• Resolved

TPA3251: Drop outs when driving speakers when high phase angle.

Part Number: TPA3251


We have a stereo amplifier product with TPA3251. Driving 4 Ohms resistors works fine. When driving 4 Ohms speakers and music playback there is drop-outs in sound for about 220mS at high playback level. I added Schottky diodes on chip outputs to GND and it did a improvement, but problem still not solved. I inspected 10 pcs from MP and they all have this issue but the threshold level for drop-outs has some variations. Drop outs happen even with cold heatsink and using high end thermal paste does not solve the problem. When Schottky diodes is added the amp can output full power in 2 Ohms with one ch. driven. So it's not current limiting issue. Power supply is 36VDC and very stable, newer drops below 34V. Below is measurements of Loudspeaker Current and Voltage when music playback and a typical 4 Ohms rated speaker, (B&W 704S2) It seems that high current when Voltage is low is causing the issue. See attached.  Any suggestions? Thanks. Michael

  • Michael,

    If the device works fine with a 4 Ohm resistor load but has issue with the 4 ohm speaker, the speaker impedance may be dipping too low at higher music transients. Do you happen to know the speaker impedance variation with frequency and power?


  • In reply to Adam Sidelsky:

    Hi Adam,

    The speaker is minimum 3.1 Ohms. Using 3 meter cables and current shunt I guess amplifier load is not lower then 3.5 Ohms. Looking at the Zoomed view measurement I attached, the current (red curve) newer exceed 8A peak. This far from the 15A limit in TPA3251. Even at 24V supply voltage (PVDD) it still make drop outs and the peak output current is below 5 Amps and far from limits in TPA3251. Also far from thermal constrains. If only one speaker attached there is no drop outs, so this rule out current limit problem.
  • In reply to Michael Kragelund35:


    Do you have an EVM for this device? It may help to check the result using that board.

    On your board, please check the PWM waveform (the Pre-filter output node) and send a picture including the current waveform as above. This way we can check if the current waveform is result of the device skipping pulses.

    Please also send me your schematic and output Inductor datasheet: adamsidelsky(at)ti(dot)com


  • In reply to Adam Sidelsky:


    Problem solved. GND power plane to chip was not optimum. Soldering wires to GND pins on chip did fix the problem.  

    Thanks for your support!