Servo load sensor

[ombrastein]

Im working on a hexapod. I need to know if any of the servo motors(standard hitech 422 and 485) get heavly loaded. Im thinking to use this to sense if any legs crash into an obstacle, kinda like a bumperswitch. Pluss it would be good to measure automaticly exactly what load the servos are pulling.

The idea i have is to use a 0.1 ohm resistor between motors and ground, thus generating a small voltage drop over this ersistor. the servos never pull more then one or two amps, so teh voltage drop is realy insignificant.

Then i can use an A/D, either directly or thru a differential amplifier to measure digitaly the load each motor is sensing (basicly measuring the current).

Im thinking i will need to measure diferentialy, and stick the wires as close to the resistor as posible, so the internal resistanse of the wires dont contaminate my emaruments.



Any thoughts or ideas about this? Anyone ever done anything similar?

Cheers,
Magnum

 

[jpanhalt]

Servo drain is a concern for lots of reasons in model aircraft and essentailly what you describe is standard practice. Even units for in flight testing are available.

Those systems are based on looking at the total current drain to the servo. That is, there is no modification or wiring done inside the servo case, as current to the amplifier part of the circuit is relatively constant. John

 

[ombrastein]

do you know of any that is lightweight and would be easy to integrate? If i can save myself the trouble of doing it myself, that would be nice ...

I was thinking to add 12 01. ohm resistors, then dif amps on all of these, using a MUX or somthing ... not quite sure yet how to do that part ... need to think on it some more

One alternative is ofcourse to do loadsensing on all the 12 servos at once, and then just test to see wich one is overloaded ...

Im thinking i can reasons that based on the gait of the robot also (obviusly a leg that is standing still is less likley to suddenly experince increased load then a legg that is actualy moving). I will have to match the load readings to the gait anyways, as when robot is pushed forward the load will obviusly increase.


If you know of any ready madde solutions, that would be easy to interprit with a microcontroller, that would be great.

 

[jpanhalt]

Unfortunately, I cannot give you a specific literature citation for the load testers. The two magazines where I would have seen them are Radio Control Modeler (RCM), which may no longer be publishing, and Model Aviation, which is published by the Academy of Model Aeronautics (AMA) in the USA. You may have better luck finding an index for Model Aviation; although, neither magazine is very well indexed, and that statement is particularly true for advertisements. I have seen advertisements in Model Aviation in the past year for system monitors with real-time or stored data. Some of the motives behind monitoring are to detect control binding or excessive flight loads, which seem very similar to what you want to do. The ads would most likely have been in the section for new products.

I did find one construction article by George Steiner in RCM, November, 2001, pages 42-46. That project was a recorder for servo loads. At the time, his e-mail address was gspprod@aol.com. My guess is that that address has changed, but George Steiner is (was?) a frequent contributor of electronic projects for model aircraft, so you may be able to find a more current address for him with Google. There may be other construction articles. I just don't have them indexed.

As far as an opinion, I think monitoring each servo would be preferred to monitoring all collectively, and then looking for the one drawing an excessive load. Assuming you have 12 servos, one over loaded servo, say with only three operating, may not add anything to the total current compared to having 8 servos operating normally. John

 

[ombrastein]

I think i will try to build the load sensors myself. Should be easy enough.

Anyone have any good diferential amplifier ic's to recomend?