Show HN: PJON – Communications bus system for Arduino | Hacker News Reader

Show HN: PJON – Communications bus system for Arduino

github.com

44 pointsgioscarab15 comments

Hi there I wrote in years of trial and error PJON library for my home home automation project. It is a multi-master, single wire, addressed and checked communication protocol and is designed to be an alternative to i2c, 1Wire, Serial and the other Arduino compatible protocols with the addition of all the functionalities you need to fastly bring working a multipurpose embedded devices network. It was a really enourmous physical and mental effort of years to bring together all the knowledge and code skills to let all this work, so I hope could be useful for all Arduino users. Please feel free to suggest and crytic if necessary and if you can TEST IT would be an enormous pleasure!!!

Have a nice summer! :)

Not having looked closely at your code yet - how would you say it compares to something like: https://github.com/ClusterM/clunet ?

Thank you, this may be exactly what I need for a project!

Its a pleasure!!! Please try it and write me an email if you have some problem gioscarab@gmail.com

If I understand correctly, PJON is not directly a replacement for the 1-wire bus as it does not specify pull-up/-down hardware that would be necessary for using VCC as the physical communication medium, right?

(Which means that I would need to define/implement this part myself.)

Hi raphman_2 yes there is not the functionality of 1wire you are speaking about! I never had time to dig further and work on this functionality. In any case would be amazing to add it in PJON. But should be taken into account the duration of 0s and probably a capacitor to rectify input voltage. I generally use a single pulldown resistor on the bus to reduce interference.

RC4? Please don't. Use e.g. NORX for an embedded-friendly cipher that isn't dangerously insecure.

Hi creshal, I started from the RC4 implementation but remains not much of it and also i further fuzz information using a initialization vector after RC4 like procedure.

You think is so holed? Considering what u can do with an ATmega 328. (Most you can do with this computation power can be in any case holed fastly with a laptop)

> One of the first tests I suggest to try with PJON is to let two arduinos communicate through your body touching with the left hand the port of the first and with the right the port of the other arduino. It's stunning to see more then 90% of accuracy for this digital communication doing all that path inside a biological body. This opens the mind to possible creative solutions; generally the average reaction is like: "lets use the car frame to let all the digital embedded systems to communicate together" and so on...

Hmmm... I guess I'm not the first to think that my home plumbing would be the perfect transmission medium to gather data from temperature sensors.

It's absurd to see the level of advance in technology we can use and work with and the level of advance of our houses and transportation systems.

Thank you for your suggestion. Take a look if you can to the crypt() function, I use on top of a modified version of RC4 the use of 1 byte initialization vector to further fuzz information (the iv is trasmitted with the string).

I am not so much into cryptoanalysis so, I really should have some suggestion in this case.

This is really awesome. I tried to make something like this once since I needed multi-master with like 50-60 attiny chips with poor time consistency etc, but never really got anywhere... This looks like exactly what I would have liked

Thank you earlz. Try it if you can and contribute if you want to expand the features contained in PJON. I am really happy to have this type of feedback from the hacker community :) thank u guys/girls really!

I know its not very important for automation, but what are the defining factor to arrive at that bandwidth?

(If I were you, I wouldn't even maintain the two modes - the bandwidth difference is not that big)

Hi! I have a test sketch, it runs for 10 seconds and a transmitter send a string made by ID + LENGTH + 10byte string + CRC + ACKNOWLEDGE. The transmitter counts every correct and mistaken reception by the receiver:

What I mean with practical bandwidth is the byte sum of all the correct packets received that are useful (so the ID, LENGTH and ACKNOWLEDGE are not counted) / 10 (seconds of test) and so you get how many correct and useful bytes/s you can send in the fastest setup (for or while). The absolute banwidth is the total byte count wihtout no exception.

You are right the difference is not enormous. fast_mode is quite near to the limits of delayMicroseconds(), micros(), function execution time. I left a standard_mode with a really solid behaviour to let people use PJON also with boards without an external oscillator (like using internal PPL clock to 16mhz on ATtiny) or equipped with ceramic resonator that is less precise in timing.

Reply on news.ycombinator.com