Probably time to re-think the communications layer- master/slave with the Wixel works well but only has a 50ft or so range. Xbee is actually cheaper and offers a point to multipoint network out of the box. Duh. Also, if you want to spend the money, the PRO will do a freaking MILE in range! Sounds almost too good to be true, eh? We shall see.

So with that in mind, I’m also thinking of switching to the AVR 1634. It has two built-in usarts and extra i/o pins, however it is only available in a SMD package. Not terrible but a little more difficult to breadboard than a PDIP package.

Also found a nice free schematic drawing package from these guys:

But their PCB software kinda sucks (I like autorouting) so I’ll be sticking with these fellows:


I’ve been working on a home-made set of widgets to convert this locomotive to R/C and battery power. It’s a chance to do some circuit design, soldering and a bit of embedded software engineering for fun. The idea is to completely replace the insides of the locomotive with inexpensive, off the shelf components. I will use those to drive the main motive power- a 15v 3500mah Nimh battery array, and tap off that to provide stereo sound and lighting control.

While the motor controller was easy- I’m using the Pololu 18v7, the sound and R/C switches for the lights and the reversing power relay was where the fun was so I designed my own board. It uses an Atmel Attiny84 microprocessor, two mp3 player cards, a 3A power relay and two power mosfets.

Anyhow, the innards of the Aristocraft RS3 were not impressive. Complicated rats nest if you ask me. I have no interest in track power nor DCC so I just gutted the entire thing down to the leads on the motor trucks. I did save the two small circuit boards that plug into those but everything else with the exception of the 7805 power regulator got tossed.


I have the sound/relay card almost finished, as things progress I’ll post more details. Here it is stripped all the way down:


And a picture of the motor truck, the colored wires are the motor leads, the two black wires go to the track pickups:



Here is everything I’ve been playing with for the past few months in preparation for the RS3 battery conversion project. The hacked up R/C transmitter is in the back, note the pocket for my android tablet. The two buttons will be for sounds, the toggle switch is the reversing switch for the power relay.

The 3D print fellow is in front of that, I have a sitting version of him to replace the out of scale cartoon figure in the cab. The soundcard/relay/mosfet board is to the left and the pololu 18v7 motor controller is on the right in front of the switch harness. The power relay is wired and working off the intercepted servo signal of channel two.

The smoke generator is the black box at the far right, I think it needs its own hack. Sync’d to the motor control would be cool? I’m not sure about this one, not even sure it’s worth it to put it into the model at all. Obviously I need some testing on that first.

The Pololu USB AVR programmer/ttl serialport/sloscope widget is at the front with the red clip connected to the back. This is my ‘scope’ probe. An amazing board for the price! I tell you what, I am absolutely sold on the AVR series of chips and this programmer. The ATTiny84 does not have a serial port, yet I found a nice software uart for free at AVR freaks. I can hook one pin of the micro up to the serial input on the Pololu and that combined with an LED on one of the port pins gives me a decent set of debugging tools. Super cheap too, ATtiny84s are like $3 in singles, way less in qty.

Anyhow, the sound algorithms need to be fleshed out, the mounting of the boards is not right yet and I need to put together a solid battery pack. But I’m getting there 🙂


I’m working on a new design, a sound card for a G scale diesel locomotive. Since I will be running this with R/C, I need a sound card that accepts servo pulses from a standard R/C receiver. I think I’ve finally settled on a design, so here is a pic as I’ve laid it out on a Radio Shack breadboard. On it are two stereo mp3 boards, an Atmel ATTiny84 microprocessor, a 4 channel mixer based on a LM3900 and a 8 pin power amp.

Here is a refactor of my rail editor experiments to make a more generic control interface for my wireless servo widgets. This one is ‘uncompiled’ (if that’s a word) so you can take a look at the source if you like. It’s mostly javascript controlling the html5 canvas with some PHP thrown in to do the file save/load. I’ve tried to make it relatively easy to modify so you can add your own tiles and symbols. Note that this is not meant to be a session based editing tool, it saves one file and is meant to do realtime control on a closed web system with tornado (Raspberry Pi)- so don’t try to hack on my server please. (note: I’ve disabled file saving and there is no tornado web socket server on this box so you will see some errors in the web console if you try that)

You can try out the actual program by clicking ON THIS LINK


To use this, click a symbol from the palette on the right and then click where you want that symbol to be placed. To erase a symbol, left click on that spot. Since this is meant to be a control surface for a real time web implementation, the ‘wixel’ button at the top is used to set the number of the slave widget (0-63) and the r/c servo that widget controls (0-5). Note this is not implemented completely in this demo, however I will be using the tornado code base I’ve posted in my other posts here, realtime control of servos, see the side bar for a link to that post.

UPDATE: October 28, 2013

I’ve released all of the wixel and html5 code for this project, you can find details here: HERE

FINALLY, the power returns and I have a chance to test my breakouts! First test is great. I have the Raspberry Pi driving the master Wixel via USB. This slave Wixel, soldered into my breakout board, sits on the other side of my shop. A 5v wall wart provides the power. This is a nice mix of micro servos, minis and three standards. Works great!

In case you have not followed all my posts, what I have here is a master wixel plugged into the Raspberry Pi. Using the built in packet addressing mode of the TI processor (check out the Data Sheet), it sends servo commands to each slave in a round robin sort of protocol. Each slave can control up to 6 servos and return 3 analog inputs to the master.

I’m getting a really good update rate of about 4ms per slave. Once started, it just keeps running, the Raspberry Pi, via the USB to the master, just has to write a servo value into a table and on the next turn for that slave, it’s updated. This frees the Pi from any work at all in maintaining the network, it’s all auto-magic.


Here are some more pics of the widget.


I’ve decided that standing it up at 90 degrees like in other pics is not worth it. Much easier just to solder it in like this:


Here is a LINK to a previous post about the breakout boards. And ANOTHER that has some s/w details. Note, I have NOT released the internal code apps for the Wixels yet, I’m not sure if I will or not, depends on the results of my next project.

Work progresses on the control interface. After fooling a bit with an Android native app, I decided that was way too overkill. So here is the same idea implemented in HTML5. It will run on pretty much anything, or I hope it will as it’s browser based.

Here is a link to a basic version. There is no save or load, it’s just the framework-


The workflow is that you click a symbol, then click where you want it on the grid. This allows you to build a route that resembles the control for any railroad layout. The right mouse button clears a symbol from the map. It still needs a load and save option and finally, I will need a run module page that will load this file and allow you to control vs editing.

The path is a web page load from the RPi, with tornado passing down data to the master Wixel. From there the master will communicate with all the respective slaves over wireless, allowing me to touch a switch icon on the tablet and have the servo at the far end of the chain move the turnout. The HTML5 logic will then redraw the route in color to indicate where the train is going to go. That’s the plan anyhow. There are lots of possibilities with this protocol model, anything that needs R/C servos controlled over 2.4ghz wireless from the Raspberry Pi. Or any master for that matter.

I’ve already tested the basic code on the RPi via lighttp to my Android tablet and it works great. More logic to put in but the idea here is to enable the editing and control on any device, laptop, desktop or tablet. So far I’m very happy with this project and I’m getting close to completion.

Later I’d like to actually track the trains with RFID (or something) and animate that somehow on the tablet. However, that is down the road a bit, I’m going to limit my scope here on this one so I can finish it up.