Whenever an array element receives a signal that is not reflected from its intended target but generated by neighboring elements either by mechanical or electrical means ( I suppose it is a piezo - based sensor array) we can consider it crosstalk. Crosstalk can be electrical or mechanical in this case but the immediate effect is affecting measured data with spurious patterns.

To measure it you can have a simple standardized target for your array, with a well-known characteristic shape and density distribution and try to measure it to the best of the equipment's abilities. Measured shape irregularity -specifically edge blurring or other imaging artifacts that can be detected by image analysis can be attributed to crosstalk. A well defined edge or square of a certain density in a mass of another density is in this case a good target to test for crosstalk and spurious signals.

To further delve in the matter would imply using a single externally excited transducer of the same kind as those in the array- and moving it closely to it in various positions while it emits/vibrates and reading spurious data from the array that has a well known target in front of it.

Hope it helps, best of luck,

Marius

I would say for you application and taking into account two of your definitions, your'e mainly concerned how to truly measure the effect of the cross-talk on your ultrasonic array. 

Being said that, if you consider voltage and vibration as a source of cross-talk for your transducer, I might consider doing a vacuum reference test (1 element excited, the rest just measuring). During this test, at least you can exclude ( or almost exclude) the cross-talk coming from vibration, since there should not be any vibration transmitted. 

With the data coming from the vacuum test, you can repeat the same reference test in ambient known pressure and check for the difference in cross-talk, I would assume the difference is coming from the vibration cross-talk then.

I am not fully sure if this is really applicable to your transducer, but at least with this approach you can try to isolate the sources (vibration and induced voltage) creating cross-talk.

Thanks for your reply Alexandru. To be honest I had never think of this problem from imaging viewpoint, and your definition is really interesting for me.

I think the proposed solution is difficult to implement in my case. If I understand correctly you propose to excite with a single element transducer (similar as what you have in array), and receive with array very close to transmit element. For medical applications, normally the array elements are very small (in order of a few hundred micron) and making single element transducer with that size is almost impossible! On the other hand, because of the directivity of single element, many elements will be excited together. So the received signal from adjacent element is a sum of received beam and crosstalk vibration of neighbouring elements.

Thank you for your answer Dario. You have suggested a good approach to measure crosstalk. I have some comments on it. The behaviour of transducer in vacuum and in water (or any other ambient material) is different from each other. Even the electrical impedance of piezoelectric element in air is different than that when you load it with water. With doing the experiment in water you could only measure summation of crosstalk inside the transducer and vibration effect in water, but you can not clearly distinguish them from each other.

What do you think about the sources of crosstalk and which source is more dominant?  

Then perhaps if there is no return beam, the only signals to be read will be due to these unwanted phenomena, the electrical and mechanical crosstalk...Anyhow to accurately measure sensor performance is an entirely different matter from measuring the entire system performance data. A commercial system should have in-built corrections for such spurious signals and various mitigation technologies; the data used to design these comes from, most likely, testing on the bare sensor made by the producer . There is no simple way around it- if the entire system is what you have, it is very hard to discern actual sensor performance from corrected /overall/system performance without taking it apart and insulating a target element or element group and exciting them and measuring output of neighboring cells. If your installation is a prototype then you can and should do these tests anyhow (bare sensor driven by generators, measuring output etc). To measure each effect in turn you need to impose your will on the sensor controller and make it work like this: you excite one particular bunch of transceivers and you measure the output around that particular zone in concentric circles.. using elements from the same sensor.This without having any valid target in its operational range. The signals will then constitute the noise part of the business(electrically and mechanically induced). The output should be raw and not processed by the imaging device in any way because it starts adding (unknown yet) corrections to your signals.

To separate between the two components you should think of how you can mechanically excite the entire sensor at the proper frequencies and amplitudes WITHOUT using any electricity (tricky, I know but we're after purely mechanically induced effects now). Perhaps a metal beam that is impacted with a very small mass impact hammer could be calculated such that it propagates vibrations of the desired parameters (higher order harmonics could reach those frequencies). Under these conditions measure sensor output. That should be mechanically induced noise only. If you can devise such an apparatus the output should be clean and electrical noise-free. Do not give in to temptation of using a normal vibration table ( lots and lots of either coils or other piezoelectric parts , induced currents, etc ). Maybe a piezostrictive resonant assembly sufficiently far away and electrically shielded from the sensor could be a more appropriate excitation source. The output from the resonant cavity can be driven through a long tube in the vicinity of the sensor and that should insulate the electrical part enough.To easily ascertain vibration levels some optical methods should provide a contact-free approach (perhaps interferometers?).

hope it helps ,

Marius..

Dear Maysam,

I strongly agree with your aiming to cross-talk in measurable quantity. Having the measured quantity of cross-talk it could be progress in definition of quality of array transducer. If there no measurable quantity then the "cross-talk" is only wording. 

I would support your 1) and 2) definitions. But during measurement, in my opinion, array transducer must be surrounded by working conditions. Working conditions which are in practical use of array transducer (water, soft tissues). I guess, "cross-talk" is acting as source of artefacts during imaging. The pulses arriving trough "cross-talk" channel could be separated from pulse-echo pulses using time window. In other words: the dimensions of water bath in which array transducer is submerged, must be longer that longest separation of "cross-talking" elements in array transducer.