RepRap: Blog

Recently, I've been working on the 'Generation 3' electronics based on the Sanguino / atmega644p chip. This new generation of electronics sets out to solve many different problems with the current electronics: modularity, simplicity of wiring, microprocessor power, as well as adding many new features: sd card support, lcd support, interface support, and much more. This next generation of electronics is a large topic, which I will cover in future blog posts, but for now I'd like to focus on one particular feature: the data bus for the modular extruder system.

If you remember back to the days of the Generation 1, PIC based electonics we had a token ring system where each part of the system had its own microcontroller and was connected via TTL serial. This system suffered from many flaws, including noise in the serial lines connecting the various microprocessors.

The Generation 2, Arduino based electronics fixed this by swinging to the opposite end of the pendulum: all the electronics are controlled by a single microprocessor and no serial data connections are required (except to the computer). The main disadvantage to this is that the Arduino is pretty underpowered to drive a complex system such as a RepRap machine. We managed to do it, but we had to take quite a few shortcuts. This basically means we're also pretty limited in the things we could achieve.

The Generation 3, Sanguino based electronics will swing back towards the middle, using a 'best of both worlds' approach. It uses a more capable microprocessor which solves the memory and pin problems. However, even with more pins its is still tricky to drive more than 2 extruders, or one extruder and an LCD screen / button interface. Thus, we need a modular system to allow expandability.

This long winded intro brings me to the subject of this post: RS485. This is a serial communications specification that uses differential signaling. The awesome thing about this is that it offers many cool things, one of which is improved noise immunity. As the wire contains the same signal twice (once normal, once inverted) When it gets to the other end, and the inverted signal is transformed into a non-inverted signal and combined with the original signal. Any electrical noise that was picked up is then cancelled out due to wave mechanics. Cool, huh?

I'm not an electrical engineer, so this stuff is all pretty new to me. Fortunately, we have many smart people participating in the RepRap project with experience in many different areas. One of those people is Nophead, who you may remember from his excellent HydraRaptor machine. He might deny it, but he's an electrical genius and he gave me a chip number and a circuit for doing RS485 communications. Today I built and assembled it.

In the picture you'll see the eventual setup I had going. I wired up two RS485 circuits to two Arduinos, and wrote a quick sketch to get them talking. The connection was a very long long, unshielded, twisted pair cable from an old ethernet cable. I hooked all these up and started them communicating. It worked great! I also did a test to try and inject some noise into the system by operating a powertool right next to the wires. It was very unscientific, but the RS485 connection didn't skip a beat.

There was one problem that I ran into, which was easily fixed. The Arduino only has one hardware serial port, which was being shared by the RS485 chip as well as the FTDI usb -> serial converter chip that is on all Arduinos. The problem was that when uploading a new sketch to the arduino, the RS485 would also be sending and receiving data which would seriously mess with the uploading process. Luckily, the RS485 chip has both a transmit enable and a receive enable pin. I used the appropriate pullup/pulldown resistors to default these both to off so that when the Arduino was reset for the upload process, the RS485 was disabled. This means that I had to wire up an additional two pins on the arduino to control the RS485 receive/transmit status. This is not a big deal because using RS485 frees up a lot more pins than it uses, so there will be plenty of pins available.

Anyway, today was a good day of hacking. Hopefully later this week I'll be able to wire up the SD card and run it through its paces. Luckily for me, there are schematics and code already written for the Arduino all over the internet, which means that it will be fairly trivial to wire it up and getting it working on the Sanguino.