Putting robotics at your service™

Free shipping on orders over $200

External reference feedback

Print view Share :
Previous topicNext topic

Page 1 of 1 [ 9 posts ]

New ( offline )
Posts: 3
Posted: 2009-05-28 20:56 
I have questions about using two servos to control two devices to nearly the same position from the same control signal. The devices to control are multiple turn and I will use external reference feed back and servos modified for continuous rotation and external feedback.

I am building a remote antenna tuner for amateur radio purposes. I need to control two roller inductors. These are 27-turn devices (9,720 degrees) and they need to end up close to the same position (+/- 30 degrees or so). I bought a cheap hexTronic HX12K and converted it for continuous rotation. I geared a new 10-turn 5K pot to the drive shaft for position feedback. That servo now tends to oscillate (hunt) near the commanded position a lot.

Before I spend a fortune I would like advice as to the hunting situation. Is this a cheap servo problem?

The HS-805BB may be powerful enough. The 425BB is not as it almost stalls.

When I move to two devices under control, what other problems should I expect?

Thankyou for the bandwidth.

Mike


Guru ( offline )
Posts: 2251
Posted: 2009-05-28 23:06 
Be advised that if the pot is one of the small multi turn trimmer types, it is probably only rated for ~200 cycles. It may wear quickly. I started to try the same thing with a 15 turn pot until I noticed the low duty cycle.


Guru ( offline )
Posts: 2158
Posted: 2009-05-29 00:16 
when you say "geared a new 10-turn 5K pot to the drive shaft for position feedback" what degree of mechanical slack is there when you change rotation direction? and zoomkat has an excellent point... you should be using a precision multi-turn potentiometer rather than a multi-turn trimmer potentiometer. an example.


New ( offline )
Posts: 3
Posted: 2009-05-29 01:34 
I am using a precision Bourne 5K pot and I believe no slack in the gearing.
It is obvious that the hunting comes from the cheap servo. The servo acted this way without being connected to any gearing anyway. So before I shovel out a bunch of money as high torque servos can be pricey I am wondering if this is just a cheap chinese servo problem.

Mike


when you say "geared a new 10-turn 5K pot to the drive shaft for position feedback" what degree of mechanical slack is there when you change rotation direction? and zoomkat has an excellent point... you should be using a precision multi-turn potentiometer rather than a multi-turn trimmer potentiometer. an example.


Guru ( offline )
Posts: 2251
Posted: 2009-05-29 02:09 
The servo you have apparently is the former TowerPro MG995 which I think has had some negative reviews in the past (search the forum). It is important that the antenna design is balanced such that the servos don't have to continously support weight. The below link has some info on the innerds of a servo that might be of interest. If you get the HS-805BB, post up its inside arrangement as I'm interested in a similar multi turn setup using it.

http://www.seattlerobotics.org/encoder/ ... 3003C.html


User avatar
Guru ( offline )
Posts: 4122
Posted: 2009-05-29 02:20 
If you use a 10 turn pot, and the A/D in the R/C servo is perhaps 8 bits (?), then you have about 15 degrees per count right off the bat. Add a little backlash to that...

10 turn pot, but 27 turn roller inductors? Then you have even more gear ratio (2.7) , so maybe 40 degrees per count?

Perhaps you can find a linear pot?

Check to see if there is any modification info available for your servos. perhaps you can "open up" the dead band. That should reduce the hunting.

There is an open-servo project. Perhaps it uses an A/D with more resolution, or you can use an external A/D chip to increase your resolution.

An encoder on the motor shaft would be better, you could graduate to a full servo system with an encoder, geared motor and a servo driver. But that's going to cost you a little more...

There was an auto-tuner article in a QST magazine of a few years ago. Might be some ideas in that.

What band?

Alan KM6VV

_________________
Visit:
http://groups.yahoo.com/group/SherlineCNC/
http://tech.groups.yahoo.com/group/HexapodRobotIK/


Guru ( offline )
Posts: 2158
Posted: 2009-05-29 07:41 
JMike wrote:
I am using a precision Bourne 5K pot and I believe no slack in the gearing.
It is obvious that the hunting comes from the cheap servo. The servo acted this way without being connected to any gearing anyway. So before I shovel out a bunch of money as high torque servos can be pricey I am wondering if this is just a cheap chinese servo problem.

Sorry, you didn't / don't mention the specific servo you hacked and the comment in your OP about cheap servo was followed by mention of two fairly respectable servos. If you are using, as zoomkat suggests, an MG995 or any other variant of the same manufacturer of that servo then you've found your problem. Alan's comments about a/d resolution directly apply if you are using a digital servo, however even if you are using an analog servo controller the increased angle range will make even a decent quality analog servo more susceptible to jitter in the pwm control signal (which could result in oscillation).
Also, digitally speaking, even a 13-bit a/d converter will resolve to more than a degree per bit. It is unlikely any off the shelf servo has a 14-bit or more a/d converter so unless you are able to wire up an analog circuit you might be better served to use a small microcontroller with a built in a/d and wire up a small h-bridge to run the motor. There are several modular products avalable that can help reduce the amount of "assembly" required if you search about.


New ( offline )
Posts: 3
Posted: 2009-05-29 09:07 
Thank you for the replies. I hacked a HexTronik HX12K which I believe is the same as the MG995. I am glad to hear that that particular servo may be part of my problem. I have built my own servo controller using the NE555 timer chip, so I am analog. My position feedback pot is linear. There will be some gear backlash however I can't detect any in the current set up. Being analog does not totally remove the position accuracy issue it just puts me in I have to try it and see I guess. Optical encoded feedback would be good to try.

The antenna tuner is a Richard Measures design balanced L to be used on all bands from 80M to 10M.
Mike (AE6WA)



KM6VV wrote:
If you use a 10 turn pot, and the A/D in the R/C servo is perhaps 8 bits (?), then you have about 15 degrees per count right off the bat. Add a little backlash to that...

10 turn pot, but 27 turn roller inductors? Then you have even more gear ratio (2.7) , so maybe 40 degrees per count?

Perhaps you can find a linear pot?

Check to see if there is any modification info available for your servos. perhaps you can "open up" the dead band. That should reduce the hunting.

There is an open-servo project. Perhaps it uses an A/D with more resolution, or you can use an external A/D chip to increase your resolution.

An encoder on the motor shaft would be better, you could graduate to a full servo system with an encoder, geared motor and a servo driver. But that's going to cost you a little more...

There was an auto-tuner article in a QST magazine of a few years ago. Might be some ideas in that.

What band?

Alan KM6VV


Guru ( offline )
Posts: 2158
Posted: 2009-05-29 09:28 
heh, I see you mentioned "HexTronik HX12K" in your OP but honest to god I just figured it was a reference to some ham radio gear I knew nothing about. Sorry. :oops:


 

All times are UTC - 5 hours [ DST ]. It is currently 2014-09-30 11:50
Feedback Form
Feedback Form