Genesis
In the beginning, when people wanted to measure things, they grabbed a specific measurement device that was expressly designed for the particular measurement they wanted to make. For example, if they wanted to measure a length, they grabbed a scale, or a tape measure, or a laser range finder and carried it over to where they wanted to measure a length. They used that specific device to make their specific length measurement. Then they walked back and put the device away in its carrying case or other storage, and put it back on some shelf somewhere where they originally found it (assuming they were tidy).
If you had a set of measurements to make, you needed a set of matching instruments. Occasionally, instruments did double duty (a chronometer with built-in compass), but fundamentally there was a one-to-one correspondence between the instruments and the measurements made.
That sort of arrangement works fine when you have only a few measurements to make, and you arenât in a hurry. Under those circumstances, you donât mind taking the time to learn how to use each sort of specific instrument, and you have ample time to do everything manually, finding, deploying, using, and stowing the instrument.
Figure 1-1 Manual measurements
Things went along like this for many centuries. But then in the 20th century, the pace picked up a lot. The minicomputer was invented, and people started using these inexpensive computers to control measurement devices. Using a computer to make measurements allows measurements to be made faster, and it allows measurements to be made by someone that might not know too much about how to operate the instruments. The knowledge for operating the instruments is encapsulated in software that anybody can run.
With computer-controlled measurement devices, you still needed a separate measurement device for each separate measurement. It seemed fortunate that you didnât necessarily need a different computer for each measurement. Common instrument interface buses, like the IEEE-488 bus, allowed multiple devices to be controlled by a single computer. In those days, computers were still expensive, so it helped matters to economize on the number of computers.
And, obviously, using a computer to control measurement devices through a common bus requires measurement devices that can be controlled by a computer in this manner. An ordinary schoolchildâs ruler cannot be easily controlled by a computer to measure a length. You needed a digitizing caliper or some other sort of length measurement device with a computer interface.
Things went along like this for a few years, but folks quickly got tired of taking all those instruments off the shelf, hooking them up to a computer, running their measurements, and then putting everything away. Sloppy, lazy folks that didnât put their measurement instruments away tripped over the interconnecting wires. Eventually, somebody came up with the idea of putting all these computer-controlled instruments into one big enclosure, making a measurement system that comprised a set of instruments and a controlling computer mounted in a convenient package. Typically, EIA standard 19â racks were used, and the resulting sorts of systems have been described as ârack-em-stack-emâ measurement systems. Smaller systems were also developed with instruments plugged into a common frame using a common computer interface bus, but the concept is identical.
At this point, the people that made measurements were quite happy with the situation. They could have a whole slew of measurements made with the touch of a button. The computer would run all the instruments and record the results. There was little to deploy or stow. In fact, since so many instruments were crammed into these rack-em-stack-em measurement systems, some systems got so big that you needed to carry whatever you were measuring to the measurement system, rather than the other way around. But that suited measurement makers just fine.
On the other hand, the people that paid for these measurement systems (seldom the same people as using them) were somewhat upset. They didnât like how much money these systems were costing, how much room they took up, how much power they used, and how much heat they threw off. Racking up every conceivable measurement instrument into a huge, integrated system cost a mint and it was obvious to everyone that there were a lot of duplicated parts in these big racks of instruments.
Modular Instruments
As I referred to above, there was an alternative kind of measurement system where measurement instruments were put into smaller, plug-in packages that connected to a common bus. This sort of approach is called modular instrumentation. Since this is essentially a packaging concept rather than any sort of architecture paradigm, modular instruments are not necessarily synthetic instrumentation at all. In fact, they usually arenât, but since some of the advantages of modular packaging correspond to advantages of synthetic system design, the two are often confused.
Modular packaging can eliminate redundancy in a way that seems the same as how synthetic instruments eliminate redundancy. Modular instruments are boiled down to their essential measurement-specific components, with nonessential things like front panels, power supplies, and cooling systems shared among several modules.
Modular design saves money in theory. In practice, however, cost savings are often not realized with modular packaging. Anyone attempting to specify a measurement or test system in modular VXI packaging knows that the same instrument in VXI often costs more than an equivalent standalone instrument. This seems absurd given the fact that the modular version has no power supply, no front panel, and no processor. Why this economic absurdity occurs is more of a marketing question than a design optimization paradox, but the fact remains that modular approaches, although the darling of engineers, donât save as much in the real world as you would expect.
One might be tempted to point at the failure of modular approaches to yield true cost savings and predict the same sort of cost savings failure for synthetic instrumentation. The situation is quite different, however. The modular approach to eliminating redundancy and reducing cost does not go nearly as far as the synthetic instrument approach does. A synthetic instrument design will attempt to eliminate redundancy by providing a common instrument synthesis platform that can synthesize any number of instruments with little or no additional hardware. With a modular design, when you want to add another instrument, you add another measurement specific hardware module. With a synthetic instrument, ideally you add nothing but software to add another instrument.