0
\$\begingroup\$

I have the following setup:

2 piezo transducers are placed 0.7 meter apart from eachother in water facing eachother. 1 piezo (PA) acts as the sender, the other piezo (PB) acts as the receiver.

PA sends periodically a pulse that the receiver picks up. The pulse gets generated by a microcontroller switching a MOSFET according to the following circuit.

send circuit

With the send pulse:

sendpulse

PB receives the pulse and is connected to the following amplifier circuit.

receiver circuit

With the amplified receiving trace:

receivetrace

The two circuits are only connected to eachother through earth (microcontroller + send-circuit are supplied by a 230V AC to 5V DC converter, the receive-circuit is supplied by 2 benchsupplies).

Extra logic is added so I can detect if the pulse arrives or not. This works, however I seem to get coupling on my scope image from PA onto PB. For now I filter this away in my code but I rather fight the cause instead of the symptoms. As you can see, the send pulse also is visible on the receive trace (the left most pulse). Both traces are captured with the same scope, though at different times and with only 1 probe connected.

Is there a way to surpress this signal? Later on, the setup will consists of more piezo's as well as both sending/receiving capabilities and I rather keep crosstalk (?) to a minimum.

EDIT:

Following the questions in the comments:

1) Are you sure that the coupling is electrical and not, say acoustic ? => Yeah I'm pretty sure that it's electrical. If I connect both probes the left most receiving pulse is at the same time as the send pulse. If you take the time difference between both pulses (around 476us) and the speed of sound in water (1480 m/s), you can calculate the distance between the 2 piezos x = 1480 * 476e-6 = 70cm

sendreceive (upper is receiving, lower is sending)

2) Why is the send a pulse and not a sinusoid ? => I'm still tweaking the send pulse circuit. I send a single pulse as the setup is to detect if something (fish) goes through the circle (only 2 piezos at the moment at the diameter). Hence receiving a pulse is nothing in between, not receiving a pulse is something in between. The setup is NOT for transmitting data.

3) Is the output of the micro controller on the send side a single pulse or a a sequence of pulses tuned to the resonant frequency of the piezo transducer ? => It's a pulse of 40us at the moment.The pulse gets created on the down slope.

pulse

4) Are they immersed in water ? => Yeah they are immersed in water. Not deep at the moment but later on this will be in the meters.

setup (Left is receiving, right is sending)

5) Are the leads to the transducers well insulated from the water ? => No! The leads are not shielded. So maybe some electromagnetical interference? I also think they are not twisted, it has a housing around it. I will pry one open to see further. Also the water is rainy water, with apparently some algaea too.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ 1) Are you sure that the coupling is electrical and not, say acoustic ? 2) Why is the send a pulse and not a sinusoid ? There seems to be very few cycles in the send pulse. 3) Is the output of the micro controller on the send side a single pulse or a a sequence of pulses tuned to the resonant frequency of the piezo transducer ? \$\endgroup\$
    – AJN
    Apr 13, 2021 at 12:20
  • 2
    \$\begingroup\$ 4) Are they immersed in water ? Can we see a photograph of the setup ? Perhaps the first pulse is the real pulse and the second pulse is a reflected pulse from one of the water tank walls or even the top surface of the water ? 5) Are the leads to the transducers well insulated from the water ? Perhaps the water is conducting the current. \$\endgroup\$
    – AJN
    Apr 13, 2021 at 12:23
  • 1
    \$\begingroup\$ How deep are the transducers in the water? If the wire insulation is in direct contact with the water, there will be capacitative coupling between the wires and the water. Depending on conditions, the transmitter wires may be coupled to the receiver wires. Is the water distilled? Sea water? Chlorinated? If you are able to experiment, does changing the conductivity of the water affect the strength of the unwanted signal? Are the lead wires twisted pair? \$\endgroup\$ Apr 13, 2021 at 12:47
  • \$\begingroup\$ @AJN I've updated my question with answers to your questions \$\endgroup\$
    – Swedgin
    Apr 13, 2021 at 13:16
  • \$\begingroup\$ @MathKeepsMeBusy I can't play with the conductivity of the water. For now it's a simple setup in rainy water. Later it should work in the sea. \$\endgroup\$
    – Swedgin
    Apr 13, 2021 at 13:17

2 Answers 2

1
\$\begingroup\$

Addressing possible electromagnetic radiation from the transmitter:

Assuming that the received, unwanted pulse is coincident with the transmitted pulse, coupling would be speed-of-light electromagnetic coupling. Suppressing this radiation involves reducing the area of transient current in the MOSfet switch. An isolating inductor has been added to block AC current path from influencing +5V line:

schematic

simulate this circuit – Schematic created using CircuitLab
A pull-up resistor is added for a different reason. Piezo internal parallel capacitance should be discharged between pulses, so that anode of D1 rises back up to +5V after the MOSfet becomes idle (open-circuit).


Note that taking these precautions may not solve the problem of coupling between transmitter and receiver circuits. The nasty path may be through ground, or too-open a layout of receiver circuitry. Too little information of OP's setup to address this - a difficult one to deal with.

\$\endgroup\$
3
  • \$\begingroup\$ Hey, this can help as I think now it's indeed possible due to electromagnetic interference as nothing is shielded as of now. The guy here who made the transducer already added a resistor in parallel with the piezo itself, this is also for damping the pulse to not let it resonate too long. (I also updated my question following some questions in the comments) \$\endgroup\$
    – Swedgin
    Apr 13, 2021 at 13:27
  • \$\begingroup\$ 100uH, is that experience or calculated? \$\endgroup\$
    – Swedgin
    Apr 13, 2021 at 13:28
  • 1
    \$\begingroup\$ @Swedgin Its a guess - a reasonably high impedance at piezo resonant frequency. \$\endgroup\$
    – glen_geek
    Apr 13, 2021 at 13:32
1
\$\begingroup\$

You will have to find the mode of ingress by experiments.

  1. surface waves faster than water medium waves.
  2. ground shift wire resonance.
  3. ground conduction crosstalk (shared currents)
  4. radiated differential loop crosstalk
  5. radiated common mode crosstalk

example:

a) If not already, use STP wire with common shield ground at the uC for send & receive circuits. so they don't share ground current but rather define a 0V reference closest to Agnd for ADC and unterminated shield at remote ends.

b) Test for radiated crosstalk by orthogonal wires outside the water and expect the least coupling when orthogonal then vary by insertion to tank. Also Check for surface waves here just below surface.

c) Suppress CM noise with hi-mu ferrite toroidal or C-clamp CM chokes to absorb CM fields.

d) Ensure Digital ground is separate from Analog ground for signal measurements so that digital current does not pass thru A gnd path.

d) The incident wave voltage appears to be 1.5 MHz so a burst pulse of more 1.5MHz cycles may be better , depending on Q of the Piezo to ramp up more energy than a simply a ~ 2.2us square pulse. Energize Piezo then short quickly with the minimum cycles necessary but at the series resonant frequency.

Then report results to negate each potential interference or ingress above in your question.

\$\endgroup\$
5
  • \$\begingroup\$ Hey, thanks for your write-up. Gives me some hold-on for stuff to check and further reading. I will start testing. Thing is that now it's still in the early develop stage and I'm using perf boards and an existing board we have with an atmega64m1 to use the analog comparator for now. It's to have a simple circuit and a working example. Though after reading the comments and answers I guess I'll start with a prototype PCB as soon as possible and keep digital, analogue, receive, send part seperated. \$\endgroup\$
    – Swedgin
    Apr 13, 2021 at 13:23
  • \$\begingroup\$ Proto board is ok if you use STP or cat 5 and be careful about gnds and decoupling Vcc. Keep small and foil gnd plane if nec., maybe. \$\endgroup\$ Apr 13, 2021 at 14:18
  • \$\begingroup\$ I have talked to the maker of the transducers and we are going to replace the cables tomorrow. I will report back. \$\endgroup\$
    – Swedgin
    Apr 13, 2021 at 14:25
  • \$\begingroup\$ Have you thought about the pulse burst pattern I suggested? e.g. gate 3 ~5 pulses at 1.5 MHz or fo with critical damping or sinX/X with suitable half bridge \$\endgroup\$ Apr 13, 2021 at 15:15
  • \$\begingroup\$ I wonder if the fish will mind. \$\endgroup\$ Apr 13, 2021 at 15:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.