k-Wave User Forum » Topic: frequency shift @ second transducer

k-Wave User Forum » Topic: frequency shift @ second transducer http://www.k-wave.org/forum/topic/frequency-shift-second-transducer Support for the k-Wave MATLAB toolbox en-US Wed, 13 May 2026 03:45:26 +0000 http://bbpress.org/?v=1.0.2 <![CDATA[Search]]> q http://www.k-wave.org/forum/search.php Vali on "frequency shift @ second transducer" http://www.k-wave.org/forum/topic/frequency-shift-second-transducer#post-8062 Thu, 11 Feb 2021 09:34:16 +0000 Vali 8062@http://www.k-wave.org/forum/ Thanks so much for your time! You have totally right, I had forgotten to cut off the input-signals. In the previous example the frequency spectrums of both sides are equal now and centered around the main (center) frequency. But if I increase the total number of elements for Tx and Rx (e.g. 128) both spectrums are getting a downshift of approx. 1 MHz in center (from 5.2 to 4.2 MHz). Does this phenomen occur in case of frequency dependent attenuation with the depth? A correct TGC should compensate the frequency shift or? (I have tried it with the TGC settings from the B-mode example, but there is no much difference in-between.) >Sensor 2 has a lower amplitude because of the directional response of the finite sized sensor which you're averaging over. --> The sensor elements do have the same size in horizontal and vertical direction. By default the sensor elements have a omnidirectional sensitivity. I don't understand the difference between the amplitudes. I have also tried to align the sensitivity direction. Would you be so kind to explain that more in detail? Best, Valentin Bradley Treeby on "frequency shift @ second transducer" http://www.k-wave.org/forum/topic/frequency-shift-second-transducer#post-8049 Sat, 06 Feb 2021 14:46:59 +0000 Bradley Treeby 8049@http://www.k-wave.org/forum/ You should trim the transmitted signal from the sensor data so it just contains the received signals. For example, set: <pre><code>% remove the transmitted signal sensor_data_trim = sensor_data1(:, 400:end);</code></pre> Then if you plot the two spectra, they cover essentially the same bandwidth. For example: <pre><code>% amplitude spectrum of average signal data testChannel1 = mean(sensor_data_trim(1:16,:)); testChannel2 = mean(sensor_data_trim(17:32,:)); [f, as1] = spect(testChannel1, 1/kgrid.dt); [~, as2] = spect(testChannel2, 1/kgrid.dt); % plot figure; plot(f * 1e-6, as1); hold on; plot(f * 1e-6, as2); legend('Sensor 1', 'Sensor 2');</code></pre> Sensor 2 has a lower amplitude because of the directional response of the finite sized sensor which you're averaging over. Hope that helps, Brad Vali on "frequency shift @ second transducer" http://www.k-wave.org/forum/topic/frequency-shift-second-transducer#post-8039 Wed, 03 Feb 2021 10:29:08 +0000 Vali 8039@http://www.k-wave.org/forum/ <pre><code>%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %frequency-shift % %description: %Transmitting a plane wave from the horizontal elements and receive it %with the vertical elements (RX2). Where does this channel dependent shift %come from? (In compare with the spectrum of the horizontal channels, %where no shift occurs.) %note: With an increasing number of channels e.g. 192 the frequency-shift %will also increase. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% clear all; close all; %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %ground-settings %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% GPU = false; %if GPU is available meanSpeedOfSound = 1540; %[m/s] transducerFreq = 5.2083; %[MHz] wavelength = meanSpeedOfSound/(transducerFreq*1e6); %[m] %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %define kgrid %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% PML_size = 20; PMLInside = 'false'; %PML is set to 'outside the grid' dxy = (wavelength)/4; Ny = 243 - 2 * PML_size; %checkFactors() ... for computational-time Nx = 243 - 2 * PML_size; Nx = 243 - 2 * PML_size; Ny = 243 - 2 * PML_size; kgrid = kWaveGrid(Nx, dxy, Ny, dxy); %make grid of given size %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %define medium with a scatter %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% c_tissue = 1540; %[m/s] density_tissue = 1000; %[kg/m^3] medium.sound_speed_ref = 1540; medium.alpha_mode = 'no_dispersion'; medium.sound_speed = c_tissue * ones(Nx, Ny); medium.alpha_coeff = 0.54 * ones(Nx, Ny); %[dB/(MHz^y cm)] ... soft tissue medium.alpha_power = 1.5; %= y medium.density = density_tissue * ones(Nx, Ny); %[kg/m^3] inclusion.dimension = [Nx/2 Ny/2 10]; disc_c1 = round([inclusion.dimension(1,1) inclusion.dimension(1,2) inclusion.dimension(1,3)]); disc1 = logical(makeDisc(Nx, Ny, disc_c1(1), disc_c1(2), disc_c1(3))); numbOfPix1 = length(medium.sound_speed(disc1)); lowerLimDensity = density_tissue / 1.25; upperLimDensity = density_tissue * 1.25; randDensity = zeros(numbOfPix1,1); randDensity = (upperLimDensity - lowerLimDensity).*rand(numbOfPix1,1) + lowerLimDensity; medium.density(disc1) = randDensity; lowerLimSos = c_tissue / 1.25; upperLimSos = c_tissue * 1.25; randSos = zeros(numbOfPix1,1); randSos = (upperLimSos - lowerLimSos).*rand(numbOfPix1,1) + lowerLimSos; medium.sound_speed(disc1) = randSos; figure(101); subplot(1,4,1); imagesc(medium.density); title('medium-density'); subplot(1,4,2); imagesc(medium.sound_speed); title('medium-SoS'); %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %define input-signal %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% height = dxy*Nx; max_time = 2*height/max(medium.sound_speed(:)); kgrid.t_array = makeTime(kgrid, medium.sound_speed, 0.3, max_time); sample_freq = 1/kgrid.dt; signal_freq = transducerFreq*1e6; num_cycles = 3; toneBurst_envelope = 'Gaussian'; bandwidth = 80; %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %define source (16 elements) and sensor (16 elements each): %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% halfDis = round(Ny/2)-8; source.p_mask = zeros(Nx, Ny); source.p_mask(20,halfDis:halfDis+15) = 1; sensor.mask = zeros(Nx, Ny); %1st RX sensor.mask(20,halfDis:halfDis+15) = 1; %1nd RX sensor.mask(30:45,halfDis+30) = 1; subplot(1,4,3); imagesc(source.p_mask); title('source-mask'); subplot(1,4,4); imagesc(sensor.mask); title('sensor-mask'); truesize; %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %tone-burst-offset %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% phi = 0; toneBurst_offset = zeros(nnz(source.p_mask(:)),1); %time-delay for elements toneBurst_offset = sind(phi)*[0:numel(toneBurst_offset)-1]'*dxy/meanSpeedOfSound/kgrid.dt; input_signal = toneBurst(sample_freq, signal_freq, num_cycles,'SignalOffset', toneBurst_offset, 'Envelope', toneBurst_envelope); source.p = input_signal; %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %run simulation %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% input_args = {'PMLInside', false, 'PlotPML', false, 'DisplayMask', source.p_mask, 'PlotScale', [-0.25, 0.25]}; if GPU, input_args = {input_args{:},'DataCast','gpuArray-single'}; end sensor_data1 = kspaceFirstOrder2D(kgrid, medium, source, sensor, input_args{:}); if GPU, sensor_data1 = gather(sensor_data1); end %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %channel selection, frequency-shift at RX2 (vertical elements) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% ch1_RX1 = sensor_data1(1,:); bandwidth = 80; filteredSigRX1 = gaussianFilter(ch1_RX1, sample_freq, signal_freq, bandwidth, true); title('channel 1, RX1-horizontal'); ch1_RX2 = sensor_data1(32,:); bandwidth = 80; filteredSigRX2 = gaussianFilter(ch1_RX2, sample_freq, signal_freq, bandwidth, true); title('channel 1, RX2-vertical'); ch8_RX2 = sensor_data1(28,:); bandwidth = 80; filteredSigRX2 = gaussianFilter(ch8_RX2, sample_freq, signal_freq, bandwidth, true); title('channel 8, RX2-vertical'); ch12_RX2 = sensor_data1(24,:); bandwidth = 80; filteredSigRX2 = gaussianFilter(ch12_RX2, sample_freq, signal_freq, bandwidth, true); title('channel 12, RX2-vertical'); ch16_RX2 = sensor_data1(17,:); bandwidth = 80; filteredSigRX2 = gaussianFilter(ch16_RX2, sample_freq, signal_freq, bandwidth, true); title('channel 16, RX2-vertical'); %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %FFT-spectrum, mean value of RX1/RX2 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% testChannel = mean(sensor_data1(1:16,:)); nfft = length(testChannel); ff = fft(testChannel,nfft); fff = ff(1:nfft/2); fs = 1/kgrid.dt; fff=fff/max(fff); fn = fs/2; df=fs/nfft; xfft = 0:df:fn-1; figure(1001); subplot(1,2,1); plot(xfft,abs(fff)); xlabel('frequency/[Hz]'); title('FFT mean value of RX1, normalized to 1'); testChannel = mean(sensor_data1(17:32,:)); nfft = length(testChannel); ff = fft(testChannel,nfft); fff = ff(1:nfft/2); fs = 1/kgrid.dt; fff=fff/max(fff); fn = fs/2; df=fs/nfft; xfft = 0:df:fn-1; subplot(1,2,2); plot(xfft,abs(fff)); xlabel('frequency/[Hz]'); title('FFT mean value of RX2, normalized to 1');</code></pre> Bradley Treeby on "frequency shift @ second transducer" http://www.k-wave.org/forum/topic/frequency-shift-second-transducer#post-8010 Tue, 19 Jan 2021 20:05:49 +0000 Bradley Treeby 8010@http://www.k-wave.org/forum/ HI Valentin, Sounds puzzling. Can you post your code? Brad. Vali on "frequency shift @ second transducer" http://www.k-wave.org/forum/topic/frequency-shift-second-transducer#post-7982 Mon, 14 Dec 2020 10:06:35 +0000 Vali 7982@http://www.k-wave.org/forum/ Hi, I've defined two linear transducers (2D) for plane-wave propagation. The first one is defined in the horizontal and the second one in vertical direction. If I transmit a plane-wave with the horizontal transducer and receive the signals with both, then I have a large frequency-shift in the spectrum. Input-Signal: Gaussian-envelope, 3 cycles and centered at 5.2MHz (max supported frequency of the grid ~8.3MHz) Spectrum of horizontal transducer: Gaussian-curve centered at ~5.2MHz Spectrum of vertical transducer: Gausian-curve shifted to 1.5 - 2.5MHz (channel dependent) Where does this large shift come from? Any suggestion? Thanks, Valentin