He Flaps like a Bird: Destructive Resonation in Robot's Arms
He Flaps like a Bird: Destructive Resonation in Robot's Arms
by se7en
Mon Aug 01, 2005 12:49 pm
My robot has developed a symptom where his arms terribly resonate/vibrate when they move even the slightest bit. Also the symptom can be induced by flicking his hand with my finger when the robot is standing in home position. The resonation can be quite strong and makes the robot very unstable. I initially blamed this on his gyros but I removed them from the system and he still vibrates. It seems to be caused by the servo's own internal feedback, like the arm is bouncing around within the backlash in the system. I wonder if anybody knows what servo parameters I can modify to possibly control this? My first fix was to tighten the free-hones so they had a small amount of drag thus dampening the system. this worked well, but not enough, he still flaps. These strong resonant vibrations will damage the robot if allowed to persist. Has anybody else had to deal with this problem?
by Guest
Mon Aug 01, 2005 7:09 pm
A small vibration is normal. However strong vibration happens when the battery is about to be dead. When battery if full and the robot is standing in his home position for example there is still a fine vibration but nothing specially noticeable, you cannot see arms or legs moving as you describe, although you can hear the sound and can feel the vibration if you touch the robot.
If the vibration you feel is stronger than that I would avise you to replace your battery or try to charge it with another charger.
If the vibration you feel is stronger than that I would avise you to replace your battery or try to charge it with another charger.
by se7en
Mon Aug 01, 2005 8:17 pm
No, it is not as you describe. battery is fully charged. The slight background vibration you mention is normal, I agree. This is resonance, the hands shake like wings of a humming bird. the robots stability is affected by the strong forces. I just got him walking and this started around the 10th time I tested the walk. Problem gets worse with a stronger charge. 
by inaki
Mon Aug 01, 2005 8:39 pm
Strange behaviour. My KHR1 stays absolutly static all the time, except when battery is getting very low.
You see the same behaviour on both hands so this likely is not a failure on wire connections.
I would try the following: disconnect all other servos while keeping the arms connected; is it still resonating ?
You see the same behaviour on both hands so this likely is not a failure on wire connections.
I would try the following: disconnect all other servos while keeping the arms connected; is it still resonating ?
Video of Arms Flapping - must see!
by se7en
Mon Aug 01, 2005 11:58 pm
I made a great little video of the arm resonance:
http://robosavvy.com/Builders/se7en/KHR1ArmShake.avi
As you can see it is quite severe. Inaki, I took your advice and disconnected all servos but the arms. He still Flaps
I remaved all but servos 2 and 8 ( the shoulder servos that raise the arms to the side) and he still shakes. These 2 servos are the problem. my kit is fairly recent and has KRS-786-ICS servos, maybe they have a problem? A tech from ko-propo suggested a setting I should modify in the servo software that might help. (as a side note, he confirmed that they sell a cheap digital servo that uses the same case and gears as the kondo servos, different pc board though.)
My video shows me starting the resonance with a flick of the finger. The resonance does begin on its own as soon as the robot starts moving. He battles through it and its severity is affected. Still, it causes him to fall even with gyros. before the flapping started, the walk was near perfect.
http://robosavvy.com/Builders/se7en/KHR1ArmShake.avi
As you can see it is quite severe. Inaki, I took your advice and disconnected all servos but the arms. He still Flaps
I remaved all but servos 2 and 8 ( the shoulder servos that raise the arms to the side) and he still shakes. These 2 servos are the problem. my kit is fairly recent and has KRS-786-ICS servos, maybe they have a problem? A tech from ko-propo suggested a setting I should modify in the servo software that might help. (as a side note, he confirmed that they sell a cheap digital servo that uses the same case and gears as the kondo servos, different pc board though.)
My video shows me starting the resonance with a flick of the finger. The resonance does begin on its own as soon as the robot starts moving. He battles through it and its severity is affected. Still, it causes him to fall even with gyros. before the flapping started, the walk was near perfect.
by inaki
Tue Aug 02, 2005 1:42 am
Oh my! The effect is worse that I expected. Nothing to do with the 'low battery' effect.
I use the 784 servo model in all places but the head, where I have a 786.
The 786 has some features that 784 does not have so I wonder whether the RCB is causing some problem here.
I suppose you have not tweaked these motors with the ICS interface ?
By the way, what is that circuit on the chest of your KHR1 ? Just curious.
I use the 784 servo model in all places but the head, where I have a 786.
The 786 has some features that 784 does not have so I wonder whether the RCB is causing some problem here.
I suppose you have not tweaked these motors with the ICS interface ?
By the way, what is that circuit on the chest of your KHR1 ? Just curious.
by Meltdown
Tue Aug 02, 2005 12:15 pm
I have had the "flapping" as well, seems to come and go.
Maybe changing the dead-band of the affected servos to a wider setting will help, haven't tried myself yet.
I have had these symptoms with rc car racing a lot.
It was always a potmeter problem. It seems like the potmeter of the servo can't find it's center anymore.
Cleaning the potmeter sometimes helped but i always wound up exchanging it in the end.
The carbon strip in the potmeter could also be damaged resulting in sending out erratic resistance values.
Or the tiny contacts which go over this carbon strip could be slightly bend resulting in the same effect. Remember that the main chaft is connected directly to the potmeter. There's no bearings to stiffen up the sideway forces subjected to it. It can easely bend.
Then again it could be something completely different.
Try to exchange the servo for another one and see if the shoulder still flaps.
If it does it's not servo related.
Maybe changing the dead-band of the affected servos to a wider setting will help, haven't tried myself yet.
I have had these symptoms with rc car racing a lot.
It was always a potmeter problem. It seems like the potmeter of the servo can't find it's center anymore.
Cleaning the potmeter sometimes helped but i always wound up exchanging it in the end.
The carbon strip in the potmeter could also be damaged resulting in sending out erratic resistance values.
Or the tiny contacts which go over this carbon strip could be slightly bend resulting in the same effect. Remember that the main chaft is connected directly to the potmeter. There's no bearings to stiffen up the sideway forces subjected to it. It can easely bend.
Then again it could be something completely different.
Try to exchange the servo for another one and see if the shoulder still flaps.
If it does it's not servo related.
Problem Solved
by se7en
Tue Aug 02, 2005 1:35 pm
The deadband is what the tech had me adjust first; it had no effect. then I played with the punch which did nothing as well. Then I noticed a choice between two types of dampening. I chose the type with a slight overshoot and this completely eliminated the vibration! I did this to all of the arm servos and now it runs nice and smooth. I think I will change this setting on all of the remaining servos.
Inaki, the circuit on the robots chest is just his gyro. I didn't want to squeeze it in the back cover. if you look under his right arm you see the other gyro.
Inaki, the circuit on the robots chest is just his gyro. I didn't want to squeeze it in the back cover. if you look under his right arm you see the other gyro.
Heavy shaking servos
by noDNA
Thu Mar 23, 2006 3:21 pm
Hi all,
yes se7en, you are right. I had the same very heavy shaking of arm servos.
Makes you getting afraid that the servos will break because of this resonance movement.
The only way I know to ged rid of it is to use the ICS Servo Manager and the ICS PC serial cable with two ends, (both same pins), one end supplies power ( I use a lab power supply, 6V, black is ground, red is 6V, instead of RCB-1 power) and the other end goes ito the servo that should be reprogrammed.
After selecting the COM port you can read the settings from the servo and store them for future use in a file.
Then select the DAMPING curve with the overshoot (1) instead of the preset slope (2).
That should solve it immediately. Did so with mine. By the way, happened to me only with the new KRS786 (red) servos, not with the old 784(blue) servos.
Disadvantage is you need the ICS Servo programmer kit. No way to solder your own cable. Has some chep electronics in there but too complicated for a user to assemble it himself. Need to be experienced. I have images of the internal PCB if someone is interested.
regards
noDNA
yes se7en, you are right. I had the same very heavy shaking of arm servos.
Makes you getting afraid that the servos will break because of this resonance movement.
The only way I know to ged rid of it is to use the ICS Servo Manager and the ICS PC serial cable with two ends, (both same pins), one end supplies power ( I use a lab power supply, 6V, black is ground, red is 6V, instead of RCB-1 power) and the other end goes ito the servo that should be reprogrammed.
After selecting the COM port you can read the settings from the servo and store them for future use in a file.
Then select the DAMPING curve with the overshoot (1) instead of the preset slope (2).
That should solve it immediately. Did so with mine. By the way, happened to me only with the new KRS786 (red) servos, not with the old 784(blue) servos.
Disadvantage is you need the ICS Servo programmer kit. No way to solder your own cable. Has some chep electronics in there but too complicated for a user to assemble it himself. Need to be experienced. I have images of the internal PCB if someone is interested.
regards
noDNA
by noDNA
Wed Apr 26, 2006 8:27 am
Perhaps we can get down to the real problem of the Flapping.
Meltdown, if you exchange the servo cables of the shoulder and hand
does the flapping occur still in the shoulder or is it in the hand?
The question is: is it really a servo problem or a servo controller problem? In my case, I assumed it was a servo problem because using the Servo Manage I was able do damp it away. But now I am not too sure anymore because it seems to occur with different KHR-1s of different users and it is (always?) in first and/or second arm servo (shoulder rotate and up/down)
kind regards
noDNA
Meltdown, if you exchange the servo cables of the shoulder and hand
does the flapping occur still in the shoulder or is it in the hand?
The question is: is it really a servo problem or a servo controller problem? In my case, I assumed it was a servo problem because using the Servo Manage I was able do damp it away. But now I am not too sure anymore because it seems to occur with different KHR-1s of different users and it is (always?) in first and/or second arm servo (shoulder rotate and up/down)
kind regards
noDNA
Oh! it is frustrating that Some KRS-784 have these problem,
