Noisy output when playing recorded line in signal

I am getting to know the teensy board and the audio shield and observe some noise I cannot explain.

My setup is simple:

- I have plugged a headset--call it headset A--into the 3.5mm plug of the audio shield

- I have taken apart another headset--call it B--, disconnected one of the left ear piece, and wired the left and ground signals from that ear piece to the left line in and ground of the audio shield. I connected the jack of this headset to my computer, which I use to stream audio. This way, I can listen to the audio streaming through the right ear piece and stream the left audio to the left line in port of the shield.

My observations:

- If I run the Example->Audio->HardwareTesting->PassthroughStereo, the playback sound quality through headset A is perfect, as good as what comes out of headset B

- If I run the Example->Audio->Recorder project and use the mic for the input, the sound quality on the playback is ok; it is what I would expect the microphone is capable of producing

- If I run the Example->Audio->Recorder project but use the line in for the input (myInput=AUDIO_INPUT_LINEIN instead of AUDIO_INPUT_MIC), the sound quality on the playback is poor: there is a constant background noise quite noticeable when the signal output is low. (That noise is much greater than the noise picked up by the microphone in the previous experiment.)

Is there an explanation for this noise?

Thanks

all things being equal the line in should be higher quality than the mic in, I would have thought. So maybe your line in signal is really noisy. Maybe you need to buffer the input signal. What are you feeding in? What load is it presenting? are there electrical appliances nearby? A radio station? You aren't running a mic into the line in are you??? thats what it sounds like to me .... Don't do that, and use a preamp. for a start, mics put out very little voltage ... then there is the impedance mismatch, and then there are other unspecified weird issues. use a preamp

> all things being equal the line in should be higher quality than the mic in, I would have thought.
> So maybe your line in signal is really noisy. Maybe you need to buffer the input signal.

Indeed. My expectation too.

> So maybe your line in signal is really noisy.... You aren't running a mic into the line in are you???

No. I wasn't clear enough about my second experiment: I should have clarified that in my second experiment I was using a mic and feeding the sound to the ADC of the audio shield unlike by first and third experiment that do not involve a mic or the ADC at all.

In my third experiment, I am just plugging the headset B in my computer from which I stream music and wiring the output at the ear piece to the line in of the audio shield.

There could be a number of things I do wrong along the way _but_ "If I run the Example->Audio-> HardwareTesting->PassthroughStereo, the playback sound quality through headset A is perfect, as good as what comes out of headset B"

This validates that I receive a signal of high quality on the line in since the pass-through test works properly. It is only when that I divert that signal to be recorded on the SD card and read it back from it that I lose on the quality. And since I don't get the same degradation in quality in my second experiment, I have also validated that the path to record audio to the SD card works well.

In short: This is what puzzles me: first experiment confirms the line in -> line out chain works well with no noise, second experiment confirms the SD record and read works well with no noise other than the expected noise from a mic, and yet third experiment has noise issues, (much more noticeable than the noise level generated by the mic).

Ok. i understand now.

It is strange.

If Example->Audio-> HardwareTesting->PassthroughStereo uses analog switching, then the issue could be with how the ADC reacts to your signal .... If you are using the signal from earbuds, I wonder if there is either a grounding issue (common ground ... seems unlikely), or a massive load difference (my headphones are like 150Ohms or something ... nothing like the 10k for line level).

Maybe try a real line level signal instead of one from earbuds. ...

maybe its enormous overload. I know that when I use the "earbuds out" from my ipod it overdrives the inputs on my 'pro' gear (+4) line inputs, creating the worst distortion imaginable .... not a level drop though...

Could be massive crosstalk and cancellation.

Other than that, maybe there is a deep mystery in the audio code, but I doubt that since it has been tested by a lot of people.

Do you have a 'real' line output to test??

Sorry for the loooong delay. It took me a while to get back to this project.

Since the last exchange, I did try with a "real" line output from a recording device I had like you suggested, _and_ with a mic that was followed by a preamp bringing the signal to line level. In both cases, I get the same behavior I reported before: a very high noise level.

I should add that the signal itself is played back at a reasonable level, i.e. the recording that plays back is the signal I recorded played back to me at a "reasonable" level; it is just simply very corrupted with noise.

I also experimented with the AudioControlSGTL5000 control lineInLevel to no avail: the value I pass in to lineInLevel does affect both the level of the noise and signal being recorded but the SNR remains unchanged.

If I send no signal to the line in of the audio shield and play back the recording, there is no audible noise. I.e. the noise isn't present when no signal is being streamed through.

Last, I tried with both the line in L and R entries and observe the same behavior.

I am not sure what to try next.

well you have eliminated a lot of potential issues .... it could well be a hardware issue with the audioshield out.... maybe use the audiosheild based sine generator and see what that throws up ....