Issues with external mic in Line In 1

Hi,
I’m Valentin, I’m working in a research chair in Montreal specialized in in ear technologies.
I’m writing you after a problem we “solved” with the Tympan rev D hardware.

Our goal was to use our earpieces (using Analog Microphone Electret Condenser) connected to the Tympan Platform, especially the input 1 on the headers.

The first thing we did was to verify the sensitivity of our mic powered by the Tympan power supply (1.8V). We placed our mic in a Sound Calibrator Type 4231 at 94dB and 114dB, we measured the true RMS value and we compared our results to the microphone documentation. The results were correct.

Then, we decided to compare the signal produced by the microphone when connected to the IN1 and the same signal produced by a function generator connected to the IN1. We measured the voltage with a National Instrument card at the input pin and we noticed a big difference between those signals. The mic signal was clearly heavily distorted at 114dB.

We had the idea that something was pumping too much current for the microphone. So we decided to remove the capacitors C37 and C39 in order to only have our mic connected to the codec. We did the same measurements as before and we noticed the majority of the distortion was gone.

Our conclusion was the resistor R32 and R33 (1K) directly connected to the GND creates an impedance that is too low for our mic.

Did someone run into the same issue?

The data in the plots shows the 2 mic signals (recorded by the Tympan on an SD card) when in the calibrator at 114 dB before and after removing capacitors C37 and C39.

What a great post! Thank you so much for your analysis and for sharing your findings with all of us.

I looked at the schematic for the RevD: Tympan_Rev_D_Hardware/TympanRevDschem.pdf at master · Tympan/Tympan_Rev_D_Hardware · GitHub

It took me a while to find R32 and R33 because they are not obviously associated with the audio codec or with IN1…instead, they’re associated with the audio output of the Bluetooth module. But, you are right, since the Bluetooth audio output is connected to the IN1 (so that you can digitize and process the bluetooth audio), these 1K resistors do affect IN1. That’s a great find!

These resistors are there to attenuate the audio coming from the bluetooth module. This is a very old part of the design. Now that we know more, I do not believe that we need to attenuate the bluetooth audio quite this much. I do not think that we can eliminate R32 and R33 from the design, but we can increase them in future versions.

For you, if you do not need bluetooth audio, your modification is perfect.

This is such a great find. Thank you.

Chip

image

Issue created on GitHub: Alter audio-attenuating resistors around the Bluetooth module (R32 and R33) · Issue #6 · Tympan/Tympan_Rev_D_Hardware · GitHub

Thank you very much for your consideration.
Best regards,
Valentin