Sound sensor reading increase over time

bben.pub · January 20, 2017, 2:54pm

Hello,

I’ve tried to use the sound sensor on a GrovePi+ but I fear it is faulty, as you can see by those readings over 24 hours:

The test protocol was to read every minute from the sound sensor and log the value (actually an average over 8 values) and draw the graph. The graph represents 24 hours, starting at 9:00 AM.

As you can see, there is probably an issue with the sensor. What can I do to fix the issue?

The source code is available here, for inspection:
logger.py (537 Bytes)

Thanks!
Ben

graykevinb · January 21, 2017, 6:26pm

I’m not exactly sure but something tells me its in your code.
Specifically in the for loop that takes the readings.
Perhaps simplify it and remove the sampling? I can’t try your code because I don’t have a grovepi.

Hope its the code, if not its most likely the firmware or grovepi.py. I highly doubt the hardware broke; not that it is not a possibility. Just highly doubt it.

bben.pub · January 22, 2017, 3:57pm

Hi,

I remove the sampling loop and I get similar results:

The updated script is here:
logger.py (401 Bytes)

Any idea?

karan · January 24, 2017, 4:35am

@bben.pub,
Can you follow the tutorial here to generate a log and upload the log here. This would help us to find if there is any problem with the firmware or the software.

Also, which image are you using. Is it a Dexter Industries Raspbian for Robots or something else.

bben.pub · January 24, 2017, 4:40am

Hi @karan,
Thanks for the reply. It’s a base raspbian (not a raspbian for robots). How do I get the tool to generate debugging information?

Thanks,
Ben

karan · January 24, 2017, 4:42am

@bben.pub,
Can you go to this folder on the Raspbian image. Make complete_test.sh executable and run it:

sudo chmod +x complete_test.sh
sudo ./complete_test.sh

Choose option 5 in there and there should be a log on the desktop by the end of it.

bben.pub · January 25, 2017, 2:36am

Thanks @karan!

Here is the log:
log.txt (5.8 KB)

I hope this helps. Did you have the chance to reproduce the issue on another GrovePi + sound sensor?

Thanks,
Ben

bben.pub · February 9, 2017, 1:33am

Hello,
Do you think the sensor could be faulty?

graykevinb · February 9, 2017, 4:34am

Maybe let it run a little longer. It seemed to level out. Perhaps it just gets a little static when it starts.

karan · February 10, 2017, 7:29am

@bben.pub: the log looks Ok. Nothing bad in there. Does the sensor respond to loud noises, like does the average increase. Can you post a picture of your setup.

iteachPi · February 11, 2017, 5:36pm

It seems that I am having the same problem as you. Did you ever find a solution?

I had my students make a noise sensor and buzzer program in scratch. When a certain noise level is reached the buzzer goes off then after a few seconds it starts detecting noise again. However we have to keep asjustIng the noise level higher and higher otherwise the buzzer keeps going off. Then if the pi is restarted or the sensor is unplugged and then reinserted the noise level has to be lowered again.

bben.pub · February 12, 2017, 3:55pm

Hi @karan,

Yes, the sensor responds to loud noise (the spikes on the graph above correspond to trucks passing by).
Here is a picture of my setup:

Could it be an issue during the A-D conversion? Is there another analog sensor I should try with? FYI, I’m using the potentiometer and it doesn’t drift.

Best regards,
Ben

karan · February 13, 2017, 4:14am

@bben.pub,
I think it might just be the sound sensor’s behavior. If the potentiometer doesn;t drift at all, then the AD channels and the GrovePi firmware are fine. I have seen a couple of sensors which require some conditioning time for their readings to stabilize. Do you have another sound sensor by any chance just to check if it has the same behaviour.

bben.pub · February 14, 2017, 2:43am

I found one to borrow. I’ll post the results in a few days.

bben.pub · February 16, 2017, 4:47am

Hi,
I did a 24-hour test with another sound sensor.

It’s much better, but still, there is a gradual increase of the base value.

karan · February 16, 2017, 6:31am

@bben.pub,
It does look like these sensors have a conditioning time after which the values normalize. What project are you using the sensors for and would this work for you.

bben.pub · February 17, 2017, 4:49am

I’m trying to monitor ambiant sound in an office space. My sensor has a conditioning of 6-12 hours, which is too much for such a project.

karan · February 20, 2017, 4:41am

@bben.pub,
Can you make a filter to remove the extra values in there. Can you post a log with values for 12-24 hours. I can try to make the filter for it.