RC Aircraft Failsafe Servos

[dknguyen]

I've been trying to squeeze RS485 serial interfaces, current feedback and all that kind of stuff onto a PCB that is small enough to fit inside an RC servo so that it is more suited to a computer's control than a human's radio transmitter. BUt it's proving to be very difficult, not to mention you have to open up the servo and sometimes the PCB Is glued inside and not removable.

But then I started thinking...do you really want that on something like a plane? Like do you want a motor on the plane to cut-out if you are driving it too hard? It's not like a car where the vehicle would just safely stop.

The problem with an RC servo is you tell it where to go, and it tries to go there with all its might. But the servo's control loop is stuck inside the servo and the computer cannot read if the servo was actually ever able to get to that position or not (the main reason I wanted to hack it in the first place). THen there's all the stuff about current limiting etc.

But suppose you did detect the servo wasn't able to generate enough torque to reach the desired position, what would you have the computer do in such a case? Or if you detected that the servo was drawing too much current? Would you just shut it off to let it cool down? Or would you keep pushing it to make sure the plane doesn't go out of control? After all, not a ground vehicle, but a plane that can crash, and an improper attempt to save a servo might result in the loss of the plane.

 

[jpanhalt]

But then I started thinking...do you really want that on something like a plane? Like do you want a motor on the plane to cut-out if you are driving it too hard? It's not like a car where the vehicle would just safely stop.

No.

But suppose you did detect the servo wasn't able to generate enough torque to reach the desired position, what would you have the computer do in such a case? Or if you detected that the servo was drawing too much current? Would you just shut it off to let it cool down?

No. In addition, may servos are "coreless," which means their resistance to motion when "shut off" is almost nil.

Or would you keep pushing it to make sure the plane doesn't go out of control?

Yes. The first rule of all of aviation is "aviate." That is, fly the airplane first come hell or high water. Communication, navigation, everything else is secondary.

Automated systems need a reasonable alternative, or they can create more problems than they cure, as you surmised.

Piper used to have an automatic landing gear in its Arrow (it may still) to keep a pilot from landing gear up. Problem was that in the right conditions, it could increase risk. The override was manual (at least there was an over ride), but that also distracted the pilot. It has not been a very popular addition in the overall market. Warning lights/buzzers are preferred.

Good luck. John

 

[dknguyen]

My job just got easier I guess. THe only thing left is whether I want a serial interface enough to modify it or stick with the regular PWM interface.

 

[jpanhalt]

A serial interface might be worthwhile. A buddy and I have considered doing that, but really have not seen a pressing need to do it. If electrical noise, length of wiring and weight are not problems, though, PWM works just fine, as you already know. Unless you build new servo electronics, you will still be stuck with PWM at their end anyway.

John

 

[Nigel Goodwin]

A properly designed and constructed plane shouldn't have such problems, the servos have more than enough power to move the control surfaces. If the servos can't move them, there must be some damage, so I don't see as it makes any difference worrying about what you do with the servos, the plane is probably crashing anyway.

What is more important is a fail safe on the radio system, so if the RC batteries go flat, or the radio link fails, it brings the plane down to the ground in a reasonably 'controlled' fashion - commonly by fordcing the rudder hard one way, and spiralling the plane down.

 

[Pommie]

Are you aware of the **broken link removed**.

Mike.

 

[jpanhalt]

Pommie: No, I wasn't aware of that project. Thanks for the link.

Nigel: Your description of what happens when a servo fails is a bit dire. On rare occasions, I have had servo failures and controls get stuck or become floppy (usually caused by a gear failure in the servo). The only time I have lost an airplane is when that happened to the throttle and elevator in the same flight. In other instances, despite the stuck control and inoperative servo, I was able to land safely and reasonably close.

As for failsafe, they are a bit more sophisticated today than the scenario you describe. A spiral will more likely damage an airplane than a crash with wings level. I would recommend cutting power (if possible -- I fly mostly gliders) and make a shallow, flat turn so the wings will be level at the time of impact.

Best regards. John

 

[mashersmasher]

***nub alert!***

maybe you could rig a parachute to go off with a bb gun cylinder (they are under 3''). stick it in a peace of pvc sticking out the back and trigger it with a servo.

 

[jpanhalt]

I don't think I would want that to happen every time there is a momentary loss of radio signal.

I think the real value of ballistic 'chutes is when an aircraft gets into an un-recoverable attitude. That rarely happens with models. In my experience, "lost" airplanes usually do fine, unless they hit a tree.

Some modelers plan ahead and put a plastic trash bag in the tail cone --makes it easier to carry the pieces back to the car. I'm not such a pessimist.

John

 

[dknguyen]

Yeah, I've heard of OpenServo. Openservo doesn't run on RS485 though, which is what I need.

I was going to have a computer translate the PWM signals from the receiver and modify them since I need to modify the PWM signals anyways (heading gyros and stuff), it's not much more effort to just modify the signal all together. THe work lies in modifying the servo.

masher: I don't think a parachute would be too safe on a chopper lol.

 

[blueroomelectronics]

Why RS485, are you running long cables?

 

[dknguyen]

Why RS485, are you running long cables?

No, but everything else is going to be RS-485 anyways (compass, IMU, horizon sensors, throttle motor driver, and it'd be nice to onyl have one bus running everywhere.

Or a more solid reason that has just arisen is that I want to use some dsPIC blocksets for simulink and due to the cost of buying it, I probably can't buy the blockset that has a large number of, or no limit on the number of pins used. RS-485 lets me use 3 pins for servo control (which is already combined into all the other module communications anyways) while PWM would require me to use an extra 5 pins.

THe quick and dirty way would be to make an RS-485 motor controller, but you know, why take the extra effort to go half way when you plan to go all the way at some point.

 

[blueroomelectronics]

How big is this plane?
You might like this fellows stuff.
https://www.thesiliconhorizon.com/

 

[dknguyen]

Right now the most promising platform is a T-Rex 500 from Align. It's a 1.5-2kg electric helicopter platform.

RIght now, I'm trying to gauge my ability to learn the quaterion math so I can use gyros to keep track of the ground's position. If not, there's no point in going on. Horizon sensors are really easy but I think a helicopter is far too nimble for them to work on their own due to their response time (like in some fixed-wing airframes designed for stability). A lot of that leads to whether or not I can use Simulink to do the quaterin math for me and maybe make things easier. One of the primary things is the blockset right now. A pin limit of 10 costs about $150 euros which I think is exactly what is needed for the IMU. No pin limit is $970...quite a bit. I dunno. THat's what I'm trying to decide right now. I suppose you could argue everything else costs so much anyways that it makes little difference to just get the complete thing.