How Your Electronics Work

Subversive Guide to Engineering

2009-03-26T15:14:00.003-04:00

I've started a new blog called the Subversive Guide to Engineering. Its a sarcastic look at what to expect in engineering school, how to succeed (or more likely, survive), life in eng, and tips on picking majors and careers. The address again is http://subversiveguidetoeng.blogspot.com.

If you have any tips or stories to share, I'd love to hear from you in the comments of the site. Likewise, if you know someone that is in engineering or is looking to go into engineering, to give them a heads up about the site. I definitely would have loved to have a source like this back when I was starting out.

Thanks again!

Fiber Optic Gyroscopes

2008-09-16T11:16:00.006-04:00

Introduction

Gyroscopes are key components of the navigation systems that keep vehicles on course. In fact, gyroscopes can be found in anti-skid systems in cars, satellites, the Space Shuttle, airplanes, ships, and missiles. The first practical gyroscope instruments were used to create “artificial horizons” for airplanes. The first gyroscope was invented in 1852 by Leon Foucault, however it was Elmer Sperry who made the first practical instruments in 1910 [1]. The first gyroscopes were mechanical, but they’re now made with fiber-optics, ring lasers, and even with solid state MEMS devices [1], [4].

Navigation Systems

Gyroscopes are fundamental building blocks of the navigation systems used to guide airplanes, ships, and spacecraft [1]. A gyroscope is essentially an instrument that measures rotation [1]. By itself, gyroscopes are useful in applications such as measuring the wing angle in aircraft in flight and the roll of ships in heavy seas. However, combined with accelerometers, they are used to create inertial guidance navigation systems [4].

An inertial guidance navigation system is essentially a form of “dead-reckoning”. If the starting position is known, keeping track of the acceleration and rotation that a vehicle undergoes can be used to determine its current position [4]. Such a guidance system is constructed with 3 accelerometers and 3 gyroscopes, each measuring acceleration and rotation in a single axis. Such a system can measure an aircraft’s position, velocity, acceleration, attitude, and heading to a high degree of accuracy [4].

Gyroscopes: Principles of Rotation Measuremens

Before considering photonic solutions to the problem of rotation measurement, it is important to consider previous methods. An important form of gyroscope, still in use today, is the mechanical gyro. The invention of the first mechanical gyroscope is credited to the French experimental physicist Leon Foucault in 1852, who planned to use it to measure the rotation of the earth [2]. As shown in figure 1, a mechanical gyroscope is created by suspending a rapidly spinning rotor inside three frictionless rings, called gimbals.

Figure 1 – [Left] A mechanical gyroscope contains a massive spinning rotor with an angular momentum pointed along the spin axis. If the gyroscope is tilted, [right], the rotor maintains its spin due to conservation of momentum. The tilt of the gimbals gives an indication of the tilt of the instrument [3]

Due to conservation of angular momentum, the spinning rotor maintains its direction in space even if the gyroscope is tilted. Therefore, tilting the gyroscope will cause the gimbals to reorient themselves to maintain the rotating mass in its original spin direction. The angle of rotation of the outer ring about its axis is proportional to the rotation of the gyroscope about its spin axis [3].

Principles of Operation of Fiber-Optic Gyroscopes

Fiber-optic gyroscopes (FOGs) are based on the Sagnac effect. Sagnac interferometers are based on the principle that if the interferometer is rotating, light waves traveling in opposite directions in a loop will acquire a phase difference, resulting in interference. The phase difference is due to the fact that light travels with a constant speed, c, as shown in figure 2. If the interferometer is rotating counter-clockwise (CCW) with an angular velocity, by the time the clockwise (CW) traveling wave (in red) reflects from mirror M1, it will have moved slightly closer to the wave. Likewise, mirror M1 will have moved slightly away from the CCW traveling wave (in blue).

Figure 2 – An illustration of the Sagnac effect. In an interferometer rotating counter-clockwise , the counter-clockwise propagating beam (red) will experience a shorter path length while the clock-wise beam (blue) will experience a longer path length. The difference in path length leads to a phase shift between the two beams and hence interference.

Therefore, the CW traveling wave will have a shorter path length while the CCW wave will have a longer path length. The result will be a net phase difference between the two waves, causing interference.

Interference Fiber-Optic Gyroscopes

The interference fiber-optic gyroscope (IFOG) is based on detecting the phase shift difference that occurs in an interferometer due to the Sagnac effect. Unfortunately, the Sagnac effect is relatively weak, so to overcome this problem kilometers of fiber optic cable are used to increase the path length of the instrument.

An example of the extreme sensitivity needed by gyroscopic instruments is the fact that an inertial guidance system must be capable of detection rotations of 0.01 degrees/hour. Using light with a wavelength of 1 micron, 1 kilometer of fiber and a coil diameter of 30 cm, the resulting phase shift is only 10^-7 rad, which is at the detection limit of current instruments [5].

In order to increase accuracy, a practical IFOG system would measure interference effects at the same port as the input light, such that both lightwaves experience two reflections. This eliminates the extra phase shift from reflection that would otherwise be introduced into the measurement. A polarizer is also placed at the input to eliminate the polarization that is introduced to the light in the fiber. This extra polarization is introduced when light travels through the fibers since optical fibers are birefringent to some degree. Birefringence refers to the fact that light of different polarizations is refracted differently. This occurs in optical fibers most commonly because of small defects in the fibers themselves.

Commercial Examples of FOGs

FOGs are commercially sold by a number of manufacturers, for civilian, military, and space applications. An example of an inertial guidance system is the IMU 200, developed by Northrop Grumman [6]. The IMU 200 contains three gyroscopes and three accelerometers and is designed for high-performance applications, especially for weapons guidance systems. The system can withstand accelerations of 12 g’s, has a long-term stability of 0.5 degrees/hr and misalignment error of 0.1 mRad [6]. A version of this guidance system is currently used in the National Missile Defence interceptor missile [6]. An important point is that in many countries, including the United States, it is illegal to export highly accurate gyroscopes, because of their potential uses in weapons [1].

Future Trends: Photonic-Bandgap Fibers

The prime limitation on the accuracy of FOGs are parasitic effects that occur inside the silica fibers. These include Rayleigh backscattering, the Kerr effect, the Faraday effect, and thermal effects [7]. As mentioned previously, Rayleigh backscattering is caused by impurities in the fiber and leads to large random errors in measurement due to spurious signals. The Faraday effect causes a change in the birefringence of the fiber on the application of a uniform magnetic field. The Kerr effect, meanwhile, is a non-linear optical process that changes the index of refraction of the fiber with small variations of input power of the two beams. This causes a “drift” in the measured rotation rate [5]. Lastly, uneven thermal effects in the fiber can cause unwanted phase change. Air-core photonic-bandgap fibers reduce the thermal effects by a factor of 3-10 and the other effects by a factor of 100-500! [7].

A photonic bandgap material is one in which certain wavelengths of light are unable to propagate. This is very similar to an electronic bandgap, where certain electron energies are not allowed. The photonic bandgap is crated by a periodic microstructure. This effect is also seen in nature, for example in butterfly wings. Butterfly wings contain a fine structure, forming a photonic bandgap. Light with wavelengths in the bandgap region is strongly reflected, forming the bright colours that pattern the wings [8].

A dielectric periodic lattice will exhibit the photonic bandgap effect. For example, as shown in figure 3, this effect is exhibited by a diamond lattice [9]. To create a photonic bandgap fiber, a periodic lattice of airholes is formed in the fiber, creating the photonic bandgap material. A defect center, the central air-core is then introduced. Light propagating in the air core will not be able to leave the fiber because of the photonic bandgap material surrounding it [8]. The structure of a photonic crystal fiber is shown in figure 4.

Figure 3 – [Left] An illustration of a diamond lattice structure. [Right] The resulting bandgap structure [9].

Figure 4 – An illustration of a commercially available photonic bandgap fiber manufactured by Crystal Fibre [8].

The accuracy improvements found in air core fibers are mainly due to the advantageous properties of air over silica. The Kerr constant of air is about 800 times smaller than in silica. Likewise, the Faraday effect is about 500 times weaker. Rayleigh scattering is theoretically lower in air. Unfortunately, in current air-core fibers, Rayleigh scattering effects are actually higher than in silica fibers. This is mainly due to small dimensional fluctuations in the wall of the fiber introduced in the manufacturing process [7].

Researchers at Stanford University have created the first air-core photonic bandgap based fiber-optic gyroscope. The FOG was built with 235 m of commercially available fiber manufactured by Crystal Fiber. The minimum sensitivity of the gyroscope was 2.7 degrees/hour and the long term drift was 2 degrees per hour. To compare the performance with regular fiber, the crystal fiber was replaced with 200 m of silica single-mode fiber. The resulting minimum sensitivity was 7 degrees/hour and the long term drift was 3 degrees per hour [7].

The gyroscope created was essentially a proof-of-concept design. Currently, much work is left to be done in the development of photonic bandgap fibers. Current fibers have high loss (~19 dB/km for high quality fiber) and scattering mechanisms. Ultimately, research and development into improved photonic bandgap fibers will lead to FOGs with improved long-term stability, simplified design, lower cost, and higher reliability [7].

References:

[1] Fischetti, M., Gyroscope Guidance: Hidden Guides, Scientific American, Vol. 286, n. 6, June 2002, pp. 96-97
[2] Greenslade, T., Gyroscope, Instruments for Natural Philosophy, 2006
http://physics.kenyon.edu/EarlyApparatus/Mechanics/Gyroscope/Gyroscope.html
[3] Hyperphysics, Gyroscope, Hyperphysics, 2006
http://hyperphysics.phy-astr.gsu.edu/hbase/gyr.html
[4] King, A.D., Inertial Navigation – Forty Years of Evolution, GEC Review, Vol. 13, no. 3, 1998. pp. 140-149
[5] Bergh R. et al., An Overview of Fiber-Optic Gyroscopes, Journal of Lightwave Technology, Vol. LT-2, no. 2, 1984. pp. 91-107
[6] Northrop Grumman, IMU 200 Product Brochure, Northrop Grumman Navigation Systems Division, 2000
http://www.nsd.es.northropgrumman.com/Html/IMU200/BrochureIMU-200_Inertial_Measuring_Unit.pdf
[7] Kim H.K. et al., Air-Core Photonic-Bandgap Fiber-Optic Gyroscope, Journal of Lightwave Technology, Vol. 24, no. 8, 2006. pp. 3169-3174
[8] Crystal Fibre, Technology – Air Guiding Fibers, Crystal Fibre, 2005
http://www.crystal-fibre.com/technology/technology_tutorial4.shtm
[9] Johnson, S., Photonic Crystals: Periodic Surprises in Electromagnetism, MIT, 2004
http://ab-initio.mit.edu/photons/tutorial/
[10] Sherman, R., Star Wars Programs, FAS.org, 2006
http://www.fas.org/spp/starwars/program/000708-nmd_ift5_8.jpg
[11] Sensormag, Sensors, Sensormag.com, 2006
http://sensorsmag.com/sensors/data/articlestandard/sensors/142006/318761/0900_101f.gif
[12] Northrop Grumman, Navigation Systems Division Brochure, Northrop Grumman Navigation Division, 2006
http://www.nsd.es.northropgrumman.com

Milestones: Flash Memory [1984]

2008-08-10T12:05:00.003-04:00

Note: This article is a lead-in to an excellent story about the invention of flash memory, and its inventor, Dr. Fujio Masuoka, titled “Unsung Hero” at Forbes.com

Introduction

Today, Flash memory is ubiquitous, compromising a $76 billion dollar a year market, largely cornered by one chipmaker: Intel. Flash memory is integral to a large number of consumer and industrial applications such as:

PCs and notebook computers
Solid state music players
Cell phones and PDAs
Security systems
Embedded systems
Networking products
Medical devices

The key aspect of Flash that makes it unique is that the memory is non-volatile and re-writeable. Traditional RAM memory must be constantly refreshed every few milliseconds in order to maintain its contents, without power the memory is reset. Meanwhile, traditional ROM memory is non-volatile, but is not easily re-written with new information. Devices used to store information more permanently, such as hard disk drives, meanwhile, are not solid-state devices but instead have many moving parts. Therefore, they tend to consume a lot of power, are sensitive to impacts and vibrations, are bulky, and suffer from a higher rate of failure. Flash memory would turn out to be a proverbial silver bullet, addressing all of these concerns.

Flash was invented at Toshiba in 1984 by Dr. Fujio Masuoka. Dr. Masuoka, pictured on the left, got to his senior research position at Toshiba by making incremental improvements to Toshiba’s bread-and-butter product, DRAM memory. However, the idea of creating non-volatile solid state memory drove him. He worked on the idea largely working on his own, without the blessing of upper management. After introducing the first prototypes of the memory in 1984, he received a bonus, worth a few hundred dollars from Toshiba. Intel, convinced the new type of memory would be an important cornerstone of the memory market invested heavily in commercializing the new technology, while Toshiba gave Masuoka a few part-time engineers.

In the end, Toshiba was embarrassed by its failure to capitalize on Flash, and Toshiba spokesmen have even tried to claim the memory was actually invented at Intel. Toshiba, meanwhile, allegedly tried to demote Fujio Masuoka, who was regarded at the company as not a team player and insubordinate due to his independent work on Flash.

Without further ado, the rest of Fujio Masuoka’s fascinating story is available here: Unsung Hero at Forbes.com.

Interesting Links:
Unsung Hero: Forbes.com
Samsung’s Solid State Disk Drive

Milestones: The First Telephone Call [March 1876]

2008-08-05T13:38:00.005-04:00

Introduction

On March 10, 1876, Alexander Graham Bell made the world's first telephone call, speaking the words "Mr. Watson, come here, I want to see you." to his assistant in an adjoining room. The communications landscape in 1876 was much different than it is today. Communications were carried out largely through mail or telegraph, with telegraph lines increasingly criss-crossing the developed world. However, the mail system was slow, and telegraph messages had to be laboriously tapped out in Morse code, making them necessarily short and limited to urgent communications. In other words, the world was a much larger and less connected place.

Development of the First Telephone

As often happens, the telephone was the result of research and development by a large number of innovators, however it was Alexander Graham Bell that filed the first patent, and hence is widely credited for its invention. The first telephone, with which Bell transmitted his historic message, used a liquid microphone, invented by Elisha Gray. Gray's liquid transmitter consisted of a diaphragm attached to a metal needle placed just barely into contact with a conducting liquid (any liquid with free ions will do, such as a water/acid mixture). As the diaphragm vibrated, the needle would dip into and out of the liquid, resulting in a variation of current passing to the receiver.

Bell, meanwhile, had been working on a similar concept, however using the movement of a reed in a magnetic field alone, rather than in a liquid. His first devices using this concept, in 1875, however, suffered from a large amount of interference and tended to have the reeds stick to the electromagnets. Bell developed his first bi-directional telephone, and the one used in the famous demonstration with Watson, using Gray's water transmitter idea. The water transmitter was not ideal however, and Bell's later experiments focused on perfecting the electromagnetic transmitter.

Bell soon created an improved version of the electromagnetic transmitter. The mouthpiece consisted of a stretched membrane to which was attached an iron armature. An electromagnet was placed behind the membrane. As someone speaks, the soundwaves emitted by their voice cause the membrane to vibrate. The vibrating iron armature, in turn, induces a current in the electromagnet. This current was transmitted along a conductor to the receiver, which converted the varying current to sound using the exact opposite process (i.e.: A varying current in an electromagnet causes a diaphragm to vibrate, producing sound). The patent application submitted by bell is shown on the right.

The First Commercial Instruments

The rest, as they say, is history. Bell would continue to make pubic demonstrations of his telephone through the late 1870s. Bell and Watson made the longest call yet, a distance of two miles, on October 9th, 1876. More inventions would be needed before the telephone became a commercial device. In 1877, Thomas Edison filed a patent for an improved transmitted, that used an independent power source, rather than just the energy contained in the user's voice, to power the device. This sparked a competitive race between Edison and Bell.

The spread of the telephone was slow at first. The first exchanges began in Britain, where by the end of 1878 there were 200 subscribers. Initially the telephone was the purview of the rich, costing 20 pounds a year (when the average yearly wage was 80 pounds or so).

The first telephone companies were set up, consisting of public "exchanges", where one could come to and use the telephones if a subscription fee was paid. By the 1890s, automatic exchanges started cropping up, as well as more sophisticated telephones, as pictured on the right. It would only be a matter of time before the telephone became a staple of modern life.

Perhaps one of the most interesting aspects of the invention of the telephone is the shear amount of litigation involved. Decades-long legal battles were fought by Elisha Gray and Alexander Graham Bell (who beat Gray to the patent office by several hours, although this is hotly contested) and between Bell and Edison, among others. While Bell was a visionary inventor, his legacy stands above the others partly due to the fact that he was also an articulate and astute businessman, who was able to secure funding and had well positioned business partners.

Interesting Links
Library of Congress: Alexander Graham Bell
Telephone History
Connected Earth: The Telephone

Alessandro Volta [1745 - 1827]

2008-07-27T16:43:00.005-04:00

Introduction

These days, its difficult to go for more than a few hours without using some sort of electronic device powered by a battery, whether it’s a cellphone, laptop, mp3 player, or a remote control. Batteries are without a doubt a critical part of our modern world, but whom do we have to thank for their invention? It turns out that the letters of gratitude can be sent to an Italian physicist, Alessandro Volta (that is, if you could send letters back in time, Volta passed away in 1827).

While the battery is important now, it was perhaps even more critical when Volta first introduced it in 1800. His first batteries, or ‘voltaic piles’, provided a steady, constant supply of electricity that proved to be crucial in some of that era’s most important electrical experiments that established the field. It’s important to realize that at that time, accessing electricity was not a trivial task. There were no power plants or electrical power lines criss-crossing the countryside. In fact, electricity in general was a relatively recent discovery and was still poorly understood.

Childhood

Young Volta had the advantage of a high birth, his parents were members of the Lombard aristocracy, who had close ties with the church. At an early age, Volta did not show any extraordinary ability, but quickly began to distinguish himself as he progressed in his schooling. Tragically, his father died when he was seven, and his raising was entrusted to an uncle. Alessandro Volta’s uncle and teachers attempted to persuade him to join the priesthood, however he resisted, and elected to become a physicist. At the age of 14, Volta finished grammar school and ended his formal education, instead undertaking study electrical phenomena that were sparking interest in scientific circles at the time.

Scientific Career

The beginning of Alessandro Volta’s is an interesting one. He began exchanging scientific correspondence with some of the major authorities on electricity at the time at the age of 18, without any university preparation. His first, officially published work was his 1769 dissertation, entitled “On the Forces of Attraction of Electric Fire”. He continues writing, and eventually becomes a Professor of Experimental Physics, at Como Grammar School. Finally, in 1778, he is appointed to the chair of Experimental Physics, at the University of Pavia, a position he would occupy for the next 25 years, where he would develop some of his most important scientific contributions.

The Voltaic Pile

Volta would not develop the voltaic pile, until 1800, when he was 55, and near the apex of his scientific career. The voltaic pile was born out of a heated debate between Volta and Luigi Galvani, who discovered that the muscles of dead frogs twitch when exposed to a spark. Galvani and his supporters termed this “animal electricity”, and believed that electricity came from the muscles, and was not separable from a biological organism. The debate began in 1792, when Volta argued that electricity was a physical phenomenon, and the frog legs simply acted as a detector.

Volta started his work by replacing the frog legs with brine-soaked paper, through which he was able to detect the flow of electricity using instruments he had developed earlier in his career. He would continue his experiments, ultimately discovering that electricity can be generated by separating two dissimilar metals by an electrolyte (a substance that contains free ions). After much trial and error, he discovered that the most effective pair of metals to produce electricity was zinc and silver.

Volta would go on to create the voltaic pile, or the first electric battery. He did so by alternating layers of brine-soaked cardboard sandwiched between copper and zinc electrodes, as pictured on the right. By attaching a wire to either end of the pile (the terminals), the battery created a steady source of current.

Volta’s invention would make him famous world-wide. He would demonstrate the invention to some of the most famous people of his day, including Napoleon Bonaparte. Volta would end up receiving many honours for his ground-breaking discovery, including being made a count in 1810 and a Professor of Philosophy at Padua University in 1815.

Personal Life and Legacy

Perhaps the most striking thing about Volta was his humble and unassuming personality. Indeed, the fact that he gradually accumulated his research, positions, and honours speaks of a man who was humble and methodical. Volta would not marry until he was nearly 49, but would end up having three children. By 1813, Volta had ceased his research work, because of attachment to family and increased political involvement. In 1819 he finally retired to his country home in Camnago, where he died on March 5th, 1827 at the age of 82.

In 1881, the unit of electric potential was named in Volta's honour, being designed the volt (V).

Interesting Links
Alessandro Volta Biography (many interesting pictures of his inventions)

Positron Emission Tomography: Part III: Applications

2008-07-12T17:08:00.005-04:00

Introduction

This article is Part III of a three-part series. See Part I, Part II, or Part III for more.

Positron Emission Tomography (PET) is a nuclear imaging method that creates
three dimensional maps of processes in the human body. PET can be used to track the rate of flow of blood, study the uptake of sugars to identify tumours, show a 3-D picture of brain activity and can be used to diagnose a variety of diseases, including Parkinson’s Disease and Alzheimer’s Disease.

Part I of this series covered the history and physical principles of positron emission tomography. Part II covered the devices and methods used to detect gamma rays, and exactly how 3D images are reconstructed from a series of measurements by a PET scanner. Part III covers the applications of PET in medicine and future trends in research.

Applications of PET in Medical Imaging

In medicine, positron emission tomography has wide applications in oncology (the treatment of cancer) and neurology (disorders of the nervous system). In general, to create a PET scan image of the human body, a radioactive isotope is chosen and bound to a tracer compound that has a high rate of uptake in the organ or tissue of interest. The tracer compound is injected into a patient who then passes through the PET scanner’s detector ring. The PET scan typically takes about an hour to complete.

Advantages of PET over MRI and CT

The advantage of PET scans to other forms of imaging, such as CT scans, is that a PET scan images the metabolism, while CT scan and MRIs image anatomical features. For this reason, PET is often called “functional imaging”. The PET scan can detect a small metabolic change long before it produces a noticeable change in anatomy.

PET in Cancer Treatment

PET is interesting to doctors trying to treat cancer since malignant tumours have a different metabolism than normal cells, specifically a much faster uptake of sugar. To detect a tumour, the radioactive isotope 18F is bound to fluorodeoxyglucode and injected. A PET scan is used to detect the presence of a tumour by contrasting the rate of uptake of glucose between the tumour and surrounding normal tissue. PET scans can also be used to measure the rate of proliferation of tumour cells, and hence assess their grade (or severity) non-invasively. An image showing the contrast between a CT and PET images of a tumour is shown below, with the CT scan on the left.

[image: see ref. [1] ]

PET in Neurology

In neurology, PET scans are an important tool for diagnosis of Parkinson’s disease, epilepsy, and Alzheimer’s disease. Radioactive tracers in these studies include 15O bound to water, 11C bound to flumazenil, and 18F bound to fluorodeoxyglucose. The rate of uptake in the brain is a good indicator of brain metabolism, which is believed to be correlated with neurological activity. A PET image, therefore, has characteristic signatures that enable the diagnosis of these diseases. A PET scan showing diagnosis of Alzheimer’s disease is shown below. Normal brain activity is depicted on the left, with an Alzheimer’s patient activity shown on the right.

[image: see ref. [2] ]

Other Applications of PET

PET scans have also been used to measure the rate of flow of blood and other fluids in the body, protein synthesis, the presence of abnormal proteins such as neurofibrillar plaques, and to determine the number of receptors and transport proteins on cell membranes.

Future Trends in Positron Emission Tomography

Positron emission tomography is a constantly evolving field with much research being performed both in its application in medical imaging for the diagnosis of diseases and the design of PET scanners. A promising new imaging technique is the fusion of CT and PET scanners to produce composite images. This is a natural evolution since the PET detection camera is also suitable for the detection of the x-rays used in CAT scans. PET-CT images show a superposition of both the structural and biochemical details, allowing for quicker and more accurate diagnosis of diseases.

Research also continues into improving the capabilities of PET scanners. Recently, researchers at the Israel Institute of Technology have developed a method to increase the resolution of PET scanners by superimposing and filtering four slightly shifted images of the same area. This technique leads to what is called Super-Resolution PET.

Further studies in PET imaging also include incremental corrections for errors introduced by patient movements. Since a PET scan takes upward of an hour, the patient who is being scanned will never remain completely motionless. Researchers at the Royal Prince Alfred Hospital in Sydney, Australia have developed an optical motion tracking system capable of correcting for patient head motion in six degrees of freedom. Motion correction leads to increased spatial accuracy and reduced distortion in the resulting images.

Conclusion

PET is a versatile tool that is of great aid in the diagnosis of disease. PET is also an important research tool, especially in neurology research. Continued technical improvements are researched in PET scanners that include improvements to gamma detectors, error correction methods, spatial resolution of the scanner itself, and image reconstruction speed and accuracy.

Other parts of the series: Part I, Part II, Part III

References

[1] Miles K.A. et al., 2006 “Blood flow–metabolic relationships are dependent on tumour
size in non-small cell lung cancer: a study using quantitative contrast-enhanced computer tomography and positron emission Tomography” Eur J Nucl Med 33:22-28

[2] Various, “Pet Scans of Alzheimer’s Disease”, Thompson Cancer Survival Center, 2006.

Positron Emission Tomography: Part II: Operation

2008-07-12T15:50:00.007-04:00

Introduction

This article is Part II of a three-part series. See Part I, Part II, or Part III for more.

Positron Emission Tomography (PET) is a nuclear imaging method that creates
three dimensional maps of processes in the human body. PET can be used to track the rate of flow of blood, study the uptake of sugars to identify tumours, show a 3-D picture of brain activity and can be used to diagnose a variety of diseases, including Parkinson’s Disease and Alzheimer’s Disease.

Part I of this series covered the history and physical principles of positron emission tomography. Part II covers the devices and methods used to detect gamma rays, and exactly how 3D images are reconstructed from a series of measurements by a PET scanner.

Detecting Gamma Rays

In order to map the annihilation events occurring in the human body, a typical PET scanner uses a ring of gamma ray detectors around the subject being imaged, as shown in the figure below.

[image courtesy of UW Introduction to PET Physics]

Detecting gamma rays from an annihilation requires timing circuitry accurate to the order of nanoseconds. If two gamma rays are detected within this extremely short time window at opposite detectors, they are deemed to come from an annihilation event, since the emission of two gamma rays traveling in opposite directions is characteristic of electron-positron annihilation.

The detection of two anti-parallel gamma rays produced as a result of positron-electron annihilation is called a coincidence event. By keeping track of coincidence events, a PET scanner filters out gamma rays from annihilation with other gamma rays that may be produced in the decay of the radioactive isotope.

Noise and Sources of Error in PET

Unfortunately, several physical factors prevent a PET scan from recreating a noiseless 3-D image. The first is that positrons travel approximately a millimeter in the human body before slowing enough to annihilate. Therefore, the annihilation event will not take place in exactly the same region as the radioactive decay. Furthermore, because of variation in the momentum of the annihilating particles, the emitted gamma rays will not travel in a direction perfectly anti-parallel to each other. As a result, there is a certain angular uncertainty at the point of gamma ray emission. This translates to a position uncertainty of 2-3 mm in a typical PET scanner with a camera diameter of 1 meter. These factors limit the spatial accuracy of a PET scanner.

A further source of error results from the interaction of gamma rays with the human body through what is known as Compton scattering. The highly energetic gamma ray photons scatter off of electrons present in the atoms that make up the human body. This effect results in large changes in direction with a small decrease in energy. Compton scattering ruins the resolution of the image since the gamma rays will no longer be detected at anti-parallel detectors, giving a false position to the annihilation event.

Finally, a phenomenon called random coincidence can occur, when gamma rays from two different annihilation events land at detectors within the same time window. Random coincidence is particularly problematic for radioisotopes that have a high rate of decay. The errors described above lead to a decrease in contrast of the resulting image as well as an overestimation of the tracer concentration if they are not corrected. An illustration of the possible error sources is shown below:

[image courtesy of UW Introduction to PET Physics]

Camera Design

At the core of a PET machine is the camera: a block arrangement of detectors oriented in rings about the subject being scanned. The camera is designed to operate in one of two modes: 2-D mode or 3-D mode. In 2-D mode, each ring is separated by thin slats (or septa) of a material that does not transmit gamma radiation, such as lead or tungsten. With the camera operating in 2-D mode, coincidence events are only recorded in closely neighbouring rings. In 3-D mode, the slats are removed, and a coincidence event can be recorded in any ring.

The advantage of 3-D mode is that the camera is much more sensitive to detecting coincidence events and can produce a higher resolution image. However, the downside is that it also becomes more sensitive to random and scattered coincidences, which in turn requires more sophisticated filtering techniques. A diagram of the modes of the two modes of operation is shown below:

[image courtesy of UW Introduction to PET Physics]

The 3-D mode also requires more processing power in the image reconstruction phase, conversely however, the camera becomes less costly to manufacture.

Detector Design

The detectors most commonly used in PET cameras are scintillation detectors. In a scintillation detector, a high-energy photon such as a gamma ray interacts with the electrons in the scintillating crystal through processes called Compton scatter and photoelectric absorption. These interactions produce visible light, which is then multiplied several thousand times in a photomultiplier tube, before finally being turned into an electric signal. The most commonly used scintillating crystals in PET machines are Sodium Iodide (NaI) and Bismuth Germanate (BMO).

The limitations on detector size however, add statistical noise to the measured signal.
In an infinitely sized detector, the incoming gamma ray would eventually deposit all of its energy in the detector through multiple scattering, resulting in a peak about its energy. In a finitely sized detector, the gamma ray may escape before depositing all of its energy, resulting in a so-called Compton edge, as shown in the figure below. Also, all detectors

have a limit as to the rate at which photons may be detected. After a scintillator detects a gamma ray, there is a finite period of time for which it will not be able to detect another pulse. The amount of time for which the detector is not active is called dead time, and adds statistical noise to the measured counts.

Image Reconstruction

Once the number of coincidence counts have been acquired and filtered to reduce the effects of random and scattered counts as well as statistical noise, the most daunting task in the PET process begins: image reconstruction. In image reconstruction, the data collected by the detectors has to be turned into a coloured 3-D image that is accurate and as noiseless as possible to allow for its use in the diagnosis of diseases.

There are a multitude of algorithms that have been developed to tackle the problem of constructing a three dimensional image from acquired count data. The most common reconstruction algorithm is Filtered Back Projection (FBP), which is relatively straight forward to implement, however has a tendency to amplify the noise in the resulting image. Extensive research has been done in PET image reconstruction resulting in algorithms in a multitude of algorithms to reduce noise and increase computation speed.

The simplest reconstruction possible in PET imaging is with the camera operating in 2-D mode and reconstruction with the filtered back-projection (FBP) algorithm. The 2-D mode essentially gathers planar slices of the subject. The line that connects two detectors at which anti-parallel photons from a coincidence event have been captured is called the line of response (LOR). In 2-D mode, the LORs are arranged into a set of parallel projections, forming a set of intensity profiles. An estimate of the original source can then be created using a technique called back projection.

In back projection, the intensities originally recorded in the forward projection step are summed over the image plane, interfering constructively where the original source was located. Back projection results in star-like artifacts around the reconstructed source, which are then smoothed using a filter. The FBP algorithm is repeated for each image “slice” and reconstructed to form a three dimensional PET image.

Conclusion

This article covered the techniques used to acquire gamma ray emissions from positron annihilation, camera and detector design, as well as image reconstruction techniques to create a 3D image from PET measurements.
Part III of the series will cover applications of PET in medicine, as well as future trends.

Other parts of the series: Part I, Part II, Part III

Positron Emission Tomography: Part I: Principles

2008-07-12T12:26:00.008-04:00

Introduction

Positron Emission Tomography (PET) is a nuclear imaging technique that creates
three dimensional maps of processes in the human body. PET can be used to track the rate of flow of blood, study the uptake of sugars to identify tumours, show a 3-D picture of brain activity and can be used to diagnose a variety of diseases, including Parkinson’s Disease and Alzheimer’s Disease. PET scans give doctors a unique tool to quickly and accurately diagnose cancers and keep track of the effectiveness of cancer treatment. A typical PET scanner used in a clinical setting is shown below, this model having been developed by Philips Medical Systems.

[Image Courtesy of Philips Medical System]

A PET process consists of several steps: a radioactive isotope is bound to a compound taken up by the human body, such as glucose. It is then injected into a patient and begins to emit positrons as the substance decays. The positron emissions are detected using a ring of detectors and a computer reconstructs this data into a three-dimensional image that is used by a doctor to make a diagnosis.

History

Positron Emission Tomography is based on some very important consequences of relativistic quantum physics. The existence of a positron was first suggested by Paul Dirac in the 1920s when he successfully explained the origin of the electron’s spin and magnetic moment. However his theory had one major drawback: it required the existence of an antiparticle for every particle. At that point, the existence of antiparticles had not been thought possible. It wasn’t until 1932 that the positron was experimentally detected by Carl Anderson, who received a Nobel Prize for his discovery. The foundation of PET techniques was laid; however it wasn’t until computers became powerful enough to process the large amount of data required that they became a reality.

Physical Principles

In Positron Emission Tomography, a positron emitting radioactive substance, called a radioisotope, is injected into a patient. The radioisotope is typically bound to a chemical compound that is readily taken up by the tissue of interest, more so than other surrounding tissue. As the nuclei in the radioactive substance decay, they emit positrons: particles with the same mass and rest energy as electrons, but with opposite charge.

As the emitted positrons travels through human tissue, they gradually give up their kinetic energy through Coulomb repulsion with the electrons of all the other compounds present in the body. When a positron slows down to what are called thermal energies (approximately 2200 m/s or less), it annihilates, recombining with an electron in a violent process that converts their mass into pure energy in the form of two gamma ray photons. The gamma rays each have energies of 511 kilo-electron volts and travel in opposite directions.

It is the detection of these gamma rays using appropriately configured detectors which allows PET scanners to accurately map the location of positron-electron annihilation in 3-D space, and hence the location and concentration of the radioactive tracer. A typical PET image of the human body is shown below:

[Image Courtesy of [1]]

Conclusion

PET imaging is a versatile tool used in hospitals around the world to diagnose diseases, as well as in research to better understand the functioning of the human body. Part 1 of this three part series covered the history and physical principles of Positron Emission Tomography. Part II and part III cover the details of PET scanner design and applications where it is used in medicine, as well as some future trends.

Interesting Links:
Philips Medical Systems
UW Introduction to PET Physics

References:
[1] Dittmann, H. et al., 2003 “¹⁸F FLT PET for diagnosis and staging of thoracic tumours” Eur J Nucl Med Mol Imaging 30 1407–1412

How Digital Signal Processors (DSP) Work

2008-07-08T19:57:00.004-04:00

Introduction

Everyone is familiar with computers. Since the early 1990s, a computer can be found in practically every home and business. Indeed, it is often difficult to go more than a few hours without using a personal computer (PC). Likewise, the names of companies such as Intel and AMD have become household names, and most consumers are familiar
with the speed of their PCs (typically several gigahertz) and that at the heart of every PC is a chip called the microprocessor.

Embedded Processors

However, there is an entire class of computers running in the background in practically every electronic device we own and operate, plus an even greater amount in systems that facilitate the devices we do run (such as cellphones, telephones, televisions) that we never see. In fact, the number of these embedded systems greatly outnumbers the number of PCs in existence today. Over the years, engineers have developed entire classes of specialized microprocessors that they use for specific tasks. These processors are typically smaller, slower (in terms of clock speed), and cheaper than the general purpose chips found in PCs. However, by specializing on specific tasks and utilizing special and optimized architectures, they often are much more efficient at the tasks they were designed for than even a high-powered PC.
[Image Courtesy of Wikimedia Commons]

Digital Signal Processing Chips

One such class of chips is the Digital Signal Processor, or DSP chip. DSPs are specifically designed to process signals. Since practically everything we do with electronics involves sending, receiving, and acting on some sort of signal, you can easily imagine that these chips are used in a huge number of devices. What sorts of signals do our devices process?

Audio signals, such as playing music on an MP3 player or playing an audio CD
Digital television (eg: HDTV)
Telephone and internet data
Fax signals
Wireless communications (cell phones, wireless internet, etc.)
Radar, sonar, etc.

DSP chips are therefore found in a wide array of devices, such as:

Cellphones
Digital TV/Digital Cable/Satellite TV receivers
Digital cameras
Sound cards

This also explains why engineers who specialize in DSP are in high demand today.

DSP Chips: Principles of Operation

Now we can answer the question: why are DSP chips much faster at signal processing than more generalized microprocessors? There are two main reasons: one, they have been designed to be very efficient at high speed arithmetic operations (addition, subtraction, multiplication, division) which are the backbone of all signal processing algorithms. Two, these chips are designed such that they can perform multiple arithmetic operations at a time, for every program instruction. Also, the chips are designed to address multiple sites in memory at once, allowing them to fetch and store lots of data at once. This is a key feature, since signal processing involves dealing with a lot of data.

DSP chips also typically have many small bits of internal memory, right in the processor, called registers that can store data temporarily. This is again important since many signal processing algorithms require the designer to know a previous value of a calculation to calculate the new value. Having so much memory right on the chip cuts out the time needed to transmit information back and forth from external memory, like RAM modules, which can significantly delay processing speed due to the length of time (or latency) that RAM has compared to on-chip memory.

DSP Chips Manufacturers

There are several key manufacturers of digital signal processing chips today. One of the biggest, and a pioneer in the field, is Texas Instruments. Other important companies include Freescale Semiconductor, Analog Devices, and Microchip. These companies are directly involved in designing and optimizing the individual DSP chips which are later utilized by engineers at other companies that embed them in the devices they are working on. Therefore, there is a symbiotic relationship between these companies, as the chip makers are influenced by the applications their chips can be used in, while application engineers can use innovations introduced by the chipmakers to develop newer and more efficient devices.

Interesting Links
Texas Instruments: What is DSP
Analog Devices: A Beginner's Guide to DSP

Building Blocks of Electricity: What Substances do We Use?

2008-07-06T20:41:00.008-04:00

Let’s start at the basics and answer the question: what is electricity and how does it make our world work? Don’t stop reading just yet if you think this article is below you, you might just learn something new or see what you already know in a different light.

Charge, Atoms, and Electrons

Quite fundamentally, electricity is caused by the movement of charge. To understand this, we have to delve to the very core of chemistry and physics. Every object in our universe is made up of atoms. These tiny microscopic objects combine in various forms to produce the oxygen we breath, the salt we put on our food, and the metal wires we use to conduct electricity.

Each atom in turn, is made up of a mixture of smaller charged particles. Roughly, we can picture an atom as consisting of a big core of two types of particles: protons and neutrons. Indicated by the red and blue spheres in the image to the left. Neutrons have no charge, while protons have a positive charge. Around this condensed core orbit much lighter charged particles called electrons. Just how small? Protons are roughly 10 000 times more massive than electrons.

Each different type of atom, or element, has a different number of electrons (and protons and neutrons) orbiting around the core in different configurations. At this point, its best to think of the whole thing as a tiny version of our planetary system, although this isn’t quite true, but it definitely helps to visualize it, as shown in the image on the left. All of the different possible elements known to mankind are enumerated in the periodic table of the elements. [Image courtesy of Wikimedia Commons]

Movement of Charge

Some of the electrons in an atom are lightly bound and can become detached from the core. The movement of these free electrons, or bits of negative charge, are what makes all electronics work. At this point, you might be asking why we can only use certain substances for electronics, and not others?

Conductors, Insulators, and Semiconductors

We can group all of the different substances into three broad categories: conductors, semiconductors, and insulators. Conductors have an enormous number of free electrons, or those that are no longer bound to the nucleus. Because they move around in a substance (such as a copper wire) easily, we can use them to send electricity jetting off to wherever we desire with a minimum of fuss. That’s why power transmission lines and wires are made up from metal. Insulators have few free electrons, and because of this, don’t conduct electricity very well (no moving electrons, no electricity).

The most important class of conductors for our purposes are metals. An amazing thing happens with metals that we can use to our great advantage. Due to the details of how the metal atoms chemically bond with each other, as they combine into a solid their outer electrons become detached from the individual atomic cores. This results in an enormous number of free electrons, constantly bouncing around at incredible velocities inside the metal. Its best to picture the free electrons as a sea of particles, ready to conduct electricity when needed.

In between are semiconductors. They have some free electrons, but not enough to be as good at conducting as metals. Seems rather boring and useless right? However, semiconductors make up the most interesting and useful class of materials. That’s because by selectively combining semiconductors with other elements, we can do some pretty amazing things with electricity. We can use them as tiny electronic switches, from which come all of the digital circuitry power our computers. We can also use them to make other electric signals stronger, or in other words as amplifiers. That’s how the tiny electric signal caught by the antenna on your stereo becomes a booming bass sound as it comes out of your speakers. An image of a particularly important semiconductor, silicon, is shown below:

That’s essentially the gist of it, although there are entire books that could be (and have been) written on this subject. When you get down to the details, some very amazing things happen with materials and electrons that require us to delve into quantum mechanics. If we really want to understand what makes up our charged particles, we’d have to pick up a book on particle physics and learn about the quarks and other strange sub-atomic particles out there. Don’t worry though; you don’t need an encyclopaedic knowledge of the minutia to understand how electricity works.

How Bluetooth Works

2008-07-06T10:55:00.009-04:00

Introduction

Bluetooth devices are fast becoming ubiquitous. Almost every new laptop sold is equipped with Bluetooth. Likewise, Bluetooth-enabled wireless mice and headphones are springing up everywhere, as are the wireless Bluetooth headsets for headphones (style notwithstanding on the last one). But what exactly is Bluetooth? In short, it’s a form of wireless network designed for transmitting data (such as voice, messages, etc.) over short-ranges between Bluetooth-enabled devices. Its main goal is to replace the cables connecting portable and fixed devices.

Key Features

The key features of the Bluetooth system, and the reason it is so useful, is that the technology was designed to be very robust, consume little power, and be low-cost. This allows Bluetooth communication to be embedded in devices without drastically increasing their cost or consuming a lot of power. This is important both from a standpoint of battery size, battery life, and heat generation. Bluetooth was also designed from the bottom-up with security in mind.

Origins of Bluetooth

But how exactly did Bluetooth come about? Bluetooth is simply a standard, an agreed-upon technology developed by a group of companies interested in adopting it for their own products. Standards are important in the world of electrical and software engineering, since they lower development costs for everyone involved and increase the likelihood that the technology will be adopted by others. The founding members of the Bluetooth standard were Ericsson, IBM, Intel, Microsoft, Motorola, Nokia, and Toshiba.

What does Bluetooth even mean? Strangely enough, the technology was named after Harald Bluetooth, the tenth-century king of Denmark and Norway. The makers of Bluetooth drew a parallel between Harald, who unified the various warring tribes of Denmark and Norway, and the Bluetooth technology, which was designed to unify different devices through wireless networks. An esoteric choice, to be sure, but one that also highlights the strong influence of Scandinavian countries on the development of wireless technology.

Bluetooth vs. WiFi

It is almost impossible to talk about Bluetooth without mentioning WiFi (or less poetically, Institute of Electrical and Electronics Engineers (IEEE) Standard 802.11). Anyone who has a wireless internet connection at home is familiar with WiFi. Like Bluetooth, WiFi is a wireless networking technology designed for transmission of data. However, WiFi was designed strictly as a replacement for wired computer networks. Therefore, WiFi equipment is much more costly and consumes far more power. However, it has a greater range and can transfer a larger amount of data each second.

Transmitting Data Wirelessly: The Radio Link

At the risk of stating the obvious, the key to Bluetooth’s operation is transmitting digital data to other devices wirelessly. While this might seem obvious, the task of encoding digital data into a radio wave is not a trivial task at all. There are many ways to do so, but Bluetooth technology utilizes a procedure called Phase-Shift Keying Modulation (PSK). Since this topic is so involved, it warrants an article to itself.

Forgetting the minute engineering details of wireless transmission, we can still talk about just how Bluetooth technology works. First, let’s assume a bunch of Bluetooth devices are in proximity to each other, let’s say your cell phone and your wireless headset. The two devices will establish a physical radio link between each other. The devices will send out radio waves at a frequency of 2.4 GHz.

What exactly does 2.4 GHz mean? If you look at the radio wave pictured below, it means that one period of the wave (denoted by the greek letter lambda) will fly past in 4.1x10^-10 seconds. That might seem blazingly fast, but radio waves actually oscillate fairly slowly, visible light waves oscillate many orders of magnitude faster, let alone real fast movers like x-rays and gamma rays. Of course, all electromagnetic waves, including radio waves and visible light travel at the same velocity: the speed of light, or approximately 300 000 000 m/s.

[image courtesy of Wikimedia Commons]

Frequency Hopping and Frequency Bands

The choice of Bluetooth's transmission frequency, 2.4 GHz is an important one, because the government assigns different frequency bands to different purposes. Putting a device on the wrong band can cause a lot of unwanted interference, not to mention drawing the ire of regulators. Bluetooth operates in a frequency range called ISM, which is allocated for industrial, scientific, and medical devices.

Once the Bluetooth devices connect, they become synchronized with each other. Once that happens, Bluetooth has one more neat trick to play. It employs a strategy called frequency hopping. The connected devices will “hop” through different frequencies together in a synchronized pattern, around the main 2.4 GHz frequency. Bluetooth does this in order to minimize possible interference that might occur from other devices emitting in the same range.

Grouping Together: Piconets

Once two or more Bluetooth devices are within range and detect one another, they group together into a network. The architecture of the network is such that one of the devices acts like a master, regulating the behaviour of the other devices, called slaves. This type of network is called a piconet. Going back to our cellphone/headset example, the phone will act as a master, with the headset being the slave.

[image courtesy of Bluetooth.com]

Sending Data: Packets

Once the Bluetooth devices discover each other, then can begin talking. Now, we already talked about the physical problem of establishing a radio link between devices, and the naïve approach would be to just starting sending bytes of data in a stream between the devices. But things are never so simple, and engineers always have to consider the possible ways that their systems can fail. In this case, what happens if some data is lost, due to interference? Also, what if a device connects to several other Bluetooth devices, like in our piconet, and has to receive information from multiple sources?

This is where a concept from information technology, called packets, come into play. A packet is basically a block of data. It includes two main parts: a header, which describes what the packet is and where it is going, and a payload, which includes the actual data being sent. This has several advantages. The network becomes much less prone to interference, several devices can be serviced at once, and it is much easier to implement error detection and correction. The structure of a typical Bluetooth packet is shown below.

[image courtesy of Bluetooth.com]

Conclusion

Well, that about covers it. Hopefully this article was useful in explaining the basics of Bluetooth wireless technology. Bluetooth is fairly intricate, and we had to touch on several topics that could span entire books in their own right, like radio transmission, information technology, and networking. Bluetooth is a fast growing technology, especially as consumers continually look for new ways to cut the wires restraining their technology.

Interesting Links
Bluetooth.com

Milestones: The First Transistor [1947]

2008-07-06T10:44:00.004-04:00

The invention of the transistor changed the world and sparked the electronics revolution we are still experiencing today. Many of today's modern accomplishments can be traced back to the transistor, such as computers, the internet, manned space flight, and portable music players, among countless others. Today, billions of transistors are manufactured every week.

To fully understand the significance of this accomplishment, we first have to understand what a transistor is. Briefly, it can be thought of as an electronic switch. A controlling voltage applied to one terminal of the device determines wether an applied current will flow across it. This explanation is a bit simplistic, since a transistor can operate in other modes (for example to amplify voltage and current), but the switch description is how transistors are operated in digital circuits, such as computers.

Transistors replaced bulky devices such as vacuum tubes and electromechanical relays. Transistors were orders of magnitude more reliable, consumed far less power, and could be made much smaller. How much smaller? Vacuum tubes and similar components were typically at least several centimeters in size. Modern transistors on integrated circuits such as computer processors are often tens of nanometers in size. So, transistors can be about 10 000 000 times smaller. An image of a vacuum tube and modern discrete transistors are shown below, the difference is quite striking:

[images courtesy of Wikimedia Commons]

The first transistor is widely attributed to have been created in Bell Labs on December 16, 1947. The invention is attributed to three engineers: William Shockley, John Bardeen, and Walter Brattain. The first commercial devices using transistors, small transistor radios, started appearing on the market in the early 1950's.

Interesting Links
PBS: Point Contact Transistor
Bell Labs: Top Ten Innovations

How Noise Cancelling Headphones Work

2008-07-05T09:27:00.004-04:00

Noise cancelling, or Active Noise Reduction (ANR) headphones have become ubiquitous. First developed to protect the hearing of pilots, they can now be purchase from several companies, including Bose, Sony, and Sennheiser for a few hundred dollars. The technology may seem new, but the first papers and patents about the technique appeared in the 1930’s. However, it wasn’t until the 1980’s, when electronics technology had become sufficiently advanced, that the first practical devices started appearing. This article traces out the theory and implementation of noise cancelling headphones, their limitations, and just why it’s not quite yet possible to have a noise cancelled house.
[image courtesy of Bose]

Sound as a Wave

The first piece of the puzzle in understanding how noise cancelling headphones work comes from understanding the wave nature of sound. Just like the sinusoid pictured below, sound is a wave. However, a sound wave is formed from a series of compactions and rarefactions of air molecules that transmit the sound energy from the source to our ears.

Likewise, just like the sinusoid, sound waves have a frequency. We are accustomed to hearing sounds with a low frequency as a bass-y hum and sounds with high frequencies as a high-pitched squeal. Just think of nails scratching across a blackboard (high frequency) versus the hum of your car’s engine (hopefully low frequency, if it isn’t, its time to see a mechanic!).

The picture becomes a bit more complicated when we consider the fact that most sounds consist of a mixture of sound waves, each with different frequencies and amplitudes. To really make things a mess, we also have to think about the fact that sound waves are typically three-dimensional and interact in complicated ways with the environment. The above reasons are actually some of the problems keeping active noise reduction technology from truly exploding into the mainstream, but we’ll come back to that later.

Cancelling a Sound Wave

The next piece of the puzzle in how noise cancelling headphones work is answering how sound waves can be cancelled. Let’s go back to our simple sinusoid model of a sound wave. Now, consider if another sinusoidal sound wave, exactly 180 degrees out of phase with the original, that is, the peaks overlap the troughs, suddenly overlapped our wave. Just as in the figure below, the result would be the two waves summing to zero. This phenomenon is called destructive interference.

Perhaps surprisingly, the same principle applies when trying to cancel out unwanted sound. If a speaker produces the same sound, but perfectly out of phase with the original, the two sound waves will cancel, as along as they also perfectly overlap in space. This is the founding principle on which devices such as active noise reduction headphones are built. Of course, what sounds simple in principle is often very difficult to implement in practice, as is the case here. In general, ANR headphones can only cancel out a narrow band of frequencies, typically in the lower end of the audible sound spectrum.

Inside Noise Cancelling Headphones

A typical noise cancellation system found in ANR headphones consists of a few key components. The first is a microphone inside the headset that records outside sound entering the headphones. The audio signal recorded is processed by a microchip, such as a digital signal processor, which determines the anti-noise signal that must be produced to cancel out the incoming sound wave. Finally, this signal is reproduced by a speaker, cancelling, or more typically reducing, the original noise.

Limitations and Difficulties

Most people that strap on a pair of noise cancelling headphones immediately notice that the technology doesn’t filter out sounds such as other people conversing or an ambulance siren. This occurs because the technology found inside ANR headphones is typically only really capable of cancelling out low frequency sounds, like the sound of an airplane engine when travelling by air.

The reason is that it’s very difficult to reproduce complex soundscapes with today’s technology. Remember, sound is three-dimensional, and it is very difficult to reproduce an anti-noise signal that is exactly the right shape and overlaps perfectly with the incoming sound, especially with only one speaker. Likewise, speakers themselves are not perfect, and often distort sounds. It is also difficult to sample the incoming noise in the case where it is structurally complicated. This would require a much larger number of microphones. There are other factors at play that make a good implementation of the technique difficult, such as the fact that echos can come into play, and there is a time delay in the electronics processing the incoming sounds. Needless to say, an engineer’s job isn’t easy!

ANR technology typically works best for sounds that are as close to one-dimensional plane waves as possible. The sounds that are the closest to this in reality are relatively constant, low frequency noises. That is the reason why your noise cancelling headphones won’t filter out speech. Fortunately, ANR headphone manufacturers have a trick left to play. High frequency sounds are actually dampened down quite well by passive noise cancellation, such as applying foam padding around the earmuffs. Active and passive noise cancellation techniques are brought together to manufacture noise cancelling headphones.

iPods and other MP3 Players

2008-07-03T17:23:00.011-04:00

Portable MP3 players, such as the Apple iPod, Creative Zen, and other music players, have dramatically changed how we listen to music. Fortunately, all MP3 players function in a very similar way under the cover, which makes this article broad in focus. To answer how iPods work, we’ll have to answer several deep questions: how is music produced, how is music stored digitally, and how it is reproduced, moving from digital memory to your headphones.
[Image courtesy of Wikimedia Commons]

Music as a Wave

First, we have to answer the question: what is music? Music is simply a sound wave, with certain instruments (and our vocal chords) emitting sound waves made up of the frequencies and intensities pleasing to our ears. Sound waves are formed when an object (such as an instrument) creates a sequence of compressions and rarefactions of air, or in other words, areas of high and low air pressure. (Its best to think of areas of high air pressure as having lots of air molecules squeezed together, those with low pressure have air molecules relatively spread apart).

Capturing a Sound Wave Digitally

The next step in the puzzle is figuring out how to capture a sound wave in a manner that can be stored on a computer, that is, digitally, as a sequence of 1’s and 0’s. This feat is accomplished through sampling.

The first step in sampling is turning a sound wave into an electrical signal that can be processed and manipulated by digital circuitry. This is typically done with a device such as a microphone. After some noise reduction and amplification, the signal is ready to be sampled.

In the end, sampling boils down to reading a continuously varying analog signal, like the voltage signal coming from a microphone, and taking a series of snapshots of its value at a certain time interval. This task is performed by a circuit called an Analog-To-Digital Converter (ADC). The value of the signal voltage coming into an ADC is turned into a binary number, which we’re free to store for later use, or process further. In this way, we turn a signal coming in from a microphone to a series of binary numbers. An example of sampling is shown below. The original analog signal is shown in grey, and the digital values captured at each sampling time are shown in red.

ASIDE: Quantization Error

Unfortunately, there’s one downside to analog-to-digital conversion. Analog to digital converters have a limited resolution with which to store the values at their input. This is limited by the number of bits the ADC uses to store the information. The end result is there is a certain amount of round-off when storing signals, resulting in an error between the stored value and the actual analog signal. This is called quantization error and can degrade the quality of the recorded signals, for example music. On the flip side, digitized music has certain advantages over its analog counterpart, such as nearly error free copying and reproduction.

We’re almost done, the last piece of the puzzle is encoding. In the previous paragraph, I mentioned that sampled music consists of a series of stored binary values. However, if we record any meaningful length of sound (or video for that matter), the amount of storage space required skyrockets. That’s why compression schemes are used to reduce the amount of information stored. Encoding schemes, such as MP3 and WMA, reduce the storage space required by discarding certain bits of information that are the most redundant via a compression algorithm.

The Music Player: Reproducing a Digital Sound

Now that we understand how music is stored digitally, we’re now ready to consider how to turn that digital information back into sound, to pump through our headphones and speakers. This is accomplished through a process called Digital-To-Analog Conversion (DAC), which is the heart of how MP3 players operate.

Recall that in sampling our signal, we took a snapshot of the value of our original signal at a certain time interval. To recreate the original signal, we just have to turn our digital data back into a voltage signal, and somehow fill in the missing space between samples. In practice, most DAC chips simply hold the voltage steady between successive samples (although other schemes are possibly). The mathematical details behind all this can be fairly intensive, and lead engineers to better methods to store and recreate signals, however for the purposes of this discussion we don’t need to concern ourselves with the nitty-gritty.

The recreated signal would look similar to the figure below. It’s not perfect, but if the samples are spaced closely enough together, it’s “good enough”.

Inside an iPod

Now that we understand how music is stored and reproduced in a digital device such as a portable music player, the contents of an iPod should be fairly easy to understand. One of the most important parts is the component to store the music. Now, there are different classes of devices, some use a small hard-disk drive, while others use solid-state memory, such as a flash chip (i.e: an iPod Video contains a hard disk, while a Nano contains flash memory).

Next, the device will typically contain a small microprocessor, very much similar to the chips driving your personal computer. However, these are typically smaller, slower, and much less power hungry. They also typically contain specialized architectures to more efficiently handle embedded tasks such as this, often including built in DSP capability right on the chip. The microprocessor will serve as an interface between the user and the device, as well as coordinating the actions of the other components.

The next very important chip will be the audio codec chip. Remember when I mentioned that audio signals are typically stored in a compressed format such as MP3 or WMV? Well, this chip will uncompress the encoded data being pumped to it from memory by the microprocessor. This is called decoding. Once the audio file is decoded, the codec chip will also perform the Digital-To-Analog conversion, and the signal is ready to send to your headphones!

Biographies: Harry Nyquist [1889 - 1976]

2008-07-03T17:14:00.006-04:00

Harry Nyquist was a Swedish born electrical engineer whose pioneering work laid the foundation for modern communications technology. Nyquist was born in 1889, at the dawn of the first commercial telephones and when the electric telegraph was only around 50 years old. Harry was born to relatively modest circumstances. His family owned a farm and had eight children, of which Harry was the 4^th. During his school years, Harry was encouraged by his teacher to immigrate to America, where there were more opportunities for someone with his skills and interests. He finally did so in 1907, and started an electrical engineering degree at the University of North Dakota in 1912, where he successfully earned a Bachelor’s and Master’s degree.

In 1915, Nyquist moved to East Haven, Connecticut, to start his Ph.D degree at Yale University. During his education, he supported himself by teaching and various summer jobs. Upon graduation in 1917, he started to work for the AT&T Department of Development and Research Transmission, which would later become Bell Labs. Bell Labs would go on to become one of the most important research institutions in the world. Groundbreaking inventions such as the transistor, cellular telephone technology, solar cell, laser, the touch-tone telephone, UNIX operating system and many others were pioneered there. Nyquist would stay at Bell Labs until his retirement in 1954.

Some of Harry Nyquist’s most important work was done in the 1920’s, inspired by the then current problems in telegraph communications. He is one of the core founders of the field of information theory, which seeks to explain how information can be transmitted and is the founding theory learned by electrical engineering and computer science students as they study information storage and transmission technologies.

One of Nyquist’s most important contributions was the so-called Nyquist Theorem, which explains how often a signal (such as sound, video, etc.) must be sampled by a digital system in order for it to be reproduced error-free on the other end. To put his originality and creativity into perspective, this landmark work was developed in the late 1920’s, many decades before computers and other digital devices were invented.

Nyquist was also instrumental in the development of the first fax machine, the first prototype machine being built in AT&T’s labs in 1924. He also made lasting contributions in the understanding of noise, and how it is introduced in circuits and communications channels. This topic is of prime importance to engineers as they design communications networks.

Harry Nyquist continued to work as a consultant after his retirement, before passing away in 1976. During his lifetime, he received numerous awards, including the National Academy of Engineer’s Founder’s Medal. Harry Nyquist is described by contemporaries as being quiet and reserved, but a great associate for close friends and colleagues [1]

[1] Hendrik W. Bode, “Obituary Statement: Harry Nyquist”

IEEE Transactions on Automatic Control, vol. AC-22, No. 6, December 1977

Milestones: The Integrated Circuit [1958]

2008-07-03T12:26:00.003-04:00

The invention of the integrated circuit ranks as one of the most important technological achievents of the Information Age. Prior to the invention of the integrated circuit, electronic components were manufactured as discrete components and then connected with wires. The resulting circuits were invariably large, even with the transistor replacing the vacuum tube. What's more, the cost of making more complicated circuits was prohibitive, consumed a lot of power (and hence also generated a lot of heat) and was a nightmare to maintain.

This was all changed by a young, newly hired engineer at Texas Instruments named Jack Kilby. Kilby was hired at Texas Instruments in the summer of 1958. Being a new hire, he did not have the priviledge of a summer vacation, and hence was left relatively alone while the rest of the staff were gone for the summer. He had the idea that instead of making electronic components such as resistors, transistors, and capacitors from different materials, one could make them all from the same material that transistors were made from: semiconductors. By integrating all of these components on a single piece of semiconductor (such as silicon), one could make the chips with the same methods as transistors. What's more, they could be much smaller. Since everything was on one chip, the reliability problem was solved since little external wiring was necessary. Working over the summer, Jack Kilby made the world's first integrated circuit. The first integrated circuit was a simple oscillator, outputting a constant sine wave. An image of the first integrated circuit is shown below:

Kilby demonstrated his invention to Texas Instruments managment in September, 1958. The integrated circuit would go on to revolutionize the world, giving us devices such as the computer microprocessor. He was later rewarded with the Nobel prize, one of the few award holders that did not hold a PhD degree.

(image courtesy of Texas Instruments)

Introduction

2008-07-02T18:48:00.001-04:00

Electronics are all around us. From the MP3 player currently plugged into your ears, to the computer you’re reading this webpage on, to the cell phone likely sitting in your pocket. The electronic revolution that started at the turn of the century with the wide spread of phones, electric light bulbs, and electricity generating stations accelerated widely after the invention of the first transistor in 1947. That revolution has allowed engineers to pack billions of microscopic switches, each smaller than the wavelengths of visible light, onto tiny chips that now aid humanity in every endeavour we undertake.

This website is the story of that revolution, and its purpose is to tell the remarkable tale of the amazing natural feats and engineering ingenuity harnessed into devices that we now take for granted. Besides just explaining how consumer electronics work though, we’ll go deeper and reveal the even more interesting parts that work behind the scenes that most people don’t even see. After all, for every cell phone call, a wireless signal is transmitted through perhaps dozens of cell towers set to relay that information. Behind every bit of information we see on the internet, data travels through hundreds of miles of fiber-optic cables, repeaters, and servers, routing it through the dense web that makes up the information superhighway.

While we’re at it, we’ll also look at just how electricity makes our devices tick, and what technologies, components, and tools engineers use every day to make the latest batch of technology. I hope you enjoy your visit and come back often to read more about how electronics work, and take advantage of our updated content.