0
\$\begingroup\$

The Issue

I've made a project centred around the XS3868 module which is based on OVC3860. The circuit works fine and I was able to get the microphone to work except for just one thing, the mic input is noisy as hell. The audio input is clear enough when I'm speaking something to the mic, but it gets noisy in an echoing manner whenever there is silence.
I want to get rid of this distracting noise. Is there any way to do so? If so, then please help me in making my project reach closer to perfection.

My Setup

Here is the schematic which I followed:

The schematic

The schematic is simply based on the manufacturer's application schematic. I'm a beginner with digital signals and hence I simply combined multiple schematics into one. However, I'm ready to learn and improve this schematic.

The details about my microphone connections are as follows:

  • My circuit uses the microphone of TRRS earphones.
  • The mic input is fetched through the MIC and GND ring and sleeve.
  • Since XS3868 has different GND pads for audio and power supply, therefore, instead of grounding one of the mic terminals according to the schematic, I connected it to AGND, or the Audio-Ground pad, of XS3868. AGND is at around 0.6V voltage level.

    This connection to AGND is indirectly made because I also use the earphones inserted to the TRRS jack for audio output and earphones intrinsically connect MIC- to GND. This has been done to prevent shorting AGND with GND.

What I've Tried

I tinkered around with the circuit and here's what I've observed:

  • Connecting an external condenser microphone according to the schematic(which is based on the manufacturer's schematic), i.e. with MIC- grounded properly, I still heard the noise, which subjectively felt slightly lesser in magnitude.

Recording of the noise: https://clyp.it/gap2kc0z

\$\endgroup\$
3
  • 2
    \$\begingroup\$ It's possible that you're experiencing some sort of automatic level control function that is built into the module. When you're not speaking, it gradually increases the gain until you hear all of the background noises. It is also possible that this is configurable -- look for it in the module documentation. \$\endgroup\$
    – Dave Tweed
    Commented Mar 15, 2019 at 16:55
  • \$\begingroup\$ @DaveTweed I've checked the datasheet for OVC3860(couldn't find XS3868's datasheet) and it doesn't contain any info on automatic level control. I also believe that the noise is due to automatic level control since it resembles the ambient noises. However, I'll double-check that today. I've also looked up the PSKeys documentation for OVC3860 which lacks any info about auto level control as well. Hence I don't believe that to be configurable. Is it possible to somehow nullify the low-intensity signals in the circuit? \$\endgroup\$ Commented Mar 16, 2019 at 7:40
  • \$\begingroup\$ I double checked the noise, and it isn't in way, resembling the ambient noise. \$\endgroup\$ Commented Mar 16, 2019 at 9:31

4 Answers 4

2
\$\begingroup\$

Electret mics often have the rear open with an open weave cloth to keep out dust. It varies a lot between mics that look the same as balancing the front and rear air pressure to cancel is hard to match perfectly.

This helps to cancel far-field sounds like echoes off walls and distant noise that degrades the near-field voice signal to noise ratio.

Check your mic quality for this feature and test for far-field cancellation. Attenuating the input with the exact amount of cloth might help but is hard to balance. So often, mic selection is trial and error, unless they say this has excellent distant noise cancellation.

Also if you are using a speaker instead of a headset, positive feedback with full duplex audio will sound like echoes unless it has echo cancellation technology. ( side tone cancellation)

\$\endgroup\$
3
  • 2
    \$\begingroup\$ I don't think the noise is due to the microphone since I've recorded very good quality audio using that. I suspect it is being caused by the mic input's wrong processing in the module. Also, regarding the positive feedback, I've ensured that there is no feedback while testing by using an earphone. \$\endgroup\$ Commented Mar 16, 2019 at 7:46
  • \$\begingroup\$ Noisy as hell is not as descriptive as aliasing , saturating or fuzzy \$\endgroup\$
    – D.A.S.
    Commented Mar 16, 2019 at 8:03
  • \$\begingroup\$ I tested the circuit with another mic(another earphone, say Earphone-2) and there was no noise in that case. I've recorded the noise(in case of Earphone-1) and pasted the link to it in the question. It is weird that there is no noise in Earphone-2 because both microphones are good in recording in a wired manner. Also, I'd like to add that even though both earphones lack any noise when connected in a wired manner, Earphone-2 records in better quality. I don't think the noise is due to the background. \$\endgroup\$ Commented Mar 16, 2019 at 9:31
0
\$\begingroup\$

The electret microphone should be powered. Please connect a 4.7k resistor between the microphone # 1 terminal and the positive capacitor terminal (1.8V). It works for me.

\$\endgroup\$
1
  • \$\begingroup\$ seems that pin 4 of the chip is a DC power rail. Without it the OP would get no signal at all. \$\endgroup\$
    – danmcb
    Commented Apr 22, 2021 at 7:52
0
\$\begingroup\$

Documentation is sparse for the interface chip, but as there is no gain control for the mic, most likely there is AGC. So when you have no signal, gain goes to max and the self noise of the preamp becomes audible. If you clearly hear the noise "pump" as signal is applied and removed, this is what is going on. Not much you can do, unless there are some commands you can find to suppress the AGC. But this will bring other issues, it's there for quite a good reason.

EDIT : of course there is something you might try, which is to have your own (variable gain) mic amp, and mix it with the line level input to the chip. Extra work and circuitry but it might be that AGC is only applied to the mic input - this would make sense.

\$\endgroup\$
0
\$\begingroup\$

A guess:

Add to the mic signal some continuous, but unaudible signal. Test if it fools the AGC to keep the gain low enough. A pure sine subbass, even few millivolt DC through a resistor to the right end of C4 can do the job. OVC3860 datasheet gives no hints does it work or not.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.