PC instrumentation through the ages
Events beyond technology helped shape the way engineers use computers to automate measurements.
Martin Rowe, Senior Technical Editor -- Test & Measurement World, 9/15/2001 2:00:00 AM
| In the dim and distant past, engineers recorded measurements with pencil and paper—a slow and error-prone method. Many engineers wanted some way to use computers to automate measurements and analyze data. Today, 20 years after the introduction of the IBM PC, two types of instruments—inboard and outboard—take measurements and move data into a host computer.
PC technology has become the backbone of automated test and measurement systems. PC-based instruments began as boards that plugged into minicomputers and were configured with switches and required direct register calls to run. Now, they have drivers and graphical user interfaces and take advantage of the Windows operating system, the PCI bus, and the universal serial bus (USB). Despite all the changes in PC technology, though, a vast number of outboard instruments still use the venerable IEEE 488 bus. Rather than give you a timeline of how technological advances in the PC affected automated measurements, I’ll focus on some of the events that shaped PC-based test and measurement. I’ll include important events based on Apple computers, too. Although data-acquisition and IEEE 488 cards already existed for the Apple II, IBM’s 1981 introduction of the Personal Computer became the most significant event pertaining to computer-based measurements. For the first time, engineers could use a consumer-grade computer (at consumer prices) to control instruments, take measurements, and analyze data. No longer were engineers tied to expensive minicomputers. First instruments
Scientific Solutions and National Instruments (Austin, TX) also introduced IEEE 488 interface cards for the PC. Hewlett-Packard (now Agilent Technologies, Palo Alto, CA), Capital Equipment (Billerica, MA), ines (Hannover, Germany), and IOtech (Cleveland, OH) soon did the same. Early data-acquisition and IEEE 488 cards for the IBM PC and clones required users to know how the computer’s expansion bus worked. Jumpers or switches set a card’s base address, interrupts, and direct-memory access (DMA) channels. To avoid bus conflicts, a user had to know what other cards resided in the computer. “Plug-and-play” cards were years away (some people claim that the PC-measurements industry still hasn’t achieved that goal). Purchasers of early data-acquisition cards may have received the card and a manual, but that was all, according to Fred Molinari, president of Data Translation. Some cards contained no drivers, no sample programs (except in hard copy), and no ready-to-run software to let you test your card. To control PC-based instrument cards, engineers wrote programs in Basic or Fortran. At the very beginning, programs wrote control codes directly to a card’s registers. The control codes, usually shown in hexadecimal format, told a data-acquisition card how to set its gain, set its scan rate, and acquire data. Instrument-card manufacturers quickly realized the need for drivers that users linked into their applications programs to read and write to card registers. Today, cards come with drivers that dynamically load as needed. Dynamic-link libraries, ActiveX controls, and other software layers ease programming so you can use English-like commands and write to the cards with meaningful numbers rather than with hex codes. By now, you can assume that just about every engineer uses a PC for instrument control. That didn’t occur overnight. For years, “real” engineers shunned the PC. They considered it a toy. Often, these “real” engineers used Hewlett-Packard’s Unix-based scientific computers to control instruments through IEEE 488 and RS-232 ports, and they programmed their systems in HP Basic, also known as “Rocky Mountain Basic.” Many engineers still use this language, as well as HT Basic from TransEra (Orem, UT). Unlike the interpreted GW Basic and Quick Basic that came with PCs, HP Basic could compile programs, which ran faster than interpreted code. HP computers also handled graphics better than did DOS-based PCs. HP Basic’s popularity also came from its ability to control instruments with less programming than either GW Basic or QuickBasic. Minimal software Data-acquisition-card manufacturers knew how to build cards, but they placed little emphasis on software or drivers to make the cards run. Recognizing the software limitations of card manufacturers, Labtech (Andover, MA) introduced Notebook in 1981. Labtech Notebook controls a myriad of data-acquisition cards and instruments through IEEE 488 ports and serial ports. The software relieved the programming burden so users could focus on data acquisition and analysis. To use the DOS version of Notebook, a user set up a card by entering its configuration data into a series of tables. By 1984, Labtech Notebook established the market for third-party support software. Labtech’s early success came from its extensive library of data-acquisition card drivers. The company established a common programming interface for numerous cards. Thus, the company could support cards from numerous manufacturers with one top-level program. In September 1984, IBM announced a series of products aimed at placing a PC on every engineer’s desk. The products included: • a Fortran compiler, • a data-acquisition and control adapter card, • an IEEE 488 adapter card, • a graphics development toolkit, • a graphics terminal emulator, • a graphics CRT display, and • a graphics adapter card. IBM contracted with Cyborg to build its data-acquisition cards and with National Instruments to build its IEEE 488 interface cards. Soon after launching the product line, IBM realized that its sales people didn’t know how to sell equipment into the engineering market nor did they want to. IBM also realized that the business and consumer market dwarfed the engineering and scientific market. According to two former Cyborg employees, the company’s management assumed that the IBM deal would reap huge profits. So, the company invested heavily in production equipment and started producing cards. When IBM pulled the plug on its engineering product line, it left Cyborg with thousands of boards and millions of dollars of unused production equipment. Unfortunately for Cyborg, it had difficulty selling the data-acquisition cards on the open market. To make matters worse, all the boards displayed the IBM name. The company had no drivers for its cards, either. By 1987, Cyborg was out of the data-acquisition business. A remnant of the company still lives on as a division of another company. National Instruments management, on the other hand, didn’t bet the company’s future on IBM. Its IEEE 488 cards already had DOS drivers when IBM started buying them. National Instruments didn’t have a large inventory of cards marked “IBM” because it sold the cards under its own name, not just to IBM. NI soon introduced more powerful cards and drivers, which established it as a major player in the IEEE 488 controller market (Ref. 2). IBM wasn’t the only maker of consumer and business-computer products to enter and exit the engineering market. Lotus Development (now part of IBM) recognized that engineers and scientists could use spreadsheets to analyze much of their data. But moving data into Lotus 1-2-3, the dominant DOS spreadsheet, required engineers to save data in a format that 1-2-3 could read. In October 1986, Lotus introduced “Measure,” a package that could move data directly from instruments into 1-2-3 through an IEEE 488 port or serial port or directly from a data-acquisition card. When it launched Measure, Lotus established relationships with Metrabyte for its data-acquisition cards and with National Instruments for its IEEE 488 cards. Like IBM, Lotus developed a product for engineers, but the technical computing market was too small for it to take seriously. Lotus couldn’t get its people excited about Measure. In 1988, Lotus sold Measure to NI. When the PC world moved from DOS to Windows, 1-2-3 succumbed to Microsoft Excel; NI reintroduced Measure for Excel in October 1995. Getting the picture Although spreadsheets and graphics cards enhanced the PC’s popularity in instrument control and data analysis, the PC still lacked serious graphics capabilities. In 1984, Apple introduced the Macintosh in it’s now-famous commercial during the Super Bowl. The Mac’s graphics capabilities placed it far ahead of the PC for ease of use, which could have established it as the dominant computer for technical applications. Unfortunately, early Macs lacked expansion slots, and Apple chose to keep its inner workings proprietary, initially limiting third-party companies from developing measurement cards. That helped boost the PC’s popularity among engineers.
Windows 3.1 changed everything for PC-based instrumentation. With it, PCs had a graphics capability to rival that of the Mac. Engineers could view their data more easily than they could in DOS, plus they could build GUIs. Perhaps most important, Windows’ multitasking ability let engineers’ computers take measurements while doing something else. No longer were computers “locked up” while taking measurements. Between 1992 and 1995, numerous graphical-programming schemes appeared on the market, but many players have since fallen by the wayside—most notably, Tek TMS from Tektronix (Beaverton, OR) and Wavetest from Wavetek (which the company subsequently sold to National Instruments). Hewlett-Packard made its entry into the PC-based graphical-programming market when it ported its Visual Engineering Environment (VEE) from Unix to Windows in 1993. Agilent VEE remains the only serious graphical-programming competitor to LabView. Despite the popularity of graphical programming, many engineers still prefer text-based languages. According to our recent survey, Visual Basic, C++, and C hold positions 1, 2, and 3 among programming languages, respectively, although LabView closely follows (Ref. 4). Where’s NT? Although most instrument hardware and software makers jumped on the Windows bandwagon, few jumped onto Windows NT. According to Bendrix Bailey, CEO of Softwire Technology (Middleboro, MA), it’s harder to write drivers for Windows NT-based operating systems than for Windows 95/98. Windows 95/98 gives drivers direct control over I/O ports, but NT drivers require permission from the operating system’s kernel to gain access to I/O ports. In addition, Windows NT required more powerful computers with more memory than Windows 3.1/95/98. Windows 95 made installing I/O cards much easier than DOS, Windows 3.1, or Windows NT did. Combined with the PCI bus, Windows 95 put an end to manual configuration of I/O cards. Not until Windows 2000 did the NT line of operating systems conform to the PC industry’s “plug-and-play” standards. Standard interface fails While PCI bus and Windows 95 made the installation of I/O cards far easier than the ISA bus, cards from different manufacturers still require their own drivers because each card manufacturer uses a unique set of function calls in its drivers. Twice in the last 10 years, makers of data-acquisition cards tried and failed to create a standard programming interface for data-acquisition cards. In 1992, several makers of data-acquisition cards and software formed a group called Windows Science, Engineering, and Manufacturing (WinSEM). Part of the group’s mission was to specify a software layer that would separate you from the manufacturer-specific I/O. Instead, the layer would present a standard programming interface regardless of the card it controlled. The concept seemed good, but no manufacturer was willing to force its customers to rewrite working code to accommodate new programming commands. The companies took the attitude, “We should standardize on one programming interface—mine.” The effort collapsed (Ref. 5). The second attempt to standardize data-acquisition programming started in 1998 when Data Translation, Hewlett-Packard, Labtech, Omega Engineering (Stamford, CT), and Strawberry Tree (now part of IOtech) formed the Open Data-Acquisition Association (www.opendaq.org). Although the association still exists, its membership remains too small to bring about change. Too many card manufacturers opted to stay out of the association to make it effective. Bus Transfers Plug-in data-acquisition cards and IEEE 488 interface cards move data through the PC’s I/O expansion bus, but that bus has taken several forms over the years. In 1984, IBM introduced the PC-AT. AT (Advanced Technology) computers significantly increased computing power over PC-class and XT-class machines through its 80286 16-bit processor. Just as significant, IBM expanded its I/O bus from 8 bits to 16 bits. The wider bus could then transfer 12-bit measurements from a data-acquisition card to its host computer with one transfer rather than two. That significantly increased measurement speed. The AT bus became the standard for other PC manufacturers, hence its current name, the Industry Standard Architecture (ISA) bus. Many ISA-bus computers still take measurements and control processes, and instrument-card manufacturers still produce cards for the ISA bus. Two other I/O buses, EISA (extended ISA) and Micro Channel, never gained wide acceptance in the PC market. IBM developed Micro Channel, but kept it as a proprietary bus. The PC industry balked at that attempt by IBM to regain control of the PC market. Compaq and others rebutted by introducing the EISA bus, but a fragmented PC market caused both buses to fail. Instrument card manufacturers developed cards for both EISA and Micro Channel, but engineers preferred to stay with the mature ISA bus, primarily because legacy systems could use it. While many ISA-bus cards are in use today and many instrument companies still manufacture those cards, many users have moved to the PCI bus. Developed by Intel, the PCI bus greatly increased I/O bus bandwidth and gave cards plug-and-play capabilities. It succeeded because Intel and Microsoft backed it. The PCI bus and Windows 95 now make it easier to install PC-based instrument cards. Test and measurement moved toward the PCI bus in 1995 when Data Translation introduced the first data-acquisition card for the PCI bus (Ref. 6). The PCI bus evolved into CompactPCI, and the telecom industry quickly adopted it. The CompactPCI platform also seemed like a good one for instrumentation.
Geotest (Irvine, CA), has introduced a line of 6U PXI cards and a chassis to house them. PXI stands poised to move in on PC-based measurement systems, especially if the slotless PC emerges as the standard computer for home and business use. You’ll have no choice but to use outboard instruments or instrument cards in a chassis. Which bus will carry data between a host PC and an instrument-card chassis? Any of several could emerge as the standard, but the winner most likely will be the bus that becomes the standard on all PCs: a high-speed serial bus such as USB 2.0 or IEEE 1394b. Ethernet also stands to make gains in instrument control. PC-based test and measurement has come a long way since engineers set jumpers and programmed with hexadecimal codes. It will continue to evolve, adding new technologies based on advances in the PC industry. The goals, however, remain the same: get more data faster, more accurately, and with less work than ever before. T&MW 1. Scientific Solutions’ - Company Information, www.labmaster.com/company/aboutcompany.html. (In 1991, T&MW gave the Test Product of the Decade award to the LabMaster. At the time, we stated it was manufactured by Tecmar. A representative of Scientific Solutions recently told us that the two companies were sister companies, and that early scientific products were marketed under both names.) 2.Schreier, Paul G., “Riding the tiger,” Personal Engineering and Instrumentation News, August 1987. p. 11. 3. Rowe, Martin, “Labview Comes to Windows, Sun,” Test & Measurement World, October 1992. p. 96. 4. Titus, Jon, “Readers sound off: PC uses, needs, & frustrations,”Test & Measurement World, September 2001. p.12. 5. Rowe, Martin, “A Noble Effort Fades to Black,” Test & Measurement World, October 1993. p. 27. 6. “PCI Bus Gets First Data-Acquisition Card,” Test & Measurement World , February 1995. p. 76. For more Information Strassberg, Dan, “Despite threats from USB and Firewire, IEEE 488 ain’t down yet,” EDN, July 16, 1998. archives.e-insite.net/archives/ednmag/reg/1998/071698/15df1.htm. Acknowledgement Thanks to Paul Schreier, founder of Personal Engineering and Instrumentation News, for his insight, wisdom, and a box of old press releases that covered some early PC-based measurement products. Martin Rowehas a BSEE from Worcester Polytechnic Institute and an MBA from Bentley College. Before joining T&MW in 1992, he worked for 12 years as a design engineer for manufacturers of semiconductor process equipment and as an applications engineer for manufacturers of measurement and control equipment. E-mail: m.rowe@tmworld.com. |
Talkback
No related content found.
- 0 rated items found.
Datasheets.com Electronic Parts & Inventory Search
185 million searchable parts
- Part Number
- Description
- Inventory
- Products
- Manufacturers
Sponsored Links
View All Webcasts
