We have a design using SSM3302 class-D power amp driving a piezo transducer via step-up transformer and matching Ls. The device is configured as a single channel mono amp (MOMO pin 17 pulled up to AVDD) with outputs paralleled. Input is differential AC coupled via 1.0 uF caps into pins 11 INL+ & 12 INL-. The design is well wrung out and on its 2nd spin of the PCB layout.
Recently as part of work characterizing the piezo transducer in the system we noted a behavior of the SSM3302 I want to ask about. Normally the signals sent this amp are in the 9-15k range and are windowed such that the amplitude builds over several cycles to full scale and then tapers off in amplitude at the end prior to shutting off. The circuit has also been extensively tested with full scale 10k Hz sine waves that turn on and off immediately at full scale. The recent test calls for sending the amp a 10k Hz sine wave (non-windowed, i.e. starts at full scale) for 20 mS. When that signal is sent the SSM3302 outputs a cycle or two at full scale then amplitude decreases until by 2 mS output signal is in the noise. Having previously in our development cycle driven the SSM3302 into over current shut down we tried reducing the amplitude of the signal, the output amplitude followed that amplitude reduction however the correct amplitude for a cycle or two followed by amplitude decrease until off by 2 mS remains. Thus we conclude it is not due to exceeding the 6 A overcurrent threshold. When the test signal is modified to ramp amplitude up over the first 10 mS to full scale for the 10 to 20 mS time period the SSM3302 outputs the desired full power amplitudes rather than tapering off.
Could the un-windowed 10k Hz sinewave be triggering the SSM3302 POP suppression? Are there other explanations for what we are observing?