Last year I made this interactive music installation for the Electric Castle festival. It’s a RPI that plays a script in order to compose music. The UI buttons are touch sensitive sensors controlled by an ATMEGA328. Each song got an unique code – displayed on a 12×2 characters LCD driven by the ATMEGA328 – that could be used to download it from a custom micro-website.
The hole project (panel, graphics, electronics, coding and the micro-website) was done in only 3 days in which I probably slept a total of 8 hours.
This is my first printer so I expected some challenges but overall I have to say that it’s better to start with an actual 3D printer than a kit like I did.
When starting from a kit you must already have some knowledge about electronics, arduino/atmega programming and grbl/gcodes. The quality control at Tevo is very bad and the printer is not entirely of metal as they advertise! The 3D printer software familly is great but there’s too much to learn before starting to use them. The kit must be build and then configured and calibrated. This is critical and a lot of time must be invested, especialy if you are a beginer.
I already own a CNC so for me was a normal transition. The first issue that I got was routing all the electronics and starting the actual machine. The connectors that are provided in the kit are not the best on the market and some connectivity issues might appear. Prepare to tingle them until you get the correct readings and values on the screen. The Repetier firmeware has great error management that help a lot in debuging, and prepare to debug them.
The bottom line, Tevo Tarantula is a bad printer and I don’t recomand it.
Been busy for some time and working very hard, so I didn’t had the time to post the works I’ve done.
3D printing has it’s limits when it comes to interlocking parts and lots more for moving parts. The grouves a printer makes are great when you do interlocking parts because they add friction and hold them in place better than their CNC couterparts. But it’s not that simple! You have to compensate for the shrinkeage you get when the part is cooled! Usually I use 200um in between! Any less and the pieces will not interlock, any more it’s ok, but alot more will mean that they will come loose and fall apart. For moving parts I add some microns to this, maybe up to half a millimeter. In this case the more you add the better they move but if you’re building models keep it under one millimeter.
Here is my Malaxa LDM 12 at HO scale. After the LDH 1250 project, I found out that is better to develop in the scale you’ll be printing than doing everything big and scale it later. You get first hand the problems (dimensions and walls) that will be impossible to make and you can correct them right there. This was the path I chose for the LDM 12: I did everything at 1:87 scale.
I want it to be able to move so I left room for a small engine and put a lid on the undercarriage in order to have access to the wheels axes. As I already mentioned, I used 200um gap for any interlocking parts as well as 200um gap to the wheels axes – in my case I will be using wheels bought from a model shop, not printed ones, so I hope they will provide the clearance I need for them to move.
After the LDH, I realised the holes that a 3D printer makes are never round, so instead of putting all those holes in the chassis I decided I will simply mark them and drill them later. It’s a manual and boring job but it’s a printer limitation you have to account for.
Also I took the time to think how to print every piece! They may look positioned correctly but some where designed from the top-down and some the bottom to the top, mimiking thier printing position. I had some surprises and had to redo some of the work but it guaranties me that the printing process will be done right.
The railroad barriers that came with the Eaglemoss magazine didn’t feel quite the actual ones. So I decided that they need some modifications. After cutting the bar bellow, it took some polishing and repainting but they now have the look I wanted.
BR 290 model with the number V90 051 and code Roco 43666. The locomotive was produced in two versions, with and without DSS (meaning it’s ready for DCC but comes without a decoder). Mine is the one without the plug so I’m stuck with it’s original board. Note that this locomotive does not have much space to add decoders, the only solution is to use the cabin!
Because ESU LokPilot did not fit in the cabin I used a Romanian decoder made by TOM. Equally reliable but more little! Testing decoder to see how it fits in the cabin:
It fits nicely so here is the complete tutorial on how to install it. Read more
Every diorama hits the block occupancy (BOD) problem: where is my locomotive and how can I track it? This is my solution using optical detection (infrared emitor-detector pair) and some artificial intelligence (distributed thinking). Basically each BOD sensor can sense when something passes over and can communicate with it’s neighbors. Than each sensor decides it’s own state. The system gets better and better as more and more BOD sensors are defined. Here is a quick video of the functioning:
The locomotive runs on a simple oval track with only one switch (only the external oval is used for this test). There is no centralized controller or app, every sensors has the capability of choosing its state and acts accordingly.
The sensors are connected to a RGB led that indicates their current state:
Green = Free Magenta = Pending Blue = Detection
The sensors are plug&play. They don’t need any configuration and the distributed algorithm can handle any track arrangement automatically. The algorithm also deals with the rail switches as a single block.
When I designed the LDH 1250 I used the correct plans but I did not respect any scale. I just draw on the plans. Everything was fine in 3D and the details where great, unfortunately before printing I had to scale down my model. This posed some issues I didn’t thing trough.
Some details were lost due to scaling; when scaling, meters becomes millimeters and millimeters becomes microns. Remember that any printer has a minimum resolution, usually somewhere between 0.2 and 0.4 microns. I my case, the details in the order of millimeters scaled down to less than 0.2. So they simply vanished from the model, even though they were showing in the Cura printing application. So be ready to lose them or just make them bigger than they really are.
Another problem I’ve encountered was that the walls were too thin. So I had to add supports to them or make them thicker.