High Quality 16 bit AM PWM Arduino TX

It won't work. I just realized the increase in resolution is due to analogue weighing of the capacitor. My audio PWM is 31.25KHz , even if I choose C so that Fc is 60KHz it still won't work as the square waves are now turned into an analog.

These are characteristics of FM

FM has a signal to noise ratio of about 60dB. That can be managed by a bit depth of 10 bits.

FM has an audio bandwidth up to 15 kHz, so a sample rate of over 30 kHz is required - 32,768 Hz is a nice power of 2 that fits the bill.

So 10 bits per sample, 32,768 samples per second, 2 stereo channels - that’s a total bit rate of 655,360 bits per second (655 kbps).

Human ear dynamic range = 90dB when apartment has other noises like traffic, AC, etc. 

Solution might be to take 2 digital pwm pulses and merge in xor gate using nand gates. 
8 bit each. Arduino has 3 timers. 16 mhz /256 = 62.5 khz pwm from each timer. The timers are fed from a single audio source. We need 16 bit. So we combine the two pwm signal in two nand gates. The other pins of the nand gates are fed from a flip flop or divide by 2 . The third pwm signal is driving the flip flop so that the clock pulses remain synchronized. The output of two nand gates now merge into one nand gate. Its output is 62.5/2 = 31.25KHz 16 bit PWM audio signal. Even if 2 bits are lost to error it doesn't matter as 12-14 bit resolution is all we can hear (99%).

Harry, Ivan please check my question

Ivan,

Question is, if we don't want an analogue output then removing the capacitor would give me digital 16 bit resolution PWM, am I right? I am trying to build an AM where the 16 bit digital signal is going to a 74hc00 on one input pin and other input pin is fed with 1600KHz carrier wave to generate 1600KHz PWM at 74HC00 output to drive a MOSFET in class D or a low pass filter feeding an antenna. I can remove the capacitor right for 16 bit digital pwm signal? Analogue has no use here. 

My only doubt.

Arduino Nano. Other one eliminated. Two separate codes have been merged now on a single Nano. Touching D6 for carrier giving a light hiss on radio. Obviously the signal is very very weak and not travelling even a foot when I touch D6 with my multimeter probe. D9 and D10 are the 31.25KHz PWM pins.

Testing results will be posted soon.

Arduino Nano 1 sketch. The code is very simple for PWM generation. Arduino Nano 2 sketch remains the same I previously posted.

Arduino 1

// adc_to_pwm.pde
// ADC to PWM converter
// takes in audio data from the ADC and generates PWM at D9 and D10
// Timer1 PWM. 16 bit, Phase Correct, 31.25kHz - although ADC is 10 bit, in real time use & high sampling freq we get 8 bit only, output D9 & D10 are added together for 8 + 8 = 16 bit resolution. Sampling frequency is 38KHz (adc clk/ 13 cycles)
It would be hard to differentiate between CD player sound and this transmitter

#define PWM_FREQ 0x00FF // pwm frequency count - 255 or 11111111
#define PWM_MODE 0 // Fast (1) or Phase Correct (0)
#define PWM_QTY 2 // number of pwms are 2, each timer has 2 counters

void setup() {
// setup ADC Analog Input Pin A0
ADMUX = 0x60; // left adjust, selects A0 as adc input pin, internal vcc
ADCSRA = 0xe5; // turn on adc, sets prescaler to 16MHz clock / 32 = 500KHz ADC speed also known as free running mode, auto trigger
ADCSRB =0x07; // t1 capture for trigger
DIDR0 = 0x01; // turn off digital inputs for adc0 or pin A0

// setup PWM
TCCR1A = (((PWM_QTY - 1) << 5) | 0x80 | (PWM_MODE << 1)); //
TCCR1B = ((PWM_MODE << 3) | 0x11); // ck/1
TIMSK1 = 0x20; // interrupt on capture interrupt
ICR1H = (PWM_FREQ >> ;
ICR1L = (PWM_FREQ & 0xff); // & 0xff is hexadecimal numeric 16, adc clock freq is 500KHz, pwm frequency 500/16 - 31.25KHz
DDRB |= ((PWM_QTY << 1) | 0x02); // turn on outputs

sei(); // turn on interrupts - not really necessary with arduino
}

void loop() {
while(1); // gets rid of jitter for smooth pwm
}

ISR(TIMER1_CAPT_vect) {

// get ADC data
unsigned int temp1 = ADCL; // fetch low byte first
unsigned int temp2 = ADCH; // fetch high byte
// ADCH and ADCL 8 bits each are represented

// output high byte on OC1A
OCR1AH = temp2 >> 8; // takes top 8 bits
OCR1AL = temp2; // takes bottom 8 bits

// output low byte on OC1B
OCR1BH = temp1 >> 8;
OCR1BL = temp1;

// top 8 and bottom 8 bits are combined together in two precision resistors at pin D9 and D10 for 16 bit of audio resolution riding on a 31.25KHz PWM modulated square wave, sampling frequency is 38KHz, good enough for an AM transmitter!
}

dare4444 wrote:... all those "too low" and "too high" values compensate each other and it becomes, suddenly, a big chain with a much better accuracy, through the law of large numbers...

I disagree. My experience is that parts from the same production lot are often "shifted" the same way. In a pack of 100K +-5% resistors there are e.g. several 101K, 103K etc., but no 99K. A precise ohmmeter is the only sure solution.

VBR Ivan

Use two PWMs, and use one pin for high bytes and another for low bytes.

Using two PWM pins and adding them together means it’s possible to add extra bits of resolution. This requires using different values of resistors on each pin. For example, using the same value of resistors on two PWM pins increases the resolution by one bit. Two pins with a resistor value ratio of 1:4 increases the resolution by four bits, and so on.

This is how it works.