Adventures in 3D printing...QiDi Tech 1

The adventure continues....

I have been blessed with a really forgiving machine for the past few weeks.  The machine seems to print well and has been a great deal of fun to work with.  As far as printed parts go, I have only had trouble reproducing one part... the rod end for the new robot arm.    The rod end needed a lot of attention and took the better part of a week to print three of them.  Such a small part and such a huge problem.  Anyway, lets look at some pictures.

The carbon fiber tubes finally arrived.  The 6mm OD tubes had to be cut to length  before they could be fitted with the ends.  Putting these together was a breeze .

The 15mm tubes also had to be cut to length, but that wasn't really and issue.  The problem was the 15mm tubes had an inside diameter of 11mm and the rod end that fit inside them was  a lot larger.  Still not much of an issue.  I went ahead and modified the rod end with TinkerCAD and down loaded the revised part.  This new part absolutely would not print.  Most of the printing problem can be attributed to the failing surface of the build plate.  The build plate surface was starting to become detached and the part would detach from the surface 3/4 through the print.  I somehow managed to get two of the three rod ends printed  before the build surface failed completely.  The last part took three full evenings to figure out how to print.   

QiDi replacement build surface sheets are in rout but in the future I'm thinking of using a PEI sheet...we'll cross that bridge soon.

This is the last of the three rod ends.  I tried everything to get this one to print... brims, rafts  and lots of glue.... nothing worked....so I cheated.   I  redesigned the part to be 3/8 inch shorter, thus reducing the print time and when I awoke on Saturday morning this is the result.  Meh, the slightly shorter part  ain't going to be an issue.

Fail, fail, fail, fail, fail, fail, fail, fail, Success!

....and here is the arm fully assembled.  This arm features miniature ball bearings on every joint and the herringbone gears provide nearly zero backlash.  Figure something like this would cost about a hundred bucks in purchased parts to complete.  I recycled some of the parts from the orange arm so my cost was a bit less.

Another shot of the arm...

Arm all wired up and running g-code.  I wrote a few macros to validate the arm has repeatable motion... and it does!  The only minor issue is the servo for the gripper is acting funny.  Once I sort ot the servo problem I'll shoot a video.

Stay tuned!

Nice craftsmanship! The carbon fiber tubes are a really nice touch in this build!

I'm still eagerly awaiting the video of this thing in action. Love the work so far.

The good news is this arm works so much better than the orange one... the bad news is I have to make a more in depth video to show how cool it is.

Stay tuned for a few more days.

In reply to Atomictaco :

Hey Dave,  thanks!

I had some time yesterday to fool around with the robot arm. The focus was mostly experimenting with different g-code command and such. Normally I would have written my own software to drive the stepper motors but using a G-code sender to communicate with the Arduino makes operation a lot smoother. The G-code commands automatically do a lot of hard stuff when it comes to accelerating and de-accelerating a stepper motor. I almost feel like I'm cheating by using G-code. At some point I'll need to figure out how to integrate the G-code driver with my home brewed software.

The arm works exactly as I expected , the movement is repeatable and smooth. You can only get this type of precision using stepper motors and not servo motors. There are a lot of servo motor based robots out there but most lack any sort of precision. The only servo on this arm is in the gripper and that is fine because the gripper just opens and closes.

One of the drawbacks of using stepper motors on a project like this is the motors need to run at 100% of their power output. On stepper motors the amount of power they use has nothing to do with the speed. Its all about torque. When a stepper motor stops, it is still drawing a E36 M3 load of power to hold the position. If the power were to be throttled back the arm would 'relax' and basically just go limp. Meh, we are only talking about 1 amp of current on this arm so its not that bad. Anyway, because to the constant power draw the arm can only operate for 30 to 45 minuets before it needs to be powered off to cool down.

One of the things I noticed was the larger stepper motor that is used to drive the main arm doesn't ever get hot, on the other hand the two other smaller motors will heat up rather quickly. I'll need to look into this.

Everybody likes pictures, so let's do show and tell now....

Stepper motors require a bit of electronics and code to work.  What I find amazing is how small the drivers have become.   The heart of any robot including 3D printers is the stepper motors.

This little chip handles all the power and sorts out a bunch of stuff so making a stepper motor spin is much simpler than in the past.

When I built this arm, I elected to use a larger motor (on right) to animate the main arm segment.  The motor is still a NEMA 17 but the additional length provides more torque.

As a side project I'm building a teach pendent of sorts.  With a bit of magic I can use this little guy to teach the robot arm new  tool paths.

And here is the long awaited video.   About halfway through the video the creepy robot shows up....  The creepy robot uses Cambot technology and other things that I wont disclose on the internet. 

Stay tuned!

Let's make some Miata parts

Ok, so todays project is to make a panel to replace the power window switches on my wind up window Miata.  The above photo is a random picture I gleaned from the internet in case some viewers don't know what a NA power window switch panel looks like.  Anyway, my Miata has a hole where the switches used to be since I converted to wind-ups.  A simple cover would fit the bill, except my car has stuff that needs to be mounted to the panel...

On my Miata there is an overheat alarm buzzer and another buzzer that is connected to the Megasquirt.   Both buzzers and a switch will need to get mounted on the new panel.

A brief search on Thingaverse.com yielded something that would make a great foundation for this project.  This thingy is a panel to convert OEM power window switches to aftermarket switches.  Meh, whatever... the thing I'm interested in is all the hard work for the pattern has been done and all I need to do is modify this design for my application.

A bit of fiddling around with TinkerCad and I had a part that would fit my application.

A short while latter the printer spit out a part that is as unique as a snowflake.  Now keep in mind this part is PLA and PLA plastic is not recommended for automotive interiors.  Something to do with the heat...  Whelp, right now its zero degrees out so this part should be fine... for a while.

Fits like a glove.... except its orange

Whelp, lets try again in black PLA

The black PLA feeds into the print head through a semi automated process.  changing colors is easy peazy, however  most of the time a color change will require the part be reprocessed  through the slicer program to change the temperature profiles.   In this case the black PLA  runs a fair bit cooler then the orange so a few adjustment were needed.

Through the magic of 3D printing a new part appears.  Notice the odd patterns on ether side of the part.   These are support columns that are used to support  the part during the print.

The supports snap off with very little effort. So far this has been the only effort in building this thing...

The part fits perfect and the surface looks fine, but we can do better.  What we need here is simulated carbon fiber!

So here we go,  Got some 3M carbon fiber wrap off amazon and soon we will have a carbon fiber part... sort of.

Fast forward a few min and the finished part is installed.  Of course this experiment  will need to be done again, but next time I'll use ABS plastic.

Hmmm, Looks like this area needs a Raspberry Pi with a 5 inch touch screen display.   

Stay tuned, coming soon is the Rasberry Pi information center for a NA Miata..

Alright...alright...I watched the video....and the Miata parts. Now we are having some fun!

Btw...can you build a whole Miata?!?

Doh!

I'm officially jealous of your printer. 

It's 5 minutes of fumbling around wishing for a third hand then a bed leveling for me to change colors. 

LOL, thanks for the comments.

A quick update... the filament for the white robot arm was actually ABS and not PLA... this sort of explains why I had such a hard time printing some of the parts.   I need to  go back and try printing some white parts with the correct settings.

Stay tuned!

Hey Doc,

What software did you use to run your robotic arms?

In reply to revhard :

The software for the robot arm is sort of a hodgpodge mix of free stuff that works on a windows XP system. 

There are two components to the software. There is the Arduino stuff and a Java based Universal G-Code sender.  The Java based G-code sender took some fiddeling to get it to work with XP.  You need to install the latest Java platform on your system.  In my case the latest Java platform is not compatible with XP  .... but I installed it anyway.  Once you have Java installed the rest is sort of easy, but can be frustrating.

For the arduino stuff Link to robot arm on thingaverse.  I used the GBRL files that are linked on that page.  The page also has a link to an inscrutable page that has a video on how to install the GBRL files on the arduino.  The GBRL files are unique because they have been written to use a servo for the fourth axis.

Be warned, there is a language barrier and the video is hard to understand.  I think I watched it 10 times to figure out what the guy was doing.  I also think there may be a mistake in how the video indicates where on the CNC shield the drivers plug in.

The universal G-code sender that is linked on the above mentioned page was NFG for my application.  

Here is a link to the Universal g-code sender V 1.09 that I used.

I believe it took three or four evenings to sort out everything. 

I would recommend using an Ardiono UNO with the CNC shield and plugging in the motors and get them working before you install the motors on your project.  

EDIT..

I guess I should explain how it all works.

The GBRL code configures the arduino as a slave to a PC.   The arduino waits until it receives a g-code instruction from the pc and then processes it.  The universal G-code sender has four modes... Command line, File mode, Macro, and machine.   These modes are self explanatory, however you will want to use the command line mode to do some setup stuff.  Depending on your application, the acceleration slope will probably need to be changed to prevent the motors from missing steps.

Anyway, if you need any more help or clarification feel free to ask.