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

Update: Line adapter for Ozito Blade Trimmer

Update (Dec 2021): If you access to a 3D printer, I would now recommend this solution , which makes it super easy to replace the trimmer line. I have been using it for a few months now with zero issue.

Line adapter for Ozito Blade Trimmer

This is an adapter for Ozito 18V battery trimmer (and possibly some Bosch trimmers as well) that uses a plastic blade for cutting. It lets you insert a 2.4mm trimmer line (about 8cm long) and use that for cutting. Simply cut a length of trimmer line and briefly heat up one end with a lighter so that a little bulb is formed. Then insert the trimmer line into the adapter and slot that into the trimmer as per normal. Make sure the trimmer line is not so long that it touches the safety guard. If that is the case, simply trim off any excess with a cutter or scissors. This part is best printed using PETG, which is a tougher and more flexible material. PLA is more rigid and breaks more easily. However, even with PETG, it will still break when it hits something really hard. Since this takes only 0.5m of material and 15 minutes to print, I will usually print a batch of nine at a time at very little cost. The blades that they sell do not break when it hits a hard object, but ...

Cooling mod for the X96 Air

I realized after my Ugoos box died that overheating is a big problem with cheap Android TV boxes. A teardown of the Ugoos box shows that it does not have any heatsink or fan at all!  The X96 Air does have a heatsink, but the heatsink is located at the bottom of the casing with no ventilation. In this default configuration, with the ambient room temperature at 25c and playing a 1080p video, I was seeing the CPU temperature at 67c. I drilled a couple of holes at the bottom of the casing. The CPU temperature fell to 59c with the box raised about 2cm with plastic blocks. I retrieved an old 5V laptop fan: Then cut and strip away a spare USB cable: Solder the red and black wires on the fan and the cable: Secure the fan to the bottom of the casing with double-sided tape, then plug the fan into the box's USB connector. Here's a view of the box with some 3D-printed risers installed at the bottom to give the mounted fan sufficient clearance: The CPU now runs at 43c, a huge drop from the ...