Dear Vasuki,

what you describe is the time-of-flight measurement. This has nothing to do with doppler as mentioned in this forum. You have to shot an ultrasonic signal against and with the flow by measuring the time of flights t_upstream and t_downstream. The difference of these times  - called delta t - is proportional to the flow velocity. That seems to be your setup.

As I understand you use clamp-on technique for not getting the PZT`s in contact with the fluid? What`s your problem with the setup? If you use ultrasonic pulses or e.g. coded signals (for correlation technique) you should be able to calculate the time of flights. BUT - of course - if your liquid doesn`t move (zero flow) - you`ll get times with the same value.

Or do I have to understand you that way, that you see ultrasonic pulses on an oscilloscope at the receiver side in the same moment when you transmit a signal? Then you just see cross coupled signals. Your real incoming signal should have a delay regarding to the distance of your transducer and the velocity of sound in your medium.

Regards,

Marc

Did you try to transmit sinusoidal burst (few cycles) to see directly the time of flight between the signal transmitted and received? I'm not sure what is the dimension of the whole system but probably the time difference you are looking for could be in the order of micro-seconds or less. Did you perform any preliminary calculation about the time difference?

Probably, you do not have any propagation in the liquid but in the thick wall of your tube only. You shoud ensure, that signal you have comes from the luquid. As Salvatore Pullano said, the estimation of arrival time could be of help.

Do you look for the Doppler effect? In this case water is not the medium to use because there are no or little reflecting particles.

hi

the doppler technique has been described by an austrian physician

you must have some refelectors

why u think this is best technique

flash mri uses similar techniques

for us you get quantitative measures

the doppler formula is very important ok

please respond

ferdinand

Your description of the problem is to short to understand your question and underlying technique. Using Ultrasound the Doppler technique ist  most pupular for measurement of flow velocities in tubes and vessels. Therfore my answer.

If your water has small particles or microbubbles you can use Doppler ultrasound. The transmitter (frequency ft) and received (frequency fr) are mounted close to each other and 'interrogate' the vessel at an angle theta. There will be a Doppler shift in the frequency df=ft-fr = 2*v*ft*cos(theta)/c where c is the velocity of sound in the liquid and v is the velocity you want to measure. if you do a spectral analysis of df you will have a histogram of the df (as different parts of the liquid will have different velocities) corresponding to a histogram of the velocities. Read more at EMITEL: http://www.emitel2.eu/emitwwwsql/encyclopedia.aspx

Not sure of your application. My experience is only in vascular and cardiac ultrasound. I agree with Fernando, that Tx and Rx need to be adjacent, to measure Doppler frequency shift, and the vector of the Doppler signal needs to be as parallel to the flow as possible to reduce the effect of cos(theta) on the calculation of c. If Tx and Rx are directed across the tube, i.e. perpendicular to flow, cos (90) = 0, so there will be no measured velocity at all. If your water doesn't contain evenly distributed reflectors, the only signal you'll get is from the tube. Also, if there is any distance between the crystal and the steel within the wall of the tube, your major reflector will be the tube immediately adjacent to the crystal. In both cases, there will be no phase or frequency shift as the reflectors are stationary relative to both Tx and Rx.

Just a correction on what Michael Gilbert wrote: There WILL be a frequency shift as the particles or microbubbles (scatterers) are moving in relation to to the transducers.

Dear Vasuki,

what you describe is the time-of-flight measurement. This has nothing to do with doppler as mentioned in this forum. You have to shot an ultrasonic signal against and with the flow by measuring the time of flights t_upstream and t_downstream. The difference of these times  - called delta t - is proportional to the flow velocity. That seems to be your setup.

As I understand you use clamp-on technique for not getting the PZT`s in contact with the fluid? What`s your problem with the setup? If you use ultrasonic pulses or e.g. coded signals (for correlation technique) you should be able to calculate the time of flights. BUT - of course - if your liquid doesn`t move (zero flow) - you`ll get times with the same value.

Or do I have to understand you that way, that you see ultrasonic pulses on an oscilloscope at the receiver side in the same moment when you transmit a signal? Then you just see cross coupled signals. Your real incoming signal should have a delay regarding to the distance of your transducer and the velocity of sound in your medium.

Regards,

Marc

Thanks for the reply,

Michael Gerhard von Reutern and Ferdinand Frauscher I am doing it for homogeneous liquids 

Michael Gilbert, The walls are thick, What about absorption and reflection of Ultrasounds by steel? Does acoustic impedance come into picture?

thanks Marc Schmieger  and Fernando Soares Schlindwein, you are right about the setup, am getting no delta-t (Tx and Rx are same) for stationary and water flowing case. I havent used CODED SIGNALS and will surely implement it.

I am using CRO to see the output. Could you elaborate on cross coupling 

One of the problem i see in your system is that the wall are really thick and the signal travelling the tube is attenuated, then...due to the difference in acoustic impedance between the steel and water (more than 1 order in difference) there is a strong signal reflection.

The first question is, are you receiving an echo signal due to the ultrasonic signal transmitted or is some noise or disturb?

Once you verified the received signal is really the one you are looking for try to increase the level of the transmitting signal. the Time of flight as you wrote is in the order of few microseconds.

Cross coupling can occur at  t=0 coming from transmit signal because of higher transmit voltage. The echo or incoming signal shouldn`t lead to significant  crosscoupling because of the low voltage (few mV).

To ensure that your received signal comes from the liquid and NOT from the wall just flush the pipe. If your signal remains constant - congratulation - you have a signal coming from the wall.

:-)

Some time ago we had the main problem with the tube walls. Using of shear wave transducers helped us a lot :) I don't really understood why, but that worked.

Keep in mind that you have to deal with refraction of your ultrasound signal travelling trough different media with different propagation speed (Snell`s law). Incident beams going under a certain angle into tube wall will result in shear and longitudinal beams with different propagation speed. So if you choose the wrong transducer (wrong incident beam) curious things can happen. Of course in water only longitudinal waves will propagate.

Thanks Marc, I will keep that in mind

Dear Vasuki,

If you are using 500 kHz then the wavelength in steel (assuming a typical speed of sound of about 6000 m/s) will be about 12 mm, which is much larger than the internal bore of your pipe containing the fluid. Hence, your signal may not be interacting with the fluid at all. Try emptying the tube so that it is full of air - if your signal was travelling through the fluid, you should now get no signal at all, as the reflection at a steel-air interface is greater than 99%. If you do still get a signal, it has travelled around the bore, in the tube wall. Your sensors are quite large (3 mm x 3mm square) so there is probably a direct path around the edges of the bore. You need to make sure that the signal you are looking at has interacted with the fluid!

Also, the time shifts you are trying to measure will be very small. If your path length in the liquid is only 4.2 mm and your fluid is water flowing at 10 m/s, then the time shift will be of the order of 40 nanoseconds.

Best wishes,

Bill

@William: exactly what I proposed but with an good explanation.

Several good answers here.  For a minimal TOF measurement in a fluid you need to have several (more than 2-3) wavelengths  of travel in the fluid, though with correlation you can improve this at the expense of more complex signal processing.  if the tube bore is too small to permit this, then you must position the transducers such that beam reflections are used between the transmitter and receiver, i.e. single reflection V path, 3 reflection W path.   The transducer design is really important in all this.   The main points are

1. ensure a single mode (mentioned above) is transmitted into the pipe wall.   When an acoustic pulse encounters an interface between two materials of different acoustic impedance (plastic vs steel for example) refraction occurs which bends the path according to Snells law.  What also happens is that two or more vibrational modes are generated.   Most commonly a longitudinal wave (vibration along axis of propagation) will create a Shear wave (vibration perp to axis of propagation).  these wave travel at different sound velocities and can arrive at your receiver in overlapping fashion, corrupting the measurement.   With knowledge of the material properties of the pipe and transducer you can select a transducer angle that will produce only a shear wave.  this works by using a steep enough angle to exceed the critical angle of the LONG wave.   there is a calculator for this here:   http://alturl.com/qpsw9

2.  You also need to ensure sufficient energy is coupled into the pipe wall.  This means the acoustic impedance of the transducer facing material (matching layers or delay line) has a sound speed very close to the pipe material.  if not, then most of the energy will be reflected instead of transmitted at both the receive and transmit sides of the system.  

3.  You want a narrow unfocused beam so that the energy goes where you want.  a small crystal is a bad choice.  A point source is completely omnidirectional while a large surface is completely directional.  so you need to use the largest crustal that is practical for the geometries you are working with.   It is not uncommon to see 3-4" diameter transducers used in large pipeline applications for example.

4.  If you are doing time domain processing, a simple envelope detector (precision rectifier followed by an integrator) can be fed into a PLL along with the transmit (reference) signal to generate the TOF error voltage.  from there it is trivial to measure the voltage and convert it to flow.

If you are attempting to build a general purpose instrument that can handle multiple pipe sizes, fluid types, etc the complexity goes up from here.   These are just the basics required.

Thanks  Gregg Pulley for that brilliant suggestions!!

Thank you all for your guidance. 

I could implement the Ultrasonic flow meter using PZT crystals. Please feel free to ask me, and I will be privileged to share all I know