Post by thread:Re: [EE]: stepper motor in TemUR microrobot

Hi all! I don't know how to begin... Perhaps this is not even [EE] question but since I know that you guys have huge experience and thousands of projects behind let's give a shot. Here we go... It is a long story. Last months I was tinkering with very small (it is nice to think 'micro' but I know what 'micro' means) mobile robots. The heart of the creatures is PIC of course! The first robot [TeamUR the First] was moving pretty well however I was not very happy with the precision of its suspension. Two DC motors with worm gear run it. http://teamur.netfirms.com/tmr001/tmr001.htm The second robot [TeamUR the Second] was supposed to be a [my!] technological breakthrough because I decided to go for the stepper motors. I was hoping for more precise motor control. The first stepper motor control attempts were very promising. Since I am electronics engineer it was easier for me to build a robot and test it rather than try to calculate all this mechanical crap like torque, friction etc. http://teamur.netfirms.com/tmr002/tmr002.htm It looks that I was wrong. I put a sprocket directly on the motor's shaft and unfortunately the torque was not enough big to move it. Of course the first mistake I made was the size of this sprocket. Replacing with the small one helped a lot. And here comes the disappointment. The robot should have a nickname "Jerky" because of how it moves! I suspected that dynamic (continuous) vs. static (single step) stepper motor control is different but what is happening with it when the stepper runs below 10 steps per second is heart breaking! The whole robot jerks probably because the motor is not enough precise (16 steps per revolution) and makes too BIG STEPS. Sometimes the motor gets stuck and robot turns around. I know that this is cheap surplus motor but at this point of time I am not that big robot "enthusiast" to pay $$$ per motor to some Swiss company. The second problem is that the motors suck current from battery like fresh orange juice. 1. Do you think that gearhead is necessary in such cases? 2. Where can I find an explanation of the dynamics of the stepper motors without going for LONG studies? 3. How to decrease power consumption and increase torque, which are contradictive goals? 4. Is there a source of SMALL (diameter << in.) & CHEAP (price << $50) stepper motors? 5. Do you have any suggestions regarding using stepper motors in such construction? I am not afraid to take the robot to my "closet junkyard" and build a new one but I would like to avoid this type of probably basic mistakes in the future. Let me apologize if you think that it should be [OT] - I can change the prefix. I appreciate any feedback. Irek -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Hello, Neat robots! Problem number 1: Jerky You have your sprockets the wrong way round. It looks like your driving gear is 1.5" diameter, which means that each step drives the chain about 0.3". The small ones look about 0.5", so if your motor uses them then each step is smaller (maybe 0.1"). Also, you should research the techniques of half-stepping, which should allow you to achieve 32 'steps' per revolution. There is another technique called microstepping, but it is more difficult to implement. Problem number 2: Power consumption Unfortunately, the coils in the stepper motor are just lengths of wire, with a fairly low resistance, so the current is high. If you keep the current turned on you can hold the stepper motor in a fixed position, however, if your motor will not be subject to external forces (and the robot is pretty small and light weight) then you could maybe try pulsing the coils to make the motion, and then leaving them off when you know the motor has reached the next step position. I have not tried this, it is only an idea. Problem n: Other small stepper motors can be found in floppy disk drives. Vibrating pager motors are small (but they are not stepper motors) Increasing torque can be done with a gearbox or series of rubber bands and pulleys. That's it for now. Oh, did I say, neat robots! Andy -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Andy: thanks for the response! > -----Original Message----- > From: pic microcontroller discussion list > [spam_OUTPICLISTTakeThisOuTMITVMA.MIT.EDU]On Behalf Of Andrew Errington > Sent: Monday, July 08, 2002 11:48 PM > To: .....PICLISTKILLspam@spam@MITVMA.MIT.EDU > Subject: Re: [EE]: stepper motor in TemUR microrobot > > > Problem number 1: Jerky > > You have your sprockets the wrong way round. It looks like your driving > gear is 1.5" diameter, which means that each step drives the chain > about 0.3". The small ones look about 0.5", so if your motor uses them > then each step is smaller (maybe 0.1"). Also, you should research the > techniques of half-stepping, which should allow you to achieve 32 > 'steps' per revolution. There is another technique called > microstepping, but it is more difficult to implement. Yes, originally the driving gear was ca. 1". The purpose of this approach was to cover (hide) the motor. Very early I realized, that this is not clever since I have low-precision motor. I've changed the sprocket. It helped but not too much. See: http://teamur.netfirms.com/tmr002/movement_problem.htm Half stepping is something I would like to try... I am afraid that microstepping requires way more complicated circuit. {Quote hidden}> Problem number 2: Power consumption > > Unfortunately, the coils in the stepper motor are just lengths of wire, > with a fairly low resistance, so the current is high. If you keep the > current turned on you can hold the stepper motor in a fixed position, > however, if your motor will not be subject to external forces (and the > robot is pretty small and light weight) then you could maybe try pulsing > the coils to make the motion, and then leaving them off when you know > the motor has reached the next step position. I have not tried this, it > is only an idea. Another yes. This was my initial idea, to put 754410' enable low after pulsing for enough long to make a step. This way for the most of time the motors are not powered. So far to make the robot moving I disabled H-bridge enable pin pulsing. > Problem n: > > Other small stepper motors can be found in floppy disk drives. I tried to use stepper motors from floppies but the current constructions require addtional support (ball bearing) at the end of the shaft. See: http://teamur.netfirms.com/test/test_floppy_disk_drive.htm > Vibrating pager motors are small (but they are not stepper motors) > Increasing torque can be done with a gearbox or series of rubber bands > and pulleys. Yeah... I was hoping that the stepper motors can give me hardware simplicity combined with relative precise control. There must be somewhere a mistake... I am not going give up yet! > Oh, did I say, neat robots! Thanks once again! Irek > > Andy > > -- > http://www.piclist.com hint: The PICList is archived three different > ways. See http://www.piclist.com/#archives for details. > > -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listservKILLspammitvma.mit.edu with SET PICList DIGEST in the body

Hi Irek, shame about the jerky problem. You really need more gearing, like a 1:5 pair of gears, as you probably won't get the performance you want with 1:1 connecting the stepper to the chain, even with your small cog. Steppers *will* vibrate quite a bit, and even with the right gearing you might find that because you have hard shiny chains it will still not get precise location on the ground. It is fairly important to have good gripping rubber between the stepper motor and the ground. :o) My little stepperbot here: http://www.romanblack.com/deskbot.htm has 1:7 gearing on the steppers, and rubber tyres. Without the rubber tyres it slides and does not drive in a straight line or get accurate distance etc. Even with the rubber tyres it loses/gains 1mm in every 50 or so, without rubber tyres it is more like 20mm error out of every 50!! Skid city. Another thing to think about is the h-bridges, they are normally 3v to 4v (saturation) when turned on, so you may only be getting 4v on your motor and losing 4v on the bridge. Better to use 4 fets and drive the motor unipolar (has half the coil resistance and 10x less loss on the drivers). You can get tiny stepper motors in any 3.5inch floppy drives, dead drives are available everywhere. The best advice I can give is a redesign using at least 5:1 gearing, *rubber* tyres, and low saturation drivers. :o) -Roman Irek Rybark wrote: {Quote hidden}> The second robot [TeamUR the Second] was supposed to be a [my!] > technological breakthrough because I decided to go for the stepper motors. > I was hoping for more precise motor control. The first stepper motor > control attempts were very promising. Since I am electronics engineer it > was easier for me to build a robot and test it rather than try to calculate > all this mechanical crap like torque, friction etc. > http://teamur.netfirms.com/tmr002/tmr002.htm > It looks that I was wrong. I put a sprocket directly on the motor's shaft > and unfortunately the torque was not enough big to move it. Of course the > first mistake I made was the size of this sprocket. Replacing with the > small one helped a lot. > > And here comes the disappointment. The robot should have a nickname "Jerky" > because of how it moves! -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email .....listservKILLspam.....mitvma.mit.edu with SET PICList DIGEST in the body

On Mon, 8 Jul 2002, Irek Rybark wrote: >It looks that I was wrong. I put a sprocket directly on the motor's shaft >and unfortunately the torque was not enough big to move it. Of course the >first mistake I made was the size of this sprocket. Replacing with the >small one helped a lot. You can measure the required torque/force using a piece of string a plastic bag (to be filled with coins) and a smooth-edged table. Refer to a physics textbook (for school 10th grade or less) for howto. >And here comes the disappointment. The robot should have a nickname "Jerky" >because of how it moves! I suspected that dynamic (continuous) vs. static >(single step) stepper motor control is different but what is happening with >it when the stepper runs below 10 steps per second is heart breaking! The >whole robot jerks probably because the motor is not enough precise (16 steps >per revolution) and makes too BIG STEPS. Sometimes the motor gets stuck and >robot turns around. >I know that this is cheap surplus motor but at this point of time I am not >that big robot "enthusiast" to pay $$$ per motor to some Swiss company. You can add viscous dampers or friction brakes. The latter are easier to DIY. A simple one: tie a short bit of string to a short piece of elastic, wrap the string (not the elastic) once round the axle and fix the two ends of your brake to the chassis. Adjust the elastic tension as required. This is a sort of rope brake (dictionary term for this type of device). The string should be wide (shoelaces etc). It brakes more in one direction than in the other. >The second problem is that the motors suck current from battery like fresh >orange juice. You need to look into current chopping stepper controllers imho, with current settable by the CPU. >1. Do you think that gearhead is necessary in such cases? Must have. >2. Where can I find an explanation of the dynamics of the stepper motors >without going for LONG studies? Ha. Try to get a copy of B.C.Kuo's book on stepper controls. >3. How to decrease power consumption and increase torque, which are >contradictive goals? See above at current chopper. >4. Is there a source of SMALL (diameter << in.) & CHEAP (price << $50) >stepper motors? Pass. >5. Do you have any suggestions regarding using stepper motors in such >construction? Avoid them. You can get servos that small or hack your own using some small motors and wakeup clocks or windup toy gearboxes. >I am not afraid to take the robot to my "closet junkyard" and build a new >one but I would like to avoid this type of probably basic mistakes in the >future. Do the basic physics on your robot ('coefficient of friction' without motors, weight -> required starting torque reduced to motor torque by gears) etc. Then it will be easy ;-) Peter -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email EraseMElistservspam_OUTTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body

Irek Rybark wrote: > > Yeah! > Roman, thanks for the professional feedback. I was hoping that there will > be someone with *robotic* experience. It's only professional if you pay! ;o) > Ok. Then dreams about attaching stepper directly to the sprocket are really > unreal? > With the gearing the whole thing becomes too big for me. I was thinking > about building a robot with the dimensions below 1". The robots currently > have ca. 2" and in the next generation should shrink rather than grow... :o) Your motors had tiny metal cogs, if you find a larger plastic gear that matches, and glue it to wheels with rubber tyres you may get 3:1 or 4:1, and have tyres. simple and still small, maybe smaller than the chain setup you have now. The extra torque from re-gearing makes all the difference. :o) > Now I am aware that one of the problems is this "hard shiny chain" I was so > proud of. Mechanically there is no way to have this chain *not skidding* on > the floor. There is no mechanical part, which can accumulate step's energy > and smoothly release it. Probably a drive made with pulleys and rubber band > should help a lot. What do you think? Not really. Part of the problem is the acceleration and deceleration of the bot, when stopping and starting. You *really* need rubber if you want decent traction, unless the whole bot runs on a rubber tabletop. > > My little stepperbot here: > > http://www.romanblack.com/deskbot.htm > > This is a real beauty! :o) Especially impressive for me is that your robot > "has enough torque to push a full coffee cup without slowing, in fact it > will push about 1kg across a smooth table". Yep it's quite torquey, the 7:1 gearing and heavy bot weight make it like a little tractor. But again, without the rubber tyres (only plastic wheels) it won't push nothin'. It really sounds like you want to make these bots small, the best advice is to look at what the REAL experts are doing, like the "ants"; http://www.ai.mit.edu/projects/ants/ For tiny bots you can get tiny "RC servos" which have a motor and gearbox built in. They can be modded to give continuous rotation. Try the RC (radio control) web sites. Rubber wheels are a must. :o) You can also add a tiny infrared sensor to the gearboxes, giving you position feedback to the PIC at 32 pulses per wheel rotation etc. Little servos are going to use 10x less power than steppers, are smoother, more torque, etc etc. One way that appeals to me is the "angled motor direct drive" concept. You use tiny pager motors, with a tiny rubber wheel not much bigger than the shaft, say 2mm shaft and 4mm rubber "wheel" like a rubber grommet around the motor shaft itself. The gearing is provided mainly by the tiny size of the wheel. The 2 motors are angled at about 45 degrees to give some ground clearance. This gives decent start torque and good speeds too. The secret is to put a tiny IR encoder wheel on the back end of the motor, and use good PWM control of motor voltage, giving good slow speeds, good high speeds, good traction, good energy efficiency, good positional feedback etc etc. See the basic motor configuration of the Beam photovore; www.robotbooks.com/beam_solar_robot_kits.htm and improve it with a couple of optical encoder wheels from a mouse. :o) -Roman -- http://www.piclist.com hint: To leave the PICList piclist-unsubscribe-requestspam_OUTmitvma.mit.edu

Hi all! Roman wrote: > Your motors had tiny metal cogs, if you find a larger > plastic gear that matches, and glue it to wheels with > rubber tyres you may get 3:1 or 4:1, and have tyres. > simple and still small, maybe smaller than the chain > setup you have now. The extra torque from re-gearing > makes all the difference. :o) This is worth to try! Sounds like amnesty for TeamUR the Second. It will be saved for later. >> There is no mechanical part, which can accumulate step's energy >> and smoothly release it. Probably a drive made with pulleys and rubber band >> should help a lot. What do you think? > >Not really. Part of the problem is the acceleration >and deceleration of the bot, when stopping and starting. >You *really* need rubber if you want decent traction, >unless the whole bot runs on a rubber tabletop. I understand. To move, the robot needs a support (good ol' Newton!) but the friction is not everything. You must agree that after while of skidding the robot should start moving but nearly it didn't. It looks for me that there is micro-acceleration and micro-deceleration of rotation of the stepper's shaft (in my case: without gearing). I think that even with the rubber wheels this micro deceleration can be drastically decreased when the rubber band stretches "on step" and releases the energy between steps. It makes the whole movement smoother. I tried it with the "shiny chain" and the rubber band. The band really makes a difference. >It really sounds like you want to make these bots >small, the best advice is to look at what the REAL >experts are doing, like the "ants"; >http://www.ai.mit.edu/projects/ants/ I saw The Ants. Love them but look at this PIC bot: http://chanhak.com.ne.kr/robot.html/ >For tiny bots you can get tiny "RC servos" which >have a motor and gearbox built in. They can be modded >to give continuous rotation. Try the RC (radio control) >web sites. Yep! As I wrote in response to Peter's feedback: I am already experimenting with one purchased some time ago in my local hobby store: Hirec HS-55 servo "it looks very promising. Dim. 7/8"x5/8"x3/8", very light comparing to the currently used steppers, A LOT of torque: 18oz.in!." >Rubber wheels are a must. :o) You can also >add a tiny infrared sensor to the gearboxes, giving >you position feedback to the PIC at 32 pulses per wheel >rotation etc. Little servos are going to use 10x less >power than steppers, are smoother, more torque, etc etc. What you are saying is that I should leave the electronics controlling the servo and send the pulses? This is a very good idea! As I saw servo hacking somewhere in the Internet I need to replace the potentiometer with resistors to force continuous rotation. But if I will stop pulsing the servo should remain in the same position. Is it true? > One way that appeals to me is the "angled motor direct > drive" concept. You use tiny pager motors, with a tiny > rubber wheel not much bigger than the shaft, say 2mm shaft > and 4mm rubber "wheel" like a rubber grommet around the > motor shaft itself. Solar bugs... I tried this approach on the very beginning. I am not a big fan of biocore. I am rather PIC-er. Unfortunately looks for me that this type of propelling tends to be somewhat unpredictable because of high rotation speed of DC motor. Besides I would like to give to my robots an ability to climb some object since the are supposed to be desktop robots and my desk is never empty :o) Thanks for your time! I am sure that I will run in some more problems soon (with PIC for example) and it will be an opportunity to let you guys know how is going. Irek http://teamur.netfirms.com -- http://www.piclist.com hint: To leave the PICList KILLspampiclist-unsubscribe-requestKILLspammitvma.mit.edu

Peter: Thanks for the feedback. I am sorry that I did not reply quickly: for some reason your post couldn't reach my mailbox but luckily I found it in list's archive... >You can measure the required torque/force using a piece of string a >plastic bag (to be filled with coins) and a smooth-edged table. Refer to a >physics textbook (for school 10th grade or less) for howto. Sure, but the tricky part is an answer to the question: what next? It is rather difficult to evaluate how much torque I need to move the robot when it exists on a paper only (or even worse: in my mind). There are way too many variables... :o( >You can add viscous dampers or friction brakes. The latter are easier to >DIY. A simple one: tie a short bit of string to a short piece of elastic, >wrap the string (not the elastic) once round the axle and fix the two ends >of your brake to the chassis. Adjust the elastic tension as required. This >is a sort of rope brake (dictionary term for this type of device). The >string should be wide (shoelaces etc). It brakes more in one direction >than in the other. Sounds simple but I am afraid that it won't last long in my “micro-construction” :o). >>1. Do you think that gearhead is necessary in such cases? > >Must have. This is sad part... :o( >>2. Where can I find an explanation of the dynamics of the stepper motors >>without going for LONG studies? > >Ha. Try to get a copy of B.C.Kuo's book on stepper controls. Will try. >>5. Do you have any suggestions regarding using stepper motors in such >>construction? > >Avoid them. You can get servos that small or hack your own using some >small motors and wakeup clocks or windup toy gearboxes. This is probably the best possible advice. I am just trying to use hacked Hirec HS-55 servo and it looks very promising. Dim. 7/8"x5/8"x3/8", very light comparing to the currently used steppers, A LOT of torque: 18oz.in!. >Do the basic physics on your robot ('coefficient of friction' without >motors, weight -> required starting torque reduced to motor torque by >gears) etc. Then it will be easy ;-) Those equations look simple in "physics textbook". Of course that I can try to make estimates... Once again: thanks! Irek -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestTakeThisOuTmitvma.mit.edu

Hi all! Brendan wrote: > The other goal of ants was to make a robotic comunity. You'll note that > that link is part of the "retired robots" section at MIT. I am thinking about building a group of semiautonomous desktop robots. I want to give a PC ability to play a role of mother-brain. The mother could show a goal and the TeamURs should proceed avoiding obstacles and solving minor problems. Unfortunately as more time I spend playing with this crazy project as better I see the difficulties... > I found "Gaak the predator robot" > education.guardian.co.uk/higher/engineering/story/0,9840,74 > 0719,00.html > to be a more up to date AI experiment (; This is really funny story. I've read somewhere more detailed relation. Unfortunately it was just accident, lack of attention rather than AI. But you never know... :o) Irek http://teamur.netfirms.com -- http://www.piclist.com hint: To leave the PICList spamBeGonepiclist-unsubscribe-requestspamBeGonemitvma.mit.edu