hamwx
New Member
Posts: 4 
|
Post by hamwx on Jun 5, 2016 19:37:34 GMT -7
I come from China, also I did a small DOBOT myself, just a fun. I can't let it work perfectly, I found here, fully open-sourced place. So thank you very much. I have bought a ramp1.4, so I will study carefully. And I will write the detail, then post here. Thank you again!  |
|
|
|
Post by Max on Jun 6, 2016 7:27:46 GMT -7
NP. Happy to help. If you you could share how you built it yourself (materials and tools used, the process, etc.) that would be helpful and encouraging for others.
It would be also interesting to find cheaper copies that the dobot team referred to, and, hopefully, better copies, if you know of any. I'd very much like to see this niche filled with inexpensive robots to bring unjustified prices for accurate robots down and enable more people to get involved (as I hoped would be with dobot, which didn't happen because of their greed and incompetence).
|
|
|
|
Post by mikef522 on Jun 6, 2016 14:04:56 GMT -7
I second this. I might have spent $500 for another Dobot (wouldn't be terribly happy about that though given a good 3D printer from China costs $350 with shipping), but I definitely won't be spending $900 (not in my budget anyways). I'm fortunate to have access to a really nice machine shop at my university, so I might try and build it myself. I do have the CAD files they supplied sometime last year. I'll post a link to them in a separate thread for anyone who wants them.
|
|
|
|
Post by Max on Jun 6, 2016 14:36:11 GMT -7
Oh, you do? Please create a project on github. I'm sure it will be cloned quickly by many people. I will clone it for sure. I'll redesign it to eliminate backlash. I've been thinking of that for a while now.
|
|
nad
New Member
Posts: 13
|
Post by nad on Jun 6, 2016 14:44:49 GMT -7
I was doing cheaper. I cut with a laser from the plexiglass. Total cost is around 200$ but my steppers are terrible. My dobot is cheap quality  |
|
|
|
Post by Max on Jun 6, 2016 14:48:12 GMT -7
|
|
|
|
Post by mikef522 on Jun 6, 2016 15:17:45 GMT -7
Not sure if you're talking about putting the CAD files (I linked to these in another thread on this forum) or build instructions on github. In any case, between work and school, it might take me a bit to post build instructions (weeks to months most likely). That's my naive estimate anyways. The arm doesn't look like it has that many parts, and they look simple enough to machine out of aluminum very quickly. The only problem is I'm not sure what some of the parts are named (though these might be detailed more in the CAD files) or more importantly, where to get them. Also, I don't know exactly how some of the pieces fit together (the bearings on the base and the right/left stepper motor shafts for example). There is an instructable here that might be really helpful though: www.instructables.com/id/Build-a-Laser-Cut-and-Soldering-Dobot-arm/?ALLSTEPS (haven't looked through all of it in detail, but I do know the bill of materials is incomplete and they removed the CAD files [posted by me elsewhere on this forum]) If you or someone else can post a bill of materials with the names of the parts (and preferably suppliers), that would speed up the build process for me significantly. Also, if you redesign it, you might want to try and make the base stepper motor more accessible for a rotary encoder and/or make a design compatible with higher torque stepper motors (my opinion anyways). I'd try and build the new design if you do. I don't see any reason I couldn't ship you the machined aluminum parts (at cost, no profit) for the new design (note: I can't do this for other people as I would get overwhelmed). |
|
|
|
Post by Max on Jun 6, 2016 17:07:00 GMT -7
The CAD files. Thanks! I've copied them over here github.com/maxosprojects/dobot-arm-CADI'm going to give it a try with timing belts some time. Those should have no backlash (or at least not that obvious). The base must be redesigned completely. The only reason I'm going to redesign it is to fix the stuff I've paid for already. It most probably won't be good, but I want to make use of that stuff and not just throw it out. If I didn't have it I would just design it from scratch, which would have very little in common with dobot. I haven't done this kind of engineering ever before (and not much RT software either), most of this stuff is first time for me. I'm going to try to design and 3D print the lacking parts some time to fix existing dobot first. For that I will of course create BOM. As for the complete hardware project, as I said, I would rather design it from scratch than make people go through machining unnecessary stuff and putting together something ugly. If the timing belt idea works out well I may give it a shot and design it from scratch. All that is way beyond anything I was ever planning to do. Do you need more robots, Mike? |
|
|
|
Post by mikef522 on Jun 6, 2016 17:30:39 GMT -7
Not much RT software? Really? Could have fooled me, but I'm also new to all of this robotics stuff. I do need more robots because there are lots of lab tasks that I would like to automate. I settled on using an arm instead of some xyz gantry like most 3D printers use because 1) it's flexible, 2) it has a small footprint (yet relatively large working space), and 3) the arm from dobot was available for a mere $500. I will try and machine another robot arm eventually (though not certain when). It may be another Dobot or it might something else. Whatever it is, it will have to be cheap. I'll have to search the internet for a good design. The Makerarm ( makerarm.com/) looks like a much better design, but I'm no expert. It's too expensive for me though. I might try 7bot ( 7bot.cc/ ) when they make it available to the public for purchase, assuming they keep the price at $350. Apparently it has a 1kg payload, which is twice that of Dobot. In any case, whatever the design, the code written for Dobot should work for most other robot arms with relatively little modification. The only parts that would really have to be rewritten are the inverse kinematics, the firmware section controlling the motors, and maybe a few other minor things like the limits used for invalid move checking. |
|
|
|
Post by Max on Jun 6, 2016 18:29:30 GMT -7
Makearm is much better designed but only for the tasks Dobot was advertised mostly for - moves in XY plane. In Z axis Makearm is very slow by design (as most SCARA robots). I assume this is its real velocity ksr-video.imgix.net/assets/004/679/030/bbd37a22eece4e1095ccae28a2a7e658_h264_high.mp4Makearm most likely has timing belts in the links, and a couple of them per link, for reduction. That is what I'm planning to try with dobot. As for 7bot, I think even if the 1kg statement is true, it won't do too well with it. It is based on servo motors, which are closed-loop systems and must be pretty sophisticated to pull a load and not oscillate. Servos are based on DC motors. You can see the trembling on those videos. If trembling is not a problem for you, then you could use one of: - www.ebay.com/itm/MeArm-Robotic-Arm-DIY-Kit-Laser-Cut-Components-GUI-Motion-Control-Arduino-uArm/151663873547- www.ebay.com/itm/Fully-Assembled-6-Axis-Mechanical-Robotic-Arm-Clamp-for-Arduino-Raspberry-mor/221982953382- www.ebay.com/itm/6DOF-Mechanical-Robotic-Arm-Clamp-Claw-Servo-Control-Palletizing-Mount-Robot-Kit/252259300694Though the load they can carry is probably not that much. I had an extremely bad experience with an arm like this www.ebay.com/itm/1-Set-6-DOF-Aluminium-Mechanical-Robotic-Arm-Clamp-Claw-Mount-Robot-Kit-Black/321973857176After that I got to reading and was very excited when Dobot campaign started as it seemed to me, not experienced at all, as a pretty solid design, and open-source was the deal breaker as I new then already that software is no less important than the hardware itself. That is my second bad experience. But now, I think, I have close to enough knowledge to design something that could actually work. Dobot can't mill, can't solder properly, can't print because of the backlash. I have just looked at the PCB soldering video again www.youtube.com/watch?v=pfrdBs931_0 and can now see the backlash and how they tried to hide it. If you look carefully you can see how the tip slips a bit when touching the spot to solder and note how they chose the angle and focus to make it look like it actually didn't solder three neighboring traces together. On top of that, the video is at least 5x the real speed, you can say by the smoke going up. So it is just a fraud. So those over a million bucks they raised (on KS + inidiegogo) never went to creating something good and open-source. |
|
|
|
Post by mikef522 on Jun 6, 2016 19:05:59 GMT -7
Yea, I was worried about the z axis speed on Makerarm. Thanks for the videos. Yea, I saw the shakiness of 7bot in the videos. I wasn't sure whether it was due to the servos or the software. I think there are high resolution servos, so I was wondering if they used those. Ultimately, the stepper motor based design of Dobot (for the increased resolution) was the reason I backed them instead of 7bot. And yea, the 1kg payload claim surprised me given that the arm is servo based. Thanks for the links. This arm that you linked: www.ebay.com/itm/Fully-Assembled-6-Axis-Mechanical-Robotic-Arm-Clamp-for-Arduino-Raspberry-mor/221982953382?rmvSB=true is interesting. The design doesn't really seem any different than 7bot. I wonder how well it works. For most tasks, I definitely need high resolution and smooth movement, but for some tasks I might not. I definitely don't plan on milling, soldering, or 3D printing with Dobot, haha. I think there are plenty of open source robots that are cheap enough that are dedicated to those tasks and will do them better (don't know about soldering though). For example, I bought my own Prusa I3 3D printer for $350 (including shipping from China). I love it, it works plenty well enough for my purposes. Most of the tasks I need dobot to do involve moving objects from point A to B (with good accuracy and resolution) and moving precise amounts of liquids around. Most everything that I need to move is lightweight, so the current payload capability of Dobot isn't terrible for me, but 1kg would be a lot better. I do recall seeing several open source stepper motor based robot designs on sites like instructables, or some other sites that I forget the name of, where people had posted CAD files and build isntructions. I'd have to go back through them to see how feasible they are. Anyways, definitely share any good redesigns for Dobot you come up with. I'd be very interested in seeing them. |
|
|
|
Post by Max on Jun 6, 2016 22:37:43 GMT -7
Looked through 7bot again and there is contradiction even on their campaign page. In one place they claim 1kg payload, but on in the "7Bot Arm Technical Specifications" it says 500g According to the motor specifications and after getting approximate dimensions from the only picture I could find, the calculations for the potential payload with the front arm horizontal and the rear arm vertical (not the worst case) are: - front arm ~20cm - front arm lever ~4cm - motor torque ~42kg*cm - force at lever end = 42/4 = ~10kg - force at front arm end = 10/(20/4) = 2kg This is not accounting for the weight of the arm itself and dimensions are very approximate. However, I suspect that when applying a force to compensate the maximum torque the arm would not only fall down, but will start oscillating long before reaching the desired angle, which is quite common with servo motors. I saw that with the first arm I had. Plus, with such a load the motor overheats and breaks pretty fast. I am, of course, speculating, but given the motors are unknown, as well as the team making that arm, chances are good that I'm right, because with such setup (nothing is known and we are not in a fairy tale, so no miracles) it is a pretty common outcome (as we've also seen with dobot). You mentioned precise servos. Dynamixel, for instance, are pretty good servos, but look at the price tag. They are pretty expensive for me to just try out, so I can judge only by videos and specifications (which I would trust in this case). Let's take the cheapest from here www.trossenrobotics.com/c/dynamixel-pro?sort=PriceAscThe backlash is pretty decent, the resolution is great, but even with the gears the torque is about 12 times lower than you get with the geared motor on dobot, and at that price! Sharp moves are going to make it to oscillate a little bit, though with appropriate planning (e.g. proper acceleration/deceleration) should be fine once you figure what to do with the torque. You could improve torque by adding reduction, but then it is going to get slower. Plus, if it is a geared reduction, then you would get backlash again. A great comparison to see the oscillation www.youtube.com/watch?v=nw6s71-SIZQThis www.trossenrobotics.com/p/mx-106t-dynamixel-robot-servo.aspx to this www.trossenrobotics.com/dynamixel-pro-l42-10-s300-r-robot-actuatorSo there are not many alternatives to steppers with a timing belt reduction really (for budget robots anyway), which is the conclusion I came to a while ago. I'd be interested to find a ready-to-print design myself to not waste time on that, but so far I haven't found a simple enough to be printable, and precise enough to not suffer from backlash. All I found is either not printable, or suffering the same problem - backlash because simple gears are used for reduction. |
|
|
|
Post by mikef522 on Jun 6, 2016 23:06:58 GMT -7
Awesome, thanks for info!
Damn, those are some expensive stepper motors servos. That resolution is insane though! 501,900 steps/rev! Holy crap!
lol, yea, I don't think 7bot used anything with that high of precision.
|
|
|
|
Post by Max on Jun 6, 2016 23:18:33 GMT -7
Those Dynamixels are servos, not steppers, I think.Steppers shouldn't oscillate.
|
|
|
|
Post by mikef522 on Jun 6, 2016 23:27:53 GMT -7
that's what I meant; was a typo.
|
|
