* Eddy Current Technology, Inc., 201-A Horace Avenue, Virginia Beach, VA 23462; (757) 490-1814; fax (757) 490-2778; email: email@example.com.
advantage over multiplexing.
A multiplexing, multifrequency eddy current instrument is relatively easy to understand. Imagine if you had a single frequency eddy current instrument and merely switched the frequency control between two different frequencies. You would have a multiplexed, two frequency eddy current instrument. Obviously, you could not conduct testing at a very fast rate, but if you used electronics to switch between two frequencies, you could increase this to a rate adequate for practical NDT.
A simultaneous injection multifrequency eddy current instrument is almost as easy to understand. Imagine if you took two single frequency eddy current instruments and wired them together at the probe, so that you combine the two frequencies that are generated within the two eddy current instruments and send them out on one probe. Splitting up the signals that come back from the probe adds a little complexity, but this can be easily accomplished by using bandpass filters, each one tuned to one of the frequencies sent to the probe. Another way to split the frequencies is to use good quality detectors for each of the frequencies, followed by low pass filters. The “sample and hold” circuitry that was used in early single frequency eddy current instruments designed for heat exchanger tube testing is not suitable for multifrequency applications. The problem is that a “sample and hold” detector is equally sensitive to all frequencies being used; therefore, it cannot separate the frequencies.
Obviously, if only need to merely take a single frequency eddy current instrument and switch between two frequencies, you have low cost, but there are two primary disadvantages. One is that this method will be relatively slow because after switching from one frequency to the next, you must wait several cycles of your test frequency before you can sample the data. The speed problem becomes even greater at lower frequencies. A more significant disadvantage is that since the probe is continuously moving through a heat exchanger tube, the samples at two different frequencies are taken at different times and, therefore, different locations in the tube (Fig. 1). This will be discussed in more detail below.
Simultaneous Injection Advantages and Disadvantages
The obvious disadvantage is cost, as it would take two eddy current instruments, not one, to build a two frequency eddy current instrument, four instruments to build a four frequency instrument. Another, although less obvious, disadvantage is noise. To look at it simply, when you wire together two eddy current instruments to one probe, you will essentially end up with twice as much noise. Double the number of frequencies to four and you get twice as much noise again. This problem, however, can be taken care of with careful circuitry design to minimize noise in the first place.
Simultaneous injection has two significant advantages. One is that it can be very fast. There is no need to limit speed by waiting several cycles after a frequency has changed before taking a sample, as the frequencies are on the probe simultaneously (Fig.2). Another obvious advantage is that when the output signal is digitized with an analog to digital converter (as it is with all modern, computer based, multifrequency eddy current instruments), the samples can be taken at precisely the same instance in time if there is a dedicated analog to digital converter for each horizontal and vertical channel for each frequency. Also, with simultaneous injection, it is much more practical to design a multifrequency eddy current instrument capable of the low frequencies needed for remote field testing and near field testing of ferromagnetic heat exchanger tubes.
MULTIFREQUENCY EDDY CURRENT INSTRUMENTS
The first multifrequency eddy current instrument that I am aware of was the subject of a patent by Renken and Meyers in 1965. Designed with vacuum tubes, this instrument used 10 kHz as a testing frequency and 100 kHz to compensate for liftoff from the testing frequency. This unit was used for surface testing. The design incorporated bandpass filters to separate the two frequencies. The frequencies were not adjustable, as it is difficult to adjust bandpass filters.
The second multifrequency eddy current instrument was the Intercontrolle IC3A and was designed specifically to test steam generator tubes in French nuclear reactors. This unit also used bandpass filters to separate the frequency, but the components that controlled the frequency of the bandpass filters and the oscillator were built on to a submodule, which could be changed within individual frequency channel modules. This made it possible to change frequencies, but you needed a module for each frequency that you intended to use. Obviously, it was a bit time consuming to slip out a module, slip out a submodule and then reassemble. This came on the market around 1978.
The first simultaneous injection multifrequency eddy current instrument in which the frequencies could be changed from front panel controls was the Eddy Current Technology ect 3000, which came on the market in early 1980. The lowest available test frequency was 1 kHz or, on special order, 100 Hz. This unit used a high quality detector followed by low pass filters to separate the frequencies. All simultaneous injection multifrequency eddy current instruments today use this design.
The first multiplexed multifrequency eddy current instrument that the author is aware of is the Zetec MIZ-12, which came on the market in late 1979. Presumably, it was a modification of their EM3300 in which the testing frequency was switched. The lowest available testing frequency was 10 kHz.
A typical four frequency eddy current instrument today runs all four frequency channels in differential and absolute modes simultaneously, which provides a total of eight channels (not counting mixers). As each channel has both a horizontal and vertical signal, there are 16 signals which must be sampled and then converted to a digital value by an analog to digital converter. The best way to do this is with a dedicated converter (preferably 16 bit) for each of the 16 analog channels. This means that you need 16 16-bit analog to digital converters. Obviously not the cheapest way to do it, but the best.
Alternately, you could use an analog multiplexer, which would select each channel in sequence, followed by one 16 bit analog to digital converter. The analog multiplexer selects one channel and waits while the converter converts the signal to digital; then the second analog channel is selected, and so on. Obviously with this approach (which is not simultaneous sampling), each analog signal is sampled at a different point in time, and that means at a different probe position. The performance would be similar to a multiplexed multifrequency eddy current instrument.
Most eddy current instruments today are digital; therefore, the continuous analog eddy current signal is sampled from time to time and converted to digital format in an analog to digital converter and the resulting value is stored in RAM and eventually saved to a hard drive. Obviously, if you pull the probe through a tube at 300 mm/s (12 in./s) and then sample only once a second, you are going to miss just about everything. So you need a higher sampling rate.
A sample rate of 400 samples per second is generally accepted as an adequate rate for probe speeds of up to 1 foot per second. This works out to 0.8 mm (0.03 in.) per sample. Today’s eddy current instruments have relatively high sampling rates thanks to modern electronics and high speed computers. Most can sample at 10 kHz, so if you multiply 10 000 samples per second times 0.8 mm (0.03 in.) per sample, you get a probe speed of 7.6 m/s (300 in./s), a number which many manufacturers prominently note in their specifications. Unfortunately, the sampling rate is not everything: bandwidth is also important.
In the mid 1970s, before there were commercially available multifrequency eddy current instruments for heat exchanger tube testing, the ASME code recommended a probe speed of 300 mm/s (12 in./s) and an absolute maximum speed of 350 mm/s (14 in./s) in order to allow some error in probe speed. The single frequency eddy current instruments in use at this time, such as the Nortec NDT6 and the EM3300, had a bandwidth of 100 Hz, although it was not specified in their sales literature. At the time, it was believed that it was probably impossible to perform testing at a higher speed. This belief came about because of the approximate 100 Hz bandwidth of these two machines, which did, in fact, limit the probe speed to about 300 mm/s (12 in./s) while obtaining quality signals. (Another eddy current instrument available at the time was the ProboLog 700, which had a bandwidth of about 50 Hz.) Eddy Current Technology, Inc., introduced the ect 3000 in 1980; this was an oscilloscope based analog multifrequency eddy current instrument with a bandwidth of 200 Hz, which had the capability to test at 0.6 m/s (2 ft/s), although it had been a common belief that it was impossible to test faster than 0.3 m/s (1 ft/s).
To explain the concept of bandwidth, it is useful to refer to stereo equipment. As a human can hear from about 20 Hz to 20 kHz, this is the specification that manufacturers and users of “high fidelity” stereo equipment would like their equipment to meet. Today, the vast majority of it does. If you listen to the older recordings, you will note that all music has some noise in it. Even a CD that you are listening to today may have been recorded some time ago with older equipment. If a tape recorder was involved, there will always be an audible hiss. This is high frequency noise. You can get rid of this if you have a graphic equalizer by turning down the highest frequency control. You could think of the hiss as being information in the original signal (even though it was not wanted information); but by turning down the high frequency control, you limit the bandwidth, getting rid of this information. When you do so, you will also reduce some wanted information, specifically, the high frequencies that you may have heard mostly in cymbals or a snare drum. By limiting the bandwidth, you have limited some of the information.
If you were to take one of these older eddy current instruments with a bandwidth of 100 Hz and pull a probe past discontinuities in the tube at about 300 mm/s (12 in./s), you would get the frequency content as the discontinuity passes the probe coils would be significantly below 100 Hz, such that all of the information from the discontinuity gets through the low pass filters. If you then change the probe speed to 600 mm/s (24 in./s), you would find that the eddy current signals get altered; specifically, they decrease in amplitude. This is because the frequency content of the eddy current signals now exceeds the bandwidth. To test at 600 mm/s (24 in./s) is a simple matter of changing the bandwidth of the eddy current instrument to 200 Hz. In the ect 48, the bandwidth is adjustable with the low pass filter, which has a maximum setting, or maximum bandwidth, of 1 kHz, which would translate to a maximum probe speed of 3 m (10 ft/s) without getting signal degradation. Most manufacturers of multifrequency eddy current instruments do not specify bandwidth. Sometimes you can get this information if they have an adjustable low pass filter. The maximum setting of the low pass filter is the maximum bandwidth of the eddy current instrument.
Actually, there are two reasons. As with in discussion above about reducing the bandwidth of a stereo amplifier in order to reduce hiss, the same applies to eddy current signals. There may be situations in which you want to decrease the bandwidth in order to reduce noise — for example, if you are looking for extremely small, low amplitude discontinuity signals. The second reason is that if your test frequency is ever lower than the bandwidth, then the frequency itself will end up as a noise signal. (This often looks like a circle or similar pattern on the screen.) Actually, you see this noise even when the frequency is slightly above the setting of the low pass filter, because the low pass filter does not stop absolutely every signal that is slightly above its setting, but instead there is a slope above the stated frequency of the low pass filter that attenuates signals significantly higher than that.
In the case of testing ferromagnetic tubes with remote field technology, test frequencies are typically lower than 1 kHz and often even lower than 100 Hz; therefore, if an eddy current instrument is going to be used to test both nonferromagnetic tubes at a high speed and also be capable of testing ferromagnetic tubes using these low frequencies, then you have to be able to adjust the low pass filter in order to accommodate these low frequencies. Also note that as the test frequency decreases and you decrease the setting of a low pass filter, it will also be necessary to reduce the probe speed due to the lower setting of the low pass filter.
Note further that if you decrease the low pass filter, you may also decrease the sampling rate in order to reduce the filesize of these eddy current signals. A sampling rate four to five times the setting of the low pass filter seems to be adequate.
The bandwidth is rarely specified for an eddy current instrument; however, the value of the fixed low pass filter or adjustable low pass filter is occasionally specified and this can be taken to be the same as bandwidth.
The ASME code requires a sample every 0.8 mm (0.03 in.) as stated above. A typical eddy current probe may have two differential coils with a length along the axis of the tube of about 1.6 mm (0.06 in.) and spacing of about the same. Therefore, the distance between the center of one coil and the center of the other coil is only 3 mm (0.12 in.). With a sampling rate of 0.8 mm (0.03 in.) per sample, then you might get a sample at the peak of the signal followed by two samples along the central portion of the signal and a sample at the following peak. This breaks the entire diagonal portion of the eddy current signal (which is used to measure angle and, hence, discontinuity depth) down to only three segments (Fig. 3). If instead of sampling exactly at the peak, the samples were taken a little before and after the peak, you would actually lose about one third of the amplitude of the signal and the corresponding angle may change as well. We do not consider this to be acceptable.
We believe that a sampling rate twice as fast (every 0.4 mm [0.015 in.]) gives a good result and a sampling rate 2.5 times as fast (every 0.3 mm [0.012 in.]) gives an excellent result. This translates to a sampling speed of 800 or 1000 Hz for a probe speed of 300 mm/s (1 ft/s). Therefore, we believe that an eddy current instrument with a sampling rate of 10 kHz can give an excellent result at probe speeds of up to 3 m/s (10 ft/s).
SO WHICH IS BEST?
There is no doubt that simultaneous injection has the speed advantage over multiplexing. Furthermore, multiplexing is not suitable for low frequency testing, such as is needed for remote field or near field testing of ferromagnetic tubes and is not even well suited to the low frequencies required to test copper and brass tubes. Also, since with simultaneous injection it is possible for the sampling and analog to digital conversion to take place simultaneously for all channels, you will get better support plate cancellations (or other mixes) and if you compare angles and amplitudes measured at one frequency to angles and amplitudes measured at other frequencies, you will be comparing the same points. Therefore, simultaneous injection is superior on all points important for speed and accuracy during the testing of heat exchanger tubes. On the other hand, multiplexing should cost less. Therefore, if you have a limited budget and can afford the degradation in speed and accuracy, multiplexing may be the solution.
Figure 1 — Four frequency signals: multiplexed signal.
Figure 2 — Four frequency signals: simultaneous injection signal.
Figure 3 — A sampled eddy current discontinuity signal.
Figure 3 — A sampled eddy current discontinuity signal.