kazobond
Greeting all, It has been a long while since I posted here. I have been working on my Inmoov whenever I can find time, which is not a lot, but slowly, very slowly progressing.
I now have a fully functional and completed torso with left arm and head. I am now starting the real fun; programming ! Yaaa !
The problem I am encountering at this time is the servo release() command. First, from what I read, on power-up, I must first set servo position, then servo speed. Then I can change servo position and/or speed at will. So far so good. Now, on some of my servos, I would like to release them to stop the "humming/buzzing" noise. I use the Release() command, but after using it in scripting, there is nothing I can do to get the servo to move at all. I know I am doing something fundamentally wrong, but can not figure it out.
Here is the sequence of action:
- Power-up EZB-V4
- Connect via wifi
- Set servo limits in init script (connect)
- Then, using the script console (or within the init script, or any scripting method), run the following commands:
- Servo(d10,130) # Set servo position
- ServoSpeed(d10,2) # Set servo speed These are required on startup. At this point, servo works fine. I can change position and speed at will without problem.
- Release(d10) # Releases the servo. No noise
- ServoSpeed(d10,3) # servo speed stays at 2, does not change to 3 (GetServoSpeed(d10) )
- Servo(d10,75) # servo does not move (stays at previous 130 position). But can hear the "buzzing noise" No matter what I do, I can no longer move the servo position. Tried setting new servo position and servo speed, but no luck.
I have captured the screen and attached it.
Any idea anyone ?
RR, Simply use high quality use cyanoacrylate (CA) glue (Super Glue). Don't use the cheap stuff. It will give you a burnt look.
Also, don't glue your fingers together and hide it at bed time if you and your better half has had a fight. eek
Wow, I was kind of affraid to ask this relase() control issue question. But I see it is an issue not that easy to solve. A good challenge for me actually. I personally hate the buzzing noise. Plus, my audio sound system is not quite loud enough to completely burry the servos noise. Pictures coming next.
@Dave Don't quite understand your solution. I would like to.
@Aerius Really don't understand your solution, but really want to. And it really seem to works perfectly. and btw, such an awesome Inmoov. You guys (including @Richard) are putting the bar way to high for me
You guys are all correct;
Must keep servo active to hold position sometimes depending on mechanical design/load. I actually think it's quite funny to see the head dropping down after a Release(), like he is pouting or falling asleep (hihihi). I think in my next prints, I'll make sure there is a lot of friction in the gear to keep the load in place (hihihi, just kidding). It's good to have a sense of humor while developing this Inmoov.
But with servos driving warm gear, the gear will hold the load no problem. Beside the magical trick that @Arius and @Dave are doing, I was thinking of moving the servo 1 or more degree opposite to the moving position, just enough to remove loading factor. Did not try that yet.
My current solution (or more patch) is to use the AutoPosition control and set the servos to 0 position in the Frame configuration. That's ok, but I would like to have more control using scripting.
One way I have found to relieve some of the load on various joints is to use a spring to counter it. For example, the robot (not an InMoov) I am currently working on had a problem holding it's head up due to extra weight I had placed on it. Worse yet, the weight was an over-hanging load which magnified it's effect. There was some buzzing as well in certain positions.
Adding a spring to pull it up cured all the problems. Although this also meant that the servo controlling the up-down (pitch) head movement had to actually drive the head down. Before, gravity took care of that. Overall though, the spring, while not perfect, did the job. No buzzing. No droop, even when turned off. Like you, the head used to drop when power was cut. With the right spring tension it now tends to hold in place no matter where it is stopped.
I'm also experimenting with a spring on the shoulder rotation joint. It has, however, proven to be more problematic to get just right. Nonetheless it does help quite a bit.
@Kazobond
Tanks for the compliment, your Inmoov is coming pretty well to.
I believe a compression spring on the neck piston will be a very good solution. It have been experiment before with success by rentenprita
https://synthiam.com/Community/Questions/8555
Now back to servo release script. As discussed previously this solution would be useful where the mechanical unit can hold position when the servo is release.
I'll be glad if you wish to help you to adapt the script setup to your Inmoov.
I just realize you have upload your Inmoov script. Let me have a look at it and if I could prepare something with explanations.
Aerius, I downloaded your no buzz project. Thanks for making it available for us. Nicely done. I'm having a little trouble figuring out how some it ties into a bigger project. I was watching your no buzz video and see it working. Is there any way you'd consider uploading the project you were useing in the vedio so I can see how it all works together? I'd also love to see how some of your other scripts look and flow. It looks like a very impressive project and has a lot of things to offer that would aid my B9 project. Thanks in advance.
@Arius, Thanks, I love your compression spring solution. Very simple. I'll try it for sure.
I'm gonna have to spend more time trying to figure out your no buzz solution. It really seem to do the trick on your video.
@Kazobond... The problem with the servos on inMoov buzzing is compounded by the fact most of the large 805bb have been modified to work through a gearbox or a jack screw (in the bicep and clavicle) for the purpose of a mechanical advantage... That means the there is also more friction in the joints as well. So when the servo gets close to a required position it sometimes stops just short of perfect position due to friction (and of course any load)... The servo buzzes because it can quite get to where it's going... Not enough juice to over come that last bit of needed movement... FYI pwm to the servo drops accordingly as it approaches it's target... Near the end there just isn't enough pwm to overcome that little bit of extra friction in inMoovs joints and gearboxes... I get buzzing with one or two shoulder servos even though inMoovs arms are at rest (no load) at his side... The friction in the piv (not so much the rot) gearbox prevents perfect positioning so the servo buzzes even with no load... Perfect positioning with no load will silence the servos...
The servos in my inMoovs neck also (for "yes" and "no" head movement) buzz a lot and not due to excess load either, but due to that extra friction preventing them from accurately positioning... Yes, I have sanded and greased them meticulously... So I am going to try adding springs as mentioned here to the neck servo to help the servo overcome the little friction still in the plastic joints... I am also going to try and use @Aerius's method to try and silence the side to side head movement servo as you guessed it, mine buzz even without load...
OK Aerius, I've been studying your No Buzz script and I'm starting to understand it. Very sweet coding. I love the logic behind it. Looks to me that the autopos control is starting a script that is estimating the time it will take a servo to reach a set spot and then sending a release command for that servo. That's simplified by me but am I close to understanding this?