What about a servo tester?

[jperch]

Has anyone seen the GWS servo tester? I have only seen it once, briefly. But it looks like it can both generate PWM with an adjustable setting or generate a signal automatically to exercise a servo. It also can read the pulse width from an Rx.

I was thinking of doing this and add the ability to measure the current pulled by the servo. It could probably sample the current and display the peak.

What do you guys think about this? I have some LCD displays laying around here somewhere. I could probably also use a simple LED numeric display.

Thoughts?

Joe

[Andy W]

It's been discussed here before.. you can use a 555 timer to generate the PWM needed to test a servo.
What I'd like is the latter part of what you describe - something that indicates full deflection each way.. perhaps a series of lights that can move as the pulse width varies..
..a

[Mr.RC-CAM]

http://www.rcgroups.com/forums/showthread.php?t=188882
which evolved into this:
http://www.ejberg.dk/servotst2/

RC-CAM

[Mikael Pedersen]

Ahhh, somebody remembered it :)

BTW. I know that I still have a few loose ends to tie up before it is finished. But it is summer and there's so many things going on right now...... Oh well. This is just to let you know that I haven't abandoned the project. Just placed it on ice for a few months.

[jperch]

Andy,
Thanks for the input. I know you can use a 555 timer for this. But I wanted something with a little more capability.

Mr. RC-CAM and Mikael,
This is more what I was looking for. I have not read through the entire thread nor have I looked through Mikael's web site. But I can see right off that this is more in line with what I was thinking. I also want to read through the manual for the Hitec tester. I had not thought about measuring deadband. But this sounds like a cool feature. I will have to learn how to do it.

I see that someone mentioned connecting a servo to some sort of force meter to measure torque. What I was thinking was to control the servo back and forth between extremes and then load it with a weight. If you know the length of the moment arm and the weight on the arm, you can get pretty close. You could then monitor the current draw of the servo while it is running like this and get an idea of its capability.

Anyhow, thanks for the ideas. If you don't mind Mikael, I will read through your design and probably use some of the ideas.

The company I work for makes MCU's so I will use one of these instead of a PIC. The PIC's are very powerful and inexpensive. But the ones at work are also very powerful and are free. :D

Joe

[Mikael Pedersen]

A servo with a known load on. Hey, that's a good idea.
My tester already has the continuous sweep between endstops, and it also measures the current, though not very fast. But that probably can be dealt with.
Please do read it through, and use whatever you can. That is what it's there for. And just ask if you have any questions.

BTW. My design isn't built around a PIC. A PIC could probably have been just as useful for such a project, but they are just not "me". I like AVR's better ;)

[jperch]

Mikael,

I noticed that it was not PIC based after I wrote that, sorry. I work for a company called Silicon Laboratories. At the end of last year, we purchased a company called Cygnal Integrated Products. They made a line of 8051 based MCU's. So, now my company makes MCU's. I have used one of them in one or two projects for the division I work in. They are very handy. One of the things that makes them nice is that we integrate a number of mixed signal items into the chip with the 8051 core.

To measure current, I could simply put a small resistor in the power line of hte servo and connect it so that the built in ADC measures the voltage across that resistor.

I have not put too much thought into it. But I was thinking I could sample that value about 1000 times a second and display the peak value. While doing this, I could be sweeping the servo from one end point to the other.

One question. When you sweep the servo, do you simply send a pulse width of one extreme for a bit, then send pulses of the other extreme? Or, do you actually sweep the pulse value smoothly from one extreme to the other? If so, how do you decide at what rate to sweep the pulse value?

thanks,
Joe

[Mikael Pedersen]

When measuring the current with the known load on the servo, you probably need to discard the measurements around the extremes. Changing direction on a servo is something that really pulls a lot of current, and it will most likely be your peak value unless filtered away.

I do sweep the pulses smoothly through appropriate values. The rate is determined by the user, together with the endstops. I simply increase (or decrease for the other direction) the pulse value by a user configurable constant for each generated pulse.

[jperch]

That is a good point. I had not thought of that. I suppose one way to do it would be to let the servo stop at the end points for a brief time. This would let the current settle before you try to reverse direction.

If that didn't work, I could simply throw away those samples that were taken when the servo was near its endpoints as you suggest.

Thanks for the ideas.

Joe