Our new Internet web-site is now operational. Please try it! You may want to add it to your browser’s list of favourites.
In the site you will find:
All back copies of Innovation News are now published on www.cambridge-en.com and all new copies will be published there too. This issue of Innovation News will be the last to be circulated fully by mail. But you can register your e-mail address and contact details with us and we will tell you by e-mail when the next issue is published on the web-site - see the middle pages for more information.
Cambridge Ultrasonics and Cambridge Software are different arms of Instrumentation Innovation Ltd. Another arm is called Rapido and it is featured on the web-site too. Rapido is responsible for:
Rapido is also a vehicle for some innovative products for the Cambridge tourist industry. It has pioneered pre-printed adhesive labels for postcards. If you are like me then you dislike writing postcard messages - most men share my opinion but most women do not. Rapido message labels are just perfect for the likes of me because you can finish-off ten postcards in little more than the time it takes to write ten addresses.
We have put some of Rapido’s products onto www.cambridge-en.com for you to read and enjoy. The humorous messages all relate to some better-known and little-known facts about Cambridge University and its history. You can download some of the messages and use them free of charge or we can supply you with a selection of the 24 different messages in any of several different languages at low cost. Also free to download is an icon walking map of Cambridge - the icons show you what you ought to see and where to eat or drink.
Former Cambridge University students with
PhDs in physics and engineering have a reputation for being good
at solving a wide range of problems outside their area of experience.
Their skills in 
mathematics
and their ability to draw upon their knowledge of science has
made them desirable for working in areas such as:
What is better than a Cambridge Science PhD? The answer is a Cambridge Science PhD with ten or twenty years of postgraduate experience in the commercial world.
Did you know that a good place to find Cambridge
Science PhDs with commercial experience is - living around Cambridge!
As well as employing Cambridge Science PhDs Cambridge Software
makes use of them through an informal association of Cambridge
graduates, some with PhDs but not all, who mostly live around
Cambridge. If you want to use Cambridge PhDs with experience as
consultants then we should be able to help you. You can telephone
directly on +44 (0)1954 231 494 or look at our web-site on http://www.cambridge-en.com
and use the menu to get to the Cambridge Network.
The list of names under the Cambridge Network is expected to expand to include skills other than those already mentioned. We expect soon to add: journalism, animation skills, art and film making, US lawyers; we already have skills in food and wine. The common thread is Cambridge excellence.
Figure caption: A crater in a copper
target created by the impact of a hardened steel sphere. The equation
was developed to model the surface of the crater and numerically
fitted to experimental data to find the volumes of craters to
2% accuracy.
Instrumentation Innovation Ltd is the wrapping for Cambridge Software, Cambridge Ultrasonics and Rapido. Together they offer a wide range of services.
Mathematics and software
Hardware development
Ultrasonic engineering
Promotion and marketing serices
Cambridge Ultrasonics and the Institut für Massivbau of the Technical University of Darmstadt and other partners will apply to the European Commission for support for a project to develop a second generation of ultrasonic inspection equipment for concrete, steel or wooden structures. Potential partners are still sought that are either owners of concrete or steel or wooden structures or inspection service providers.
For details of the projects offered see
www.cambridge-en.com and use the menu to navigate to the Collaboration
section. There are three projects described but there will probably
be only one application to the European Commission.
Several organisations have already stated that they want to join the consortium, including industrial concerns and universities from several countries around Europe. We would welcome other industrial partners with suitable skills and facilities to join the consortium.
There have been unexpected delays in the announcement of the call for proposals in the Framework V round. But the deadline for proposals is expected to be in May 1999, with projects starting later in the year.
The project is known as the second generation project because it will build upon the results of two previous projects in which Cambridge Ultrasonics and the Institut für Massivbau have taken leading roles. The best aspects of each of these successful projects will be developed to a stage where prototypes are ready for commercialization as instruments and as services in the hands of inspection service providers. The instruments resulting from the project should find application in a wide range of structures including: offshore oil production, nuclear power stations, dams, bridges, roads, telegraph poles, cat cracker linings and liquid petroleum gas confinement slabs.
Please send an e-mail as soon as possilbe
to 
or to with your
details and your reason for interest.
Please send your e-mail address to and we will notify
you by e-mail when new copies of Innovation News are posted on
our web-site. You will also be informed of any other news, such
as collaboration opportunities, new services and products.
We would be grateful if you would also kindly send us your latest contact details so that we can keep our records up to date.
The essential information we need is:
We would also like to have the following information:
Our database records are regularly cleaned to remove old or inactive contacts. We will be cleaning it again in April. Cleaning makes use of the last time you contacted us. If you want us to continue to keep you informed of our activities and services then please send us an e-mail with your contact details.
Our database is registered with the Registrar for Data Protection, which is compulsory in Great Britain. British laws are amongst the most stringent in the World now. Roughly stated we follow these principles with our database:
In the last issue of Hot Hints we dealt with Phase Disrupted Wavefronts (PDW) and how they can cause loss of signal in ultrasonic inspection systems. The essence of the article was that the ultrasonic wave must be spatially matched to the receiving transducer. In this article we consider how to do matching of the time variation of wavefronts instead of their spatial variation.
Just as there is something to be gained from spatially matching the receiver profile to the received wave so there is something to be gained in matching the signal in the time domain. There are two things to consider:
Very often the same transducer is used for both transmitting and receiving, guaranteeing matching in the time domain. In some cases where harmonic or sub-harmonic generation in the propagating medium is to be detected, for example bubbles being excited in water, it is not desirable for the transmitter and receiver to have the same centre frequency or bandwidth. In this case the centre frequency depends upon what harmonic is to be detected - if it is the third harmonic, for example, then the receiver should have a centre frequency three times the centre frequency of the transmitter.
Figure 1 caption:
A linear sweep frequency chirp signal is synthesized for transmission in an ultrasonic test. Note the variation in period as the frequency sweeps from a low frequency to a high frequency.
A transmitter responds to a sharp electrical
spike with its impulse response - usually a small packet of waves
not a sharp spike. In simple ultrasonic systems the receiving
equipment is like an oscilloscope with a rectified signal display.
The operator learns to recognise the impulse response of each
transducer and learns to ignore random noise. She/he will also
focus on certain pulses on the screen and ignore others. The operator
is performing some sophisticated signal processing but it is possible
to use computers to perform some of the processing more effectively,
leaving the operator to concentrate on the final interpretation.
Matched filtering is a process known to be the optimum linear filter for detecting a known signal in random noise. The computer responsible for the processing of signals has a copy of the signal it should detect. It time reverses the signal for detection and cross-correlates it with the received signal. You can think of it as a form of pattern recognition.
Figure 2 caption:
The impulse response of a transducer.
The figures show some Matlab examples of a chirp and matched filtering. The copy of this newsletter posted on our web-site (www.cambridge-en.com) has a more extensive range of signals. If you find this subject interesting then look at the web-site.
It is possible to continue to drive the
transmitter with a spike but the matched filter should use the
convolution
of the spike with the impulse-responses of the transmitter and
receiver as the signal fragment to reverse in time. This sounds
complicated but in practice it usually only requires a reference
test to provide this signal.
Figure 3 caption:
An ideal delay impulse at 1000 time steps. It is the objective of the ultrasonic system to detect this pulse and to measure its position accurately.
It’s possible to use a spike but it
is better to use a synthesized signal, usually called a chirp.
Linear sweep frequency chirps are one example of chirps that we,
at Cambridge Ultrasonics, have used extensively in developing
inspection systems for concrete. The chirp might start at 50 KHz
and sweep (with a linear change of frequency with time) up to
200 KHz in a time of 100 us. This chirp is almost 500 mm long
in concrete: it could be bigger than the sample under test! It
is certainly not the short pulse that we like to see as a result
of a pulse-echo test!
Don’t worry, the process of matched filtering not only improves signal-to-noise but it also compresses the duration of the chirp to a peak of width 2/B (where B is the bandwidth of the original chirp). With the chirp described above it should be possible to get pulses after matched filtering of about 13 us. This is much better! Now we are dealing with a pulse that has an apparent length in concrete of only 60 mm. Since concrete is heterogeneous and can contain aggregate particles as large as 40 mm in size we now have a pulse that couldn’t be much shorter without running into severe scattering.
Figure 4 caption:
The convolution of the driving chirp and the impulse response of the transducer. Note there is little or no need to put a windowing function (such as Hamming etc) on the driving chirp because the impulse response of the transducer acts as a windowing function - it is bandwidth limited. Usually the chirp should make use of as much of the available bandwidth as possible because the peak width after matched filtering depends upon the reciprocal of the bandwidth of the chirp.
Matched filtering has some other attractive
benefits too. Its good for portable systems because the signal-to-noise
ratio after matched filtering depends upon the energy of the chirp.
With a pulsed system about the only way to increase the signal-to-noise
ratio is by increasing the pulse voltage, hoping that the transducer
doesn’t get damaged. But by using matched filtering much
lower drive voltages can be used with relatively long chirps.
It is the much longer duration of a chirp that gives it high energy
and hence good signal-to-noise. The benefit of low-voltages is
simpler circuits. 
There is also scope to use different chirps for different conditions. This is what bats do. Tone bursts are good for detecting doppler shifts (moving targets) but are poor at giving good spatial resolution. Wideband chirps are best for pin-pointing scatterers. Presumably bats like long chirps because it saves them from chirping so loudly. I’ve read that bats don’t cross-correlate so their matched filters may be somewhat different to those described here.
Figure 5 caption:
The convolution of: the drive chirp, the impulse response of the transducer (as transmitter), the delay impulse and the impulse response of the receiver (assuming the transducer is both the transmitter and receiver and reciprocity applies) with added random noise. This would commonly be the received signal without any processing. Note that the noise is not gaussian noise but is instead uniformly distributed over the interval [-1000, +1000].
So far we have talked about software approaches
for matched filtering but there are hardware approaches - build
the matched filter into the transducer. An example is Surface
Acoustic Wave (SAW) devices. Instead of having a simple pair of
electrodes to receive a surface pulse on a crystal they have a
complex pattern of inter-digital electrodes. These are arranged
spatially to create the equivalent of a 
matched
filter and achieve very high signal-to-noise ratios because of
the inter-digital pattern.
Figure 6 caption:
The time reversed chirp is used to create a matched filter for this system - note this is not the signal used to create the optimum matched filter for this system. A better matched filter would be derived from a reference test and would include the impulse responses of the transmitter and receiver. The advantage of using the synthesized chirp is simplicity - there is no need for a reference experiment.
What can someone do who is involved in designing an ultrasonic inspection system and who wants to optimise its performance? Here are the recommendations we make to many clients:
reference signal in the matched filter.
Figure 7 caption:
The result of matched filtering (filter in figure 6) the received signal (figure 5).
As can be seen, optimising an ultrasonic system requires consideration of many elements in the system. It helps if you can see the ultrasonic waves too.
For more details of matched filtering a
personal recommendation goes to:
P.A. Lynn “Radar systems” Macmillan 1987.
Figure 8 caption:
The envelope of the filtered signal - created using the magnitude of the analytic signal (Hilbert transform). This signal would form the basis of interpretation. Note that the echo peak is not exactly at 1000 time steps - it is slightly later. The delay is caused by propagation through the transmitter and receiver.
The following section is not included in the printed version of Innovation News 5.
We will now compare the performance of three possible ultrasonic systems using the same transducer as both transmitter and receiver:
Figure 9 caption:
Case 1 above. Spike impulse used to drive the ultrasonic transmitter. Note the amplitude level is 100. Compare this with the amplitude of the chirp used earlier of 1.
Figure 10 caption:
Case 1 above. Signal received after reflection
from 
an
ideal reflector. Drive signal was a spike impulse.
Figure 11 caption:
Case 1 above. Signal received after simple
rectification and level detection. Note that the single echo has
resulted in three peaks - this complicates the task of interpretation
for the operator.
Also the noise, although absent away from the peaks, is superimposed
upon the peaks, adding to the difficulty of interpretation.
Figure 12 caption:
Case 2 above. Spike impulse drive followed
by matched filtering using a reference signal collected in an
ideal (no noise) test. Envelope detection is done using the magnitude
of the analytic signal. 
The
advantages of this method are: conventional drive circuitry with
a single peak for simpler interpretation.
Figure 13 caption:
Case 3 above. 
A
linear sweep frequency chirp is used to drive the transducer and
the matched filter is developed from a reference test. Compare
this result with figure 8. The difference is that a reference
signal is used here whereas the synthesized chirp is used in figure
8, which shows some sign that the main peak is not as well defined.
The result here is better. Compare this result with figure 11
and the peak here is much simpler to interpret. Compare this result
with figure 12. The peak width is narrower in figure 12 but the
signal-to-noise ratio is better here by about +6 dB. Also the
side lobe suppression is better here than in figure 12. Note also
that the process of creating the reference signal has removed
the transducer delays seen in figure 8. A practical reference
experiment may not always give such a good result.
Conclusions