Last time I tried printing with the raspberry pi I had only one machine to try with now I have two. Lets see if the Pi can handle two instances of AtCore and control two 3d printers at the same time. This is a follow up to AtCore takes to the pi. So please read that for more about the RPi setup. This post is in video form, please enjoy.
Since I find myself with a bit of down time. I have decided to update you all on the progress of atcore and atelier. We should hopefully be ready soon to start the process of releasing AtCore 2.0. I only have a few things I really want to get merged before I start that process. In the mean time here is what has been landed to AtCore since my last post.
Atelier is getting more and more polish. Here is what has been landed to atelier since my last post.
It has been sometime since I’ve written about our progress with AtCore and now that I find myself with a bit of down time Its time to give you all an update. Since the end of May we have landed 32 commits from 4 contributors. I would like to first thank our newest contributor Leandro Santiago for taking time to contribute to AtCore.
Our changes include:
For Atelier we have 44 commits and 3 contributors for this time and changes include:
You can find our more about Atelier by reading Lays’ Post about Akademy 2018
There are way to many changes to put here in one post for these two projects since my last update. I will very quickly just go over some changes. With AtCore we now detect plugins better. Provide a more flexible deployment configuration and and have a Homebrew script for Mac in the KDE-mac tap. (log) AtCore is getting ready for a new release v2.0 this will be required to build Atelier. This will be the topic of my next post. Where plan to cover the journey and changes from AtCore 1.0 to AtCore 2.0.
Atelier has gotten the lions share of the work since the last update at least a diff a day! These include using all AtCoreWidgets for the common items shared with the test gui. Supporting drag and drop to open files. Having the 3d View tracking on the gcode files open the in editor so you can preview the open file. Improvements to the gcode editor. Quite a few more that I am not remembering all while cutting down on the code where possible and making parts leaner. (log)
It has been over a month since my last progress update. Here is what I’ve done.
D10813,D10845,D11185,D11252,D11702,D11565,D11381, D11403, D11345, D11024,D10299,D11772
AtCore will now deploy libAtCoreWidgets. Sd card support is working for repetier and marlin. I can not assure you it will work for any other firmware correctly. Command injection is now done (D11024).
D12024,D12054,D12021,D12012,D12020,D11996,D11873,D11730,D11837,
D12030,D12046,D12076,D12008,D12009,D11995,D12010,D12011,D11476
General UI Improvements.
A few users have asked for Sd card support and hopefully the basic version of that will be merged soon. This feature appears to be a more firmware dependent so it may take some time before each of our target firmwares has full support for SD Cards. The other feature that is being worked on is command injection. I have been thinking about adding this for a long time and I finally have a decent set of changes to support this feature in atcore. I can hear some of you out there asking about what exactly I mean by command injection is so I’ll explain. This feature will allow you to place specific commands for atcore into the gcode file. There is no need to worry about the commands breaking your gcode file since they are added in a way that makes them invisible to other hosts. You can trigger temperature changes, speed changes, pauses and more in this way. Using a few injected codes I was able to modify my test code that I print into a print that was made of abs with a tpu core. I didn’t have to touch atcore’s gui other then to resume after the auto initiated pause(s) and I can make prints with multiple colors and materials.
The Raspberry Pi3 is a small single board computer that costs around $35 (USD). It comes with a network port, wifi , bt , 4 usb ports , gpio pins , camera port , a display out, hdmi, a TRRS for analog A/V out. 1GB of ran and 4 ~1GHz armv8 cores Inside small SOC. Its storage is a microSd card they are a low cost and low power device. The Touchscreen kit is an 800×480 display that hooks to the Gpio for touch and dsi port for video. To hold our hardware is the standard touch screen enclosure that often comes with the screen if you buy it in a kit.
Most of the time it is sufficient to run Raspian the official distro that is made for the raspberry pi. raspian is a fork of debian as much like debian it has some old packages. In order to build AtCore (with the gui) you need at least qt5.8 raspian comes with 5.7. This is not a show stopper but compiling qt does take quite some time and can easily take a week if we were to build it on the rpi itself. Setting up a cross complier will save you a lot of time but even on my fastest machine it will still take some hours to build qt for the pi. I decided it would instead be faster to install a completely unsupported OS on the device. I was going to install install arch linux on the rpi. Arch is a rolling distro that ships with qt 5.10.
Arch linux on the raspberry pi is what you might call “doubly unsupported” That is that the raspberry pi foundation does not support running arch on the pi. You only get support from the arch community. Arch Linux is not officially supported on platforms other then x86_64 the rpi running on arm. Worry not for the Arch community has some forks such as archlinux-arm this is what we will be installing. Like most things arch you get some instructions. After getting the os installed its still arch so you only get a cli interface on your first boot and are left to set up the rest of the system.
For my system I installed plasma-meta , xf86-video-fbturbo-git and sddm. With those I was able to have a working plasma shell. I also installed dolphin, kate and konsole as well as the few atcore dependencies. I then added my user to the uucp group so I was able to r/w to serial devices and restarted. When all the dependencies are installed you can proceed to build atcore. Since this is arch we can simplify the building part just by using atcore-git from the aur. Im Happy to say it took only minutes and 48 seconds to build atcore-git on the rpi. I then used pacman to install the atcore-git package and it was time to see how our gui was looking on such a small screen.
At first the contents of the atcore-gui spill off the screen due to the way our docks are tabbed, however since they are docks you can just adjust them to make a quick usable interface. After adjusting the gui AtCore worked exactly the same as it does on other systems. Below you can see a picture and video.
There is lots of room for improvement. Atcore-gui does not save any settings. You’ll need to move the docks around each time you launch the program. Atcore-gui can not automatically connect to a device on launch. Again for reasons of not saving settings This gui is of course not really made for use on a touch screen and it does seam to work good enough to use. Some non atcore changes on the system itself would be nice. A few services running on the pi would improve this over all. Mjpeg streamer or another method of streaming video from the attached camera on my printers extruder. Uploads are done via stfp that’s not always easy for everyone. The pi does not launch atcore-gui on start up . Small stuff like this could easily be adjusted. I was really just playing around to see how atcore ran on the rpi. I was very happy when this worked well.
Today I would like to announce the release of AtCore 1.0.0. This is the first stable release for AtCore. Since its the first release and we have not written a “real” client for it yet we include our test GUI. If you own a 3D Printer you are encouraged to try AtCore for at least one print job.
Using the AtCore test gui is really easy for Linux you only need to make the AppImage executable then run it, On OsX you run the .app file and for Windows run AtCoreTest.exe. To do anything in the application you must first connect to a printer. Use the connection tab in the top area left area of the window. Select your Device and Baud then click connect. If for some reason your firmware is not able to be autodetected or you wish to speed the connection up you can select your firmware from the drop down list. If you don’t know what firmware you have try using repetier. You can find more information on the Test Gui if you view its User Guide. If you are intrested in hacking on/with AtCore you may want to check the Api Docsumentation.
Be sure to Report Bugs that you might find.
Its been some time since I’ve posted any progress for AtCore. Some may wonder what we have been up to ..
We would like to get some info from the printer while printing we need to have some method of flow control. AtCore now uses a command queue to send all commands. With the command Queue in place we have to handle stop and Emergency stop differently . For example Emergency Stop should skip the queue and be sent as soon as the button is pressed.
After you connect to a FW plugin every 5 seconds a m105 will be send to the printer and the temperature results are put onto a pretty graph that Patrick made. In order to make the graph work we need read the M105 return and extract the data from it. While our current methods of parsing this info work its very specific to each firmwares return . Since these string can be differnt between fw versions it can crash so we are working on a way to do this better.
The test clients GUI needed some love. All the widgets now live within a dock and the docks can be arranged how ever the user likes. I’ve also added a status bar to show the status of AtCore. We’ve also added a print job timer and remaining print time estimate. A Seperate axis control for relative was asked for by a user and I’ve added one that Lays wrote some time ago . It works well and allows for movements on 1, 10 or 25 units
Lays has been working hard to get atcore buildable via craft. Now we can build and deploy AtCore (and testgui) from craft. After building we had a problem of not finding the plugins on windows and them not deploying to the correct path.I added some instructions for deploy and tweaked plugin detection on window. Everything is now working well.
When connect to a 3d printer with your compuer you are really just connecting to a serial device. There are some commads you can send it in the form G# and M# commands . These commands do all kinds of stuff every thing from homing the axises to feeding filment and moving the head around. When you slice your model and generate that gcode file you are making generating basicly a long list of comamnds for the printer to follow. The gcode files are plain text and have to be sent out thru a serial device to the printer. Its not complicated to parse the file you just send a command and wait for the printer to return a message indicated its finished the comamnd and then send the next command. We have had this working for some time using a QEventLoop and a while to keep the loop going until the printer is ready for a new command. This was working wonderfully Untill we realized that we were having some blocking problems when printing. After some discussion tomaz , patrick and I decided the best way to fix this is for us to split the printing to its own thread so it can no longer block other parts by hyjacking the main event loop while printing.
Since I have been done the print function to begin with I figured that I should be the one to move it to a thread. The first thing that I did was move the print function to a new class based on Q_OBJECT. This is the new object that will be our thread. Moving to a thread ment we had to redo how the firmware plugin lets us know the printer is ready for a command. All the communication between the threads has be to done with signals /slot connections so the firmware plugins now send a signal when the printer is ready for a command. This means that we can now process the temperature returns when the machine is not printing. The rest was not to hard. I moved the logic to its new thread and it had to be rewritten to no longer use the QEventLoop but instead use Signals and Slots between the main thread and the print thread. The main logic for print was broken up in to a few smaller functions. The parts were connected with the various signals and slots and thigs begain to work.
Not everything was perfect the first time. It took time to get this all tested and not crashing. This is my first time multithreading so I was a bit conserned about getting it right. I had some typos that made me loose way to much time looking in the wrong place. Quite a few other crashes and problems that were just a result of me being new to threading. However I was finally able to get everything working in the same way it was before only now each print job will have its own thread.
This blog needs a better title 