Help supply clock signal from Arduino (to debrick AVR)

Using Arduino Microcontrollers
1

In error, I set a fuse bit on my ATtiny85 to expect an external clock (lfuse:w:0x62:m) instead of a crystal (lfuse:w:0x6f:m).

That locked me out of running or reprogramming it, but I imagined I could reprogram it if I simply supplied the desired clock signal, using PWM from another Arduino. However, what I set up isn't doing the trick. Anybody care to help troubleshoot? (Or is the whole concept faulty?)

The setup

I'm using the Arduino UNO as ISP (that's also what I used to brick the ATtiny85). I'm using the Arduino MEGA to supply a pulse for the clock.

The sketch on the Mega (for supplying the pulse) is as follows:

void setup()
{
  setup2560();
}

void setup2560()
{
  // OC0A is PB7, i.e. D13
  pinMode(13, OUTPUT);
  // PWM mode, non-inverted
  TCCR0A = (3<<WGM00) | (2<<COM0A0);
  // use Output Compare Register A, no prescaling
  TCCR0B = (1<<WGM02) | (1<<CS00);
  // Output Compare Register A
  OCR0A = 4;
}

void loop()
{
}

My command to avrdude is:

avrdude -p attiny85 -P com8     -c stk500v1    -b 19200 -U flash:w:main.hex

The error message

avrdude.exe: Device signature = 0x000000
avrdude.exe: Yikes!  Invalid device signature.
             Double check connections and try again, or use -F to override
             this check.

The error message is exactly what it was when the ATtiny85 had a crystal hooked up instead of the Arduino MEGA. (And same as when nothing was connected for clock.)

I didn't do any math to ensure that my MEGA's pulse frequency should equal the 16MHz which I instructed the ATtiny85 to expect (back when I bricked it). But I don't suppose that matters so much as the fact that the pulse is well over four times as fast as the programmer's Overriding Baud Rate of 19200 (which I set in the args for avrdude).

2

I'm assuming the hook-up is correct, but you really should check it against the data sheet for the device.

Did you look at that clock on a scope? It should have the same duty cycle for "on" as for "off." If "4" is your compare value, the PWM should run with TOP=8.

Also, when running with a slower clock, you should use a higher delay value in the ISP. You may need to delay as much as 1 millisecond between each bit. With avrdude, this is the -i option. Start with -i 100 and see how it goes.

3

I uploaded that and got no output. However this outputs 4 MHz:

void setup()
{
  setup2560();
}

void setup2560()
{
  // OC0A is PB7, i.e. D13
  pinMode(13, OUTPUT);
  // CTC, toggle output
  TCCR0A = (1<<WGM01) | (1<<COM0A0);
  // no prescaling
  TCCR0B = (1<<CS00);
  // Output Compare Register A
  OCR0A = 1;
}

void loop()
{
}
4

Vaselinessa:
In error, I set a fuse bit on my ATtiny85 to expect an external clock (lfuse:w:0x62:m) instead of a crystal (lfuse:w:0x6f:m).

Which core are you using that you have to set the fuses "by hand"?

(Or is the whole concept faulty?)

The concept is not faulty. It works well. Supply a reasonable clock (I typically use 1 MHz for no particle reason) to the XTAL1 pin and you're good to go.

I didn't do any math to ensure that my MEGA's pulse frequency should equal the 16MHz which I instructed the ATtiny85 to expect (back when I bricked it). But I don't suppose that matters so much

It doesn't. Any clock should work.

In addition to Nick's code, Lady Ada has published a version of ArduinoISP that outputs a "recovery clock". If Nick's code doesn't help and you have trouble finding Lady Ada's sketch, send me a Personal Message; I have something that works.

5

I have another stand-alone sketch for the Atmega328:

// For Atmega328

void setup ()
  {
  // set up 8 MHz timer on pin 9
  pinMode (9, OUTPUT); 
  // set up Timer 1
  TCCR1A = _BV (COM1A0);  // toggle OC1A on Compare Match
  TCCR1B = _BV(WGM12) | _BV(CS10);   // CTC, no prescaling
  OCR1A =  0;       // output every cycle
  }
  
void loop () {}
6

Wow! Thank you all for your assistance. (NIck's code worked great for de-bricking to ATtiny.)

jwatte, I have no oscilloscope. Perhaps that will be on my wish list ere long.