Fullscreen HTML5Med Res (??MB)HTML4

By PETER SMITH Any PC with a sound card can function as a digital oscilloscope, spectrum analyser, voltmeter and signal generator. It’s just a matter of installing the right software, much of which can be downloaded for free or at low cost from the Inter­net. E LSEWHERE IN THIS ISSUE, we describe a simple adapter that provides a safe and easy way of connecting test probes to your sound card. Below we introduce some of the important principles of sound card-based digital instrumentation and follow up with a quick rundown on some of the more popular software packages that are available. Digital instrument basics Digital instruments have many advantages over their analog cousins. Take the oscilloscope, for example. Digital ’scopes can store waveforms in memory or on disk for www.siliconchip.com.au comparison with “live” waveforms, or play them back at a later time for detailed examination. And once a waveform is stored digitally, it is easily manipulated (level-shifted, filtered, transformed to the frequen­cy domain, etc) for display in a variety of formats. The downside is that digital instruments are often expen­sive, standalone devices with a CRT or LCD display, multiple CPUs and complex data capture electronics. However, much cheaper solutions based on the PC platform are now readily available. These utilise the existing processing power and graphics capabil­ities built into all PCs, thereby greatly reducing costs. PC-based solutions range from small external data acquisi­tion pods (we reviewed such a device in the June 2000 issue) through to add-on cards that plug into a free expansion slot. These are cheap by comparison to the standalone devices but still too costly for those of us on tight budgets. PC sound card The simplest and cheapest solution utilises hardware that already exists in all multimedia PCs – the sound card. It might seem unusual that a PC sound card could be used as the basis of any data acquisition system. However, one of the main components of all sound cards is an analog-to-digital (A-D) converter, a core function of even August 2002  7 amplitude of that signal. The digital values are then processed by software to drive on-screen voltmeter displays or to plot waveforms on an oscilloscope X-Y grid. Let’s look at some of the more important aspects of the analog-to-digital conversion process in a little more detail. Resolution Fig.1: Oscilloscope 2.51 uses large vertical sliders for program­ming the sweep, gain, trigger level and delay settings for both channels. A-D converter resolution is characterised by the number of digital bits that it takes to represent the results of a conver­sion. Older sound cards, such as the classic SoundBlaster 2.0 and SoundBlaster Pro, have only 8-bit resolution. All recent PC sound cards have 16-bit (or higher) resolution, which equates to 216 (65,536) possible discrete values. To make some sense of this, we need to know the upper and lower limits of the voltage that can be sampled by the converter. Sound cards typically have a 0-2V input span. Applying some simple maths, we find that 2V divided by 65,536 gives 30.5µV – a respectably small slice indeed. Bandwidth Fig.2: real-time signal spectra can be examined in FFT (Fast Fourier Transform) display mode. In this shot, Oscilloscope 2.51 plots a simple 4kHz square wave (the large spike) and its har­monics. Fig.3: many ’scopes support an X-Y mode, where the amplitude of the second channel is plotted against the amplitude of the first. Here we’re using Oscilloscope 2.51’s delayed sweep function and X-Y mode to get a new perspective on our measurements! the most expensive digital instruments. In simple terms, the A-D converter’s job is to periodically sample the incoming analog signal and come up with a digital value that represents the instantaneous 8  Silicon Chip Equally important as the signal amplitude is its frequency. Sound card converters operate at a known (programmable) sampling period under control of a crystal-locked clock. All this means is that each conversion cycle occupies a precise period, easily measured and manipulated by software to glean the signal frequen­cy. Naturally, the faster the input signal can be sampled, the more accurate the displayed result. Imagine, for example, a sinewave signal with a period of 100µs. If the A-D converter samples at, say, 20µs intervals, then what you would see on an oscilloscope display wouldn’t look much like a sinewave. Instead, it would look like an ascending and descending “staircase”. It follows that the faster the signal is sampled, the smaller the steps will be and the less visible the staircase effect. When the sampling rate is much higher than the signal frequency, software interpolation removes virtually all traces of this “digitisation” and the waveform looks much the same as it would on a traditional analog scope. Apart from affecting how waveforms appear on an oscillo­scope display, the sampling frequency is important for another reason. It must always be at least twice the frequency of the signal being measured, otherwise a problem called “aliasing” occurs (see Fig.12). Most sound cards have a maximum sampling rate of 44kHz, so the highest frequency you can expect to measure accurately is 22kHz. The minimum frequency that can be measured is about 20Hz, due to AC-coupling at the sound card inputs and a high-pass filter in the A-D block. This can be an annoying limitation when the signals you’re measuring contain a DC component but a good multimeter will usually fill in the gaps. Storage As each A-D conversion is completed, software reads the result and stores it sequentially in a block of memory www.siliconchip.com.au Fig.4: this view shows TrueRTA’s oscilloscope and signal generator. The signal generator and righthand toolbar can be detached from the main display if required. (or “storage buffer”). As mentioned above, each cycle occupies a precise period, so the storage locations also act as time mark­ers. Once the buffer is full, oscilloscope software can be used to read the contents and plot the traditional amplitude versus time waveforms. Of course, this is a very simplified description of the process. Depending on the particular software and the active instrument, mathematical calculations may need to be performed on all or part of the buffer contents before the results can be displayed in the appropriate format. For example, voltmeter software might need to extract average, crest factor and RMS values from the raw data. Triggering The storage buffer is in effect “circular”. Once full, old data is overwritten with new as the entire cycle repeats. Howev­er, although the A-D converter may be sampling and converting at full speed (called “free-running”), the software does not neces­ sarily immediately write the results to the buffer. Rather, it monitors the incoming data for predefined trigger conditions. On most sound card-based software, the triggers are programmable to specific voltage levels – either positive (rising) or negative (falling). Without some means of synchronising the signal with the beginning of the buffer, even simple waveforms would be impossi­ble to comprehend on an oscilloscope; they Fig.6: the “front panel” of Osci – it’s almost as easy to use as rotary dials and switches! www.siliconchip.com.au Fig.5: plot of a 2kHz square-wave on TrueRTA’s spectrum analyser. This is the “level 3” version of the product, which allows up to 60 frequency bars (1/6 octave) across the horizontal axis. Display update speed can be traded off with accuracy to suit the speed of your hardware – necessary because this analyser is a real CPU hog. would appear to jitter and jump across the horizontal axis. In fact, most digital in­struments need reliable triggering in order to make accurate measurements. For greater versatility, many digital ’scopes provide a variable pre-trigger (or “delay”) time. This simply means that a certain number of samples are written to the buffer before the trigger condition is met. Generating output So far, we’ve only talked about instruments that utilise sound card inputs. Not surprisingly, a good deal of software is available that makes use of the outputs as well. In operation, an analog output voltage is generated by the sound card’s digital-to-analog (D-A) converter. Software feeds a stream of 16-bit digital data into the D-A block, where it’s converted to analog, filtered and then amplified before appearing on the card’s output sockets. As you can see, this process is similar to the signal input side, except in reverse! Digital signal generators are the most-used output Fig.7: Osci’s waveform display can be dragged away from the main window and resized as required. The dark lines on the display are measurement cursors that we used to determine the frequency and amplitude of the sinewave. August 2002  9 What’s A Spectrum Analyser? Most of our readers will already be familiar with the oscil­loscope. These instruments display signals in the time domain. The horizontal axis is graduated in time and the vertical in amplitude. This format is ideal for determining time, phase and amplitude information. On the other hand, spectrum analysers display signals in the frequency domain. Frequency is displayed on the horizontal scale, and is divided into bands, or octaves. The lower frequency bands are on the left, with progressively higher frequency bands to the right. The scale is logarithmic, such that each octave or fractional octave is equal in width. The amplitude of the signal is displayed on the vertical axis, which is graduated in deci­bels. A spectrum analyser enables us to see certain information that is just not visible in the time domain. For example, a sinewave may look good in the time domain but show visible distortion in the frequency domain. Also, a noise signal may look totally random in the time domain but in the frequency domain one frequency may be dominantly present. In audio frequency work, the spectrum analyser is commonly used to measure signal-to-noise instru­ments. Together with the instruments we mentioned earlier, they provide a convenient means of analysing a host of analog circui­try. Just like their analog counterparts, digital signal genera­tors provide the usual sinewave and square-wave outputs with programmable frequencies and amplitudes. Some even include sweep generators and noise sources for frequency response and distor­tion analysis. The software OK, now that you’re familiar with some of the basic terms, let’s examine a few of the more popular digital instrument pack­ages that are available on the Internet. We have selected five quite different software packages that we think demonstrate the capabilities of sound cardbased instrumentation quite well. These are all listed in Table 1, along with the links to their download sites. Some are free for non-commercial use, while others are offered on a shareware basis. Many more are available – too many for us to seriously evaluate here. Our advice is to shop around and be sure to “try before you buy”! For additional software, point your browser to www.google.com and search for sound card oscilloscope software. Oscilloscope 2.51 The first package we examined is titled simply “Oscillo­ scope 2.51”. It runs on Windows 95 & 98 (a Windows 3.1 version is also available) and requires only an 80486 processor and an 8-bit sound card. In common with most packages, it includes a dual-trace storage ’scope, as well as a real-time spectrum analyser. All major functions are controlled via a series of “clickand-drag” sliders on the righthand side of the display (see Fig.1). Vertical trace position, gain, and trigger level are all independently programmable for left (Y1) and right (Y2) channels. In single trace (YT) mode, two trigger delay sliders pro­ vide coarse and fine adjustment of the amount of 10  Silicon Chip ratio, distortion, intermodula­ tion distortion and frequency response. A mathematical process called Fast Fourier Transforms (FFTs) is used to convert signal information from the time domain to the frequency domain. To complicate matters further, analyser software often includes several complementary FFT “windowing” functions. Unfortunately, a detailed explanation of windowing, or FFTs for that matter, is well beyond the scope of this article. However, plenty of information on the subject is available on the Internet and in printed form. pre-trigger data written to the buffer. In dual trace and X-Y mode, these sliders vary the time delay between the Y1 and Y2 channels. The gain settings control software gain only; hardware gain must be varied via the Windows audio mixer software. In addition, the product of the left and right sliders is used to determine gain when in spectrum analyser mode. This provides a convenient, faster-than-linear adjustment rate. The sampling rate can be set to 11.025kHz, 22.05kHz or 44.1kHz. On all but the slowest (486) hardware, it makes sense to sample at the maximum supported frequency for best accuracy. The buffer (and hence display) refresh rate can be programmed to any realistic value, with the default of 330ms being too slow for smooth updates. High-performance PCs will support a much faster rate than this. Frequency measurements are made by left and right-clicking on the oscilloscope display. Oscilloscope 2.51 measures the period between the selected points and displays the result on the status line. In Spectrum Analyser mode, it’s only a matter of running the mouse over the area of interest, as the frequency of the plot at the cursor position is displayed in real time. A snapshot of the buffer can be saved to disk as an ASCII file for use by other CAD packages. Of course, you can also cut and paste the 8 x 10 graticule display into your favourite graph­ics program for documentation purposes as needed. All up, this tidy little package uses few resources but offers a lot. However, it lacks a means of calibrating the input signal levels, so all amplitude measurements should be considered relative rather than true. TrueRTA TrueRTA (Real Time Audio Spectrum Analyser) is a complete suite of audio test instruments. The latest release (V2.0 as we write) is available in four distinct levels. The levels differ in functionality and price, with level 1 www.siliconchip.com.au Fig.9: if you need a simple no-cost generator, check out Sound­Arb. You can even roll your own waveforms! Fig.8: WaveGen can generate just about any waveform you care to name. Digital signal generators are quite frequency-accurate but distortion at the high and low ends of the scale must be considered. available free of charge. TrueRTA runs on all 32-bit versions of Windows and requires a Pentium 200 (or equivalent) processor with 64MB of RAM and a 16-bit sound card as a minimum. Included in the package is a spectrum analyser, oscilloscope, digital multimeter and signal generator. This package is aimed squarely at the audio test and devel­opment area. The built-in signal generator (with digital sweep function) and the spectrum analyser enable quick evaluation of audio circuit performance. In addition, support is provided for a calibrated microphone input, enabling loudspeaker and acoustic environment testing. As with most packages, the input sampling rate can be set to any one of the standard sound card values between 8kHz and 48kHz. User controls are intelligently arranged and clearly la­belled (see Fig.4) and the instruments are dead easy to drive. You simply select between the oscilloscope or spectrum analyser instruments and then fine-tune the settings using buttons and drop-down menus on a detachable toolbar. The digital ’scope provides both single (left or right channel) and dual-trace (left & right channel) modes, as well as channel addition and subtraction. The horizontal axis can be programmed from .05ms to 200ms per division, while the vertical axis ranges from a low 100µV per division right up to 5V per division. The ranges are variable in the traditional 1-2-5 steps via rows of buttons on the toolbar. The digital voltmeter displays the RMS value of the refer­ence channel in the top left corner of the graticule. If you decide to purchase one of the higher-level versions, you get additional voltmeter readouts of dBu, crest factor in dB (ratio of peak to RMS level) and crest factor in mV/V. The reference channel, by the way, is the one that you select as the trigger source. Triggering is automatic – no provi­sion has been made for varying the level or polarity. A single click on the toolbar switches to spectrum analyser mode (see Fig.5). The graticule is now displayed in logarithmic format – the vertical axis in dB and the horizontal axis in bands of frequencies. The free version of the software provides only a single octave across the horizontal, which equates to 10 bands (or bars) in the 10Hz-22kHz spectrum. Pay money, and you can select from 1/3, 1/6, 1/12 or 1/24 octave (30, 60, 120 or 240 bar) displays. The horizontal scale can be expanded for detailed examina­tion of a particular frequency range by modifying the upper and lower frequency limits. Similarly, vertical scale limits can be adjusted to zoom in on an area of interest. TrueRTA’s signal generator includes both sinewave and pink noise output. The output level is variable MORE FROM YOUR EFI CAR! Own an EFI car? Want to get the best from it? You’ll find all you need to know in this publication  Making Your EFI Car Go Harder  Building A Mixture Meter  D-I-Y Head Jobs  Fault Finding EFI Systems  $70 Boost Control For 23% More Grunt  All About Engine Management  Modifying Engine Management Systems  Water/Air Intercooling  How To Use A Multimeter  Wiring An Engine Transplant  And Much More Including Some Awesome Engines! AVAILABLE DIRECT FROM SILICON CHIP PUBLICATIONS PO BOX 139, COLLAROY NSW 2097 - $8.95 Inc P&P To order your copy, call (02) 9979 5644 9-5 Mon-Fri with your credit card details! www.siliconchip.com.au August 2002  11 over the entire range of the sound card’s D-A circuitry and the sinewave frequency is variable from 5Hz to 24kHz. Higher-level versions of the product include a logarithmic sine sweep function for more accurate frequency response measurements. Unlike the previous package, this one includes comprehen­ sive calibration features. The Fig.10: if you want to do serious work Fig.11: AudioTester’s spectrum analyser professional version (level in the audio spectrum, AudioTester is a dual-channel affair, allowing direct 4) even includes a feature to offers a comprehensive range of FFT comparison between reference and test functions. signals. remove the “coloration” that sound cards inevitably add to Trigger levels are independently programmable for your distortion and frequency channel 1 (left) and channel 2 (right) via vertical sliders response measurements. and can be of either positive or negative polarity. Auto As this package focuses on the real-time aspect of triggering is also provided, as is variable trigger delay and audio work, it does not have a number of features that a single sweep mode. are handy for general electronics bench work, such as Left and right channels can be added, subtracted and/ moveable measurement cursors, triggering options and or in­verted. The inversion function could be handy if you storage capabilities. have a sound card that inverts input signals (don’t laugh, Osci we’ve heard of some that do!). Waveforms are displayed on a standard 8 x 10 gratiThis is one of the best of the low-cost digital ’scopes cule that can be detached from the main window and that we’ve seen. Although the shareware version is not resized to suit your needs (see Fig.7). Optionally, the crippled in any way, it does have a maximum use period X and Y-axis settings can be displayed right on the of 15 minutes. Once this expires, the program terminates graticule – a very handy feature for documentation and you need to relaunch it. Of course, if you like the purposes. product, you can license it for a nominal fee and get rid Osci’s display can be printed on demand or copied to of this annoyance. the Windows clipboard for pasting into your favourite Osci runs on Windows 95, 98 and NT4 and requires at application. In addition, the buffer contents can be saved least a Pentium 266 with 32MB of RAM and a 16-bit sound as an ASCII file for use in other programs. card. If you want to run a signal generator in parallel with The storage buffer generally holds more than can the ’scope, your sound card and its driver software must be dis­ played on-screen at one time, so a horizontal support “full-duplex” mode. This applies to all packages, scroll bar at the bottom of the display allows quick by the way, not just Osci. panning from buffer beginning to end. A nearby “x10 All sound card sample rates up to 96kHz are accomMag” button allows you to instantly zoom into areas of modated, as is 24-bit resolution for those cards that interest. support it. In addi­tion, Osci supports up to three sound Getting accurate waveform measurements is easy with cards, so you don’t need to dismantle your existing audio this package. Click on the start point with your left mouse setup. button, drag the horizontal and vertical rulers that appear The Osci user interface is uncomplicated and easy to to the end point and release, and hey-presto – the period, drive (see Fig.6). Vertical (Y) axis settings are variable from frequency and amplitude of the bounded area appear as 100µV per division up to 2V per division, while horizontal if by magic! (X) axis (or “timebase”) settings are variable from 20µs to Finally, all your settings can be saved as “presets” for 200ms per division in the usual 1-2-5 steps. quick restoration later. Table 1: PC Instrumentation Software Package Licence Download Link Oscilloscope 2.51 Freeware polly.phys.msu.su/~zeld/ osci ll.html True RTA Level 1 is free, Levels 2-4 are acti vated on purchase www.trueaudi o.com Osci, WaveGen, AudioTester Shareware (30-day eval uation) www.sumuller.de/audiotester SoundArb Freeware Analyzer 2000 Shareware (30-day eval uation) www.brownbear.de Freeware hel iso.tripod.com/download/ download.htm Digital Sound Generator 12  Silicon Chip www.wavebuilder.com WaveGen WaveGen is a comprehensive standalone signal generator. Tone, impulse and sweep generators are all included and accessi­ble from the front panel. WaveGen can be used in conjunction with Osci (they’re from the same author), so system requirements are the same for both packages. The maximum D-A conversion rate of the sound card deter­mines the highest output frequency. For a www.siliconchip.com.au typical 48kHz card, the highest frequency will be 24kHz. The minimum frequency is listed as 0.1Hz but this seems a bit optimistic as it dips well under the lower frequency limit of around 20Hz for most sound cards. Generator frequency can be programmed in 0.1Hz, 10Hz and 1kHz increments. As you can see from Fig.8, all the usual waveform types can be generated. In addition, WaveGen will play back any user-de­fined WAV file. Output levels can be adjusted with either “analog” or “digital” control buttons. As far as we could determine from the documentation, the 0dB to -96dB digital level adjustment is soft­ware-based, whereas the 0dB to -48dB analog adjustment controls the Windows mixer. No calibration is provided for the line output socket. Instead, when using WaveGen for frequency response and distortion measurements, the documentation suggests that you feed the right line output directly back to the left line input, so creating a “reference” channel. The right output also connects to the input of the circuit under test, while the output of the circuit under test is con­ nected back to right line input. This method allows you to com­pare the differences in the two waveforms on an oscilloscope or spectrum analyser display. The spectrum analyser instrument in AudioTester (which we mention below) expects this connection and can automatically compensate for sound card amplitude and fre­ quency response characteristics. Virtually all generator parameters can be configured via setup buttons on the front panel. We won’t bore you with all the details here. Instead, why not download WaveGen and check them out for yourself! SoundArb This is a no-frills, easy-to-drive signal generator. It runs on Windows 95 and above, requires a 16-bit sound card and only minimal PC hardware – and it’s free! A shot of SoundArb’s super-simple front panel is shown in Fig.9. Sine, square, triangle, sawtooth and white noise wave­ forms can all be generated, as can user-defined arbitrary wave­forms. These are loaded from a simple ASCII file, which can be manually created or generated by software. Normally, SoundArb outputs the selected waveform on both the left and right channels. Optionally, a synchronisation signal can be output on the right channel Fig.12: the A-D converter must sample the input signal at least twice as fast as its frequency otherwise “aliasing” re­sults. Here, the input signal (shown in red) would need to be sampled at least twice each period but instead it’s sampled only once every 2/3 period. Therefore, the frequency of the signal is erroneously calculated to be much lower than it really is (as shown in green). instead by choosing the “Right channel sync” option. The output amplitude is set by a horizontal slider and is uncalibrated. SoundArb provides simple triggering options and these include “Free run”, “One-shot” and “Burst”. The burst mode length is programmable in cycles. AudioTester AudioTester boasts just about every feature imaginable. Like TrueRTA, it includes an oscilloscope, spectrum analyser and signal generator but lacks a separate voltmeter. We’ve mentioned this package only because it originates from the same author as Osci and WaveGen. These two standalone instruments are apparently intended to replace the oscilloscope and signal generator that are part of the AudioTester suite. To date, the author has not released a standalone version of the spectrum analyser, so AudioTester is still available to fill in the gaps. Like Osci and WaveGen, the spectrum analyser in AudioTester is packed with features. We’ve run out of space to describe them here but we’ve included a couple of screen shots (Figs.11 & 12) to whet your SC appetite! UM66 SERIES TO-92 SOUND GENERATOR. THESE LOW COST IC’S ARE USED IN MANY TOYS, DOORBELLS AND NOVELTY APPLICATIONS 1-9 $1.10 10-24 $0.99 25+ $0.88 www.siliconchip.com.au August 2002  13