I2C Communications between Raspberry Pi and Arduino – Part Four

It has been two weeks since my last post but it has been out of frustration on porting the code over to the ATtiny85. The first thing that I ran into was that the Wire library is not supported on the ATtiny85. I needed to modify my code to work with the TinyWireS library. This did not seem too bad and worked once in a while. It was a bit frustrating as I followed examples and it appeared that I was doing everything correctly but that is typically how it goes when coding.

I finally took a look at the specs for the ATtiny85 and realized that memory may be my issue so I started to pare down the memory requirements. The Arduino IDE was not complaining but I recalled an posting that was published on Adafruit a couple of years ago called Arduino Memories. After rereading the article and looking at a couple of other references, I determined that I needed to tackle the memory is see if it was an issue.

At some point in my debugging, I had noticed that the examples for TinyWireS were utilizing a buffer and pointer method to do fast reads and writes. I had a significant switch statement on the request data handler so I removed that and went with the buffer option. By doing so I reserved a whopping 256 bytes for the buffer. This was a very stupid move which I realized when I took a look at the specs for the the ATtiny85. The ATtiny85 has only 512 bytes of RAM so I was consuming half of it for the buffer which did not leave much room for anything else.

I dropped the buffer size down to 32 bytes which helped a great deal. After reducing the size of the buffer, I could get communications between the ATtiny85 and the Raspberry Pi to work a few times before the communications stopped working. I further refined the code to reduce memory usage and swapped out the Adafruit DHT library for one written by Rob Tillaart for the DHT11 only.


With these modifications, I was able to get the code down to using 113 bytes of RAM and 4,918 bytes (60%) of Flash.

With these changes, the code works quite well but sometimes it appears that the ATtiny85 does not read the correct request from the Raspberry Pi. After some searching it was found that there is a known issue with the Raspberry Pi and clock stretching. It appears that there is a bug which has not been fixed yet if the slave stretches the clock at the right moment and the stretching is too short. The ATtiny85 implements I2C in software so this is going to happen at some point.One of the best articles on this issue is the Raspberry Pi I2C clock-stretching bug.

There are some suggested fixes which I need to read more to understand well enough to use. The most promising fix appears to use Python to perform I2C communication in software. The recommendation is to use the PiGPIO library.

Below is the code that I have thus far on the ATtiny85.

Here is the code on the Raspberry Pi to verify that things are working.

Running the Raspberry Pi program produces the following result.


Next step is to see if I can resolve the clock stretching issue and then connect to adafruit.io to post data. If it is not possible to address the clock stretching issue, it would be possible to identify when it occurs and reset the power to the I2C slave devices. I am trying to avoid that solution but I may need to resort to that solution.

Leave a Reply