Skip to main content

Switching ATtiny core

I was not getting reliable I2C communication between the D1 mini and the ATtiny85, starting with TinyWireS, then moving on to the more developed TinyWire. I would send out 5 bytes from the D1 mini, then try to get 16 bytes back. It will start off well, but a few hours later, something will break and the D1 mini will start getting 0xFFs from the ATtiny85.

Then I found "ATtinyCore" by Spence Konde (V1.1.5), which appears to be a more mature ATtiny core compared to the "attiny" core by David Mellis (V1.0.2). It has integrated I2C support that manifests as the familiar "Wire" library. The installation instructions are here. I did a manual upgrade to the latest git version as some I2C issues have been fixed in the V1.1.6 milestone but have not been officially released. As recommended by the instructions, I bumped up the Arduino version to 1.8.6.

Finally, happy to see reliable I2C communication between the 2 components! The communication has been going non-stop for 2 days now, with 1-minute interval between the send/receiving of messages.

From the D1 mini end, the I2C master code is:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
  byte outbuf[5];
  outbuf[0] = random(128); for (int i=1; i<sizeof(outbuf); i++) outbuf[i] = outbuf[0] + (i*2);
  Wire.beginTransmission(I2C_SLAVE_ADDR);
  int rc = Wire.write(outbuf, sizeof(outbuf));
  Wire.endTransmission();
  debug("Wire.write (%d): %X %X %X %X %X", rc, outbuf[0], outbuf[1], outbuf[2], outbuf[3], outbuf[4]);

  byte inbuf[15];
  Wire.requestFrom(I2C_SLAVE_ADDR, 15);
  Wire.readBytes(inbuf, 15);
  Serial.print("Wire.readBytes: ");
  for (int i=0; i<sizeof(inbuf); i++) { Serial.print(inbuf[i], HEX); Serial.print(' '); } Serial.println(' ');

From the ATtiny85, the corresponding I2C code is:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
volatile byte msg[5], out[15];

void requestEvent() {
  Wire.write(out, 15);
}

void receiveEvent(uint8_t numbytes) {
  if (numbytes > sizeof(msg)) return;
  int idx = 0;
  while(idx < numbytes) {
    if (Wire.available()) 
      msg[idx++] = Wire.read();
  }
  out[0] = numbytes;
  for (int i=1; i<sizeof(out); i++) out[i] = msg[(i-1) % numbytes];
}  

For testing purposes, the D1 mini sends out 5 random bytes, and the ATtiny85 basically copies the received message into the outgoing buffer, with the first byte being the length of the received message (should be always '5').

I found that the maximum number of bytes that can be sent by each requestEvent() is 15 (one less the default value of TWI_RX_BUFFER_SIZE in Wire\src\USI_TWI_Slave\USI_TWI_Slave.h. That could be increased to 32, or 64 (powers of 2 up to 256, as noted in the comment). A workaround will be for the master to make multiple requestFrom() calls, and for the slave to keep track of which block of 15 bytes to send out in requestEvent().

ESPCLOCK1 / ESPCLOCK2 / ESPCLOCK3 / ESPCLOCK4

Comments

Popular posts from this blog

Adding "Stereo Mixer" to Windows 7 with Conexant sound card

This procedure worked for my laptop (Thinkpad E530) with a Conexant 20671 sound card, but I suspect it will work for other sound cards in the Conexant family. I was playing with CamStudio to do a video capture of a Flash-based cartoon so that I can put it on the WDTV media player and play it on the big screen in the living room for my kids. The video capture worked brilliantly, but to do a sound capture, I needed to do some hacking. Apparently, there was this recording device called "Stereo Mixer" that was pretty standard in the Windows XP days. This allowed you to capture whatever was played to the speaker in all its digital glory. Then under pressure from various organizations on the dark side of the force, Microsoft and soundcard makers starting disabling this wonderful feature from Windows Vista onwards. So after much Googling around, I found out that for most sound cards, the hardware feature is still there, just not enabled on the software side. Unfortunately, to

Hacking a USB-C to slim tip adapter cable to charge the Thinkpad T450s

This hack is inspired by this post . A year ago, I bought an adapter cable for my wife's Thinkpad X1 Carbon (2nd Gen) that allows her to power her laptop with a 60W-capable portable battery (20V x 3A). A USB-C cable goes from the battery into the adapter, which converts it to the slim tip output required by the laptop. Everything works out of the box, so I didn't give much thought about it. Recently, I decided to buy a similar cable for my Thinkpad T450s. I know technically it should work because the T450s can go as low as 45W (20V x 2.25A) in terms of charging (though I have the 65W charger - 20V x 3.25A).  I went with another adapter cable because it was cheaper and also I prefer the single cable design. So imagine my surprise when the cable came and I plugged it into my laptop and it didn't work! The power manager just cycle in and out of charging mode before giving up with an error message saying there is not enough power. After much research and reading the Thinkwiki

Using Google Dashboard to manage your Android device backup

I used to use AppBrain/Fast Web Install to keep track of which apps I have installed on my phone, and to make it easier to reinstall those apps when the phone gets wiped or replaced. But AppBrain had been going down the tubes, and Fast Web Install had always been a hit-and-miss affair. Android's own "backup to the cloud" system had previously been even more unusable. There isn't a place where you can see what has been backed up. And when you setup a new phone with your Google account, you just have to wait and pray that your favorite apps will be restored to the phone. Typically all the stars have to be aligned just right for this to happen. More often than not, after waiting for an hour or so and nothing happens, you just curse under your breath and proceed to install your favorites apps manually via the Play Store. But I just looked again recently and was pleasantly surprised that things are much more civilized now. Firstly there is a place now where you can loo