#robotics Logs

Mar 15 2019

01:49 AM SleepyTurtle: Why is it I can find servos that have good torque for a los weight but not normal motors with the same torque/weight ratio?
01:53 AM SleepyTurtle: It would seem as if the latter are easier to manufacture.
03:05 AM ib0t: SleepyTurtle: servos have a gear inside of it normally I am sure that you can buy a motor and add a gear and you can reach the same torque/weight ratio for the gear and the motor as you can reach
03:05 AM ib0t: with a servo
03:06 AM SleepyTurtle: I've been looking at gearmotors
03:06 AM SleepyTurtle: and all seem to be at least 4 times as heavy
03:07 AM SleepyTurtle: I'm thinking of building a robot arm and would like the servo's specs without the internal control loop
03:07 AM ib0t: in the end a servo is nothing else as a normal motor with a gear
03:08 AM ib0t: https://www.instructables.com/id/Hack-a-servo/
03:09 AM ib0t: you can hack a servo to let it move continously
03:10 AM SleepyTurtle: I also need access to the encoder though
03:11 AM ib0t: where is the problem with the internal control loop for your project
03:13 AM ib0t: for a robot arm a normal servo which has a control loop for the position is perfect, normally
03:14 AM SleepyTurtle: That limits me to open loop control on my part
03:15 AM SleepyTurtle: and controlling every joint independently
03:15 AM SleepyTurtle: which doesn't take into consideration arm dynamics
03:16 AM orlock: also
03:16 AM orlock: you are talking about hobby servo's
03:16 AM orlock: there's a wide range of non-hobby servo motors
03:17 AM orlock: there's also the openservo control boards
03:17 AM orlock: which add feedback and stuff
03:18 AM orlock: SleepyTurtle: People have also taken servo's and relocated the pot, so instead of reading what the servo's direct output shaft is doing, it monitors the position of the joint it's actuating
03:19 AM mrdata||: hobby servos usually have 180 degree rotation
03:19 AM SleepyTurtle: orlock: openservo looks interesting
03:19 AM mrdata||: this is ideal for a rudder for a model plane
03:19 AM mrdata||: or wing flaps
03:19 AM mrdata||: and can sort of do for a lot of uses
03:20 AM mrdata||: and position is usually pulse width coded
03:21 AM mrdata||: concerning torque i took apart a hobby servo and found several gears
03:21 AM orlock: yeah
03:22 AM orlock: pololu used to sell partial-servos they had as a special
03:22 AM orlock: i think they were missing the control board and maybe pot or motor?
03:22 AM orlock: s i'd transpant the missing parts from cheaper servos when they broke
03:25 AM orlock: You can go and implement the same thing with any motor and an arduino and pot
03:25 AM orlock: but arduino's only have a 10 bit DAC iirc
03:26 AM orlock: and going by the PWM range of servo's i think they should have a higher resolution than 10 bits
03:26 AM orlock: anywya
03:26 AM orlock: home time
08:53 AM rue_mohr: SleepyTurtle, servos use tiny high rpm motors with a large ratio
08:54 AM rue_mohr: SleepyTurtle, the "encoder" is just a potentiometer that works over 300 degrees
08:54 AM SleepyTurtle: rue_mohr: Why can't I buy these in "simple motor form"?
08:55 AM rue_mohr: you can, they cost more tho
08:55 AM SleepyTurtle: all I want is a motor with an encoder, but all simple motors seem to weigh a ton and be very large for some reason
08:55 AM rue_mohr: your better modifying a standard one
08:55 AM rue_mohr: yea,
08:55 AM rue_mohr: SleepyTurtle, servos use tiny high rpm motors with a large ratio
08:56 AM SleepyTurtle: why would they cost more? It's a simpler product, that's why I'm so confused
08:56 AM rue_mohr: volume
08:56 AM rue_mohr: billions of standard servos sold, only thousands of non-complete ones
08:57 AM rue_mohr: you CAN go dynamixel
08:57 AM rue_mohr: ~$160ea
08:58 AM rue_mohr: SleepyTurtle, if you relocate the feedback pot of a standard servo you might be ok
08:58 AM SleepyTurtle: I see
08:58 AM rue_mohr: hobby servos have a 300 degree range
08:58 AM rue_mohr: but NOT over the 1-2ms thats advertized
08:58 AM rue_mohr: you go from about 0.5ms to about 2.5ms
08:58 AM SleepyTurtle: I'm fine with limited range
08:58 AM rue_mohr: k
08:59 AM SleepyTurtle: it's for an arm
08:59 AM rue_mohr: mhm
08:59 AM SleepyTurtle: Thanks
08:59 AM rue_mohr: how big?
08:59 AM SleepyTurtle: 30cm
09:02 AM rue_mohr: k
09:02 AM rue_mohr: counterbalanced or cantalevered?
09:03 AM rue_mohr: http://ruemohr.org/~ircjunk/robots/arm10/slide.htm
09:03 AM rue_mohr: http://ruemohr.org/~ircjunk/robots/arm10/slide.htm
09:03 AM rue_mohr: ^^ that is a link for you to click
09:04 AM rue_mohr: its a ~20cm robot arm build
09:04 AM rue_mohr: counterbalanced
09:04 AM rue_mohr: basically if the motors weren't there, it would prettymuch hold its posiiton
09:04 AM rue_mohr: http://ruemohr.org/~ircjunk/robots/arm10/slide.htm
09:04 AM rue_mohr: ^ click here
09:08 AM Tom_L: click
09:08 AM rue_mohr: so many people only see the last line said :(
09:08 AM rue_mohr: and dont understand to open a url sent to them
09:08 AM Tom_L: sry, been on the phone for 1.5 hrs
09:11 AM SleepyTurtle: thanks rue_mohr
09:11 AM SleepyTurtle: looks really interesting
09:12 AM rue_mohr: ok
09:12 AM rue_mohr: so, doyou understand a counterbalanced robot arm now?
09:13 AM SleepyTurtle: rue_mohr: yes, but it's very mechanically complex. I was trying to focus more on control
09:13 AM SleepyTurtle: I'd rather have a simpler body with dynamics I can "accurately" model
09:13 AM rue_mohr: ok, if you dont use a counterbalanced design, you need MUCH stronger servos
09:14 AM rue_mohr: those were standard sized servos,
09:14 AM rue_mohr: you will need large ones
09:15 AM SleepyTurtle: rue_mohr: yeah, I know
09:16 AM SleepyTurtle: I think I'll go with pololu gearmotors in the end though. I can't find feedback servos
09:16 AM SleepyTurtle: thanks a bunch everyone
09:16 AM SleepyTurtle: I've got to go
09:16 AM SleepyTurtle: cya
09:17 AM rue_mohr: uh what?
09:17 AM rue_mohr: huh
09:18 AM rue_mohr: Tom_L, think I can do 8 PID loops on an stm32?
09:18 AM rue_mohr: maybe just pd, i dosn't like me
09:18 AM Tom_L: don't see why not
09:20 AM rue_mohr: 72Mhz w/16 bit
09:28 AM Tom_L: what's the sample rate of the pid?
09:29 AM rue_mohr: havn't designed it yet
09:29 AM Tom_L: that's what determines the numbers for the p i d
09:30 AM Tom_L: btw your slide show is somewhat better than the elevator music that's been in my ear for a couple hours now
09:31 AM rue_mohr: on hold eh?
09:31 AM Tom_L: just a bit
02:44 PM Moistmelon is now known as Wetmelon