Hi,
I have tested the audio of a Sierra Wireless RC7620 device mounted on a mangOH Yellow board.
The RC7620 device has been upgraded with the “RC76xx_Release8_GENERIC_GCF_7620_SWI9X07H_00.08.07.00” firmware.
Here below the response to the ATI command:
Manufacturer: Sierra Wireless, Incorporated
Model: RC7620
Revision: SWI9X07H_00.08.07.00 0ce4c1 jenkins 2021/03/17 02:18:20
IMEI: 353634110179750
IMEI SV: 17
FSN: 7Q1085702611B0
+GCAP: +CGSM
I have configured the audio with the following AT commands:
AT!AVCFG=0,0,1
AT!AVSETPROFILE=0,0,0,0,5,0
AT!AVRXG=0,2000
AT!AVTXG=0,2000
AT!AVEC=0,0
In order to test the audio:
I activate the audio loopback with the command AT!AVAUDIOLPBK=3
I transmit a 1 kHz sine wave to the audio input
I expect to receive from the audio output a signal identical to the signal transmitted to the audio input
In such a test, the signal received from the audio output is a distorted 1 kHz sine wave.
As you can see in the attached picture, there’s an interference overlaped to the sine wave.
Note that I have acquired the audio using a 48 kHz sampling rate because my application requires wideband audio.
That interference reduces the audio quality, which is important for my application.
Please could you investigate this issue.
Thank you.
You can see when there is distortion on the input signal this is immediately mirrored, to a lesser degree, in the output.
During a normal call with bi directional speech then the audio goes through codecs which are lossy and produce an audio signal approximating the original i.e. the 1KHz signal but would be hard to differentiate with the human ear.
in the Picoscope screen shot I see that you have set a 8 kHz sampling rate, which is ok for narrow band audio but not for wideband.
In our test we use a 48 kHz sampling rate in order to acquire wideband audio, which is required for the POLQA measurment.
Could you please verify on the development board if the audio waveform is clean using a 48 kHz sampling rate?
Thank you for your support.
So I re ran the test as before and set the sample rate to 50KS/s, I then reduced the time base to 500uS and 200uS to give better granularity on the signal. Scope traces are in the file below. I can’t see anything that changes my previous statements. In fact you can especially see where the input signal (Channel A) has some noise it is reflected in the output (channel B), for some reason this is almost always at the point it triggers. RC76 audio loopback test 50Ks per second, 200us and 500us TB.zip (180.7 KB)
Also FYI I have attached the schematics of the HL dev kit in case you think the codec might be smoothing things out (they are different between the dev kit and mangOH board), this is something that could potentially make a difference. HL dev kit schematics Dev_kit_Rev_C1.6.pdf (1.6 MB)
I agree with you that probably the distortion comes from the mangOH board, which integrates a WM8944B PCM codec, and in this case we’ll have to change our hardware.
The schematics of the HL dev kit indicates the U1104 component but it does not report the name of the audio codec.
The “AirPrime_HL6_and_HL8_Series_PCM_Codec_Application_Note_Rev1_1.pdf” document from Sierra Wireless indicates that the audio codec is the W681360YG (from Nuvoton).
Could you confirm that the HL dev kit you have used integrates the W681360YG codec?
