Vectrino+ noise levels

Hi

I'm using the Vectrino+ with the standard head. I'm trying correct the measurement data to compute Reynolds stresses. The method I'm using assumes that, and only works, if the noise level is the same for all velocity components. However, my measurements show that the noise levels aren't the same at all. The Z1 and Z2 components have a noise level that is one order of magnitude lower than the X and Y components.

Why are the noise levels different? (see attached figure) Is there a post processing of the Z1 and Z2 components?

Or:

According to the manual "USER GUIDE" October 2004, rev. C. p. 18

"This means that the Vectrino is more sensitive to the Z-velocity (the component

parallel to the transmit beam) than it is to the X- or Y-velocity. Consequently, the

Z-velocity component yields a lower measurement uncertainty."

Is it beacuse of the geometric configuration that the noise level for the Z-velocities are lower? In that case, what I understand is that I cannot use a correction method which assumes an equal noise level for all velocity components.

Regards,

Stig Grafsrønningen

Hi Stig,

Regardless of manufacturer, all acoustic velocimeters equipped a standard head like the Vectrino you are using have lower noise levels for the Z component of velocity because of the head geometry. This is because the Z component is aligned closely with the bistatic axis of the transmitter-receiver pair and is more directly measured than the X or Y components.

The Vectrino also differs from other velocimeters by having a separate channel for each receiver rather than multiplexed sampling where the receivers would share electronics. This is what allows it to sample up to 4x faster than other velocimeters at 200 Hz.

This means each receiver will have different (but very similar) noise levels. Noise levels for each receiver are reported in the header for a data file. Typical levels are 10-12 dB. Check the section called "Velocity Header" for this information.

Finally, the noise for the velocity components can be regarded as white noise (equal energy at all frequencies) and uncorrelated with the noise from another velocity component. Reynolds stress calculations (or any other covariance quantity) should be fairly robust to noise contamination after averaging because of this.

P.J.

Hi

Thanks for the clarification! I guess what I meant is that I need the noise level for each beam (and not velocity) to be the same. A transformation from XYZ to beam coordinates and a FFT should show this.

The Reynols stress calculations should come out fairly easy when taking advantage of the fact that the noise is uncorrelated. I computed the normal stress in the Z-direction (using both signals for Z) for a laminar flow, which should be 0. The variance for these signals were about 1E-6. The computed normal stress was about 1E-18, which translates to machine precision.

However, I've encountered another problem that I hope you could help me with. I have the feeling that I'm doing something wrong when transforming the velocities from XYZ to beam velocities using the transformation matrix. What I've done so far is:

beam_velocities = inv(T)*U(X,Y,Z1,Z2)-->

U[beam1,beam2,beam3,beam4]=inv[1.9949         0   -1.9963         0; 0    1.9741         0   -1.9666;0.5190         0    0.5134         0; 0    0.5022         0    0.5315]*[U,V,W1,W2].  where T is the 4x4 Transformation matrix.

By doing this I get out velocitis that look good,but which velocity is for which beam? I hope you can clarify this for me.

I also tried to compute the vector for each beam in order to find out where the data belonged. However, I must be doing something wrong. It doesn't ad up. By taking:

inv(T)*[1 0 0 0 ; 0 1 0 0; 0 0 1; 0 0 1 0] one migh(?t!??) get vectors pointing from the sampling volume perpendicular on to the beams. Nowing which beam is in the x and y direction, this should give me enough information to find the out which velocity that belongs to which beam. The result from this for my transformation matrix was:  [0.2492         0    0.9690;      0    0.2609    0.9655; -0.2519         0    0.9683;      0   -0.2466    0.9692 ] --> the length of each of these vectors is 1.0 which is a good sign.

Taking the first vector: [0.2492         0    0.9690]  This vector should point in the positive X direction and in the Z direction. (i.e. I know which beam it is)

However, this is were things doesn't ad up. The angle between the Z axis and this vector should be 30° (according to the manual), this is not through because theta=acos(0.9690)= 14.3°. Which is approx. half of the 30° it should be.

Got any tips? I have really no idea.

Regards,

Stig Grafsrønningen