means that the signal will tell me where I am to within six degrees. Now, I will also connect another 360' pot to the shaft, but via , say, a ten-one gearing. This will thus give me the option of 120 steps within each 36', for a discrimination of 0.3'. So then I would know from that one that the bearing was one of ten possible exact readings, and the one coarse pot would tell me which one it was.Is this reasonable ?Richard

[Guest PB4]

Hi Richard,This could work but when buying the pots, make sure they have a full 360 course (I don't know if these exists). Generaly so called "360" ptentiometers can run round and round but you get a linear output only on say 340 degrees...(like the simkits ones)If you found real 360 degrees pots let me know I would be interested !! (in that case you would also have glitches problems when around 360 - 0 degs... but that can be corrected by software)Benoit

[Guest SHORT360]

Hi Richard,I have been working almost one year on real instruments. Finally I sold them all to a fellow german cockpit builder because it is a very long process and I had not that time. Finding the good instrument, modifiyng it ( For an HSI it's a nightmare)and finally trimming it after having used the good program with the good controller. I gave up and choosed the Magenta new solution called Old Efis. I would still love the have a real cokpit with real analog gauge, that's thousand times more sexy than an 737 NG or Airbus 320 cockpit. Back to your question. 360

[rickalty 0]

thanks, roger

[Guest MikePowell]

Richard,There are special potentiometers that can be used for 360 degree feedback. Generally you will find them called "resistive position sensors" rather than potentiometers. A subset of these beasties are termed "resistive resolvers". They are designed to give two outputs, one proportional to the sine of the shaft angle and the other proportional to the cosine of the shaft angle. Resistive resolvers have been used for simulated instruments in the distant(?) past. The issue with pots of any type is that they wear out. The wiper continually abrades the resistance element. Most standard pots have a spec lifetime of 50,000 rotations. High end units push that up to a couple million. Sounds big, but if you use a resistive sensor for an instrument that is continually moving, like an artificial horizon, and you use your sim alot,... well, you need to stock spare parts. Current instruments, both simulated and real, use non-contact position sensing like control transformers (selsyns).That said, there are some good resistive sensors ($$$) that will not limit the position sensing accuracy, so you do not have to use a second sensor through gear reduction. The limitation in RC servos actually comes from the control electronics that are designed to prevent "hunting". Hunting is an oscillation about the loop null point. (Oh, THAT's clear, isn't it?) Uhmmm, You tell the servo to move to X degrees and it bounces around near X, but won't stay still due to electrical noise, gear backlash, or general perversity. It may be very near but it still vibrates a bit. Eats powers, wears the gears, pot, motor brushes, etc., all in all, bad news. The fix is a "dead zone". The controller is configured to accept some small region around X as close enough and stops sending power to the motor. Result: no hunting, reduced wear, but also reduced positioning resolution.Bottom line (almost): use a quality resistive position sensor and a servo loop with a small dead zone.If you are interested in servos that truly have free rotation, the simple RC servo type loop is not appropriate. You will need a more sophisticated control scheme. This is why resolvers have sine and cosine outputs.If you are interested in something that is limited to 5 or 10 revolutions, use a multi-turn pot for position feedback.Mikewww.mikesflightdeck.com