Kinetis: MCUXpresso: How to change usb audio sampling rate in generator example?

[baldurthorgilss]

Hi

I'm struggling with understanding USB audio. I am using MCUXpresso and I have my own board based on MKL46Z256 . My end goal is to send I2S stream from a digital microphone chip to PC via USB audio. 

I tried the USB frdmkl46z_dev_audio_generator_bm project and it works fine. I can hear voice repeating in  my Windows PC. It uses 8 bits per sample and sampling frequency of 8Khz and fixed data.

My application involves speech only but in a noisy environment. I like to use 16 bits and 16kHz sampling rate.

My first modification of the example project should have been just to change the sampling frequency, but I am stuck here:

In my usb_device_descriptor.c 

0x01U, /* The number of bytes occupied by one audio subframe. Can be 1, 2, 3 or 4. */
0x08, /* The number of effectively used bits from the available bits in an audio subframe.*/
0x01U, /* Indicates how the sampling frequency can be programmed: */
//0x40, 0x1F, 0x00U, /* Sampling frequency 1 in Hz for this isochronous data endpoint. */
0x80, 0x3E, 0x00U, // changed to 16MHz sampling frequency

From various documentation I found about FS USB I think the USB isochronous method checks data every 1 ms so increasing sampling frequency will mean more data generated and bigger buffers needed. If I got it right 8ksps means 8 bytes every 1ms and thus I need 8 byte buffer in the original example. This would now be 16 bytes.

This is why I changed  in my usb_device_descriptor.h the following:

//#define FS_INTERRUPT_IN_PACKET_SIZE (8)
#define FS_INTERRUPT_IN_PACKET_SIZE (16)

and 

//#define FS_ISO_IN_ENDP_PACKET_SIZE (8)
#define FS_ISO_IN_ENDP_PACKET_SIZE (16)

I have tried various combinations but all experiments results in no sound on PC side and I don't know why.

I found it very hard to find any information about this topic and would be very happy for any comment. I have looked at a lot of pdf's about usb audio but it does not help me with my example.

I guess the basic question now is:" what else do I need to change when doubling the sampling frequency in the generator example?"

Best regards,

Baldur

[Alexis_A]

Hello @baldurthorgilss,

The interrupt size shouldn't change, if you want to check an example that uses a 16-bits and other frequency I will suggest checking the example for the K66 found in the SDK.

Best Regards,

Alexis Andalon

[baldurthorgilss]

Hi Alexis_A

And thank you for answering my question. I did not notice your reply until now.

I built and downloaded the k66 sdk, but I could not find any generator example in it. I looked at the speaker example, but that was quite different.

My status is now that I can generate sound with the default settings: 8kHz sampling frequency and 8 bit resolution (pcm8). I can also generate sound with 16kHz sampling and 8 bit if I increase the buffer size

//0x40, 0x1F, 0x00U, /* Sampling frequency 1 in Hz for this isochronous data endpoint. */
0x80, 0x3E, 0x00U, // changed to 16kHz sampling frequency BÞ

and

//#define FS_ISO_IN_ENDP_PACKET_SIZE (8)
#define FS_ISO_IN_ENDP_PACKET_SIZE (16)

Next step is to go to 16 bits. Just to not increase the buffer more for them moment, I went back to 8kHz sampling, changed from pcm8 to pcm, change from 1 byte to 2 bytes and from 8 bytes to 16:

//0x02U, 0x00U, /* PCM8 */
0x01U, 0x00U, /* PCM BÞ */

and

//0x01U, /* The number of bytes occupied by one audio subframe. Can be 1, 2, 3 or 4. */
0x02U, /* The number of bytes occupied by one audio subframe. Can be 1, 2, 3 or 4. BÞ: 2 bytes*/
//0x08, /* The number of effectively used bits from the available bits in an audio subframe.*/
0x10, /* The number of effectively used bits from the available bits in an audio subframe. BÞ: 16 bits*/
0x01U, /* Indicates how the sampling frequency can be programmed: */
0x40, 0x1F, 0x00U, /* Sampling frequency 1 in Hz for this isochronous data endpoint. */
//0x80, 0x3E, 0x00U, // changed to 16kHz sampling frequency BÞ

The microphone seems to get active in windows, but no sound is hearable.

I added error messages to the start og USB_DeviceCallback and USB_DeviceAudioCallback functions in auio_generator.c

usb_echo("B: in USB_DeviceAudioCallback");//BÞ

and got this output:

USB device audio generator Pi

B: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceAudioCallbackB: in USB_DeviceAudioCallbackB: in USB_DeviceAudioCallbackB: in USB_DeviceAudioCallbackB: in USB_DeviceAudioCallbackB: in USB_DeviceCallbackB: in USB_DeviceAudioCallbackB: in USB_DeviceAudioCallbackSet Cur Volume : fe0f

B: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallbackB: in USB_DeviceCallback

(then everything is silent)

It seems like the two functions are called several times before the system gives up.

I also had changed the buffer to int:

//extern uint8_t s_wavBuff[];
extern uint16_t s_wavBuff[]; //BÞ

and expect the size FS_ISO_IN_ENDP_PACKET_SIZE in USB_DeviceAudioSend to refer to bytes rather than samples. Is that correct?

I have studied several sample setups on the net and it seems to me like these are the only things to think of. Am I correct? What else might I have to change?

I would be very happy for any hint regarding solving this problem.

Best regards,

Baldur

[Alexis_A · Accepted solution]

Hello @baldurthorgilss,

Asking the applications teams, this issue looks like is a problem with windows, if you want these changes to apply you need to uninstall the driver.

0x80, 0x3E, 0x00U, // changed to 16kHz sampling frequency BÞ

#define FS_ISO_IN_ENDP_PACKET_SIZE (32)

These changes should be enough.

Best Regards,

Alexis Andalon

[baldurthorgilss]

Hi Alexis

Turns out that you had the solution

When I managed to delete the windows audio driver properly (it appeared in two groups in device manager) it turned out that my code was working all the time.

My conclusion is:

that logic analyzers might not reflect the actual situation (the way a root hub would understand the signals) as they flip the state by a single level but USB is defined upper and lower levels. 

Wireshark turned out just as good a tool in my case. It reflects the actual understanding of the USB packages.

Also deleting a driver is not trivial, it might appear in more than one group under device manager and in this case it must be deleted in all groups.

Going from 8kHz sampling frequency to 16kHz and from 8 bit to 16 involves:

changing sampling frequency

changing endpoint buffer size

changing audio format (PCM8 to PCM)

changing bits per sample

changing effective number of bits per sample

Thanks for the help,

Baldur

[baldurthorgilss]

Hi Alexis

I'm still struggling with this problem. At the moment I have doopts that my logic analyzer is interpreting the signals the way the PC does so I have involved Wireshark software. Here is the package sequence for the original version of the generator:

53     2.981347      host   3.11.0 USB    36     GET DESCRIPTOR Request DEVICE

54     2.981576      3.11.0 host   USB    46     GET DESCRIPTOR Response DEVICE

55     2.981622      host   3.11.0 USB    36     GET DESCRIPTOR Request CONFIGURATION

56     2.981755      3.11.0 host   USB    37     GET DESCRIPTOR Response CONFIGURATION

57     2.981785      host   3.11.0 USB    36     GET DESCRIPTOR Request CONFIGURATION

58     2.982033      3.11.0 host   USB    146    GET DESCRIPTOR Response CONFIGURATION

59     2.986505      host   3.11.0 USB    36     SET CONFIGURATION Request

60     2.987241      3.11.0 host   USB    28     SET CONFIGURATION Response

61     2.987553      host   3.11.0 USB    27     Unknown type 7f

62     2.987563      3.11.0 host   USB    27     Unknown type 7f

63     2.987589      host   3.11.0 USB    27     Unknown type 7f

64     2.987599      3.11.0 host   USB    27     Unknown type 7f

65     2.990824      host   3.11.0 USB    36     GET DESCRIPTOR Request STRING

66     2.991067      3.11.0 host   USB    32     GET DESCRIPTOR Response STRING

67     2.991106      host   3.11.0 USB    36     GET DESCRIPTOR Request STRING

68     2.991257      3.11.0 host   USB    58     GET DESCRIPTOR Response STRING

69     3.008799      host   3.11.0 USB    36     GET DESCRIPTOR Request STRING

70     3.009070      3.11.0 host   USB    32     GET DESCRIPTOR Response STRING

71     3.009107      host   3.11.0 USB    36     GET DESCRIPTOR Request STRING

72     3.009263      3.11.0 host   USB    58     GET DESCRIPTOR Response STRING

73     3.009777      host   3.11.0 USB    36     URB_CONTROL in

74     3.009934      3.11.0 host   USB    29     URB_CONTROL in

75     3.009976      host   3.11.0 USB    36     URB_CONTROL in

76     3.010136      3.11.0 host   USB    30     URB_CONTROL in

77     3.010164      host   3.11.0 USB    36     URB_CONTROL in

78     3.010357      3.11.0 host   USB    30     URB_CONTROL in

79     3.010384      host   3.11.0 USB    36     URB_CONTROL in

80     3.010542      3.11.0 host   USB    30     URB_CONTROL in

81     3.010580      host   3.11.0 USB    36     URB_CONTROL in

82     3.010733      3.11.0 host   USB    30     URB_CONTROL in

83     3.010806      host   3.11.0 USB    36     SET INTERFACE Request

84     3.011046      3.11.0 host   USB    28     SET INTERFACE Response

85     3.146705      host   3.11.0 USB    37     URB_CONTROL out

86     3.147025      3.11.0 host   USB    28     URB_CONTROL out

87     3.159268      host   3.11.0 USB    38     URB_CONTROL out

88     3.159578      3.11.0 host   USB    28     URB_CONTROL out

89     3.502870      host   3.11.0 USB    36     SET INTERFACE Request

90     3.503298      3.11.0 host   USB    28     SET INTERFACE Response

103    3.517515      3.11.2 host   USB    239    URB_ISOCHRONOUS in

105    3.527517      3.11.2 host   USB    239    URB_ISOCHRONOUS in

107    3.537521      3.11.2 host   USB    239    URB_ISOCHRONOUS in

109    3.547514      3.11.2 host   USB    239    URB_ISOCHRONOUS in

(continues sending data)

When I try my own version I get the following

19     3.888895      host   3.29.0 USB    36     GET DESCRIPTOR Request DEVICE

20     3.889139      3.29.0 host   USB    46     GET DESCRIPTOR Response DEVICE

21     3.889200      host   3.29.0 USB    36     GET DESCRIPTOR Request CONFIGURATION

22     3.889425      3.29.0 host   USB    37     GET DESCRIPTOR Response CONFIGURATION

23     3.889542      host   3.29.0 USB    36     GET DESCRIPTOR Request CONFIGURATION

24     3.889864      3.29.0 host   USB    146    GET DESCRIPTOR Response CONFIGURATION

25     3.894468      host   3.29.0 USB    36     SET CONFIGURATION Request

26     3.895379      3.29.0 host   USB    28     SET CONFIGURATION Response

27     3.895561      host   3.29.0 USB    27     Unknown type 7f

28     3.895569      3.29.0 host   USB    27     Unknown type 7f

29     3.895584      host   3.29.0 USB    27     Unknown type 7f

30     3.895588      3.29.0 host   USB    27     Unknown type 7f

31     3.898641      host   3.29.0 USB    36     GET DESCRIPTOR Request STRING

32     3.898915      3.29.0 host   USB    32     GET DESCRIPTOR Response STRING

33     3.898967      host   3.29.0 USB    36     GET DESCRIPTOR Request STRING

34     3.899208      3.29.0 host   USB    54     GET DESCRIPTOR Response STRING

35     3.913608      host   3.29.0 USB    36     GET DESCRIPTOR Request STRING

36     3.913895      3.29.0 host   USB    32     GET DESCRIPTOR Response STRING

37     3.913904      host   3.29.0 USB    36     GET DESCRIPTOR Request STRING

38     3.914125      3.29.0 host   USB    54     GET DESCRIPTOR Response STRING

39     3.914244      host   3.29.0 USB    36     URB_CONTROL in

40     3.914412      3.29.0 host   USB    29     URB_CONTROL in

41     3.914420      host   3.29.0 USB    36     URB_CONTROL in

42     3.914609      3.29.0 host   USB    30     URB_CONTROL in

43     3.914614      host   3.29.0 USB    36     URB_CONTROL in

44     3.914783      3.29.0 host   USB    30     URB_CONTROL in

45     3.914790      host   3.29.0 USB    36     URB_CONTROL in

46     3.914957      3.29.0 host   USB    30     URB_CONTROL in

47     3.914962      host   3.29.0 USB    36     URB_CONTROL in

48     3.915135      3.29.0 host   USB    30     URB_CONTROL in

49     3.915152      host   3.29.0 USB    36     SET INTERFACE Request

50     3.915381      3.29.0 host   USB    28     SET INTERFACE Response

51     4.031398      host   3.29.0 USB    37     URB_CONTROL out

52     4.031765      3.29.0 host   USB    28     URB_CONTROL out

53     4.043707      host   3.29.0 USB    38     URB_CONTROL out

54     4.044069      3.29.0 host   USB    28     URB_CONTROL out

That is, the second "SET INTERFACE Request" seen in the un-modified version (line 89 in the first list) never comes. This must be a hint in my debugging. I would be happy for any advice regarding where/what to look for regarding this stop in communication.

Best regards

Baldur 

[baldurthorgilss]

In my quest to understand the problem I got me a USB analyzer. The USB protocol is of cause extensive, but I start slowly. One common thing is the frame rate package sent every 1 ms. Checking my mouse I got an image like the following

When I run the generator project on my NXP board unchanged from the example I get this every 1 ms (otherwise same setup on logic analyzer):

Almost every frame package has error!

In spite of this large amount of errors, I can still hear the generator sound in my PC.

This is the original example, I have not changed it. Why has my frame (and other) packages error (it can easily be seen e.g. in the D+ signal how strange it looks)? I have tested a number of other devices, no other device behave like this.

My ultimate goal is to understand why my 16-bit, 16kHz signal is not working as streaming audio, but that error drowns in these errors. I need to fix this first. 

I would be happy for any comment. 

Best regards,

Baldur