Calvin Stence @calvinstence - Tumblr Blog

A Comedy of Errors - Printrboard Rev F5 Firmware Editing How-To

Preface: I am writing this after the fact, and narrating events as they happened to me. I will (somewhat shamefully) admit that this took me about a day and a half to figure out. Hopefully it helps someone figure out how to do this, if I am correct it is a fairly niche problem as there is very little information for this specific board (rev F5). The process I went through to program the board starts several paragraphs down, feel free to skip around and control F, you will probably find something that is relevant to your particular situation.

As a gearhead, I am enthused by almost all things machines, and love to make. Naturally, I have a 3D printer (a Printrbot Simple Metal with heated bed from Dec. 2015).

It comes with a Ubis hotend. Over its life, it slowly became more and more averse to having filament pushed through it, and eventually rendered the printer slightly better than unusable. It was time for a change, as you can see...

Melted plastic had permanently cemented itself around the nozzle, rendering it hoopajooped. I had heard plenty about the E3D V6, so I got one and began the upgrade.

The build is fairly self explanatory, and E3D gives excellent instructions on how to do it. If you have the same Printrbot as me, be warned--the heater block WILL COLLIDE WITH THE BASE OF THE PRINTER IF YOU HOME X Y AND Z AND HAVE THE LONG SIDE OF THE BLOCK TOWARDS THE REAR OF THE PRINTER! And this WILL bend the heat brake--ask me how I know! If you scroll down you can see the orientation of the heater block that I ended up using.

With that done, it was time to modify the firmware so that the printer would recognize the new thermistor in the new hotend.

This is where things got dicey. There is almost no information on how to perform this modification with this exact board, the Printrboard rev F5. And if you’re like me (a somewhat boneheaded mechanical), this poses a great challenge. There are quite a few ways to perform these upgrades. If you just need to flash your firmware, you can use Atmel Flip to flash the Printrboard (I had done this before, there are a few guides on how to do this, such as this one https://printrbot.zendesk.com/hc/en-us/articles/202457834-How-to-Flash-Printrboard-Firmware). However, we don’t want to use a pre-baked firmware for this because we need to CHANGE the firmware. How the hell does one do that? This is about where I was at, and boy was I lost.

I had remembered watching a video by Thomas Sanladerer, an excellent YouTube content creator that makes informational videos about 3D printing. It shows the exact process of how to upgrade your firmware for a general 3D printing control board, such as a RAMPS or RAMBO board (Link: https://youtu.be/3gwWVFtdg-4). But I didn’t have one of those, so when I tried it, I uh... had some problems.

First, Tom recommends to use the Arduino IDE to flash the board. Now you might say, “This is a non problem! You can just tell the Marlin firmware to use the Printrboard Rev F in Configuration.h by specifying that it’s a printrboard and then hit upload. Easy!” Well, maybe for you, my microcontroller-familiar friend. Let me tell you a story of honest confusion, hunger induced frustration, and general disorientation as a hardcore mechanical tries to figure out how to do basic things with microcontrollers and C++ configuration files.

At first, I was confident. I had watched Tom’s video and thought that there was no way I could mess this up. And it would have been, if the Printrboard was based on the Arduino Mega like basically every other 3D printer control board. Haha, guess what, it doesn’t.

What the hell is that? And why is the Arduino IDE telling me that it doesn’t recognize a Printrboard when it’s one of the supported options? What does it mean Unknown CPU?

Later I would realize that these error messages were telling me exactly what I needed to do. But, in the moment, I was baffled and frustrated that it wasn’t working. More importantly, I was neck-deep in the names of different microcontrollers that were clearly reluctantly created by computer engineers who really didn’t care what the hell they called their new product, and just decided to use the seventh thing that came to their mind.

At this point I thought that the Arduino IDE was telling me that it couldn’t identify the printrboard. While it was true that it couldn’t identify the printrboard, it wasn’t telling me that. Nevertheless I soldiered on, and figured out that I needed to install another stupidly named piece of software (link https://www.pjrc.com/teensy/td_download.html). This let the IDE actually recognize the printrboard, and the fact that it doesn’t actually use an Atmega 328 processor, but an AT90USB1286.

So that was a revelation. At this point, for the first time in four hours, the IDE could actually see the damn board. Keep in mind that I still had made no progress on the actual errors that the IDE was spitting at me. So I set off there.

The compiler kept claiming that this was an “unknown board” and that it didn’t have the pins for the board. I didn’t understand since I had done exactly as it asked and put “BOARD_PRINTRBOARD_REVF” and that the board was explicity listed in pins.h. After trying the number of the board, 811, in exasperation, I finally read what the IDE was actually telling me, saying that it lacked an identifier. I don’t know what I expected, this being C++, but you have to not just #define 811, you have to #define MOTHERBOARD 811.

And now the momentum built (or so I thought). The verifier progress bar advanced a little bit more. Next it complained that it wanted the ”traditional marlin” pin layout instead of the “AT90USB1286.” So I looked in the pins_PRINTRBOARD_REVF.h, and once again actually read the error message, directing me to fastio.h and comment out _AT90USBxx_TEENSYPP_ASSIGNMENTS. Now I felt like I was on to something! Next was something about DAC stepper current. I am writing all of this after the fact, but it was somewhere, I think in configuration_adv.h. The change should be similar to the previous ones, but I’m not sure, it could be something else.

By the way, if you’re installing the E3D v6, the steps for modifying the firmware are well detailed in the installation manual you can find by googling “e3d v6 install.” These changes are much easier by comparison.

The verifier finally completed successfully, and with the board I was using recognized, I selected the USBtinyISP programmer in the “Tools” tab of the Arduino IDE and uploaded it to the board. I fired up the printer and to my amazement, it seemed to work. Seemed to work.

The hotend performed very well. I didn’t have much of a way to check if the temperature was correct (i.e., the firmware changes worked). I touched a piece of scrap PLA to the block at about 60c and it seemed to go into glass transition, and heated it up all the way to see if it would melt. It did, so it must have been in the ball park.

So I used the printer for a while until I was ready to try enabling mesh bed leveling (which isn’t even possible on a printrbot that doesn’t have an LCD screen). At this point i noticed the little teensy board programmer that keeped popping up saying that the “Marlin.ino.hex” file was too large. I had completely ignored this thing up until now, simply reading that the arduino IDE had “Uploaded Successfully” and assuming the teensy programmer was redundant.

...it wasn’t.

I had somehow figured out that most 3rd party AT90USB1286 programmers could only write 64 kilobytes of program to the chip, even though it has 128 kilobytes of memory. Why? I have absolutely no idea. Regardless, I remembered that I had used Atmel FLIP to install a new prefab printrbot firmware onto my printrboard a while back. You can find a link to that above in the first part of this post. So, I tried that, and it worked. I then went to test the unified bed leveling and the printer instantly crashed as I tried to raise the Z height and it crashed the print head into the bed. Sure enough, NONE OF THE PREVIOUS FIRMWARE CHANGES I HAD EVER MADE WORKED. NONE OF THEM AT ALL. So, suddenly, all of the changes I had progressively thought i had made happened at once, and all of the default settings in Marlin that were wrong showed up, and everything went to shit. But now I had completely figured out how to change firmware settings on my printrboard. And you too can figure it out if you read this... hopefully.

Something I repeated to myself a lot during this is that if you’re not making mistakes, you’re not learning. As you can see from this article, I learned a lot in the past few days. Here’s the hotend on the printer! It makes a huge difference. No more roasty Ubis for me.

Happy printing!

Email me if this doesn’t make any sense and I’ll try to help.

#3dprinting #printrbot

This was a small analog pulse rate monitor I built for my Circuits and Applied Electronics class last semester (a Mechanical Engineering elective). It wasn’t pretty, but I learned a whole lot about analog electonics and how electricity works. I have to say, I was intimidated by electronics before taking this class. I didn’t even know how a breadboard worked. But now I do, and it’s great to be able to understand my EE colleagues a bit better. I wanted to learn how to do some electronics because I feel it’s necessary for all engineers to have at least some understanding of how they work. Most modern devices and advancements have to do with software and computing hardware, and much less hardware... and though I do love mechanical things and hardware most, I have to recognize that I must be well rounded to do a good job as an engineer that works on new and exciting things. In the future, I plan to work with microprocessors to build a testing rig for linear actuation systems subject to inertial loads (in particular to characterize the torque vs. speed capabilities of stepper motors), as there isn’t much of a clear answer when you read the stepper data sheet for what sort of torque they can supply. Let the learning continue!

#electronics #3dprinting #learning

This video is of the Locomotor Control Systems Laboratory’s prosthetic leg, which I helped repair in my work at the lab. In the video, the leg walks up to 3 miles an hour as an amputee wears it, which is an incredible achievement in the robotic prosthetic space. The repairs I did helped it achieve that goal by eliminating inaccuracies in the construction and motion components. Check out the full details of my process below if you like!

Last summer, my professor, Dr. Robert Gregg, principal investigator of the Locomotor Control Systems Laboratory (it’s where I work, here’s the website, http://www.utdallas.edu/~rdg130030/locolab/, check it out!), asked me to help out with some mechanical issues the lab’s prosthetic leg had. Normally I work with the exoskeleton team, but I was interested have a chance to work with the prosthetic leg, as its design philosophy is very different. It needs to generate enormous torques (135 newton meters at the ankle, almost as much as a small car) to move joints, and uses linear actuators and levers to do so.

The problem was that the knee load cell, used for measuring the forces produced by leg’s knee joint actuator, had failed. This was strange, because it had never recorded a load higher than their rated load. So,I was tasked with figuring it out, and making the necessary repairs to make it work. To keep operations running, I machined a load cell replica out of 4140 steel at the Dalls Makerspace and gave it to them so they could continue bench testing of the leg without force measurement at the knee. Then, I set to work determining the causes of failure.

I didn’t design the leg, and the people who had were long graduated, so I had little information on the leg’s history and why it was built the way it was. This put me at square one, and I had to figure out almost everything on my own. Whenever one is trying to fix a faulty machine, it’s critical that the root cause of the problem must be found and fixed (when possible), as a band-aid fix will simply end up causing more delays and headaches in the future. So, I set to work by putting my hands on the machine, and feeling how it moved. Immediately I noticed a small wiggle when both the knee and ankle joints reversed motion. With the type of components in the assembly, I immediately knew this was bad news. I continued to isolate the wiggle, and found that it was between the screw thread and nut (this is a ball screw, so the “nut” is actually a tube of recirculating ball bearings). This allowed the screw and nut to have an angular offset of almost three degress--and even though the sine of three degress is small, at a lever arm of 0.15 meters and a force of over 2000 newtons on the ball screw, this could create huge torques at the load cell... which it isn’t rated for at all.

This issue was on top of the fact that the design constraints of the leg hadn’t allowed a system of supporting linear guides to eliminate torques in the screw thread. As I continued to work, I noticed that it wasn’t only the torque sensor had failed. It was the entire ball screw! There was a juddering motion when the knee joint rotated, and this was due to the knee ball screw being bent. And, the pivot for the base of the ball screw was offset, so it didn’t rotate on a point, but about one, which would change the support type from fixed to rounded, making it even more susceptible to moments. To put the icing on the cake, looking at the structure of the joints and videos of the leg operating in the past, there was evidence that it had crashed (exceeded its designed range of motion), which could have caused even more problems. So, my work was cut out for me. I knew I couldn’t completely eliminate moments in the system because of its dynamic and high-load nature (and because I could not take it completely apart and inspect to see if the ball screw pivots that allowed it to move with the knee flexion were properly aligned), but I redesigned a few parts to allow it to rotate inline with the axis of rotation and added hard stops to prevent it from crashing in a damaging way in the future. I also specified a preload for a set of new ball nuts (i.e., the bearing balls inside are about 2% too large for the space they have, so they always deform slightly and grip the thread of the screw tightly). This way, there could be no misalignment between the screws and the nuts, which would prevent them from generating any moments. After installing the new parts, I felt the axes of motion to determine how well it operated. The ankle joint was flawless, and it was even backdrivable (able to be moved without control input from the machine), which is an amazing feat for a robotic prosthetic device (the reasoning is a bit complicated, I will leave it at handwaving for now). Feeling that beautiful motion was incredible! But unfortunately, the knee was very stiff. It turns out that the knee ball screw had been bent, presumably in one of the crashes. This, sadly, meant a whole month of lead time. I was amazed I hadn’t seen it before--I must have assumed that the juddering was due to the ball nut not being preloaded, and not have been able to feel it well due to the lack of preload tolerating the misalignment in the system. With that fixed, the leg went back together, and now operates well, as you can see in the video. The leg no longer has crashes in operation, and is reaching the end of its planned service life. The next version of the leg is in the works in our lab’s prosthetic group, which I occasionally provide design, manufacturing, and assembly advice to as I work in such proximity to them. I must say, while I enjoy R&D work, there’s something about getting one’s hands dirty and figuring out how something that already exists works that thrills me whenever I do it. Maybe it’s because I get to put all my skills to work at once. In any case, I hope to get to do something similar soon!

#robots #problemsolving

This was a human hamster wheel desk that a team friends and I built in 24 hours for HackDFW 2015. One of my friends had seen plans for such a desk on an instructable (creds: http://www.instructables.com/id/Hamster-Wheel-Standing-Desk/), and we thought it would be crazy and fun to build it at a hackathon where software and electronics were typically developed. We were also motivated to build it because we, as engineers, often find ourselves sitting in front of computers for hours on end, and know well that this impacts our health. So, we set about redesigning the desk so that it could be built with the tools available at the Dallas Makerspace. Eleven hours later, the desk was on the back of a pickup heading down the highway to the competition. It was designed in SolidWorks, and then manufactured with a MultiCAM CNC router table and a great deal of hand labor. It was over 8 feet tall, and allowed a user to walk (and run!) on the large wooden circle while using a laptop. While we weren’t selected as an overall winner, we were selected for startup company mentorship by IStartValley, a nonprofit startup accelerator. Unfortunately, with our team being students (and me being employed in the construction of robotic exoskeletons), we did not pursue starting up a business to sell the desk, but we still had a wonderful time and learned a lot!

#hackathon #cnc #woodworking #teamwork

This is a phone mount that I designed for my dad’s IPhone 6+ and 3D printed with my Printrbot. It solved the problem that most mounting positions available for the phone were difficult to reliably operate while the car was in motion, and also put the phone in the glare of the windshield on sunny days. While it was functional, as soon as summer hit, the PLA plastic hit its creep temperature and deformed badly in the Texas heat. Lesson learned!

#3dprinting

This was the robot I built for a “robot sumo” competition in my senior year of high school. It had a computer-milled aluminum frame, which was carefully designed in SolidWorks to be attractive... but sadly, not be very functional. I learned a good deal about the rigidity materials, form vs. function, and engineering design from this project. While it worked, it was nowhere near the weight limit of the competition (as I had removed all of it when I cut its beautiful curves), and its tread drive system craved the extra weight needed for traction. And, on top of that, its control cable was not fastened well enough, and it fell off during the competition, causing me to lose. I have found, though, that losses of this sort are often much better in the long run, as they have helped me learn most.

#cnc #robots

My friends and me at the second Comets Create make-a-thon, another 24 hour competition in which we created a device which helped elderly people in their daily lives. We met with several people who lived at an assisted living center, and interviewed them about their lives and issues they may have had, and in listening to them, figured out that one of the salient issues in sudden elderly health decline is due to forgetting to take daily medications. We also found that there are many devices that already attempt to solve this problem, but that they are complicated to set up and difficult to operate with potentially arthritic fingers. The device was a simple electronic pill organizer, which would flash an LED for the section of the pill box that needed to be taken, and turn off when that section was taken. Afterwards, it would advance to the next section of the organizer. The project took home first place!

Me and my friends in our first Comets Create, where we had 24 hours and $100 to create something which would inspire middle schoolers to do engineering. With our backgrounds in mechanical and biomedical engineering, we created an upper limb exoskeleton which used an electric motor and 3D printed hypocycloidal reducer to help a person lift about 10 extra pounds. We took home first place, and learned a lot about the engineering process!

#3Dprinting cycloid cometscreate

This was the robot me and my FIRST Robotics Competition Team (Purple Haze, team 418 based in Austin, Texas) built in 2014 for Aerial Assist. Designed in SolidWorks, built in the Liberal Arts and Science Academy metal and wood shops, using 3D printers, CNC routers and mills, as well as plenty of good ol’ hand tool labor and welding. It was over 140 pounds, had a two-speed transmission that allowed it to go 14 miles an hour, and could throw an excercise ball over 35 feet.

#robots machining 3dprinting CNC

Hypocycloidal reducers, designed via Matlab, SolidWorks, and Fusion 360, and 3D printed. These are prototypes, made purely to prove the concept of the reducer, which employed eccetric rotating cams to reduce the angular velocity of an input rotation and thereby amplify torque. I hope to use them in a home-built robot arm someday!

#cycloid 3dprinting

3D printed model of SpaceShip One and its carrier craft White Knight, designed from scratch by myself and teammates in PTC Creo, printed on a Stratasys machine.

#3Dprinting

HELLO

world

Trending Blogs

Recently Viewed Blogs

Calvin Stence