PI backend Atmega328P and firmware

[Teleno]

A bird's view of the digital signal processing. It's a bit more complex than this but not much.

[Gunghouk]

Hi Teleno, could you confirm that the ICP1 signal is just the inverted OC1A gate signal drive to the mosfet? I don't have any NE556 chips on hand so am looking at using a single TC4426 dual mosfet driver for gate and audio drivers. Thanks

[Teleno]

Hi Teleno, could you confirm that the ICP1 signal is just the inverted OC1A gate signal drive to the mosfet?

Yes but ICP1 is not in use by this firmware, you may just ignore it and leave it unconnected.

[Gunghouk]

[Teleno]

New version 2.

- Median filter extended to 17 samples by use of the QSORT fast sorting algrorithm.
- Decluttered main() for clarity. Detail code moved to functions.
- Extensible model for samples and filters using arrays of structs.
- Function intToAscii32 corrected and optimized for speed.
[/CODE]

 while (1) {
  
    acquire_samples();
    if(recalculate_delays) calculate_delays();
    // Display immediately after last ADC interrupt
    //to avoid noise in the ADC by the I2C transmission
    if(USE_LCD) LCDView_show();
    update_median_filters();
    update_signal_filters();
    if (!integrators_ready) init_baseline_filters();
    else {
      update_baseline_filters();
      // Subtract consecutive samples to eliminate 1/f noise
      target = samples[0]->filtered - samples[1]->filtered;
      id  = samples[1]->filtered - samples[2]->filtered;
      int32_t target_b = samples[0]->baseline - samples[1]->baseline;
      thisSignal = target - target_b;
    }
    if (USE_SOUND) doSound();
    // process data from rotary encoder
    LCDView_encoder();
  }// end while

PI_backend_m328p_v2.zip

[Gunghouk]

I know it sounds trivial but could the audio level be set and displayed using the encoder in order to reduce components/wiring?

[Teleno]

I know it sounds trivial but could the audio level be set and displayed using the encoder in order to reduce components/wiring?

Do you read minds? I've been doing just that using this concept
https://www.geotech1.com/forums/foru...n-arduino-nano

Far from trivial.

The trade off is removing the median filter or else the MCU cannot complete all tasks within the one Tx period time budget.

[Gunghouk]

I meant the idea was trivial not the implementation

I thought maybe just altering the on (or off) time of the audio pwm timer would do it while leaving the frequency alone. I'm not suggesting target dependant volume just a variable, fixed level.

[Teleno]

I meant the idea was trivial not the implementation

I thought maybe just altering the on (or off) time of the audio pwm timer would do it while leaving the frequency alone. I'm not suggesting target dependant volume just a variable, fixed level.

It's more complicated than that. Follow the link to see how it's done.

In the end this is a nice project, with a minimum of components you get a dsp detector that has even digital volume control, only one button, lcd. All in taylored made C code for speed, with print functions that avoid printf, fast fixed point log2, fast median etc.

The front end is just as simple, just one op amp with a gain of 20, a pnp to refer the output to ground and a mosfet.

my prototype has just 3 chips: atmega328p, LM 4562 and NE556. Plus a 5v regulator and an LM317 current source. Running from 9v to 12v.

On the display you can see the levels of the ADC to adjust the delay to the minimum possible for a given coil.

And then a novel ground balance algorithm that learns the characteristics of the soil and minimizes target holes.

[Gunghouk]

Fairymuff. Thanks for sharing

[Teleno]

DIgital volume works. The problem is that the 125kHz PWM drives the 556 increasing it to 12V peak to peak to directly drive the loudspeaker, and the interference with the analog preamp is horrible.

I think the PWM has to be filtered at low level and applied to an audio amplifier as an analog signal, rather than pulsing the loudspeaker at crazy logic levels with the 556.

[Gunghouk]

I was intending to run the pwm into a TC4426 driver on 5volts and then through a 5.6mH inductor and 1uF cap to ground before the speaker, same as moodz MagpiV3 audio. Found a 75 ohm speaker in a tower system, ideal.

[Teleno]

Looking at the schematic in the first post, can anyone graphically suggest a better audio circuit so that the pwm of 135kHz gets smoothed to the loudspeaker and the pwm gets isolated/damped to the +12v Vcc that also powers the preamp?

[pito]

RC filter + LM386

R = 10k, C = 10n

[Teleno]

Here's my solution to the PWM problem. It seems to work. The LC filter has a cut-off frequency around 12kHz.The intereference is gone.