For The Agency | For The Technology Developer | For The Magazine Publisher | For The Individual

Chapter 2

Basic Channel Input Specifications

In a sense, data acquisition hardware is a commodity. All you need to make a single-channel data acquisition system, as shown in Figure 2.1, is an appropriate sensor, an analog link, an A/D converter, a digital link, and a computer.

The sensor is a transducer that "feels" the physical parameter of interest, and puts out an analog (voltage) signal whose level indicates the value of the physical parameter. That analog signal travels over the analog link to the A/D converter, which actually makes the measurement and puts out a digital word indicating the analog level and, by extension, the physical parameter’s value. That digital word then travels through the digital link to the computer. The computer, finally, absorbs the information and stores it so that somebody can, later on, do something useful with it.

Click image to see full size

Figure 2.1: Each channel of a data acquisition system includes a sensor, an analog signal link, an A/D converter, and a digital signal link into a computer that receives and archives the data in a data storage device.

That is the minimum you need for a single-channel data acquisition system. Other amenities, such as signal conditioning and data buffering, are icing on the cake to make the whole process go more smoothly, accurately and reliably.

For a multi-channel data acquisition system, you need only to duplicate the sensor, analog link, A/D converter, and digital link for each channel. Since the computer can really accept only one data word at a time, somewhere along the line you need to add in a scanner to multiplex the signals together, interleaving them so that they arrive sequentially rather than in a bunch. The scanner funnels all your data channels down to one channel feeding the computer.

All data acquisition boards essentially do the same thing: they provide the A/D converter, the scanner and the digital link. It is in this sense that DAQ boards are commodities.

The idea of a "commodity," however, also carries the connotation of sameness: one commodity of a certain type is pretty much the same as the next one. When looked at in this light, DAQ boards cease to look like commodities at all.

Data-Acquisition Specifications

As a DAQ-system designer, you don’t care at all about a lot of the things closest to the hearts of DAQ-board suppliers. An obvious example is the board’s ability to serve in many different applications. The board supplier wants all his or her boards to cover as many applications as possible because it minimizes the number of different boards that must be designed, manufactured, stocked and so forth. You, on the other hand, only care about one application—yours!

Having a board covering many applications is likely to influence its price to you, and you certainly are interested in minimizing that price. You don’t, however, care how that price gets minimized. You just want (among other things) the lowest one you can muster. If the manufacturer lowers the price by making more-general-purpose boards, so be it.

What you really care about are the board’s qualities that fit it for your particular application. It is up to you to identify those qualities that are most important for your application and put values to them. The qualities are the specifications you will look at and their values are what you want the board to meet.

Begin with an inventory of the physical parameters you want to measure. Go through your physical system bit by bit and identify everything you might want to monitor there. That will give you two things: a total input-channel count and a sense of what type each input is.

In fact, the only thing DAQ-board inputs sense is voltage. Even an input that nominally senses current really only senses the voltage that the input current develops across the input impedance. The input-type specification really just tells you that input channel has signal-conditioning circuitry that makes it work particularly conveniently with a certain type of sensor.

Of course, having a DAQ input conveniently set up to work well with, say, Type K thermocouples can be a great convenience indeed!

The next bit of information you need for each channel—that is, for each physical parameter you want to monitor—is the range of values it is likely to take on. DAQ-board manufacturers often specify gain and offset rather than range. Gain and offset acknowledge the fact that the board’s A/D converter works more-or-less linearly over a limited dynamic range of ADC-input values. To widen its range, the signal-conditioning electronics ahead of the A/D converter includes a DC amplifier with programmable gain, and a DC level shifter to provide programmable offset.

Suppose the A/D converter is designed to take voltages from zero to 1 volt positive, but you’re sensor puts out voltage in the range of 6.0 to 6.1 volts. The span of those voltages is 0.1 volts, so to take best advantage of the ADC’s dynamic range, you’d best put in a gain of 10, increasing the span to 1 volt.

But, that would give you a totally unacceptable range of 60 to 61 volts! So, you need an offset of -6.0 volts at the amplifier input (changing the range from 6.0-6.1 volts to 0.0-0.1 volts), then apply the 10X gain. The signal-conditioning electronics then translate the 6.0-6.1 volt range of the sensor’s output to the 0-1 volt dynamic range of the A/D converter’s input.

Bandwidth and Sampling Rate

After determining each physical parameter’s range, you also need to find out how fast it varies. Actually, all you care about is the highest Fourier component of interest in the signal waveform. If, say, your physical parameter is the total weight of a tree whose growth pattern you’re studying, your highest frequency component could have a period of months. If, on the other hand, you’re looking at the sound produced by a motorcycle exhaust system, the highest frequency of interest will likely be in the high audio range (perhaps a harmonic of the self-resonance of some baffle in the muffler).

That highest Fourier component of interest is usually called the signal’s bandwidth (on the presumption that the lowest component of interest is at zero Hertz). Although DAQ-board electronics, like all electronics, have a characteristic bandwidth that must be large enough to pass all the significant signal-component frequencies, the sample rate turns out to be a more important board specification.

The theoretical minimum sample rate you should use for any data acquisition application is two samples per cycle at the maximum frequency of interest. Many manufacturers recommend at least four samples per cycle. I’ve always suggested shooting for 10 samples per cycle if you can get them. It is reasonable to assume that any board manufactured to achieve a given sample rate will have signal-conditioning electronics with a bandwidth high enough to pass any frequency components that can be captured at that sample rate.

Between them, the type, range and bandwidth characterize the physical parameter that a single channel of a data acquisition system will be monitoring. Those characteristics need to match the DAQ-board specifications of input type, gain and offset, and sample rate.

Chapter 1 | Chapter 2 | Chapter 3 | Chapter 4