Readers sound off: PC uses, needs, & frustrations

You can’t do much engineering work these days without a computer, so we wanted to know what types of PCs our readers use most often for data-acquisition and instrument-control applications. It’s no surprise the majority—over 82%—use a basic desktop computer (Table 1). You can buy a powerful PC for less than $500 at a discount store, so a desktop unit makes economic sense for most computing tasks. Even if you wanted to embed a small computer in a piece of equipment, you’d be hard pressed to find an embedded PC for under $500. So, desktop PCs rule. We found 33% of respondents also use a laptop. (For this and a few other questions, percentages add to more than 100% because respondents could choose more than one answer. We also rounded values to the nearest percent.)

But off-the-shelf computers don’t always carry the day. We found 161 people who take advantage of rack-mounted or rugged/industrial computers—computers that you won’t find in a discount store or outlet. In these days of inexpensive PCs, you might think everyone would use one in some form, but 44 of our respondents built their own computers for data-acquisition and control applications. The vast majority of all the computers our respondents work with use Pentium CPUs, although 5% of our survey respondents use Unix workstations. Apple’s Macintosh shows up in the noise level—fewer than 1% or our survey participants use a Macintosh.

The Palm, Handspring, and similar “pocket” computers barely show up on our survey. But the availability of add-on hardware and inexpensive software-development tools that run under the Palm OS and Microsoft’s Pocket PC operating system make these small devices worth watching for future applications.

Even though computers are relatively inexpensive, respondents have mixed plans to upgrade their systems. Just under 50% expect to upgrade during the next year, while the remainder don’t expect to upgrade for a year or more. Although pundits think the PC market is in the doldrums, the industrial market for these devices should remain strong. Most of the upgrades, regardless of when they take place, will result from either company-wide upgrades, or a need for a faster PC (Table 2).

Look at buses, too

But the types of PCs our readers use for instrument-control and data-acquisition tasks tell only half the hardware story. We also examined the types of buses our readers use in their systems, their future system needs, and a few other hardware-related issues. Of the 440 respondents to our survey, 419 said they use some sort of bus to control instruments or acquire data. The venerable RS-232 serial bus shows up at the top of the list (Figure 1), followed closely by the IEEE 488 bus, a long-time favorite for instrument control. The PCI bus has made inroads into T&M applications, probably due more to its widespread use in modern PCs than to its suitability for instrumentation applications.

Figure 1. The individual bars in this chart represent the buses that people currently use (blue), the buses they rely on most (yellow), the buses they prefer (green), and the buses they plan to start to use in the next 12 months (pink).

We’ve also seen a trend toward stand-alone instruments and data-acquisition systems that provide USB as their only communication path with a PC. The popularity of USB should grow during the next few years, as the faster USB 2.0 standard takes effect. USB and Ethernet each have a place in instrument control: Ethernet in remote or networked systems, and USB in nearby instruments and systems that can take advantage of this PC-industry standard.

Although we haven’t seen many data-acquisition or instrument systems that provide an IEEE 1394 (Firewire) interface, 25 people have an interest in using this high-speed serial bus. So far, most applications using IEEE 1394 center on machine-vision applications in which digital cameras use the bus to communicate with PCs.

About 40% of our respondents use a remote system for data-acquisition or instrument-control applications, and they rely on standard buses for these applications (Figure 2). The CompactPCI bus has the most support, with the ISA expansion chassis and VXIbus as runners up. When we asked these people what backplane bus(es) they plan to be using 12 months from now, we found the CompactPCI bus still retains its popularity. Surprisingly, more people expect to make use of the VXIbus than the newer PXI bus. The mature VXIbus still has life left, even as the industry shows great interest in the newer CompactPCI bus and PXI buses. 

Figure 2. Backplane buses play a large role in data-acquisition and instrument systems, so we asked people which types of backplanes or plug-in modules they currently use (blue), and how they expect their backplane or plug-in module use to change in the next 12 months (yellow). These questions allowed multiple selections.

 As people design their data-acquisition and instrument-control systems, they must decide how many analog and digital I/O ports they need. But asking questions about “average” numbers of I/O lines, average number of bits of resolution, and so on, provides little useful information. After all, who has an average application, and who wants an average ADC?

But we did ask people how they get signals from the sensors and transducers they use to their computer or instrument system. Because I/O boards come with all sorts of connections, we thought it might help to find out what types of connections these people prefer (Table 3). The ubiquitous D-type connector—think parallel port—came out on top, probably because people have used this type of connector for decades. The D-type connectors are easy to obtain, and wiring flat cable or individual wires to them is a snap. Also, many products provide a D-type connector as the standard means to gain access to I/O signals.

Engineers use BNC, SMA, and similar connector types on benchtop instruments, so it’s no surprise to see these types of connectors high on a list of preferences. Of course, this preference could have less to do with familiarity than with the characteristics of the signals, the need for impedance matching, and the nature of the mating sensor or signal source. Flat, ribbon-cable connectors (IDCs) and subminiature D-type connectors have a good following, too. We were surprised that removable screw terminals weren’t more popular. From our experiences wiring instruments and sensors, we’ve found it easier to wire a terminal block and snap it on a card than to wire each sensor directly to a card through a rat’s nest of other wires.

Software plays a key role

By now, you’re probably wondering about software. Of our 440 respondents, 438 answered our question about their current operating system, and 350 gave us an indication of their plans (Figure 3). Perhaps it’s safe to assume the “missing” 88 respondents plan to stick with the OS they use now. It’s no surprise that in a world dominated by Microsoft products, Windows appears high on the list, and that Windows 2000 gathers the most responses as the OS people plan to use in the next 12 months.

MS-DOS, the original—although often upgraded—OS for the IBM PC, introduced in 1981, remains on the list, but it barely shows up in plans for 12 months from now. As newer computers replace legacy systems, DOS will disappear except in some embedded systems. Unfortunately, DOS’s demise and its dwindling support by Microsoft and other suppliers leave in the lurch those people who still depend on older systems that continue to run well. Manufacturers take note: Users of older—even obsolete—DOS-based products still want you to support them.

For all the press that Linux has received—some of it in T&MW—it surprised us to see such little interest in it as an OS for data-acquisition and instrument-control applications. During the next 12 months, though, Linux pushes above the noise level as it goes from five to 25 possible users in our sample. Will Linux show up even stronger in our next survey? The general interest in Linux and the number of Linux-related magazines and Web sites makes us think so. 

Figure 3. Computers need an operating system, so we asked what OS respondents use now (blue), and what OS they plan to use 12 months from now (yellow).

Running applications is one thing, writing them is another. We asked our participants how they programmed their data-acquisition and instrument-control applications, and what programming languages they use. We also wanted to know what languages they plan to start using in the next 12 months. We allowed multiple answers, so in Figure 4, percentages add up to more than 100%.

There’s no way around it: Engineers still write their own code. Among the 328 people who told us what languages they now use, Visual Basic, C++, C, and LabView find the most use. Fewer people—144—told us which language they plan to start using, but the the lineup looks familiar, with C++, LabView, and Visual Basic taking the top three places among new users.

A programming language can take considerable time to learn and much longer to master, so it’s no surprise that “older” languages such as QuickBasic, HT Basic, Fortran, and other dialects of Basic still rank strongly. Some of these preferences will change, though, as MS-DOS-based systems go into retirement. But it’s interesting to see there’s enough life left in an old language such as HT Basic—the successor to Hewlett-Packard’s popular Rocky Mountain Basic—that TransEra (Orem, UT) still sells and supports it.

Our ranking also includes languages such as Java, Perl, and Tcl that we normally associate with Web-based applications. Perhaps the showing of these languages—even though it’s small—foretells the emergence of more Web-based instruments and applications in the near term. Our 12-month projections show Java and Perl will gain some new users.

Figure 4. The blue bars represent the computer language respondents now use to program data-acquisition and instrument-control applications. The yellow bars show the number of people who will start to use each of the languages represented above.

Readers speak for themselves

Few people who design and build data-acquisition or instrument-control systems can say, “We did A, and that worked, so we did B, and that worked, and so on.” Things just don’t work that well in the real world. From a list of 16 options (along with one open-ended choice), we asked people to identify their three biggest headaches when they plan and assemble a test system. The top six “headaches” facing respondents were:

• no suitable or compatible software drivers for the instruments or cards I want to use;

• cost of products;

• time, personnel, staff, and training issues;

•  insufficient experience using equipment specified;

• signal conditioning, noise, cable routing, and similar issues; and

•  lack of vendor application information.

Considering the current emphasis on plug-and-play hardware and software, and attempts by manufacturers and trade groups to standardize drivers, it’s surprising to see that a lack of suitable drivers tops the list of headaches. One respondent wrote that most drivers from all manufacturers are terrible, and it’s easier to write one’s own.

Our survey also asked two open-ended questions that gave participants a chance to express themselves. First, we wanted to know what single most important challenge these people thought they would face in the next six months in the area of PC-based data acquisition and instrument control. Many answers centered on “TTL,” which stands for time, talent, and loot, not a type of circuit. And many respondents face a continuing shortage of resources and still have to solve difficult technical problems. At the same time, they must keep up with the latest technologies, upgrade software to use the newest operating system, and expand the capabilities of their test systems. Companies ask these engineers to perform the same or more tasks with less money and fewer people—and to do so faster.

In addition to stating the usual TTL problems, many people responded that they need to connect instruments to an Ethernet network so they can control instruments remotely and so they can quickly access and distribute test data. Companies now provide Ethernet connections as standard equipment on instruments, but users still face challenges getting Ethernet connections to work and adding Ethernet connections to older equipment and test systems.

Our second open question asked people what message they want to send to vendors of instrument-control and data-acquisition hardware and software. A large number of answers cited information as a key need. Many respondents complained that manuals read as if written by engineers with poor writing skills and aimed at advanced users. People complained about the lack of basic examples and the need for more application notes that show how to use a product. Overall, documentation must be clearer and more complete, and it must provide detailed information about product specifications. According to several responses, users feel suppliers don’t understand the needs of people who are new to instrument control and data acquisition. Several respondents stressed the need for more source code that they can use to better learn how to apply hardware and software.

Granted, suppliers can’t answer every question or meet every need. Given the number of detailed and forceful responses we got to this question, though, we wonder if vendors survey their customers or pay attention to what customers say. We sensed discontent among the small sample we surveyed. Multiply that across an entire market, and companies should see an economic advantage to improving their tutorial materials and to offering some of it at a more basic level. T&MW

Jon Titus has written real-time software and designed embedded systems and computer/instrument interfaces. He worked in electronics for 10 years and spent nine years at EDN magazine prior to joining T&MW in 1993. He has a BS from WPI, an MS from RPI, and a PhD from VPI. E-mail: jontitus@tmworld.com.

Martin Rowe has 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.

Our 440 survey respondents reflect a cross section of T&MW’s circulation, which in turn reflects the people involved in the test-and-measurement industry. In May and June, we contacted many of the T&MW readers for whom we have e-mail addresses and asked them to take part in our Web-based survey. We’ve based the results in this article on their answers.                

Like the electronics industry, the survey was dominated by men. We received only nine responses from women in the group of 420 people who provided information about their sex, meaning 98% of these responses came from men. The people who responded fall predominately between the ages of 35 and 54 (35–39—(16%); 40–44—(21%); 45–49—(20%); and 50–54—(15%), and the mean age is 45. Almost everyone (434) replied to our question about education, and people with higher education made up slightly over 91% of this group: 52% have a college degree, 24% have a graduate degree, and 15% have some education beyond graduate school.

Almost everyone (433 out of 440) told us about the size of the company they work for, and a large group (44%) of these people work at companies with more than 1000 employees. We calculated a median company size of 754 people. We also asked what type of end products our respondents’ companies produce (Table A).

Although no survey can perfectly reflect what everyone thinks, we feel that the results in this article do provide a useful snapshot of the needs and concerns of our readers involved in data-acquisition and instrument-control applications.—Jon Titus