CharlieMasi

How do Hall sensors work?

Wed, 30 Jul 2008 13:54:02 GMT

��The Hall effect, is a magnetohydrodynamic phenomenon where an external magnetic field affects the flow of charge carriers in a semiconductor in such a way that an electric field appears directed at right angles to both the magnetic flux lines and the current-flow direction.
��Magnetohydrodynamics (MHD) is the study of electrically conducting fluids in motion. In this case, the electrically conducting fluid is the ensemble of majority charge carriers in the semiconductor.
��For example, in P-type semiconductor material the majority charge carriers are positively-charged holes. A Hall sensor could be made from a rectangular thin film of P-type semiconductor with aluminum contacts laid down along its edges. One pair of opposite electrodes are connected to a source of dc current. The other pair connects to a sensitive voltmeter with a very high input impedance (electrometer).

��In a Hall effect sensor, a magnetic field pushes the majority charge carrier flow to one side, creating a voltage difference across the flow.

��The dc current will flow uniformly from the positive electrode to the negative electrode. This flow acts as a conducting fluid moving uniformly across the film. Since charge is uniformly distributed over the film, no voltage appears across the opposite pair of electrodes connected to the electrometer.
��If we now apply a magnetic field directed normal to the film��s surface, a Lorentz force appears whose magnitude and direction are everywhere given by the vector cross product of the local charge-carrier velocity and the local magnetic field. In other words, the moving holes are pushed across the film at right angles to the dc current flow.
��The holes therefore pile up along the edge next to one electrometer electrode (which sees a relative positive voltage), and away from the other (negative). The electrometer registers a potential difference proportional to the dc current and the magnetic field strength.
��Since such Hall effect sensors can be manufactured from virtually any semiconductor by ordinary semiconductor fabrication processes, they are ideal sensors for magnetic fields. All of the circuitry can easily be integrated along with the sensor, so very small, highly stable, and extremely sensitive sensors can be made. The chips need only four connections: two raw dc power leads (the regulation can be integrated on the chip along with the rest of the electronics), and two analog output leads. Depending on the application, it is possible to integrate a Hall sensor into virtually any semiconductor device that needs one.
��Note that the Hall sensor reports only the component of the magnetic field vector normal to the thin-film��s surface. It is quite insensitive to any other components. To report the field��s complete magnitude and direction requires three sensors oriented along three orthonormal axes.

��Three Hall sensors oriented along three orthonormal axes determine the magnetic field��s magnitude and direction.

What is a successive approximation ADC?

Wed, 30 Jul 2008 13:51:04 GMT

��Analog to digital converters (ADCs) are arguably the most common mixed signal circuit in use today. The successive approximation register (SAR) type is potentially the most accurate ADC circuit, although it is not the fastest.
��ADCs accept an analog voltage as input and provide a digital bit pattern as output. The digital bit pattern represents the value of the input after preparation by the measurement system. Preparation steps include signal conditioning, amplification, and quantization (sampling).
��Signal conditioning operations may include converting the signal from a current, frequency, or resistance value (depending on the sensor) to a voltage, limiting its bandwidth to prevent aliasing, and shifting its zero level. An instrumentation amplifier sets the dynamic range appropriately to contain the expected input signal variations. Finally, a sample-and-hold circuit stabilizes the voltage at the ADC input during successive sampling intervals.
��All of these preparation steps affect the relationship between the input signal level and the digital code put out. So, it��s important that the measurement system capture and report the various parameters, such as bandwidth, gain, and sampling interval, separately to allow reconstruction of the input value in engineering units.
��Typically, there will be a decoding matrix to convert the binary output to binary coded decimal (BCD), Grey code, or whatever the next processing step needs.
��The basic digital measurement system architecture involves signal conditioning, amplification, sampling, and, finally, digitization via the ADC.
��Thus, the ADC �� any ADC �� sees a stable dc voltage level at its input, which will be held steady for long enough for the ADC to complete its job. How long ��sufficient�� is depends on the type of ADC, its characteristics, and the required measurement precision.
��The SAR type ADC illustrates how these characteristics interrelate. It measures the input voltage by an interative process during which the ADC makes successive guesses (which it keeps track of in a digital register) at the input��s value, then compares its guess to the actual value, then makes a better guess, and so forth until it reaches a good enough approximation. At that point, it presents the register��s contents at the ADC output.
��SAR ADCs have a two chambered heart. Source: Wikipedia and Control Engineering
��SAR ADCs have a two-chambered heart. One chamber is the successive approximation register itself, which carries the system��s current best guess at the analog value. The other is a digital-to-analog converter (DAC) that puts out an analog voltage based on the register��s contents. A voltage comparator decides what the next guess should be.
��Process steps include:
��0: Set all register bits low.
��1: Set the most significant bit high.
��2: Compare the DAC output to the sample voltage.
��3: If the comparator output is high, set the current bit low; otherwise, keep the current bit high.
��4: Set the next bit high.
��5: Repeat steps 2 through 4 until either all register bits have been set or time runs out in the sampling interval.
��Notice that there are two ways to cut off the process. Either the system achieves the ultimate precision it is capable of, or it runs out of time.
��The ADC��s ultimate precision depends on the register��s length in bits. Thus, a 16-bit SAR ADC has a register 16 bits long. It therefore divides the ADC dynamic range into 65,536 levels. The ultimate precision is therefore one part in 65,536, or 15 parts per million (ppm), or 0.0015% of full scale.
��It takes, however, 65,536 iterations to reach that precision. What happens if the user can��t wait that long? He or she can speed the process up by sacrificing precision.
��The reason for cutting the process short is to achieve higher measurement bandwidth. If it takes, say, one microsecond per iteration, achieving the full 65,536 iterations will require almost 65 ms. The Nyquist�CShannon sampling theorem tells us that this will limit our useable bandwidth to about 7.5 Hz!
��Some applications, such as temperature measurement, can live with this limitation. Others, such as monitoring a 60 Hz sine wave, cannot. But then, many such applications don��t need 16-bit precision.
��Suppose we only need 10 bits. That��s 1,024 levels, which will take 1,024 iterations to achieve, which will require only about 1 ms, giving a useable bandwidth of 488 Hz. Still not impressive, but certainly enough to monitor that 60 Hz sine wave. If even less precision is needed, then we can cut the ADC off even sooner and get higher bandwidth. The rule of thumb is that you can double the bandwidth for every bit of precision you sacrifice.
��Many real SAR ADCs iterate much faster than once per microsecond, and so can achieve greater bandwidth with more precision than in this example. The principle, however, is the same. Every SAR ADC has a bandwidth-precision product (when bandwidth is measured in Hz and precision in ppm) that outlines its performance envelope.

Where can I pursue machine vision training?

Wed, 30 Jul 2008 13:48:48 GMT

��Machine vision is not a discipline in itself, such as mechanical engineering. It is a skill involving specialized equipment. How you should go about gaining that skill depends on your goals.
��If you simply want to gain proficiency with specific equipment in your shop (as is the case with the person who sent in this question), the best place to start is with the manufacturer of that equipment. Virtually all vision-system suppliers provide excellent user manuals and similar documentation in printed and online form. Some, such as Cognex and Dalsa also provide tutorial material, such as white papers and application examples, to give basic background needed to understand, interpret, and troubleshoot the system results. Some, such as National Instruments, also provide hands-on training classes that cover machine vision basics, applications, and more advanced topics.
��For more general background, consult white papers and tutorials available through Control Engineering��s Website. Control Engineering also provides articles, such as ��Measure Up with Machine Vision,�� on general vision topics important to control engineers. Find these by typing ��machine vision�� into the search engine on any Control Engineering Website page.
��If you wish to develop real machine vision expertise, start by reading one of the many good texts on machine vision, such as Machine Vision by Ramesh Jain et al, or Understanding and Applying Machine Vision, by Nello Zuech. It is best to obtain and read texts by different authors because each has specific interests and areas of expertise, which will be reflected in their books.
��Edmund Optics�� Website provides an extensive online library containing many authoritative articles covering technical aspects of machine vision. An advantage of this library is that the company has a long history retailing optical components to amateurs and professionals with a wide range of abilities and interests, and distributes equipment from several name-brand manufacturers as well as having its own world-class optics manufacturing facilities. The library thus contains articles at technical levels ranging from very basic to highly advanced and covering many application areas.
��Finally, a Google search on ��machine vision course�� turned up almost 3 million hits, including a number of online courses, such as Automated Vision Systems�� Fundamentals of Machine Vision. Other courses are provided by universities, such as the Massachusetts Institute of Technology (MIT).
��The topics you��ll need to familiarize yourself with are optics, lighting, electro-optics, data acquisition, computer image processing, and automation networking. The educational resources available are quite extensive. How to proceed depends strongly on your goals.
��In any case, it might be wise to start with free online white papers for background, along with practical experience with real equipment. From there, you can go as far as you want, even to becoming a recognized expert on machine vision.

What is a magnetic multipole?

Wed, 30 Jul 2008 13:47:01 GMT

��Mathematically, a pole is any geometric singularity. For example, the Earth��s north and south poles are places where lines of longitude meet in singularities. In other words, the longitudes of these poles are undefined. Actual, physical poles don't seem to exist. Black holes are theoretical singularities, but are saved by having an event horizon. (They may be singularities, but not in this Universe!) Electrons are saved from being singularities by wave-particle duality.
��In electromagnetism, the term ��pole�� refers to a point where field lines meet. Electric field lines, for example, meet at point charges. One point charge is called a monopole. Two charges form a dipole. Three form a tripole, and so forth. Of course, real charges are spread out over the surfaces of physical objects, so they aren't really singularities.
��Magnetic fields never have actual poles, either. They do, however, have regions of ��magnetic concentration�� that may appear like poles when observed from the outside. For example, the ends of a long solenoidal coil of wire carrying a current appear to be magnetic poles. In actual fact, these solenoid ends really only form approximately spherical concentrations with diameters equal to that of the coil. Inside the coil, the field is fairly uniform. Outside the coil, it appears to have two poles, so it is a dipole. It is useful to classify magnetic field geometries as if there really were poles.
��Physicists have found experimentally that, unlike electric fields, magnetic fields can only form shapes with even numbers of poles. That is, there are magnetic dipoles (2), quadrupoles (4), hexapoles, (6), and so forth, but there are no magnetic monopoles. This fact annoys some theoretical physicists no end.

��Magnetic fields take on only geometries with an even number of poles.

��Some years ago, I was involved in a series of development projects making useful multipoles out of anisotropic permanent magnet material. The term ��anisotropic�� simply means the material has a grain, like wood. It is easy to magnetize anisotropic magnetic material parallel to the grain (which is called the ��easy axis��), but impossible to magnetize it perpendicular. I know because we tried it. We cut a piece of anisotropic barium ferrite into a cube one inch to a side with the easy axis parallel to the sides. Then, we clamped it into a very large electromagnet with the electromagnet��s field perpendicular to the easy axis.

��Anisotropic magnetic material has a grain direction called the easy axis.

��We had to clamp it because unclamped, the cube spun to point its easy axis parallel to the field. A technician almost lost a finger trying to hold it by hand.
��Turning the electromagnet current up to several times what it would take to fully magnetize the magnetic material parallel to its easy axis had no discernible magnetizing effect across the easy axis. It just did not magnetize. The technicians continued to turn up the juice until finally the sample simply exploded! The pieces then spun to reorient their easy axes parallel to the electromagnet field, magnetized completely and flew to the electromagnet��s pole pieces.
��We used this material, as well as other anisotropic materials, such as samarium cobalt, to create a number of ring-shaped multipole structures. For example, we built magnetic quadrupole lenses to focus proton beams in the world��s first privately owned linear accelerator. We also designed dipoles to steer charged particle beams around corners.
��Anisotropic material was important for these devices because it has the unique property of steering magnetic field lines. The figure below shows a structure we designed and built for magnetic resonance imaging (MRI). Unlike conventional dipole magnets, this device concentrates all of the magnetic force lines into the ring��s interior, greatly increasing the field��s strength and uniformity. The field outside of the structure is almost non-existent.

��Carefully designing structures made of anisotropic permanent magnet material makes it possible to have exquisite control of the magnetic field shape.

How does a digital filter work?

Wed, 30 Jul 2008 13:43:51 GMT

��Digital filters capitalize on the fact that digital electronics looks at waveforms as a timed sequence of values, rather than instantaneous voltages. Digital filters start with an analog waveform, convert it to a digital waveform, perform operations on the digital values, then convert the waveform back to an analog waveform.
��The figure below shows the steps in this process. The analog waveform is a time-varying voltage called the baseband signal. Passing this baseband signal through an analog-to-digital convertor (ADC) changes it to a series of voltage measurements, which are stored as an ordered set of numbers with a time interval associated with each. A digital signal processor (DSP), which may be an ASIC, FPGA, or even a fast microprocessor, performs a Fourier transform on the data streaming in. This changes it from a series of numbers associated with time intervals to another series of numbers associated with frequency intervals �� a digital spectrum.
��The digital filter is actually an ordered set of numbers representing the filter function. That is, they can take on any values from zero to one (inclusive) and each number is associated with a frequency interval. It is important that the frequency intervals match the frequency intervals associated with the digital spectrum.
��Next, the DSP convolves the spectrum with the filter. Convolution is simply the process of taking the spectral value associated with each frequency interval and multiplying it by the associated filter value. The result is another set of numbers representing the filtered spectrum.
��In the fourth step, the DSP performs an inverse Fourier transform, which recovers a digital representation of the filtered baseband signal. Finally, a digital-to-analog converter produces a time-varying voltage that closely matches the analog signal that would result from passing the analog baseband signal through an analog filter having the same filter function as the digital filter.
��Digital filters have three major advantages over analog filters. First and foremost, digital filters can use any filter function. Unlike analog filters, digital filters are not limited to those functions attainable with resistor-capacitor-inductor networks. The filter function need not be continuous. It need not have definable derivatives. It��s just a bunch of numbers assigned to frequency intervals. This characteristic provides essentially limitless flexibility.
��The second advantage is size. Analog filters are as big as the components needed to make them, which can be quite large for complex filters operating at low frequencies. Digital filters can be and have been completely integrated on a single IC.
��Finally, reprogramming a digital filter is simply a matter of loading in a new set of numbers for the filter function. Reprogramming an analog filter generally means designing and building an entirely new filter.
��Digital filtering��s main disadvantage is speed. The calculations, especially the Fourier transforms, are computationally intensive. The filter bandwidth can be only as wide as the speed of the processor(s) allow. This is why digital filters are often implemented with FPGAs or ASICs, which do the math in fast hardware rather than much slower software.

Digital filters use a five-step process that starts and ends with an analog signal, but performs all the mathematical operations digitally.

chemistry background of a mechatronics engineer

Wed, 30 Jul 2008 13:42:08 GMT

��I mentioned chemistry as being important to a mechatronics engineer working on environmental projects in my June 9, 2008 blog entry, and for most such projects that would be true. The thing to keep in mind is that mechatronics engineering integrates development projects at a very high level. Traditional engineering disciplines focus on the piece of equipment being developed.
��An electronics engineer working on an audio amplifier, for example, works from specifications that involve electronics quantities, such as input signal level, frequency bandwidth, and so forth. It makes little difference what application that amplifier ends up in. I have seen high end stereo amplifiers that were designed for consumer high fidelity market used quite successfully to power ac motors in motion control systems.
��I have, myself, cut the tip off a Christmas tree light bulb, epoxied the remaining envelope into a high vacuum system, and used the exposed filament as a Penning vacuum gauge.
��Most marine inboard power plants are engines designed for automotive applications and repurposed for installation in boats. The modifications involve replacing the air cleaner with a flame arrestor, the exhaust manifolds with special water-cooled manifolds, and similar replacements of peripheral components bolted onto the outside.
��Geographic mapping and remote sensing satellites were originally designed as spy satellites.
��The list goes on��.
��The mechatronics engineer, however, needs to understand the end application because he or she works at the interface between the system and the application. It is impossible to organize the system correctly without knowing exactly what application it is to work in and how it is to function.
��Chemistry is a good example of a science that affects so many things that it is a good addition to a mechatronics background. To get away from the environmental applications previously mentioned in passing, let��s assume our mechatronics engineer has been asked to design an autonomous system to find and disarm terrorist bombs in buildings.
��Along with the automated guidance, robotic manipulation, and other more obvious mechatronic components, the project could use a good ��sniffer�� to pick up volatile organic compounds associated with explosives. What compounds might be appropriate? What detector technologies are available? How to compare potential candidates? Someone versed only in the usual mechatronic disciplines of mechanical, electrical/electronic, and automated-controls engineering would be ill equipped to answer these questions, but they��re duck soup for someone with a few college chemistry courses under his or her belt.
��The point is that mechatronics engineers must interface with the real world of applications in a fundamentally different way than engineers in more narrowly defined disciplines. The broader his or her background, the wider the range of projects he or she can approach with confidence.

How old is the RCA connector standard?

Wed, 30 Jul 2008 13:39:29 GMT

��The other day, Control Engineering process industries editor Pete Welander and I were talking about the evolution of various standards on the way out to the parking lot, when this week��s question came up. We couldn��t remember quite when RCA connectors, one of several competing standards for interconnecting audio equipment, appeared.
��RCA connectors are those neat little color-coded plugs found on stereo equipment, public address (PA) systems, and high-end televisions. They��ve been around so long most of us take them for granted, but few of us remember when they first appeared.

��RCA audio plugs are ubiquitous in high-end audio equipment. Source: Wikipedia

��Also called phono connectors or CINCH/AV connectors, RCA connectors have been around since the 1940s, when the Radio Corporation of America introduced them to allow connecting a phonograph turntable into a radio receiver��s preamplifier.

��Originally, the RCA connector was used to allow connecting a phonograph turntable into a radio receiver��s preamplifier.

��RCA connectors made setting up such a high-fidelity (hi-fi) component audio system much easier and safer for consumers than the terminal strips or even (Heaven help us!) Fahnestock clips previously used. All the consumer needed to do was push the male (plug) connector attached to a cable from the turntable into the famale (jack) connector mounted to the receiver��s chassis. The term ��phono�� plug probably arose from the ��PHONO�� marking stamped on the receiver��s metal chassis next to the RCA jack. No more screwdrivers to tighten connections, or potentially live wires to handle. Just plug it in, set the receiver��s mode switch to ��phonograph,�� and, voila, the record plays through the speaker!
��Audiophiles could personally integrate separately purchased state-of-the-art radio receivers, turntables, and loudspeakers. This ability gave consumers the feeling that they had created a ��custom�� system personally selected to match their tastes and the acoustic properties of their rooms. Makes you think about the value of standards.

the ideal background for a mechatronics engineer

Fri, 06 Jun 2008 11:11:33 GMT

��Mechatronics is a cross-disciplinary system-level discipline. So, a mechatronics engineer needs a very broad technical background as well as communication and interpersonal skills.
��The term system-level gets bandied about a lot, but I��m not too sure how many engineers have a good working definition of it. I came across my favorite definition long, long ago in a galaxy far, far away during a brief stint as a business systems analyst at Dupont. I snagged a book out of the department library, which I remember being Fundamentals of Systems Analysis by Jerry Fitzgerald.
��The definition (as I remember it) took a Zen approach by explaining what systems weren��t. It pointed out that physics deals with situations that have very few elements, like a ball rolling down an inclined plane or a mass attached to a spring. Statistical mechanics deals with situations characterized by enormous numbers of elements. Systems fall in between. They have too many elements to analyze efficiently the simple differential equations physicists love, but not enough elements to analyze statistically. So, system analysis needs a hybrid approach that involves solving many differential equations simultaneously to model situations with many degrees of freedom.
��Mechatronic systems add the complication that some of the elements are mechanical, some are electrical, some involve information technology (IT), and all interact with each other via control loops. This is its multi-disciplinary side, and a mechatronics engineer needs familiarity with all disciplines, as well as an IT background to keep it all organized.
��Taking a mechatronic system from specification of needs to deployment in the field generally requires too many disciplines for any one individual to do well. This means the mechatronics engineer needs a project management background as well. The projects are small compared to, say, building the Sears Tower, but there are still a lot of balls to juggle.
��But, managing a mechatronics project takes more than milestones and Gantt charts. Deliverables for all the tasks in the Gantt chart have to be tightly integrated and optimized for overall system performance, not individual performance.
��In other words, the mechanical designer wants to make the mechanical portion work well as a mechanical system, while the computer designer wants the embedded single-board computer optimized to do its thing. Optimizing either is irrelevant from a mechatronics point of view, however. The only thing that counts is how well the overall system works when assembled. There is every reason to expect that the mechanical portion of the optimum overall system will not be optimum on its own. It is the mechatronics engineer who must understand this and make it happen.
��That is why the mechatronics engineer needs communication and interpersonal skills �� to guide disparate team members into behaving differently than they might individually.
��To sum up, the mechatronics engineer needs a background that includes project management, applied mathematics, computer science, mechanical engineering, electrical/electronic engineering, and any other scientific or technical disciplines required for the project, such as fluid mechanics for an aviation-related project, or chemistry for an environmental project.
��Such individuals do exist, but they buck the trend seen over the past decade or so of increasing specialization.

Why conduction cool an embedded computer?

Fri, 06 Jun 2008 11:10:05 GMT

��Anyone who has actually put an operating laptop computer on top of their lap knows why cooling is important for any computer. Microprocessors dissipate a lot of electric power, which turns into heat, which must be removed somehow. They do this despite the fact that the metal-oxide-semiconductor (MOS) field effect transistors (FETs) making up the bulk of the computer circuitry are theoretically very efficient when operating in switched mode (saturated on or cut off).
��MOSFETs are effectively voltage-variable resistors. A voltage applied to the transistor��s insulated gate electrode modulates resistance between the source and drain electrodes. In an off state, no conduction path exists between the source and drain electrodes, hence no current flows, and no power is dissipated. Placing sufficient voltage of the correct polarity on the control gate creates a low-resistance inversion layer between source and drain. Current then can flow efficiently between those electrodes, meaning that, once again, very little power is dissipated.

��Microprocessors generally incorporate millions of MOSFET transistors.

��There are three factors that combine to foil this effect and burn a lot of electric power in MOSFET-based microprocessors.
��First, today��s microprocessors operate at very high clock frequencies.
��Second, substantial (in microscopic-transistor terms) current is needed to charge and discharge the interelectrode capacitance between the gate and ground each time a transistor switches. The higher the frequency, the more often charge and discharge cycles occur, and the more current flows.
��Third, there are millions of transistors in microprocessor ICs. Combining millions of transistors switching at high speeds with substantial current needed to switch all these transistors leads to a lot of power dissipation �� and a lot of heat. Heat management therefore becomes an important part of every computer design, including single-board computer (SBC) design for embedded control.
��Your typical computer �� whether an SBC or any other form factor �� gets heat out of a microprocessor by pressing a great whacking head sink on top of it, and blowing lots of nice, cool air across it. That air has to come from somewhere and blow to somewhere else. Those places are outside the computer enclosure, and the air gets in and out through holes in the enclosure.
��Air isn��t the only thing that can go in and out through those holes. Dust and fluids can get in, too. Dust and fluids are very bad for electronics and are especially problematical in the industrial environments where many embedded systems live and work.
��A second issue with cooling microprocessors by blowing cool air across them is noise and vibration. Few computers can survive using only natural (unforced) convection for cooling. There��s too much heat to extract by just letting cooling air waft lazily by under natural convection. You need a substantial fan to push enough air through the enclosure. Fans make noise and vibrate. Any facility with a number of computers going at once becomes noisy and vibrates.
��In response, a number of SBC vendors have introduced conduction-cooled products for embedded systems. These SBCs solve the dust-and-fluid intrusion problem by the simple expedient of closing up ventilation holes, and the vibration-and-noise problem by eliminating fans.
��Completely closing SBC enclosures, however, leaves designers with the problem of how to get rid of heat. They have solved it by placing the microprocessor package in thermal contact with the aluminum enclosure walls.
��Heat generated as the MOSFETs go clickety clack at multi-GHz speeds conducts through the IC package to the case. The case then acts as quite a large heat sink. Being aluminum, the case can efficiently conduct that heat to the system��s mechanical support structure, a large heat sink, or cooling fins on the outside where natural convection works better.
��The result is an adequately cooled SBC with great tolerance for dusty, damp, wet, nasty environments. In addition, no forced-air cooling means no fan, so the SBC is virtually silent, and no holes means superior immunity to electromagnetic interference (EMI).
��Yeah, I haven��t mentioned EMI, yet, have I? There are two types of EMI to consider: radiated and absorbed. The radiated type arises when the SBC acts like a UHF transmitter, interfering with anything around it capable of being interfered with. An SBC is also more or less susceptible to absorbing �� and being interfered with by �� UHF signals radiated by other equipment.

��EM waves gradually lose amplitude while traversing a conducting wall. The thicker the wall, the more amplitude lost.
��Metal enclosures act as Faraday cages, blocking (within limits) entry of external EMI fields, and containing (again within limits) radiated EMI. When an electromagnetic (EM) wave impinges on a conducting wall, it immediately begins losing amplitude. The energy lost out of the wave either reflects back out, or heats the material through resistive dissipation. In either case, what reaches through the wall is greatly diminished. How diminished depends on the wall��s thickness and conductivity. High conductivity leads to rapid amplitude decline. Thicker walls give provide more room for the wall to work on the wave, so it exits at much lower amplitude. Thick (several millimeters) walls made of highly conductive material can attenuate the wave by many orders of magnitude.
��Things that limit an enclosure��s ability to block EMI include material conductivity, wall thickness, and openings.
��The aluminum typically used for fanless SBC cases is an excellent conductor. Plastic enclosures often used for less rugged units generally have little or no EMI-blocking ability unless fortified with conductive fillers, such as carbon.
��Holes have the effect you would expect. They let portions of the wave front pass through the wall without attenuation. The interaction between a conducting wall and an EM wave is more complex than simply blocking some energy and allowing some to pass through, but the net effect is to let significant radiated power in, and the larger the hole, the more EMI can pass through.
��Conductively cooled SBCs also typically have thicker walls. There is less machining to do (no vent holes to cut), so designers need not thin the walls to reduce machining expense. Enclosure walls act as a much needed heat sink, which motivates designers to thicken them as well. Of course, thicker walls produce a more mechanically rugged package for industrial use.
��In the end, the reason to choose conductively cooled SBCs for industrial embedded applications is for better environmental, EMI, and mechanical characteristics.

A portable instrument to measure room air pressure

Fri, 06 Jun 2008 11:01:45 GMT

��The full text of the question is: Could you recommend some handheld instruments that can easily measure and indicate room air pressure (positive or negative, and how much)?
��The short answer is ��No.�� As a journalist I can��t recommend products. Also, I would need more information to make a selection for you. I can, however, suggest how to go about finding the instrument you need.
��You are, of course, looking for a barometer. There are a number of technologies used to create barometers, ranging from the original mercury barometer invented by Evangelista Torricelli in the 17th Century to miniaturized units fabricated using microelectromechanical system (MEMS) techniques. All barometers are specialized adaptations of dual-port differential-pressure gauges. The adaptation invariably consists of closing off one port to create a reference-pressure chamber. The instrument then reports the difference between the pressure at the open port and that in the closed chamber.

��Differential pressure displaces fluid in a hydrostatic gauge (manometer) and a diaphragm in an aneroid gauge.

��There are really only two basic pressure-gauge types: hydrostatic and aneroid. Hydrostatic guages �� generally referred to as manometers �� consist of a U-shaped tube partially filled with fluid of known density. Any pressure difference between the two sides (arms) of the tube registers as a difference in the height of fluid in the arms. Aneroid (meaning ��without fluid��) uses a flexible membrane (diaphragm) in place of the fluid. The degree to which the diaphragm flexes signals the pressure difference.

��Converting a differential pressure gauge to a barometer is a simple matter of closing off one port, which becomes a reference-pressure chamber. In the manometer, the reference pressure is a vacuum.
��QA26MAY08a2:
��Wall-hanging and desktop barometers are available in almost any hardware or department store. Such instruments are actually quite sensitive due to the small range of natural atmospheric pressure variations �� usually 0.5 in/Hg above or below the average of 30 in/Hg (or ��1.7%) according to the Challenger Building, Tips, and Technique web site. (This website is associated with the vendor of Challenger homebuilt ultralight aircraft.) Their accuracy depends on how accurately you zero them in to begin with. Linearity and calibration slope are, however, not guaranteed.
��If atmospheric pressure is an important input for controlling your process, you probably want something a bit more reliable. The best place to start looking is the Internet. Googling ��barometer�� turned up over 10 million hits. Narrowing the search to ��handheld barometer�� reduced that to 155,000. The top hits were distributors who cater to backpackers and boaters.
��Two promising vendors, however, were more scientifically oriented: the Control Company, and Edmund Scientifics. The Control Company offering is said to have ��0.75% NIST-traceable accuracy and sells for $160.50. The Edmund Scientifics unit looks suspiciously similar, but does not claim to be traceable and sells for $129.95.
��Remember that just because a unit looks the same as a traceable unit doesn��t mean that it��s traceable, too. To claim traceability, the unit must be part of a calibration program that involves a chain of calibration reaching back to the National Institute of Standards and Technology (NIST). When purchasing such a unit, you should receive a certificate evidencing this traceability chain.
��Often distributors special order units marked with the distributor��s logo from a manufacturer. Special-order instruments from the same manufacturer for different distributors usually look the same except for the logo. They are likely to be the same physically, too. One, however, may order instruments run through a NIST-traceable calibration program, while the other may not.
��My Google search turned up a third vendor, Omega Engineering (my old standby for this type of sensor), which offers a unit for which it claims very high accuracy (��0.02%) and NIST traceability �� as well as a nose-bleed-inducing $2,385 price tag. You pay for what you get with scientific instruments!
��To choose among these and other offerings, you need to consider your application requirements. How much accuracy do you need? How portable does the unit really have to be and why? How do you plan to record the measurements?
��For example, when playing around with the wind tunnels at Arizona State University, I used a mercury barometer affixed to a wall. I��d have to dig out my notes to quote its calibration specifications, but let��s just point out that it took into account variations in the local gravitational field. The unit sure wasn��t portable, but I didn��t really need it to be. I just wrote the reading down in a research notebook, and keyed it into the data acquisition program for each test run.
��There are any number of reasons why you might really need portability. Perhaps there are several separate locations within your facility that need comparative barometric measurements. Maybe you plan to check barometric-pressure sensors that are part of your process control at several locations around your facility. Maybe you want to lock the instrument in a safe location when not in use.
��All these factors need to be considered carefully when choosing an instrument.