We have designed in the SSM3302 Class-D audio amplifier into a product under development for use driving a piezoelectric transducer. The device is used in in mono mode with PVDD = 15VDC (19VDC also tried with no effect on this issue). Input is differential to INL+ & INL- pins. Output is via paralleled OUTR+/OUTL+ and OULR-/OUTL- pins to a Murata DLW5BTM102SQ2L common mode choke + three 1000 pF caps (as in Evaluation board schematic) then onto a 1:8 ratio step-up transformer, impedance matching inductors, and to the piezoelectric transducer. The circuit has been developed initially using the EVAL-SSM3302Z evaluation board and is now on the second revision of the PCB for use in the final product.
The issue that has arisen is as testing progressed to use more of the systems’ multi-tone burst signals we began seeing the SSM3302’s outputs mute off and then back on. Some of the signals play ok, others cause the behavior shown below. The cause of this behavior is a mystery I hope engineer zone forum members can help with.
Figure 1 - Example of circuit behaving correctly outputting the 1 second long multi-tone burst signal. Channels 1 & 2 show the signal across the piezoelectric transducer. Channel 3 is PVDD set to 19VDC, after the burst is output PVDD is turned off to save power. Channel 4 is one side of the SSM3302’s output at the node where OUTR+/OUTL+ are joined prior to filtering components. The scope is setup in this configuration for all the following images.
Figure 2 – Data from Figure 1 zoomed to 10x to 10mS.
Figure 3 – Data from Figure 1 zoomed to 2mS.
Figure 4 - Example of circuit behaving incorrectly outputting the 1 second long multi-tone burst signal showing dropouts.
Figure 5 – Data from Figure 4 zoomed to 10x to 10mS.
Other behaviors that might offer clues to what is going on are:
- When the multi-tone burst signal is replaced with a 10k Hz sinewave at full amplitude the entire 1 second signal drops out.
- When the amplitude of the multi-tone burst signal or 10k Hz sine wave is reduced by ~half the dropouts disappear. This is puzzling because all thru the Summer of 2014 testing proceeded using the full amplitude signals.
- Dropout issue manifests on both revisions of the PCB.