As promised I've started a separate thread on this. I had mentioned over at my Alan thread that I was thinking about using hoverboard motors for a rolling base for the full bodies of Alan and Alena. These motors are super cheap at about $25 a piece on ebay. They are 3 phase just like the 3 phase hobby motors for cars and planes. So there should be driver support for them.....well yes and kinda no.
I have been watching a motor controller called the ODrive. Its all open source (hardware and software) and its a dual motor driver. I posted a video of these motors on a grocery cart driving around using hoverboard motors. So I bought an Odrive. I ended up buying a whole hoverboard so I could use the charger, battery and some of the connectors. This one came from Frys and cost $100. Its good upto 265 pounds.
Well now the bad. The ODrive is still a work in progress and not too intuitive. But they have a good write up on how to set them up with the hall effects encoders. All went well until I tried to save out to the board. Over and over again i could not get it to work right. I posted on the forums and got no reply back as to what the issue could be. So after days I began looking for another solution.
I found a great source in china that is making opensource stuff for skate boards. The got 2 single motor drivers. Each one has a standard servo connector and i can use two of them for mixed mode (tank mode). Each board costs $85. So its a bit more than the ODrive ($150 with shipping for the 48 volt ver.)
The name of the board is the FSESC 4.12 50A Based on VESC 4.12 from Flipsky ESC. And can be found here: Flipsky ESC
This controller can be completely programmed with their really great software..automatically figures out all the poles for the phases. Set up the frequency for the pwm and even have shut offs for voltage mins etc. They have a good youtube channel with vids on how to set up everything.
I have the video of the break down of the hoverboard done. As soon as I get these control boards I will be doing tests with the ezb to control them. Then I'll put together a final video and post here.
Other robots from Synthiam community
Jstarne1's Iotiny Recon 6.0 Rovers , Educational Robots
Pabi!O's Pabi!O's Wall-E
in the meantime, I’ll continue living vicariously through you!
Brave kids with a fast go cart controlled by Dad ? Sounds Great ! Normal to me...
Thanks for starting the new thread. Are the motors normally 48v? I thought 12v.
The controllers sound interesting. I am looking forward to your progress. As soon as my work projects let up, I hope to begin with the base unit also.
Will, we have no Fry's near me here on the East Coast. Could you tell me what the Manufacturer name of the hover board you got is? This way I will be able to share my ideas for the base using the same components.
I'm using a hoverboard called the Prime R6. Its some generic hoverboard...but i got it because it was on sale and it can carry 265, which is about 40 pounds more than most. The manufacturer has it on sale for $144 reg $349. But you might be able to find a better deal elsewhere like I did.
My end goal is to build a base with navigation (hopefully based on the unit DJ will release soon).
Hopefully with the hall effects encoders there can be some good positioning. My biggest fear is not being able to control both motors equally. Meaning if one motor moves slightly faster than the other it will not drive straight. So there will be some challenges ahead!
I also need to refresh my brain about PID loops? They may help with this? Ptp may know.
I've downloaded the software and Its very very thorough and covers so many options. There is a set up wizard which i will use to program them with and then I will dig deeper to adjust settings manually.
I've already found a fix for my problem about controlling both motors and keeping them straight. There are two wires that connect both controllers called CAN. One becomes the master at ID 0 and the other becomes the slave at ID 1. You program both but the master delivers the signal so they stay in sync when travelling straight. The hall effects encoders will assure that...now bolting them straight to the board may be more of a challenge
I just clicked submit and have a hoverboard on the way to Connecticut. I paid a bit more but it's on its way.
Did you plan on a 4 wheel platform? More stable than 3. My goal was to mount my 2 batteries on the base. One will now be the one included with the hover board. Still, I need weight to offset the head, torso and arms.
Yeah you definitely going to need some counter balance. In the end I will be using some sealed larger batteries for Alan and I hope their weight in the base will help offset the motion at the top! You can spread the wheel base out but still gotta get it through the doors!
I am planning on bolting the wheels to aluminum angle then bolting the angle to a steel base plate ( more weight,)( final design) . The aluminum angle will allow me to adjust the tracking alignment. ( No shims LOL). I haven't decided on casters or omni casters? The design needs to develop. Prototype will be what ever I can find. A robot about 5 ft tall is my goal, maybe.
Here is the screen from the VESC program. This screen is the set up for PPM...you can see the mapping over Can Bus for two different controllers acting as one... also includes traction control and difference...ie as in one wheel slipping under load or on slick surface then working to get going at the same speed to go straight. Powerful hardware and software.
I also am excited about ramping. You can ease in and out of your speed. There is also braking which is important if your robot finds itself on a hill. And reverse. Alot of these controllers do not reverse, because right now they are just set up for skateboards and you really only go one direction. But this has reverse. You can set a separate reverse speed as well.
FOC:
FOC is a different method of spinning up the motors. It was originally developed for high-performance motor applications that are required to operate smoothly over the full speed range, generate full torque at zero speed, and have high dynamic performance including fast acceleration and deceleration.
I promise, i will stop now and leave you alone to play. It's after 6 here and i will finish up work soon.
Looks like boards go on order tomorrow. Poor credit card this month, especially if Dj comes through with the rest of the parts we need.
You are saying these can work as servos?
Can you imagine the massive shoulder and elbow servos using these? Instant full torque without gearboxes. Once I get the new hover board, I may buy a replacement motor to see if it can be hacked.
You can use nearly any type brushless motor with these controllers. Skateboard, hobby style like this 320KV Turnigy Multistar and you can use standard DC motors too. Below is one of those motors in a 3d printed gear housing.
Here is an old video from the author from 2015, where he is playing with the implementation of position control and a skate board motor. Here he uses a encoder to turn the motor 1:1. Jump forward to 8:00 to see him move the motor.
Slippage then is not a concern. We end up hoping the nav system will trim as we go. I had planned originally to use an Adventure Bot to test the Nav system. What are your thoughts?
5 wheels might work to stabilize your platform.
Do you know a Waldo Mason? He has some nice work.
Have you gotten the boards yet?
Ron
I do know of Waldo, but not met him personally. He does alot of body work...dead bodies etc. Very good artist from UK.
I have begun digging around for the material for the base. I was wondering if suspension should be considered even though it will be indoors. It will be a bit heavy and I don't want it rocking and bouncing over rugs, thresholds, tile etc.. What's your thoughts?
Maybe suspension on the upper torso?
Well Christmas came early. The controllers from China just got here. Hopefully I can get the wheels rolling before the weekend! Whoo Hoo!
First thoughts. The signal going in to the Flipsky from the IOTiny might not have the resolution that the controller is looking for. The resolution goes into the 4th decimal value. The more resolution the finer the control. Using the mobile joystick has a very small sweet spot to control speed, but mostly goes from full stop to full on in either forward or reverse. In the video you can see I can find that little area..sorta. There is a wizard for the input as well, with alot of choices...so another area to test other settings.
Also motor sounds a little loud and I suspect this have something to do with how the wizard detects and sets up the motor. I think there maybe an area where you can manually add the poles.
I have something going on this weekend until Tuesday. Then I can keep testing. Until then enjoy this first test.
Much more detailed video coming...this is just to prove its working.
Items:
Flipsky controller
Hoverboard motors
EZ Robots IOTiny
Synthiam with mobile interface and joystick controlling in continuous servo mode.
VESC
36V battery
Video shows 36V Battery connected to Flipsky and IOTiny connected to PWM with onboard 5 volt regulator (had to disable my battery is low). The hall sensors is connected to the Flipsky as well as the 3 phase connections. Controller is in "current mode", so there is no braking engaged. There is another mode where you can add braking at the cost of no reverse.
I'm looking forward to receiving my parts so I can play too.
I dont feel like I am getting those types of movements yet. The motor seems a big crunchy on lower RPM. I do know that the phases have been set up correctly. And the Hall effects sensors are working.
After working with the ODrive, I learned a lot obout the importance of poles. In their software there is a place to put the magnets number in, in this case it’s 30 magnets, you divide that by two to get your poles. But in VESC there is not a way to add in this value when using the wizard. So I gotta figure out how to do it manually.
Be Well, catch you next week.
The Brake or Regen mode has no reverse but by using a relay two phase wires can be swapped to change direction. I haven't played with the software yet. Hope to get some time to read up and check out videos.
Also still thinking about suspension. I wonder if an air bag with coil over shock absorbers from an rc buggy would work?
My new hoverboard is here. Tested out well and took a full charge. Cost was $ 109 with free shipping. Capacity 200+ pounds and a 4 amp hour battery. I will start the teardown soon. Next, waiting for Vesc.
My test unit will be similar in design to that in the video. I hope we get an update on the EZ nav system soon too.
I am curious as to weather or not you could cut the speed range down even more? Can't wait for my boards.
Looks good, and been following your posting on this as I'm looking for something similar build for my InMoov. I'm tried of pushing it around on a chair castor wheeled base with a chrome pole up its backside!!
I can't remember, but did you buy a complete hover board and strip out the motors, rather than buy individual hover board motors?
Was it cheaper to buy the complete hover board?
Chris.
@cem Yes I bought a whole hoverboard, because it has everything you need, battery/motors/charger/connectors. I will put up my tear down video on youtube in a day or two. I was going to wait until I had it all going...but prolly should do over several videos.
Only issue now is the bird chirpy high squeal. Not sure if that’s duty cycle or phases are still off?!
edit: found the issue. I had changed motor type to BLDC from FOC. FOC is so much quieter.
Does the motor shutter when first started at a very slow speed? (Due to sensor less mode).
Do we need or want to use the hall sensors for the Nav system if not needed for motor control.
I have a new issue that wheels will get lost in their signal and jerk crazy, but only sometimes. If I reset and use the signal positions with the rc controller it doesn’t seem to happen.
Here is a video of both working in mixed mode ( tank mode) at different speeds using the built in joystick in mobile interface. Still jerky between direction changes but pretty sure it’s that dead band size. ( this is at 50% speed setting).
I need to know the exact max (180) min (1) and stop (90) in hertz frequency for the EZB anyone know? And what is the range for the wi-fi?
I'll tell you...near disaster !!xD Impossible to run all those things together in my brain!!! My garage fridge got a dent in it!
I have this thing dialed in perfect...just need a safer speed. This is still at 50% I think 2% would be safe for a robot in your house (maybe less than 1%?).
Remember that most robots are going to be controlled with PID so you can control speed and direction with sensors etc. This is just connected to a virtual joystick in Synthiam.
I'm still waiting for Vesc. Soon I hope.
Thanks for the video on the Hoverboard break down!
Its certainly has some speed on this, I know you said you can dial down the speed to about 1% or 2% for indoors, but I think it’s way over the top, for what I’m planning to do with my InMoov. It would be such a waste of power and cost.
I think I need to look at a more low keyed setup and a lot cheaper option??
BUT I’m very jealous!!!!
Looks good! We both knew initally it would be too fast but it will be reliable.
For the full setup I think it will be a good value. Motors, battery, controllers and charger for about $ 300 I think isn't bad.
I am anxious to see the speed test then the dialed down test. The "morph" test video should be about the speed we will probably use and it seemed smooth. The battery life should end up super great.
Yeah I spent about $300 total, not including the ODrive (which I will use for another project). With other options you have to spend extra on batteries, charger, gear boxes or belts and pulleys to change ratios etc. Seems like the price could get outta control.
I should have no problem reproducing the speed to that of the morph video. As the load increases in weight then the amount of current to get that to move would go up...so figuring out that sweet spot would take some experimentation once you have your robot built and on the platform. Great thing is the UI is so intuitive and easy to use that making adjustments is quick and easy.
By the way: did you ever find out what the extra wire on the eBay motors you have was used for?
I have used a small magnet run along the perimeter of a similar motor to count magnet poles without taking apart the motor. You can feel the attraction of the magnet pole as as you run the magnet around.
Perry has a great idea there, except these stators are covered by the rubber wheel..maybe you could come at it from the top with a small magnet?
I kept the zinc frame for the wheels as well..until i build a proper frame for it and have some aluminum wheel mounts built.
My Flipsky's came in about 2 weeks. I opted for the DHL shipping, but it comes to the US as China postal service first, then arrived in LA and picked up by DHL seems like some kinda way to save them money.
One thing I will say. Its the first time I have had such great support from a company in China. My shipping was not updating, so I contacted them via their facebook page with my order number and they shot back a facebook message right away with china postal tracking to see where my package was.
I'm still having troubles with getting the mobile controls to work with VESC correctly. When I turn for example, one wheel should turn opposite the other wheel. I can do that sometimes...other times no. So in the video you see that at the beginning it was doing the opposite direction, but was unable to reproduce it when on the ground.
I would use an RC controller to test with, but none of mine have a mixed mode (differential mode/tank mode), so each stick would just control a side (L or R) so controlling direction would be a challenge.
Another configuration for a robot with these motors would be both motors moving together (forward/backwards) with a another wheel controlling turning.
Dumb question, do the boards need to be grounded together for some reason ?
It works great with an RC controller very snappy and stops as soon as i let off the controls. The ARC has about a second or more before the controls respond. My thoughts are on the frequency. I have to set up the controller in VESC and it uses the input from the EZB as its control frequency. It will detect neutral, full forward and full back. During the set up you look at the numbers and input those values. The EZB is very finicky on giving out its position with this software. It seems difficult to obtain the correct values. Maybe because my finger is on a track pad and not getting into those correct positions?
Like i said most people are not going to use their robot like this ..(driving around with RC control), but instead use PID or some other way to tell the robot where to go or things to avoid as in the case of the lidar, etc.
But...if i knew the exact frequency for the IOTiny and EZB I might be able to just input those and get a better response. (I.E max, neu, min full spectrum)
I will think about this.
Also PID loops can require a lot of playing with too. I hope I get my toys soon.
Does the movement of the joystick in the phone or laptop give the proper digital output signal in a consistent manner and time frame.
Jeremie. If you read this can you offer some info?
I was wondering, if you slave the output and put a servo on the slaved output to confirm the Io tiny works normal. It I weird that the signal hangs.
Does it do it if you load an Adventure Bot basic program and use the pad to run it? Just wondering if the joystick function is hanging?
Also I increased the speed to 10% for testing and it takes nearly a full 2 seconds for wheels to get signal to stop after i release the joystick on the mobile app. So the Flipskys are not in love with IOTiny.
I tried the Adventure bot and it’s slightly better. I changed the stop position value down to .01 and got even better results. But there are moments where it stops responding altogether. Then I have to release the joystick and try again. I found forward to be ok but reverse was hard to find . I also tried out the forward and reverse icons in the mobile interface ( full speed forward) with no movement at all.
Edit---The video you have does look like you got it on the right track if you can iron out any bugs!
I'm also reading that there is a safety out of range feature built into VESC, where the motors are not responsive if the the frequency goes out of range.
Everytime the ezb stops sending a pwm, the receiver has to relearn the signal range.
Your receiver must know the signal range - OR - specify the signal range using the continuous rotation Movement Panel direction values.
Read the ezrobot tutorial on how a servo works to understand how a servo works with pwm. And why stopping pwm for a not smart receiver has that effect.
Nice project!
I got curious ...
Q1) Why aren't you using the odrive controller ?
Q2) Do you know the wheel encoder resolution ?
The wheel encoder count is 8192.
1) finishing the wiring
2) installing and setting up VESC
3)connecting to the Flipsky’s using VESC
4) setting up the motors and finding the phases order
5) saving the settings to the Flipsky's
6) mapping the input signal from the IOTiny and saving to the flipsky
7) connecting the IOTiny
8) creating a mobile app with a joystick to control the wheels
9)creating a platform to test on!
Liam is braver than I thought. Looks like you kept your word about strapping on your son and driving him around, even if not at full speed ! Well done Liam !
It sounds like I missed all the fun. I guess you fixed the control issue. I be starting my base tomorrow. Back to work later today. I look forward to your videos. OOPS ... I just realized I didn't look back far enoughin the posts for all the answers.
I'm finishing up the video for the above 9 items and will post later today. (its a longer video about 9 min)
and the videos.
I didn't voice over which can be more explanatory, but after i watched the video several times i think all the info is there in lower thirds.
While wiring I was thankful the original hoverboard control boards identified the hall sensor wire colors which allowed me to correctly connect them to the plug. I made an assumption the white one is the "tp" wire.
My goal is to add a camera to the platform which will let me end up with a big Adventure Bot. I will load the bare adventure bot program and test using that.
">
Get my email?
( getahead)
I assume I need to build a 5 vdc supply with a capacity available to run the add-ons. Do you agree? I didn't want to depend on the Vesc. My plan is to use the Tiny for any Nav system I build. Any ideas ?
Work has become busy, so no setup or tests yet.
I hope I get to play before I have to leave for Arizona.
Thanks Will.
Yeah I'm fully expecting to have to change the % to something in the 25% to 30% range with the added weight of rider and chair.
do you know If the hover board will work with a sabertooth 2x25To control each wheel/motor?I know the max voltage is 33.6 V for the saber tooth, there’s other motors out there that would work but they only hold 60 pounds and I would like to have a lot higher Weight limit for the robot.Also I believe the hover board has an encoder built-in which would help with controlling the movements of the robot in the distance Interns traveled.
Thanks in advance,merne
there are many advantages to using the software to program them like continual consistent speed up and down hills, and to increase speed to get the mass to move at consistent speeds and even break regeneration for the battery.
Video Thumbnails Maker