Tài liệu Circuit Cellar P3 doc

Buy Online www.silabs.com Complete Evaluation System for $10.99 • Development Tools • Source Code Examples • Documentation • 2K Limited C Compiler www.silabs.com Product details: www.silabs.com/ToolStick MCUs TIMING POWER BROADCAST WIRELINE WIRELESS Fully Contained MCU Development System Not only did Silicon Labs introduce the worlds’ first USB ToolStick evaluation platform, we continued the evolution with a family of interchangeable ToolStick modules including base adapters, daughtercards and debug adapters. Now, software and hardware MCU development is easier than ever. Using only a PC with a USB port, designers can fully experience the software development environment in conjunction with the MCU on-chip debugging hardware that allows full, non-intrusive access to the target CPU, peripherals and memory. The software development environment consists of the Integrated Development Environment (IDE), editor, debugger, Flash programmer and a demonstration version of the Keil compiler. Description Price ToolStick Starter Kit (includes C8051F330 Daughter Card) $24.99 ToolStick Evaluation Kit $10.99 ToolStick Base Adapter $17.50 ToolStick C8051F330 Daughter Card $9.90 ToolStick C8051F360 Daughter Card $9.90 ToolStick C8051F530 Daughter Card $9.90 ToolStick Debug Adapter $8.90 17.qxp 3/1/2007 2:51 PM Page 1 20 Issue 202 May 2007 CIRCUIT CELLAR ® www.circuitcellar.com needed and should be removed. INTERFACE TO THE MCB2130 The signal analysis, which results in a measured magnetic field strength, was all done by an LPC2138 on an MCB2130 board. The microprocessor communicates with the outside world via a serial RS-232-compatible inter- face, does the analysis, and controls the amplifier/polarization circuit. The MCB2130 board has a breadboarding area. I built a small circuit to buffer the 3.3-V logic output of the LPC2138 to the 5-V logic level required by the optically isolated logic input of the ATtiny26 on the amplifier/polarization board. The optically isolated analog output of the amplifier/polarization board goes directly to one of the A/D inputs of the LPC2138 (see Figure 6 ). MAIN PROGRAM The MCB2130 accepts single-char- acter commands from the serial port to run the magnetometer. For exam- ple, to make a single measurement, it must turn on the polarization by bringing the output line to a logic- high level for a predetermined time, turn it off, wait for all the transients to die down, and then take the 512 samples of the analog output of the low-noise, high-gain, band-pass ampli- fier. Then, the MCB2130 must find the phaseslip in the calculated arctan- gent data. Knowing the sampling fre- quency, it must determine the preces- sion frequency, convert it to a magnet- ic field strength, and send it out the serial port. There are also commands to let it set the polarization time and set the time to wait after polarization stops and the data sampling begins. Table 1 is a full listing of the com- mands and their actions. The heart of the program is the routine to take the sample, calculate the arctangents, and then find the best value for the average phaseslip. This is the measure() procedure. The most difficult part is determining how (with the long sequences of data) to remove all the phase rollovers when the phase changes from near 2π to 0. After this has been done, I think of the phase versus time as a linear graph and solve for the least-square best fit to a straight line. The line’s slope is the mean phaseslip per cycle. It can be converted to a mean fre- quency difference to the base sampling frequency. The scatter of the data points is a measure of the accuracy of the measure- ment. It can be represented by the statistical value called the standard deviation (σ). When the calculated value of the magnetic field is sent out the serial port, the value of σ, converted to magnetic field strength units, goes with it. If the input S/N of the precession signal is better than about 10, the value of σ is typically less than about 1 nT, with the sensor described here. FIELD INSTRUMENT A complete instrument con- sists of a sensor, an amplifier/switching board, an MCB2130 board, and a laptop. It is convenient to mount the sen- sor on a framework made from plastic irrigation tubing with the Figure 4—This portion of the circuit shows the HEXFET switches and the low-noise amplifier. The polarization sequence that switches the various HEXFETs on and off is controlled by the ATtiny26 microprocessor shown in Figure 5. 2704016Koehler.qxp 4/5/2007 3:18 PM Page 20 www.circuitcellar.com CIRCUIT CELLAR ® Issue 202 May 2007 21 Figure 5— This portion of the circuit shows the optically isolated analog output to the MCB2130 board and the optically isolated digital input from the MCB2130 board. These two lines connect to the MCB2130 board through JP3. 2704016Koehler.qxp 4/5/2007 3:18 PM Page 21 22 Issue 202 May 2007 CIRCUIT CELLAR ® www.circuitcellar.com possible, the capacity of the battery will determine how long the instrument will operate in the field. The MCB2130 is powered from a laptop via the USB cable. Serial data between the Keil board and the laptop can be provided by a simple terminal program on the lapto p. IMPROVEMENTS My magnetometer is as accurate as amp lifier/switching board mounted at least 30 cm away from it. The battery operating the unit must be nonmagnetic. This means that it must be a gel cell. The battery should be located as far from the sensor as possible. Since the polarizing current is about 2.3 A and polarizing current will be flowing for more than 50% of the time, if the meas- urements are being made as quickly as Photo 2—Check out the finished prototype board for the amplifier switching circuit. Surface-mounted components were used. Some minor changes were made after the initial assembly. The circuit in Figure 4 is correct. The res- onating capacitor was omitted when using the sensor shown. Figure 6—The small buffer circuit was built onto the breadboard area of the MCB2130 board. It brings the 3.3-V logic level output of the LPC2138 microprocessor up to 5-V logic levels. 3URXGO\GLVWULEXWHV Available in 40 and 44 pin header configuration Support PIO 0-4 and Ultra DMA 3 mode. Bootable from Transflash/micro SD. Low power consumption. SD-IDE-40/44 IDE Flash Drive Carrier Board with Micro SD Interface 1.800.665.5600 www.tri-m.com info@tri-m.com tel: 604.945.9565 fax: 604.945.9566 HEAD OFFICE: VANCOUVER Intel ® Pentium ® M 745 1.8GHz, 2MBL2, ATX Based on Intel ® Pentium/Celeron ® M processor. Intel 855GME / ICH4 chipset with integrated Intel ® Extreme Graphics 2 engine 2x32 MB VRAM UMA. 2x DDR-RAM-SODIMM for up to 2GByte. EPIC/PM Fanless Intel ® 852GM Celeron ® M 600 MHz Micro PC Onboard Intel ® Celeron ® M 600 MHz with 512K L2 Cache CPU. Intel ® 82852GM Chipset. Scalable with Evalue EPIC SBC. One SODIMM up to 1 GB DDR SDRAM. EES-3723 2704016Koehler.qxp 4/5/2007 3:18 PM Page 22 www.circuitcellar.com CIRCUIT CELLAR ® Issue 202 May 2007 23 James Koehler graduated from the Australian National University with PROJECT FILES To download code, go to ftp://ftp.circuit cellar.com/pub/Circuit_Cellar/2007/ 202. REFERENCE [1] R. Green, “Faster Math Functions,” Sony Computer Entertainment America, www.research.scea.com/ research/pdfs/RGREENfastermath _GDC02.pdf. SOURCES ATmega32 and ATtiny26 Microcon- trollers Atmel Corp. www.atmel.com MCB2130 Evaluation board Keil www.keil.com LM394 SuperMatched Pair National Semiconductor Corp. www.national.com LPC213x Series of microprocessors NXP www.nxp.com Command Action “m” Make a measurement “d” Set the delay before analysis after polarization in milliseconds (e.g., “d100” sets it to 0.1 s) “b” Set the estimated local magnetic field in nT (e.g., “b55100” sets it to 55,100 nT) “t” Set the desired polarization time in milliseconds (e.g., “ t750” sets it to 0.75 s). Table 1—These commands are used to operate the magnetometer. They are sent to the magnetometer via a serial RS-232-level interface to an external computer. any commercial unit, and it can take measurements very quickly. The data- sampling period lasts only about 0.25 s. Data analysis and all of the calcula- tions take another 0.25 s or less. The time required to polarize the sensor depends on the liquid used and the degree of saturation desired. For kerosene as the sensing liquid, this could be as short as 0.75 s. Because polarizing for the next measurement can start at the same time as the analysis of the data for this measure- ment, it is possible to make about one measurement per second. A colleague and I are working on an improved version of this PPM that is a stand-alone instrument. It does not require a computer for operation. An LCD displays the measurements, and push button switches are used to oper- ate it. It features an LPC2148 micro- processor, which has a USB interface in addition to the standard features it shares with the LPC2138. We use the second RS-232-compati- ble port to monitor the NMEA output from a GPS system. As a result, we can record the exact location of the sensor at the time of a measurement. The data is stored internally on an SD card, which we can remove to transfer the data to a computer after a day of using the instrument. The software has several operating modes so the PPM can be used as a single magne- tometer or as one of a pair that operates as a gradiometer synchronized with one another by the GPS time signals. The improved system’s amplifier and switching board is similar to this one, except for a few minor differ- ences. The MCB2130 is replaced by the LPC2148 with an SD data card and an LCD-PB switch human interface. I a Ph.D. in Astronomy in 1966. He worked as a Professor of Physics and Engineering Physics at the University of Saskatchewan and retired in 1996. His field of research was in Upper Atmospheric Physics, and he designed and built several radar sys- tems for use in the study of ionos- pheric electric fields. James has writ- ten more than 60 peer-reviewed scien- tific papers on this topic and others. His current hobbies include electron- ic design (mostly RF), photography, and model aircraft. He lives with his wife of 44 years on Vancouver Island, where he enjoys the beautiful envi- ronment and relaxed lifestyle. 2704016Koehler.qxp 4/5/2007 3:18 PM Page 23 24 Issue 202 May 2007 CIRCUIT CELLAR ® www.circuitcellar.com won’t replace your favorite PCB shop, but it will save you time and money when you’re trying to get your design into hardware for the first time. It also gives you complete control over your boards. PLOTTING SYSTEM My PCB plotting system consists of four compo- nents: a PC, a used Hewlett-Packard 7440A ColorPro plotter, a CAD program for layout, and a Sharpie Ultra Fine Point permanent marker. The PC is a standard desktop running Windows 98 (it’s old, but reliable). The operating system pro- vides a “native” environ- ment for my CAD program and a driver for the Color- Pro plotter that allows you to “plot to file,” a key to refining your layouts before plotting them. I bought the plotter last year for $35, and strongly recommend the 7440A or a similar Hewlett-Packard model. It is frequently available for purchase on the ’Net, it has every fea- ture you’ll need, and most importantly it uses the HP Graphics Language (HPGL). HPGL is an ASCII-based human-readable command language that drives the W hen I discovered there were used Hewlett-Packard pen plotters for sale at bargain prices on the Internet, I decided to buy one and answer a ques- tion that had intrigued me for years. Could a plotter, with its thousandths- of-an-inch positioning accuracy, be used to draw printed circuit board (PCB) traces directly onto copper-clad? And if so, what kind of design-rule limitations would this process impose? Would the process, if feasible, be reliable? I hoped it would make it possible to build good, inexpensive, moderate-density circuit boards at home and perhaps bring surface- mount layouts within reach—but would it? In this article, I will answer those questions. Can you plot directly onto copper- clad with a pen plotter? You bet! The results I’ve been able to achieve with this design have been remarkably good. Double-sided 4″ × 7″ boards laid out on a 25-mil (0.025″) grid can be produced routinely at a per-board cost calculated in pennies. The process is slow and certainly not suited for pro- duction. But, for the developer who needs only one or two small boards to execute a prototype, it is almost ideal. A PCB can be plotted, inspected, erased (if necessary), and plotted again until a satisfactory result is achieved. Each pass costs no more than the ink used, and with experience you’ll be able to achieve good results on your first pass. Best of all, a complete system, including a plotter, software, and sup- plies should be available to the careful Internet shopper for less than $75. It FEATURE ARTICLE by Curt Carpenter Circuit Board Plotting Circuit Board Plotting 101. Curt describes how to draw PCB traces onto copper-clad with a Hewlett-Packard 7440A ColorPro pen plotter and a Sharpie Ultra Fine Point permanent marker. Photo 1a—This Sharpie holder from an old HP pen has a purpose-built “shoulder” that helps keep your pen secure and at a fixed height above the work piece. b—A scrap of copper-clad is used to set the pen height. With the pen in the “down” position, the tip is adjusted to just touch the gauge. If it is set too high, the pen won’t draw on your copper-clad; too low, and it’s likely to snag on something as it travels across your board. c—The registration jig serves two purposes: it carries your PCB materi- al through the plotter while holding it in a known position, and it allows you to plot the two sides of a double-sided circuit board “in register” with your pads and vias aligned from one side to the other. A morning’s production of corner brackets is shown as well. a) b) c) 2705015Carpenter.qxp 4/5/2007 3:10 PM Page 24 plotter. It is the second key to being able to manually fine-tune your draw- ing files before you plot them. Whatever plotter you buy, make sure it’s compatible with your soft- ware and that you can find documen- tation for it. Most used plotters seem to be offered without this important component. If your plotter did not come with documentation, manuals for many old HP peripherals can be down- loaded from the HP Computer Muse- um web site (www.hpmuseum.net), a great site that contains a bounty of user and service manuals for older HP products. If you buy a plotter on the Internet, be sure that it can pass material that is 0.125″ thick under the pen. Some can’t; they were made to move paper, not copper-clad. Also make sure you get the required power supply. My ColorPro came without a wall trans- former. (Luckily, I had one on hand.) Verify too that you can connect the plotter to your PC. While many plot- ters use a standard RS-232 interface, some use less universal connections. With my plotter working correctly, I made three minor modifications to it. I snapped off the clear plastic hood so the plotter’s pen carriage could accom- modate Sharpie pens, removed the eight-pen carrousel (it’s not needed), and drew an array of small dots on the plotter platten to aid in registration. For layout, I use TurboCAD, a remarkable general-purpose drafting program. Version 12 of this software is available at www.imsi.com, but I’m happy with Version 8. This and other versions of TurboCAD are available on the Internet for less than $10. All of them will meet your needs for this application. In my professional PCB work, I use a layout program featuring schematic capture, auto routing, design-rule checking, Gerber output, and so on. But, for plotting directly to copper- clad, you’ll want to have more fine control than programs like this pro- vide. Layouts for the plotter must make explicit provisions for top-to- bottom registration, something your PCB shop normally takes care of. TurboCAD offers crucial features, such as layouts on multiple drafting layers that can be hidden or shown (and printed) at will and mirror-image generation around a selected axis. The program also allows you to download PCB layout “foot- prints” for connec- tors, ICs, and a host of other components from many sources on the Internet in “ICES” format and copy them directly into a TurboCAD symbol library. This saves you hours of drawing time. The Sharpie Ultra Fine Point perma- nent marker is the real key to making this system work. The ink serves as resist that can stand up to most cop- per-etchant solutions, and it can be erased using mineral spirits (common paint thinner) to easily correct mis- takes. (After erasing, wash your board thoroughly and scrub it with cleanser. Even a slight thinner residue will pre- vent good ink adhesion). The secret to using the pens on cop- per-clad is to make multiple passes at a m oderate pen velocity over your work. Apply several layers of ink with a short 1- to 2-min. drying time between passes. This builds up a reli- able barrier for your etchant, ensuring that your narrowest copper traces will remain intact after the etching proces s. I’m sure alternatives can be found for any of the components, but I know the parts work well together and pro- vide the abilities needed to plot circuit boards. TOOLING Before you can plot your first PCB, you’ll need to build four small tools: a pen holder that adapts your plotter to Sharpie pens, a pen-height gauge, a “registration jig” that holds your cop- per-clad in place and carries it through the plotter, and a plotter-driver pro- gram. To build a pen holder, cut the ends off an old plotter pen, remove the fiber pellet from the inside, and run a file through the empty barrel a few times until your Sharpie fits into it snugly (see Photo 1a). Pen height is a compromise between drawing a clean line with the pen down and clearing the work piece with the pen up. The gauge helps you achieve this compromise on a consis- tent basis (see Photo 1b). My registration jigs are made from clear plastic slides, like the ones used with overhead projectors (old page pro- tectors also work) and a few bits of poster board (see Photo 1c). Using your CAD program, draw a registration target. Put the target on a separate “registration template” draft- ing layer, and save the drawing. It will be used as the starting point for all subsequent layout work. The crosshatched area at the upper- right corner of the target is used to accurately locate the copper-clad when plotting its “component” side, while the hatched area in the upper-left cor- ner locates the board when plotting its “solder side.” The small boxes drawn on the template allow you to position your registration jig accurately on the plotter. Line any three of these boxes up with marks that you have scribed onto the plotter platen, and your jig will be in a known, fixed position rela- tive to your plotter pen. Plot a few of these registration tar- gets on your plastic sheets—using your newly built pen holder of course—and set them aside. This jig www.circuitcellar.com CIRCUIT CELLAR ® Issue 202 May 2007 25 Photo 2—The registration jig with a piece of copper-clad is taped in place, ready for a component-side plot. Painter’s tape works well. Make sure the tape is flat to the surface so it won’t snag the pen. 2705015Carpenter.qxp 4/5/2007 3:10 PM Page 25 plotted on the “component” side of your PC board after they have been aligned with the right-hand corner of the registration target. Layers three and five will be “mirror-imaged” and plotted after being aligned to the left- hand corner of the target. As you gain experience, you may want to add additional layers to this stack: one for general notes, one for component-side “fill” areas, and one for solder-side fill. The last two will prevent wear on your pens. The symbols layer helps you main- tain your sanity while doing a layout. It contains things like the physical outline of your board, mounting holes, connector cutouts, package outlines, and more. I draw everything on this layer in blue pixels. (Plotting the sym- bols layer on a two-sided board is dif- ficult since the pen tends to snag on traces, but it can be drawn on single- sided boards to look like a “silk screen.”) Once you’ve defined a symbols layer, draw a rectangle on it with the same physical dimensions as your copper-clad. This rectangle is used for registration. Then draw another rec- tangle, spaced 0.05″ inside the first one, and ensure that no pads or traces fall outside of this inner rectangle. The 50-mil margin ensures that your pen won’t snag on a board edge during the plotting process. (A larger safety margin won’t hurt and it may save you some frustration.) The pad layer, which is plotted on both sides of a two-sided circuit board contains all of your layout’s “donuts” and vias, pads which are intended to create a connection from one side of your board to the other. I draw these in black pixels. How you arrange items here is the key to a good layout. Everything that should appear in cop- per on both sides of your PCB belongs on this drafting layer. Use your profes- sional layout software to do a first- pass place and route and let this guide your CAD drawings. The two trace layers are plotted on their respective board sides. I like to use red pixels for solder and black pix- els for component-side geometries. I aim to put the majority of my traces on the solder side to avoid vias. All of issues that are particular to plotting directly to copper-clad. You will cer- tainly want to adapt these comments to your own style and experience. Make a copy of your registration tar- get file and add some additional draft- ing layers to suit your style. I use a minimum of five layers: the standard registration template, symbols, pads, top-side traces, and bottom-side traces. Of these, layers three and four will be 26 Issue 202 May 2007 CIRCUIT CELLAR ® www.circuitcellar.com will let you plot 6″ × 8″ boards. With care, you can add an additional 1″ to this capacity, but at the sacrifice of the single-corner registration that this design offers. Now draw a set of six or seven 2″ × 2″ right-angle corner brackets in a sep- arate CAD file. Save and plot these on a sheet of poster board that is thinner than your PCB material (see Photo 1c). Carefully, cut these out and glue two of them on the registration corners of your template. Save the others for future use. Note that a good plastic-to- poster board glue bond can be achieved by piercing the materials a few times with a scribe. The registration jig is now ready for use (see Photo 2). The plotter driver is a short program that inputs a file of HPGL commands, modifies these slightly, and sends the result to the plotter. Your CAD pro- gram/operating system must allow you to “plot to file” to use it. In my case, “plot to file” produces PRN files. The source code, executable, and other information about my program is posted on the Circuit Cellar FTP site (PCB Plotting Files.zip). The text in the memo window describes button functions and it is replaced with an HPGL listing after a file is opened. I want to thank M. Cocco for making his Delphi-compatible RS-232 compo- nent (used here) freely available on the Internet. The program’s primary function is to read a PRN file and remove any HPGL pen-selection (SP) and pen- velocity (VP) commands that it may contain. (The ColorPro plotter sup- ports eight pens on a carrousel con- trolled by SP codes. Attempting to execute these codes when your pen is in place is disastrous! The VP com- mands control pen velocity, and you want to ensure that these always select 10 cm per second—my preferred speed for plotting on copper-clad.) It also allows you to view your plot files as text, move the pen to a convenient position for height adjustment, tap the pen to start ink flow, and draw a sim- ple test shape. LAYOUT PCB layout is an art form. All I can hope to do here is describe some Figure 1a—Donuts are drawn so the first stroke of the pen begins on the inside of the donut. This is because the ink from your pen will “bloom” slightly each time it is lowered onto the copper—and this keeps the “bloom” inside the pad, where it can do no harm. b—It is diffi- cult to get good resist coverage over a large area with the pen (your etch may cut through thin spots in the ink). If you must cover a large area with solid resist, you will probably need to resort to a broader-point pen or some form of paint. c—Wide traces are drawn as multi- ple lines or as crosshatch filled rectangles, even though your CAD program probably offers variable line widths. d—Unless you have the facility to plate through the holes in your boards (I don’t), you must allow room to solder a physical wire on both sides of your board to create vias. Component bodies, as illustrated, often hide these side-to-side connections. a) b) c) d) 2705015Carpenter.qxp 4/5/2007 3:10 PM Page 26 49.qxp 2/6/2007 1:58 PM Page 1 28 Issue 202 May 2007 CIRCUIT CELLAR ® www.circuitcellar.com my circuits are drawn on a 0.025″ grid with occasional exceptions. It takes practice to remember that you must place each of the objects you draw vertically (by layer), but this soon becomes second nature. A good symbol library helps in this regard and it is a huge asset to an efficient layout. I’ve provided a “starter set” of sym- bols, but you will certainly want to add more. When a board is inked, the plotter will draw each of your pads and traces several times to ensure a good build- up of Sharpie “resist” on the copper- clad. Fill patterns used to create ground planes (or simply hide copper surface area in order to preserve your etchant) will also be drawn more than once, but they aren’t as critical. Hence, I suggest that you place fills on separate drafting layers so you can control the number of times they are overdrawn. Here are a few tricks for doing direct-plots to copper-clad layouts that work for me. Circles take a long time to plot, so I prefer rectangular “do nuts.” All of the donuts I use are a variation of the ones shown in Figure 1a . My smallest donuts are 38 mils wide (50 mils are preferred), limited by my ability to draw them properly. You can make them smaller if your eye- sight permits. Traces should always begin and end on the inside of a donut or connector pad, again to render the ink bloom harmless. They should be drawn as a single polyline without breaks and in an order that minimizes pen travel in the “pen up” position. This takes more disci- pline than I can usually muster, but it is a good goal. The reason? “Pen- up” movement on my plotter always takes place at 40 cm per second and this can subject the PCB to considerable g-forces, causing it to shift in the regis- tration jig. (Pen- down velocity, on the other hand, is within your control.) Solid area fills should be avoided in favor of crosshatch patterns (see Figure 1b). Similarly, if you need a trace that is wider than the one that your pen normally delivers, use several parallel lines or rectangular areas filled with crosshatch (see Figure 1c). Your CAD program may offer to draw wider lines as a matter of “line style,” but I haven’t had good luck with this. Your layout must plan ahead for vias. You must make special provisions when they are hidden (see Figure 1d). Remember that you don’t have as much fine control over trace width as you do with photographically pro- duced PCBs. Line width varies with pen quality and wear, so you should keep trace- to-trace and trace- to-pad spacing as large as possible. I recommend a trace pitch of no less than 50 mils. You can violate this design rule when you want to squeeze a wire between two pads of an integrated circuit, but if you do, plan to careful- ly run a scribe between the trace and the pads. Finally, give yourself as much room as you can and appreciate that you are not going to get the kind of routing densities you may be used to in your commercial work. You’re building a prototype after all, not a production PCB. PLOTTING With your layout finished and saved in a single “master file” of five or more drafting layers, you’re almost ready to plot. First, you need to create two PRN files using the plot-to-file capability. One of your layouts should be registered and contain only those layers needed to draw the component side of your board. The other PRN file should contain only what is needed to draw its solder side. The process is a bit cumbersome to describe, but it will be obvious to you “on the ground.” It will become second nature after you’ve done it once. Use the example and figures described here as a guide. A flow chart of the entire process is posted on the Circuit Cellar FTP site. An example of a double-sided board or “master layout” (as it was drawn) is shown with all of the drafting layers in Figure 2. This file is copied and its registration and solder-side drafting layers are turned off (only pads, com- ponent-side traces, and symbols are visible). These elements are grouped into one graphics object that can be Figure 3—Here is a component-side layout properly positioned on the registration target drawing layer. Figure 2—The density of pads and traces here is on the “bleeding edge” of my process, but there’s plenty of room to advance the state of the art! 2705015Carpenter.qxp 4/5/2007 3:10 PM Page 28 . 2704016Koehler.qxp 4/5/2007 3:18 PM Page 20 www.circuitcellar.com CIRCUIT CELLAR ® Issue 202 May 2007 21 Figure 5— This portion of the circuit shows the optically isolated. the MCB2130 board through JP3. 2704016Koehler.qxp 4/5/2007 3:18 PM Page 21 22 Issue 202 May 2007 CIRCUIT CELLAR ® www.circuitcellar.com possible, the capacity