Hello,
I have an audio application that I intended to use a Colibri iMX6S for recording audio. However I recently run a couple of tests and noticed there is what seems to be a hardware noise near 10kHz, and at 650Hz (thou it is less noisy) as shown in the picture (using audacity, and showing the spectrogram).
[upload|KMIcuCh9AHExoelSsiEHrjN2C5E=]
I have tested this with 3 Colibri iMX6, and all of them have shown the same problem. I am using iris carrier board with all of them.
Note, that I also have several VF61 that do not show this problem. Picture below was taken from an audio recording from a VF61, and what I expect from iMX6 as well.
[upload|Z+ct3nUo1HesQkGtxaKnLkTNarM=]
I have used the exact same setup for both the VF61, and iMX6. I have also tried changing power source, location, nothing worked.
These audios were recorded without any microphone, thou the noise is still there and very visible when using one. So it seems this problem is associated with the hardware.
I am recording using the following command and parameters on the iMX6:
arecord --device=hw:0,0,0 -q -r 48000 -f S16_LE -c 1 -t wav -d 60 “recording.wav”
(I am recording using only one channel and subdevice 0 (left channel)).
Any idea on how to eliminate the noise?
Thank you.
hello
Which Bsp version are you using? Could you provide recorded files with and without noise? It can be also useful to have some measurements results using oscilloscope?
Thanks
I have uploaded both files. One recorded with VF61, which has no noise. And another with iMX6 with noise.
I am using Linux LXDE 2.7 (stable).
About the measurement with oscilloscope, what exactly do you want me to measure? Just to be clear, both the files were recorded without any signal (no external signal), that is, we didn’t use any microphone. We did this to try and isolate as much as possible any external factor.
I might add that to completely exclude any external factors (like some electromagnetic interference) we even tried to change locations of the recording several times. The noise was still present.
That is why we think that the noise shown seems to be a problem internal to the iMX6.
link text
I have uploaded both files. One recorded with VF61, which has no noise. And another with iMX6 with noise.
Thanks for the files.
I am using Linux LXDE 2.7 (stable). About the measurement with oscilloscope, what exactly do you want me to measure? Just to be clear, both the files were recorded without any signal (no external signal), that is, we didn’t use any microphone. We did this to try and isolate as much as possible any external factor.
I thought there might be some interferences caused by other devices, but if you isolated the problem, then it is not the case. The noise is coming from the microphone input on Iris Board, since there is no connection to ground when there is no mircrophone connected. As Solution you can shorted the Pin2 and Pin5 of Connector X9 in Iris Board, this should reduce the noise. Further usually it is not common to record a sound without having any microphone connected. Do you have a special application for that?
Perfect, that it works. You are welcome. Thanks for the feedback.
Your answer has led me in the right direction.
Our aplication does use a microphone, but we were receiving noise. And we thought it was from the iMX6, but indeed, I believe it was because of some connection issue with the ground pin. I remade the connection and the noise was gone.
Thank you for your help.
