analog – How to use two or more ultrasonic transducer drivers?

My circuit is below, it has two ultrasonic transducers, transformers and drivers. The transformer is B78416A2232A003, the transducer is MCUSD14A40S09RS-30C, the driver is PGA460-Q1. Digital interface wires isn’t shown.

^{simulate this circuit – Schematic created using CircuitLab}

Transducers is mounted opposite each other in distance 88 mm. The one can burst pulses, the other can listen. This is a waveform plot measured on B1 when B1 burst 20 pulses. Blue curve is a raw ADC samples, orange is BPF-processed waveform, green curve is an envelope.

Waveform has “swing period”¹ ~54 us, pulse generation period ~486 us, decay period.

¹If you know more appropriate term you can edit my question.

Now let’s see the waveform measured on B2 when signal is burst by separate circuit. This circuit has it’s own voltage source and doesn’t have any common wires with measuring circuit. This transducer is driven by waveform generator that burst 4 pulses with time interval 700 us.

We can see nice waveform that can be easily detected and processed.

Now let’s see the waveform measured on B2 when B1 burst signal and B2 listen.

Firstly, B2 started signal listening (see vertical red time mark). Then B1 started burst (see second vertical red time mark). This time moments are fine measured and plotted.

We can see mark “Burst received”. Time between marks “Burst received” and “Burst started” is ~51 us, that is approximately equal to “swing period” on plot 1. We also can see two humps. The second hump looks like real wave received by transducer, and it’s propagation time is ~243 us, and when speed of sound is 350 m/s, the distance is 85 mm. Nice.

But the first hump is a trouble. Why is it appeared? Since time between marks “Burst received” and “Burst started” is near to “swing period” on plot 1, and voltage levels are above the ADC measuring limits, I can suppose that the signal from driving PGA is propagated on wires to the measuring PGA. Two circuits has common HVGND wire. I’ve made modification of the circuit decoupling this wire. The new version is below.

^{simulate this circuit}

Now, when B1 burst and B2 listen, the waveform looks so strange:

The signal has pulse frequency 104 kHz. When I measure signal on transducer with oscilloscope I can see sine wave with same frequency. So, this modification is wrong.

So, is my hypothesis about signal propagation on wires right? How can I modify this circuit to make signal wouldn’t be able to propagate from one transducer to another?