I bought used Logitech Z-10 computer speakers. They can act as a USB audio device or accept line input, and they have a headphone output jack. Some music didn't sound right, so I played a frequency sweep and heard other frequencies and beat frequencies between them and the sweep. I uploaded a spectrogram of a frequency sweep. From the very start, some but not all odd harmonics are visible, indicating symmetric distortion. Then at higher frequencies there are descending images of those harmonics, indicating non-linearity and constant higher frequency noise mixing with the harmonics. The image of the 3rd harmonic crosses over the original signal at 12 kHz and reaches zero frequency at 16 kHz. That means it is mixing with 12 * 3 + 12 = 16 * 3 = 48 kHz. Because 48 kHz is a common audio sampling frequency and the highest frequency supported by the USB audio interface, I don't think this is from power supply ripple.
The harmonics and their images appear suddenly when digital input level exceeds around 74% of digital full scale. That makes me think "clipping", but it's kind of weird, with different distortion at different frequencies, not simply flat clipped tops. I'm attaching an image of the waveform at 1000 Hz, overlaid with my attempt to fit a sine wave to it. All of this is played via USB and recorded from the headphone jack. Looking at it with an oscilloscope I also see plenty of irregular high frequency noise that should be inaudible.
According to https://www.badcaps.net/forum/troubl...akers-no-sound "The USB/DSP chip is a Micronas UAC3556B" and a UAC355XB datasheett is available at multiple sources, including https://www.digchip.com/datasheets/p...C355XB-pdf.php. I suspect one of the filtering capacitors attached to the UAC3556B chip is bad. These problems happen with both analog and USB sound input, and with both headphone jack and speaker sound output. The sound probably always passes through the UAC3556B chip, and the analog sound being digitized because the chip doesn't offer an analog sound path.
The worst thing about fixing these is that the front is transparent plastic with a black backing. That plastic needs to be unglued without damaging the black backing. Then screws can be accessed and the rest should be straightforward. I put the speaker with the controls and electronics inside on a heating pad to hopefully soften the glue. I'm hoping to fix this, though if I can't, I can still use them but limit myself to the portion of digital full scale that doesn't cause this distortion.
Edit: Heating it up on a heating pad worked well for softening the glue. Then plastic putty knives worked well for separating the front, though I needed to start at particular locations where it was easier. The touch controls were glued to the front, and fortunately I saw that early enough to avoid damaging the ribbon cable connecting to them. The LCD is not glued to the front.
The harmonics and their images appear suddenly when digital input level exceeds around 74% of digital full scale. That makes me think "clipping", but it's kind of weird, with different distortion at different frequencies, not simply flat clipped tops. I'm attaching an image of the waveform at 1000 Hz, overlaid with my attempt to fit a sine wave to it. All of this is played via USB and recorded from the headphone jack. Looking at it with an oscilloscope I also see plenty of irregular high frequency noise that should be inaudible.
According to https://www.badcaps.net/forum/troubl...akers-no-sound "The USB/DSP chip is a Micronas UAC3556B" and a UAC355XB datasheett is available at multiple sources, including https://www.digchip.com/datasheets/p...C355XB-pdf.php. I suspect one of the filtering capacitors attached to the UAC3556B chip is bad. These problems happen with both analog and USB sound input, and with both headphone jack and speaker sound output. The sound probably always passes through the UAC3556B chip, and the analog sound being digitized because the chip doesn't offer an analog sound path.
The worst thing about fixing these is that the front is transparent plastic with a black backing. That plastic needs to be unglued without damaging the black backing. Then screws can be accessed and the rest should be straightforward. I put the speaker with the controls and electronics inside on a heating pad to hopefully soften the glue. I'm hoping to fix this, though if I can't, I can still use them but limit myself to the portion of digital full scale that doesn't cause this distortion.
Edit: Heating it up on a heating pad worked well for softening the glue. Then plastic putty knives worked well for separating the front, though I needed to start at particular locations where it was easier. The touch controls were glued to the front, and fortunately I saw that early enough to avoid damaging the ribbon cable connecting to them. The LCD is not glued to the front.