I2S DAC generate tone

[nameofuser1]

I am trying to produce sine wave with I2S. There are several strange things I have found I don't know how to overcome.
Here is my code:

Code: [Select all] [Expand/Collapse]

1. #include <Arduino.h>
2. #include <driver/i2s.h>
3. #include <driver/dac.h>
4.  
5. /* I2S DAC */
6. #define  DAC_CHANNEL            DAC_CHANNEL_1                   // GPIO25 channel
7.  
8. #define I2S_DAC                 I2S_NUM_0
9. #define I2S_SAMPLE_RATE         8000
10. #define I2S_BITS_PER_SAMPLE     I2S_BITS_PER_SAMPLE_16BIT
11. #define I2S_CHANNELS_NUM        I2S_CHANNEL_MONO
12. #define I2S_CHANNEL_FORMAT      I2S_CHANNEL_FMT_ONLY_RIGHT
13. #define I2S_COMM_FORMAT         I2S_COMM_FORMAT_I2S_MSB
14. #define I2S_DAC_CHANNEL         I2S_DAC_CHANNEL_RIGHT_EN        // corresponds to DAC 1 channel
15.  
16. #define SINE_SAMPLES_NUM            256
17. #define SINE_800HZ_SAMPLES_NUM      10
18.  
19.  
20.  
21. /*
22. const uint16_t i2s_sine_wave_800hz[SINE_800HZ_SAMPLES_NUM] = {
23.     32768, 53830, 65038, 61145, 43975,
24.     21561,  4391,   498, 11706, 32768
25. };*/
26.  
27.  
28. const int16_t i2s_sine_wave_800hz[SINE_800HZ_SAMPLES_NUM] = {
29.         0,       21062,  32269,  28377,     11206,
30.         -11206, -28377, -32269, -21062,     0
31. };
32.  
33. static bool prepare_i2s_dac();
34. static bool play_sound();
35.  
36.  
37. void setup()
38. {
39.     Serial.begin(921600);
40.     Serial.println("Hello!");
41.  
42.     if(!prepare_i2s_dac())
43.     {
44.         Serial.println("Failed to initialize DAC");
45.         while(true);
46.     }
47.  
48.     play_sound();
49. }
50.  
51. void loop()
52. {
53.     //delay(10);
54.  
55. }
56.  
57.  
58. static bool play_sound()
59. {
60.     static uint8_t sin_samples[SINE_800HZ_SAMPLES_NUM*2];
61.     for(uint32_t i=0; i<SINE_800HZ_SAMPLES_NUM; i++)
62.     {
63.         sin_samples[i*2] = (i2s_sine_wave_800hz[i]) & 0xFF;
64.         sin_samples[i*2+1] = (i2s_sine_wave_800hz[i] >> 8) & 0xFF;
65.     }
66.  
67.     if(i2s_write_bytes(I2S_DAC, (const char*)sin_samples,
68.             SINE_800HZ_SAMPLES_NUM*2, portMAX_DELAY) != SINE_800HZ_SAMPLES_NUM*2)
69.     {
70.         return false;
71.     }
72.  
73.     return true;
74. }
75.  
76.  
77. static bool prepare_i2s_dac()
78. {
79.     i2s_config_t i2s_config = {
80.         mode : (i2s_mode_t)(I2S_MODE_MASTER | I2S_MODE_TX | I2S_MODE_DAC_BUILT_IN),
81.         sample_rate :  I2S_SAMPLE_RATE,
82.         bits_per_sample : I2S_BITS_PER_SAMPLE,
83.         channel_format : I2S_CHANNEL_FORMAT,
84.         communication_format : I2S_COMM_FORMAT,
85.         intr_alloc_flags : 0,
86.         dma_buf_count : 2,
87.         dma_buf_len : 128,
88.         use_apll : false,
89.         fixed_mclk : 0
90.     };
91.  
92.     if(i2s_driver_install(I2S_DAC, &i2s_config, 0, NULL) != ESP_OK)
93.     {
94.         Serial.println("Failed to initialize i2s driver for DAC");
95.         return false;
96.     }
97.  
98.     if(i2s_set_dac_mode(I2S_DAC_CHANNEL) != ESP_OK)
99.     {
100.         Serial.println("Failed to set i2s DAC channel");
101.         return false;
102.     }
103.  
104.     i2s_set_sample_rates(I2S_DAC, I2S_SAMPLE_RATE);
105.  
106.     return true;
107. }

GeSHi © Codebox Plus Extension

First of all I am totally missing why signal is produced more than once. Image shows that pattern is repeated each ~30ms. Why is that so?

Signal repeated

photo_2018-11-28_12-41-17.jpg (100.22 KiB) Viewed 8859 times

Secondly I am not sure how does I2S interpret sent data, should it be signed or unsigned?

However I tried both approaches(you can see two arrays in my code). What for unsinged pattern, it produces the following output:

Unsigned DAC signal

photo_2018-11-28_12-59-12.jpg (100.89 KiB) Viewed 8859 times

And here is how it should look like:

Unsigned sequence how it should be

Figure_1.png (24.2 KiB) Viewed 8859 times

Clearly sequences of points are not the same.

Could someone clarify what is wrong here?

[ESP_Sprite]

1. The I2S needs 16-bit samples, you're stashing (repeated) 8-bit samples into it.
2. The I2S reads the samples as 32-bit words and outputs the high 16-bit first and the low 16-bit second. This means that if you want to send it 16-bit samples, you'll need to swap the even and odd 16-bit words.
3. The DAC only plays the high byte of the 16-bit sample as an unsigned value (0..255, not -128..127).

This should (=untested) work:

Code: Select all

   static uint16_t sin_samples[SINE_800HZ_SAMPLES_NUM];
    for(uint32_t i=0; i<SINE_800HZ_SAMPLES_NUM; i++)
    {
    	uint16_t samp = i2s_sine_wave_800hz[i] + 32768;
        sin_samples[i^1] = samp; //note i XOR with 1 to swap even and odd samples
    }
 
    if(i2s_write_bytes(I2S_DAC, (const char*)sin_samples,
            SINE_800HZ_SAMPLES_NUM*2, portMAX_DELAY) != SINE_800HZ_SAMPLES_NUM*2)
    {
        return false;
    }

[nameofuser1]

1. The I2S needs 16-bit samples, you're stashing (repeated) 8-bit samples into it.
2. The I2S reads the samples as 32-bit words and outputs the high 16-bit first and the low 16-bit second. This means that if you want to send it 16-bit samples, you'll need to swap the even and odd 16-bit words.
3. The DAC only plays the high byte of the 16-bit sample as an unsigned value (0..255, not -128..127).

This should (=untested) work:

Code: Select all

   static uint16_t sin_samples[SINE_800HZ_SAMPLES_NUM];
    for(uint32_t i=0; i<SINE_800HZ_SAMPLES_NUM; i++)
    {
    	uint16_t samp = i2s_sine_wave_800hz[i] + 32768;
        sin_samples[i^1] = samp; //note i XOR with 1 to swap even and odd samples
    }
 
    if(i2s_write_bytes(I2S_DAC, (const char*)sin_samples,
            SINE_800HZ_SAMPLES_NUM*2, portMAX_DELAY) != SINE_800HZ_SAMPLES_NUM*2)
    {
        return false;
    }

Sorry for the late reply! Thanks a lot for your answer. I did not test your solution since I have found that ESP has hardware cosine waveform generator which solves the issue easily. Nonetheless thanks!