This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

MSP430FR6043: Repeatability issue regarding DToF value

Part Number: MSP430FR6043
Other Parts Discussed in Thread: EVM430-FR6043

Hello,

We're currently trying to measure mass flow of a gas inside a pipe using EVM430-FR6043 and 200kHz transducers (USS_SW_Lib_v_02_30_00_03)

In steady state (means constant flow), we're currently facing repeatability issues. Most of the time DToF and mass flow are expected ones (corresponding to real mass flow) but sometimes we have spikes on DToF values.

Note : this issue is also present without any flow..

please find below screenshots with 2 different DToF measured in exactly the same configuration.

Correct DToF :

"Wrong" DToF (spike) :

Thanks and regards,

Guillaume

  • Hi Guillaume,

    I see your numbers...

    Please make a screenshot of the ADC capture, and screen shots of Configuration.Parameters and Configuration.Advanced_Parameters as displayed by the GUI. This allows me to see what could be the reason for that.

    have a nice day

         Johann

  • Hi Johann,

    Please find ADC_Capture screenshot, parameters and advanced parameters screenhsots below.

    Thanks for your help.

    Note : we're using 200kHz transducers (and we made a sweep to confirm best operating frequencies for our transducers)

    Regards

    Guillaume

  • Hi,

    as you can see from the ADC captures the beginning of the Ping is missing.

    ...a typical ADC capture has a quiet phase before the oscillations start.

    Now this could mean that you have either a superposition of the acoustic axis of body and medium, or you've got the wrong timing.

    Just compare with your theoretical calculated TOF to find out what case it is.

    When I talk about body noise I mean the shortcuts the signals could make from one transducer to the other without going through the medium.

    In the above case this would be the dashed line inside the pipe wall. It is essential to avoid those shortcuts since acoustic signals of the different paths superimpose, and cannot be separated anymore.

    Johann

  • Hi,

    as you can see from the ADC captures the beginning of the Ping is missing.

    ...a typical ADC capture has a quiet phase before the oscillations start.

    Now this could mean that you have either a superposition of the acoustic axis of body and medium, or you've got the wrong timing.

    Just compare with your theoretical calculated TOF to find out what case it is.

    When I talk about body noise I mean the shortcuts the signals could make from one transducer to the other without going through the medium.

    In the above case this would be the dashed line inside the pipe wall. It is essential to avoid those shortcuts since acoustic signals of the different paths superimpose, and cannot be separated anymore.

    Johann

  • Hi,

    Thanks for your reply.

    Regarding propagation of acoustic wave inside pipe wall, this should lead to smaller dToF, correct ? (path is shorten)

    In our case we have the correct dToF and sometimes a bigger or a smaller value (and in worst case negatives values...) without changing anything in our setup (please see screenshots below).

    If our timing configuration is wrong (means ADC capture starts too late and we're missing beginning of Ping) we should not be able to obtain correct dToF no ?

    Thanks.

    Guillaume

  • Something went wrong with screenshots, sorry for that.

    Regards,

    Guillaume

  • Hi Guillaume,

    yes you are right, the beginning of the ping is not visible. But you can easily find out how your signal develops by shortening the "Gap between pulse start and ADC capture". At the beginning everything is supposed to be quiet. Here neither the body noise not the medium noise should have made it to the receiver. Then the body noise I expect to arrive first. If you got an certain angle you might experience that the body noise exists out of several modes. E.g. a sheer wave could also be excited in the wall. Later if those waves get superimposed they cannot be separated anymore.

    In the diagram above the red signal is the excitation on CHx output; the blue one is the picked up signal at CHy input. Her show is the case when no waves are overlapped.  Reflections of the body noise are not shown.

    …. the crosstalk from CHx_out to CHy_in cannot be recorded by the EVM; it does not interfere since it is completely gone after the excitation and has not reflections.  

    Proposal: look for the beginning of the signal to find out what is going on.

    Looking forward on your results, have a good day

       Johann

  • Hello Johann,

    I tied what you proposed (means shortening the gap between pulse start and ADC capture) but even if I reduced it at its minimum value, I was not able to able to see any "quiet period". Please see screenshot below.

    Note : I also increased capture duration to able to record full signal

    Thanks and regards.

    Guillaume

  • Hi Guilaume,

    I am trying to interprete your signal...

    - at ADC_Buffer_Index #25 the excitation is already over, and RX-bias is applied...  right after this some ringing is visible that dies relatively quickly.

      this ringing is caused as a "step response", since the bias being turned on to an resonator, the transducer.

    - after index ~#70 till ~#200  I would expect a quiet phase... ringing is gone and the ping (pulse train) has not made it to the RX transducer.

    - at ~#200 the ping arrives at the RX-side. For eight pulses the signal amplitude is rising.  The envelope function of the rising part is pretty linear; this is an indication of an high channel attenuation and typical for gas. After the 8th pulse the amplitude decays with an amplitude envelope that reminds to an discharging capacitor. This is an indication of a low termination load.

    Everything looks ok except for that high noise.

    Guillaume do you have other noise sources in your setup ?

    e.g. electrical noise as CFL-inverters, Neon-Lamps, notebook to close,...    or

    e.g. mechanical noise as ferro-electric capacitors on your board or nearby that generate ~200kHz noise

    Sometimes I use a "bat detector" to find unexpected noise sources I modified that one to go beyond 250kHz

    https://de.elv.com/elv-fledermaus-detektor-fmd1-komplettbausatz-151462

    In out lab we use the pipe enclosed in an tin can to avoid electrical and acoustical interference.

    Johann

     

  • Hi Johann,

    Thanks for your explanations.

    I do not see any obvious noise source in my set-up (except maybe the laptop I'm using to power EVM and to make ADC capture through the GUI ?).

    I do not have any neon lamps, CFL inverter or such electrical devices around me.

    Set-up is very simple : EVM430FR6043, connected to 2 200kHz transducers inserted inside a 3D printed tube.

    Regards,

    Guillaume

  • Which transducer models are you using, if I may ask ?

    Johann

  • Hi Johann,

    We prefer to avoid giving model and name of our transducers supplier (sorry for that). But it's 200kHz (effective bandwidth +/- 10kHz) ceramic transducer with a static capacitance of 2000pF and electric sensitivity >= 20mV.

    Regards,

    Guillaume

**Attention** This is a public forum