In the autumn of 2001 the European Commission agreed to support a proposal for a collaborative R&D project submitted by Cambridge Ultrasonics on the subject of improved inspection methods for concrete structures. The project started in January 2002 and will last for 4 years.
Other partners in the project are: Sonatest PLC (the largest European manufacturer of ultrasonic NDT equipment), Necso (a large Spanish construction company), SP (the leading Swedish inspection company with responsibility for national and international standards), IETcc (the Spanish research institute for civil engineering) and Queen’s University Belfast (civil engineering department).
The proposal was made to the European Commission’s Framework V R&D programme. Unfortunately, submitting a proposal is a painful process and merit does not always guarantee success - this proposal was the fourth submitted by Cambridge Ultrasonics, the other three being rejected!
Maintaining a consortium of partners faced with rejections of proposals has been one of the difficulties. Many organisations have expressed an interest in the project and some participated as partners in unsuccessful proposals. To all those orgainisation who expressed an interest Cambridge Ultrasonics would like to say thank you and please keep in touch because your organisation may still be able to benefit by participating informally in the project (contact Merete Bergmann on camben@cambridge-en.com).
The objectives are to develop and validate new, advanced computer-based ultrasonic measurement instruments to: (a) create images of the interior of concrete and (b) to give early warning of significant structural changes using continuous monitoring. The work will require not only sensor and hardware development but also development of and validation of metrological software.
Improvements will be in the performance and reliability, intelligent operation, cost-efficiency and suitability for use in the field on large concrete structures. Novel and innovative ultrasonic instrumentation, including sensors, will be developed that will improve the quality of measurements for structure owners, operators, construction companies and service companies as well as that required for the establishment of the international traceability of measurements. The activities will include not only hardware development, but also development and validation of metrological software.
There is a growing problem within Europe concerning the maintenance of large concrete structures. The most notable problem is the unexpected deterioration, due to corrosion, of steel stressing-tendons, causing the tendons to reduce their load-carrying capacity. This is serious because there is no reliable, non-destructive measurement for testing the degree of corrosion. The project aims to provide for the first time cost-effective and reliable measurement systems for use in surveys and inspections of large concrete structures and for monitoring the structures during normal operations. This will enable owners and operators of large concrete structures to measure the quality of these structures; operators will be able to rate the repair needs of the structures under their control and prioritise their repair budgets for optimum safety. As citizens of Europe and users of these structures we should all benefit.
Examples of structures are: residential and office tower-blocks, bridges, nuclear power stations, off-shore oil production structures, multi-storey car parks and dams. The replacement value in Europe of such structures is several hundred billion Euros and there is, therefore, a strong incentive to prolong their working life.
Cambridge Ultrasonics is a source of ideas and innovation for its clients. Using our knowledge and experience in ultrasonics, physics, electronics and software we work with our clients’ R&D departments to fill gaps in their understanding. Our flexible approach to working with our clients means that they can leverage the abilities of their own R&D engineers. Over the last 15 years we have worked with large blue chip companies through to small emerging businesses, government agencies, research institutes and universities. Most of our clients return to make use of our services over the years.
We help our clients in the following ways:
The plan for the future of Cambridge Ultrasonics is to convert our considerable knowledge and reputation in the field of inspecting concrete into new products and services in a spin-off company. There are several business opportunities before us (Spring 2002) but to exploit them we need an injection of capital. With this capital we can grow the new business.
For example, Sonatest Plc, a large manufacturer of industrial ultrasonic inspection equipment, wants to sell our products throughout the world. The same company wants to be a major shareholder in the new business.
We want a partnership with private investors,
institutional investors and corporate investors who would like
to provide business capital and share in the opportunity to grow
the business. For more details see 
.
Tess Recordon is a Cambridge based artist.
She paints bold colour abstracts in oil, inspired by nature and
world-wide travels. She has exhibited widely, future exhibitions
in 2002/3 include the Channel Islands, Cambridge and the USA.
A proportion of painting sales from Tess's last exhibition at
ARM Holdings in Cambridge went to Milton Children's Hospice, over
£3000 was raised. Tess undertakes large commissioned works
for site specific locations.
For more information visit either the Cambridge
Network or Tess’ own site
http://www.tessrecordon.com
Cambridge is well-known as a melting pot for humour and comedians. Here is a web-site you might like to browse for free, created by former Cambridge students.
Are these to be the Peter Cook, John Cleese,
Stephen Fry, Hugh Laurie or even Blunt and McLean of the future?
http://www.lieslieslies.co.uk
A lot of things have been said about this site. "A sparkling and witty new smile on the haggard and depraved face of internet comedy" was not one of them.
In September 2001 Cambridge Ultrasonics provided demonstrations of its ultrasound visualizing experiment (see figure) for about 1000 GCSE students as a part of a 3-day event held in Cambridge called Physics at Work. Other organisations giving demonstrations of how physics is used at work included Rolls-Royce, Cambridge Consultants and the University of Cambridge. The purpose of the event was to provide motivation for children to study physics at GCSE and beyond.
With the same purpose in mind Cambridge
Ultrasonics has been helping the Institute of Physics East Anglian
branch (IoP EA) to promote physics with schoolchildren just before
the Christmas holiday. On 16th December at the Cavendish Laboratory
(Cambridge University’s physics department) the IoP EA ran
three events on the same day: a lecture on explosives and high
speed photography by Dr Bill Proud, an exhibition of Formula 1
cars and speed and a quiz game called Call My Bluff - In Physics
with a panel of experts including the head of the Cavendish, Professor
Longair and Terry Holloway a director of Marshalls Aerospace.
David Andrews of Cambridge Ultrasonics was one of the organisers
of the event and together with Suk Pannu of Videojet scripted
and ran Call my Bluff in Physics. About 350 people came of whom
about 50% were schoolchildren.
Call my Bluff in Physics is a quiz in which the audience’s-favourite Suk Pannu gave a series of demonstrations, for each of which the panelists give an explanation. At least one explanation was true and usually the other three were bluffs. The members of the audience were equipped with voting cards and could vote for the panelist they believed was giving the most accurate physics. David Andrews (the new Angus Deayton of physics?) was the chairman of the event. A free and popular raffle followed the quiz.
The Institute of Physics asked Suk and David to stage the event at the IoP’s annual congress this year due to the event’s popularity.
Poster designed by Cambridge Software.
Wanted - hardware/software engineer(s)
Cambridge Ultrasonics is expanding. We are seeking a talented engineer (possibly two) to work with us. Experience in ultrasonics is not essential but you should have a variety of hardware and software skills instead, for example: signal processing experience, Matlab, Simulink, PLDs, FPGA design, interfacing to microprocessors (PCs in particular), A/D stages, analogue signal conditioning circuits, PIC microcontrollers, DSPs, PCB design, C++, MFC, VB, signal processing, designing and testing to meet international standards and CE marking.
You will be a graduate with a good degree probably in electronic engineering or physics or computing or similar, with several years relevant postgraduate experience, possibly with a PhD in a related discipline.
See www.cambridge-en.com for more details. Send CVs to camben@cambridge-en.com.
For sale - Gage data aquisition and signal synthesis system
CompuScope 1012 - 2 channel digital sampling system, 12-bit and 20 MSamples/sec per channel with deep 512 kB on-board memory. Input sensitivity +/- 100 mV to +/- 5 V. External trigger input as well as very flexible triggering modes on the two input channels. This is an ISA bus plug-in card for a PC and comes with a DOS executable program and drivers for various C-compilers, Pascal and Basic compilers as well as LabWindows/CVI.
CompuGen 840 - 1 channel, 8 or 12-bit arbitrary waveform and digital pattern generator. 16 kB of memory and maximum output clock speed of 40 MHz. Both analogue and digital outputs can be created simultaneously. Analogue input trigger channel. Analogue output is 20 v p-p at 50 ohms. This is also an ISA bus plug-in card for a PC and comes with a Dos executable program and drivers for various C-compilers, Pascal and Basic compilers.
Cost when new £ 3,874. Price now £
800 excluding VAT and delivery. All in perfect working condition.
Contact .
The famous physicist Heisenberg told us if we want to measure arrival time accurately we must use a sharp pulse and therefore a wide range of frequencies. This seems quite reasonable at first sight but is this advice followed in practice? Read on...
How is arrival time commonly measured in commercial systems? The first time of arrival of a pulse is used almost exclusively. It’s the first acoustic wave energy to get to the receiver or the first sign of the wave appearing in the background noise (see figure 1). This implies that the signal-to-noise ratio is 0 dB or worse - does this make you feel comfortable about making an accurate measurement! No, it’s probable there are going to be difficulties. The accuracy of first time of arrival depends upon the noise distribution in the electrical signal from the ultrasonic receiver (and input amplifier) and the slope of the ultrasonic signal emerging from the noise (the smaller the slope the bigger the error).
One example of the importance of accuracy in deciding when a wave has arrived is the measurement of flow speed. A popular flow-meter uses an ultrasonic path inclined at an angle to the flow vector (see figure 2). The speeds of ultrasonic pulses are measured in the two directions possible so that in one test the waves travel with the fluid motion and in the second the measurement is against the fluid motion. There is a difference in speed that can be used to estimate the fluid flow speed.
Flow meters are sometimes required to be accurate to better than +/-1% when used for calculating consumption of fuel gas (North Sea gas supply for example). Unfortunately, this kind of difference method is prone to problems due to the propagation of errors when two numbers close in value are subtracted. The absolute error in the result is the sum of the absolute errors in the numbers and the relative error is found by dividing by the result of subtraction. The relative error can grow to be very large indeed when the two numbers are almost equal, so the engineer has to design a system with exceptionally small errors.
Let’s use some numbers; assume the ultrasonic wave speed, c, in the fluid is 200 ms-1 and the path length is 10 cm. With no fluid flowing the times to travel (t1 and t2) are 500 us (equal in both directions). It should be possible to measure the arrival time to better than +/-1 us but assume +/-1 us for now. The relative error in measuring the ultrasonic wave speed is approximately 1/500 or +/-0.2% - this appears to be very accurate and very promising. However, these seemingly tiny errors combine to give +/- 60% error in measuring a fluid flow speed, v, of 1 ms-1.
Digital sampling can be used at speeds of 10 Msamples/sec or higher to give potential accuracy of +/-0.1 us or better instead of +/-1 us but don’t be seduced into thinking that switching to digital signals will miraculously solve this problem. Look again at figure 1 - converting the signal into the digital domain won’t alter the fundamental issue. The issue, often ignored in the early stage of design, is when does a wave arrive?
Heisenberg’s uncertainty principle in physics, from quantum mechanics, states that the product of the range of angular frequencies and range of times is greater than or approximately equal to 1 for any pulse of waves (see equation below). It is derived using Fourier’s method of representing functions (signals in our case) by a superposition of sine-waves. We can accurately know the frequency of a long pulse of a single frequency of ultrasound (figure 3) but because it is so long we cannot be sure when it arrives; conversely, we can accurately know when a very short pulse arrives (figure 4) but only if it contains a wide range of frequencies, because Fourier’s theory tells us it must be constructed from a very wide range of frequencies.
Let’s put some numbers into Heisenberg’s equation: let’s assume we have an ultrasonic transmitter working at 1 MHz and the bandwidth is 200 kHz (range of frequencies making up the pulse) then the uncertainty over the arrival time of the pulse has to be about 0.8 us or longer (don’t forget it’s angular frequency in the equation).
Another effect to consider is dispersion. Dispersive materials cause waves of different frequencies to travel at different speeds. In many materials the speed decreases as the frequency increases so the first wave to arrive will have the lowest frequency and the smallest slope (and hence be most inaccurate in the common first time of arrival method!). Another problem with dispersion is that the design may call for a measurement using a specific ultrasonic frequency (of say 1 MHz for the flow-meter example) and this is usually interpreted as the centre frequency of the ultrasonic transducer. However, dispersion and the use of first time of arrival commonly causes the effective test frequency to be much lower than 1 MHz. Let’s apply Heisenberg’s equation to the conventional first time of arrival method with dispersion. Assume the true test frequency is 200 kHz (not 1 MHz) and the bandwidth is 10 kHz then the error in arrival time is +/-16 us.
Returning to the effect of making measurements at 0 dB signal-to-noise ratio, these are ideal conditions for electrical interference or mechanical noise to interfere with the measurement system. Inevitably the poor design engineer is forced to set a threshold for detection of the wave at some level well above the noise floor but this is not now the first time of arrival - it is an arbitrary time after the first time of arrival (great care is now needed to achieve repeatability). Another common approach is to use signal averaging to improve the signal-to-noise ratio but this is equivalent to low-pass filtering and so the flow-meter loses sensitivity to transients in the flow.
Have you patiently read to this point waiting for a solution? It may not be a solution but it is certainly something to think of trying. Hot Hints Issue 5 described a linear signal process called matched filtering (back issues available on www-cambridge-en.com). It’s possible for an ultrasonic transmitter to launch a known pulse of waves (for example a linear sweep-frequency chirp - see figure 5), for a receiver to collect the ultrasonic chirps then for a computer to look for occurrences of chirps using matched filtering. The result of matched filtering is to convert the (long) chirp into a sharp pulse (figure 6), it compresses the chirp, and the degree of compression is greater the wider the bandwidth of the chirp.
Heisenberg would approve of chirps and matched filtering. Radar engineers have been using it for years. Electronic engineers should approve too because now they have to measure the time of the tip of a peak in a signal (signal-to-noise ratio is maximum and slope is high). The measurement is based upon a broad range of frequencies and although dispersion can still be a problem it is possible to devise matched filters that compensate for dispersion; work at Cambridge Ultrasonics in the past has looked into this.
There are a number of other interesting benefits of using matched filtering: lower voltage electronics - chirps can be quite long and that means there is more time to get energy into them. The energy of any pulse determines its signal-to-noise ratio so with more time to drive the chirp the amplitude does not need to be as high as in a spike (conventional NDT drive signal). Sometimes the ultrasonic medium is heterogenous (rather than homogeneous) and this can give rise to multiplicative noise (rather than additive noise) - experience at Cambridge Ultrasonics suggests that matched filtering is of some help in suppressing multiplicative noise. Chirps can also be tailored to the transducer to extract the maximum bandwidth available. There are other kinds of chirps that can be used instead of linear sweep frequency chirps, for example Barker codes, which can be used to get better side-lobe suppression for those all-important peaks.
At Cambridge Ultrasonics we use Matlab to simulate and develop signal processing for our clients. We can fairly quickly compare the results of first time of arrival processing with chirps and matched filtering. Incidentally, we can also convert our Matlab programs into C or C++ code that can be recompiled into clients’ systems quickly (hardware changes take a little longer!). So converting a system to Heisenberg’s ideas does not need to be quite as onerous as it might seem at first.
Heisenberg was in charge of building an atomic bomb for Hitler during the second world war. Vilified by his peers it is now thought that he may have effectively sabotaged Hitler’s bomb. The recent stage play “Copenhagen” focused on Heisenberg’s war time activities and uncertainty about his commitment to Hitler’s aims. It’s somehow fitting that a man famous for his uncertainty principle should himself be the subject of such historical uncertainty.
Figure
1 - when does a wave arrive? Signal from a receiver showing the
first time of arrival of a wave. It is difficult to estimate precisely
when the ultrasonic signal emerges from noise.
Figure
2 - sketch of a flow meter, showing two transducers.
Figure
3 - Long pulse of primarily one frequency component. The frequency
is known accurately but the time of arrival is not known accurately.
Figure
4 - A short pulse. The time of arrival is known accurately but
a wide range of frequencies is needed to construct the pulse and
so the frequency is not known accurately.
Figure
5 - an example of a linear sweep frequency chirp that might be
transmitted by an ultrasonic transducer. Note the long duration
of the pulse.
Figure
6 - chirp in figure 5 processed by matched filtering. Note that
the long chirp is compressed into a short pulse. This pulse should
be better for time of arrival measurements. Compare with figure
4.