Electronic – Optical audio transmitter can’t handle speech

audiolaserlaser-diodephototransistortransmitter

This is a question relating to my project on an optical audio transmitter.

I wired up a very simple device in which I use an audio signal transformer (EI14) to modulate the intensity of a laser beam (a cheap 650nm, 5mW diode) according to the audio output of my phone. The laser is then received by a photo-resistor (G5528 A205) wired up to the microphone socket of my laptop. With this setup I can transmit audio between the two devices.

I then upgraded the setup, replacing the photo resistor with a photo transistor in order to improve the response time to the change in the intensity of light (12us rather than 20ms).

The quality has improved significantly; it's still not great and I never expected it to be as there's much to improve. But there is one particularly baffling property of this setup. It transmits audio just fine, but it can't handle speech. Not that it transmits it badly, it's more that it doesn't transmit it at all. In this recording, the audio is captured but the vocal that should be in full swing after around 50s is just not there. If you listen really carefully and know what to look out for, there is the slightest of remnants of the original vocal.

I thought that it's drowned out by the music and tried recording a voice message and transmitting that instead. Same result – the receiver sees nothing, the received amplitude sketch won't even twitch. Same with an audio book recording.

Any ideas about as to how such discrepancy between music and speech could arise?

Best Answer

It seems to me that you are probably using a music source that is stereo and, instead of it delivering a mono signal as you expected (L plus R), the connection of the transformer to L and R channels results in L minus R.

Given that vocals and bass are usually equally mixed left and right in a stereo mix, the transformer connection you have used will only pass signals that are either in the left channel or the right channel hence, bass and vocal are largely ignored.