There's More to Data Acquisition Than A/D

Over the years, much has been written about the specifications and usage of analog inputs in data acquisition (DAQ) systems. However, in most applications, the analog input is only a piece of the hardware puzzle. Often overlooked are the key attributes of and specifications for the other parts of the system: the analog outputs, digital I/O, and communications interfaces.

Almost all systems require one or more D/A, DIO, and COM interfaces. How they are integrated into the system often is a critical aspect of overall system functionality.

Analog Outputs

Analog or D/A outputs are used for a variety of purposes in DAQ and control systems. To properly match the D/A device to your application, it is necessary to consider a variety of specifications.

Number of Channels
Make sure you have enough outputs to get the job done. If it's possible that your application may be expanded or modified in the future, you may wish to specify a system with a few spare outputs. At the very least, be sure you can add outputs to the system later without major difficulty.

Resolution
As with A/D channels, the resolution of a D/A output is a key specification. It describes the number or range of different possible output states the system is capable of providing. This specification is almost universally provided in terms of bits where the number of output states is defined as
2(# of bits). For example, 8-bit resolution corresponds to a resolution of one part in 28 or 256. Similarly, 16-bit resolution is one part in 216 or 65,536.

When combined with the output range, the resolution determines how small a change in the output may be commanded. To determine the size of this change, simply divide the full-scale range of the output by the number of output states. A16-bit output with a 0 to 10-V full-scale output provides 10 V/216 or 152.6-µV resolution. A 12-bit output with a 4- to 20-mA full scale supplies 16 mA/212 or 3.906-µA resolution.

The standard resolution of most DAQ D/A output interfaces is 16 bits, and you also will see some devices with 12-bit resolution. Although it is common now to see analog inputs with 20-bit or 24-bit resolution, D/A resolutions of greater than 16 bits are fairly rare in applications where DC accuracy is important. However, they are common in AC applications such as in the audio output world.

Accuracy
Accuracy often is equated with resolution, but they are not the same. An analog output with 1-µV resolution does not necessarily mean the output is accurate to 1 µV. Outside of audio outputs, D/A system accuracy typically is on the order of a few least significant bits (LSBs). However, be sure to check this specification because all D/A products are not created equal.

The primary error contributors of a D/A output are output offset, gain error, reference error, and nonlinearity, as illustrated in Figure 1. Both gain and reference are shown on a single graph because both contribute to an undesired change of slope in the output diagram. Remember that these errors are additive, and to get the overall system accuracy, you must account for the contributions from all error sources.

To view Figure 1 click here.
Figure 1. Most Significant and Common Error Contributors in D/A Output Systems

Additional contributing error factors that must be taken into account are more application specific than product specific. Errors may be generated by the D/A channel's output impedance as well as IR drops induced in the field wiring since neither the current flowing in the field wiring nor the resistance of the wiring is zero.

Both output impedance and IR errors manifest themselves when the D/A channel is required to drive a significant output current. Ohm's Law dictates that the error generated will be the product of the channel's output impedance, plus the resistance in the field wiring, multiplied by the current flowing. The equation for this error is

Resistance Error = (D/A Output Impedance +
Field Wiring Resistance) (Current Flow)

In many applications, the device the output is driving is high impedance and the current is so low that this error is negligible. However, many D/A outputs can drive 5 mA or 10 mA or even more.

If your application requires output drive in the milliamp range or higher, check this error. D/A output impedances typically are on the order of 0.1 Ω. A 10-mA signal flowing through 0.1 Ω generates a 1-mV error signal. The resolution of a ±10-V, 16-bit, D/A output channel is 305 µV so a 1-mV error actually represents an error of greater than 3 LSBs.

More insidious than the channel output impedance is the IR drop in the field wiring. While many people simply assume the resistance is low enough to have no impact, this often is not the case.

Note that 26- and 30-gauge, single-conductor copper wire have resistance of about 0.026 Ω per foot and 0.105 Ω per foot, respectively. If your output is driving 5 mA and connected by 50 feet of 30-gauge wire, you'll see an IR drop in the field wiring of about 53 mV.

In our typical case of a 16-bit D/A output with a ±10-V output range, 53 mV is about 173 LSBs. Table 1 shows the IR error induced in a number of different combinations of wire size, output current, and cable length.

Table 1. Resistance of Common Solid Copper Wire SizesNotes: All voltage errors are in millivolts. Cable length is total length including both output and return cables.

There are three options for reducing this IR drop error. First, you can minimize the distance between the D/A output and the device it is driving. Second, you can increase the size of your wire to reduce the series resistance. However, it is not always possible to do either of these, which leads to option three: use a board with sense leads or connections. The sense capability is designed to automatically compensate for IR losses in the system.

Basically, the sense leads are connected in parallel with the main D/A output leads but do not conduct any current. This allows the D/A converter to adjust its output so the voltage at the load is the desired level and not the output at the D/A converter itself. Many D/A output devices, particularly those designed to drive higher currents, will have sense leads that may be used.

Monotonicity
Presumably, if you command your output to go to a higher voltage, it will do so regardless of the overall accuracy. However, this is not necessarily the case. D/A converters exhibit an error called differential nonlinearity (DNL).

In essence, DNL error represents the variation in output step size between adjacent codes. Ideally, commanding the output to increase by 1 LSB would cause the output to change by an amount equal to the overall output resolution. However, D/A converters are not perfect, and increasing the digital code written to a D/A by one may cause the output to change 0.5 LSB, 1.3 LSB, or any other arbitrary number.

A D/A channel is said to be monotonic if every time you increase the digital code written to the D/A converter the output voltage does indeed increase. If the D/A converter DNL is less than ±1 bit, the converter will be monotonic.

If not, commanding a higher output voltage could cause the output to drop. In control applications, this can be very problematic because it theoretically becomes possible for the system to lock onto a false set point distant from the one desired.

Output Type
Unlike a myriad of sensor-specific input configurations for A/Ds, D/A outputs typically come in two flavors: voltage and current. Be sure to specify the right type for your system. Some devices offer a mixture of voltage and current outputs although most have only a single type.

If your system requires both, you may want to consider a current output since the current outputs often can be converted to a suitable voltage output with the simple installation of a shunt resistor. The accuracy of the shunt resistor-created voltage output is very dependent on the accuracy of the resistor used.

Also, the shunt resistor will be in parallel with any load or device connected to it. Be sure the input impedance of the device driven is high enough to not impact the shunt function.

Output Drive
Be sure to investigate how much current is required by whatever device you are attempting to drive with the D/A output channel. Most D/A channels are limited to less than ±5 or ±10 mA maximum.

Some vendors offer higher output currents as standard. For higher output drive, it often is possible to add an external buffer amplifier. If you are driving more than 10 mA, you will likely need to specify a system with sense leads if you need to maintain high system accuracy.

Output Range
The output range must be matched to your application requirement. Like its A/D sibling, it is possible for a D/A channel to drive a smaller range than its maximum although there is a reduction of effective resolution.

Most D/A output modules are designed to drive ±10 V; however, some will directly drive outputs up to ±40 V. Still higher voltages may be accommodated with external buffer devices. At voltages greater than ±40 V, safety becomes an important factor.

A final note regarding increasing the output range of a D/A channel: If the device being driven is either isolated from the D/A output system or if it utilizes differential inputs, it may be possible to double the effective output range by using two channels that drive their outputs in opposite directions.

Output Update Rate
Although many DAQ systems set and forget the D/A outputs, many more require that they respond to periodic updates. In control systems, loop stability or a requirement for control smoothness often will dictate that outputs be updated a certain number of times per second. Also, in applications where the D/As provide a system excitation, a certain number of updates per second may be required.

Verify that the system you are considering is capable of providing the update rate required by your application. It also is a good idea to build a little contingency into this spec in case you later need to spin the outputs a little faster.

Output Slew Rate
The slew rate, the second part of the output speed specification, determines how quickly the output voltage changes once the D/A converter has been commanded to a new value. It typically is specified in volts per microsecond. If your system requires the outputs to change and stabilize quickly, you will want to check your D/A output slew rate.

Ethernet-Based Cube and RACKtangle Chassis

Output Glitch Energy
As the output switches from one level to the next, a glitch is created. Basically, the glitch is an overshoot that subsequently disappears via dampened oscillation.

In DC applications, the glitch is seldom problematic. But if you are looking to create a waveform with the D/A output, the glitch can be a major issue because it may generate substantial noise. Most D/A devices are designed to minimize glitch, and it is possible to virtually eliminate it in the D/A system. But that also virtually guarantees that the output slew rate will be diminished.

Digital Inputs

Specifying the appropriate digital input for a system often is pretty straightforward, but there are a number of issues that must be considered. It is surprising how many people take the DIO part of their system for granted only to be later pressed into panic mode when they realize the DIO specified is not the right match for the application.

Input Type
Digital inputs are available in a wide variety of configurations. Some monitor voltage; some are current activated. Some accommodate DC signals while others can sense AC and DC signals. Still other inputs indicate the status of an electrical contact such as opened or closed. Be sure to identify and categorize all of the digital inputs required by your system early on. It is surprising how many people specify and buy a DAQ system with a cavalier it's-only-digital-I/O attitude only to be bitten later.

Input Impedance
Input impedance, or input drive required, is an often forgotten and problematic specification. Some inputs, such as most opto-coupler inputs, frequently require a substantial drive current. Many outputs are only capable of supplying a very small output drive. Be careful that each of your inputs will be provided with an appropriate drive capability.

Input Range
Don't try to monitor your 24 VAC signal with a logic level input. You won't like the results although your DAQ vendor might because you will almost certainly be facing a repair or replacement charge.

Sample or Update Rate
Like every other element of a DAQ system, timing often is a critical component. Be sure your input system is fast enough to respond to signals provided within the timing required by your system.

Special Considerations
Another thing to consider is hysteresis. Basically, hysteresis is a dead zone in the switching behavior where a low-to-high transition occurs at a higher voltage than a high-to-low transition. This hysterisis zone reduces the input's susceptibility to noise.

One more common capability is input debouncing. When the actual contact in a switch or relay closes, it typically will bounce up and down one or more times before it finally settles into a fully closed position. The bounce cycle often is as long as 100 ms.

A debounce circuit slows the response of the digital input so it only appears as closed once the contact has stabilized. The chatter is sometimes only a minor inconvenience in a static digital input but can create large errors in applications where the digital input drives a counter.

Additional diagnostic capability also is provided on some inputs. The price of A/D converters has decreased to the point where some manufacturers are monitoring their digital inputs in the analog world.

A/D-based boards like UEI's DNA-DIO-448 provide the same digital information as a standard board but also offer a diagnostic voltage measurement mode. In the diagnostic mode, the actual digital input voltage is read. This information is extremely useful in identifying broken wires, short circuits, and damaged output devices.

Digital Outputs

Digital outputs require the same scrutiny and many of the same considerations as digital inputs. These include careful consideration of output voltage range, maximum update rate, and maximum drive current required. However, the outputs also present a number of specific considerations:

Relay vs. Semiconductor Outputs
Relays have very high off impedance, very low off leakage, very low on resistance, ambivalence between AC and DC signals, and built-in isolation. However, they are mechanical devices and, as a result, provide lower reliability and typically slower response rates.

Semiconductor outputs typically have an advantage in speed and reliability. Semiconductor switches tend to be smaller than their mechanical equivalents so a semiconductor-based digital output device typically will provide more outputs per unit volume.

When using DC semiconductor devices, be sure to consider whether your system requires the output to sink or source current.

Current Limiting/Fusing
Most outputs, particularly those used to switch high currents of >100 mA, offer some sort of output protection. There are three types most commonly available:
• Simple Fuse—Fuses are inexpensive and reliable but cannot be reset and must be replaced once blown.
• Resettable Fuse—Typically, this device is a variable resistor. Once the current reaches a certain threshold, resistance begins to rise quickly, ultimately limiting the current and shutting off the output. When the offending connection is removed, the resettable fuse reverts to its original low-impedance state.
• Current Monitor—An actual current monitor that turns the output off if an overcurrent is detected does not require replacement following the event. Many implementations of the controller configuration also allow the overcurrent trip level to be set on a channel-by-channel basis, even within a single output board.

Output Confirmation/Read-Back
For critical controls, it often is desirable or even required to read back the status of a digital output. This can be done by connecting a digital input to the output and monitoring it, but that doubles the number of DIO channels required.

Many digital output devices provide a way to automatically read back the state of the output. Be a bit careful with how the read-back is implemented. In some products, the read-back simply is the status of the latch or buffer controlling the output and not of the output itself. This allows the application to confirm that the correct data has been written to the device, but it does not confirm that the output actually has gone into the desired state.

The more secure systems monitor the actual output voltage or current. UEI's DNA-DIO-432/433 measures both output voltage and current, confirming the output state. This capability also provides a diagnostic capability for detecting short/open circuits as well as other suspect conditions or behavior. Many other vendors also offer some type of output-confirmation or read-back capability.

Counter/Timer Functions

Counter/timers are used for functions such as measuring frequency, pulse width or duration, counting events, and generating periodic or pulse-width modulation outputs. Counters typically can be configured as up counters, down counters, or up-down counters. Up counters are the most commonly used of the configurations and, as the name implies, simply start at zero and count up.

Down counters are most commonly used as timers or alarm generators such as watchdog timers. Typically, a preset value is loaded into the counter register, and on each rising or falling edge,
the counter is decremented by one. When the counter reaches zero, it typically generates an interrupt or reset pulse so the application knows the specific number of input events that has been obtained.

Up-down counters are used when the difference in the number of events between two inputs is important while the absolute number of events is not. Up-down counters generally are used in devices such as quadrature encoder inputs or balancing applications.

Communications

A very common requirement of many data acquisition and control systems is the capability to communicate with other intelligent devices. This communication is in addition to whatever link is used to connect the DAQ system to the host computer. These additional communications links may be acquiring data, controlling systems elements, synchronizing multiple systems, or talking to a person via some type of human-machine interface.

Serial I/O
Although people have been predicting its death for about 30 years now, the serial interface is still alive and kicking. In fact, in most surveys taken even today, it is the most popular communications interface in DAQ systems. Since it has been around for so long, the standard is pretty well developed, and most modern interfaces share very similar specifications. Things to look for include the serial standard such as RS-232, RS-422, or RS-485; maximum baud rate; support for hardware triggering; user-selectable number of data bits and stop bits; and parity settings.

Isolation, particularly channel-to-channel, can be important in many applications, especially those connecting to distant devices or equipment, and the local grounds may be different by many or even hundreds of volts.

The capability to transmit as well as receive also is an issue. Some DAQ systems only receive data while some external devices require a query to be sent over the serial port before transmitting any data.

In aircraft applications, not only is it critical to acquire data from various on-board sensors, but also to correlate that data with what's going on elsewhere in the aircraft. For example, if an application wishes to test the stress on a wing spar in flight, it's very important to tie the stress values to information in the avionics including air speed, bank angle, angle of attack, and altitude.

CAN-bus
In automotive and truck applications, it often is necessary to measure a parameter as well as integrate the data with other automotive parameters. For example, an application measuring interior noise would likely want to correlate that data with vehicle speed, engine rpm, and transmission gear. This information could be easily obtained from the vehicle's CAN-bus.

Conclusion

When specifying a data acquisition and control system, it's easy to get so tied up in the analog input system that other parts of the system are left to afterthought. However, this attitude is a path to trouble because nobody wants to implement a system that only does 85% of what's required. Once you have the A/D specs nailed down, be sure to consider the other parts of the system. When it comes time to build and install your system, you'll be glad you did.

About the Author
Bob Judd is director of marketing at United Electronic Industries. Prior to joining UEI, he was general manager and vice president of marketing and hardware engineering at Measurement Computing and previously vice president of marketing at MetraByte. Mr. Judd, who has been involved in the PC-based DAQ market for more than 20 years, holds a bachelor's degree in engineering from Brown University and a master's degree in management from MIT. United Electronic Industries, 27 Renmar Ave., Walpole, MA 02081, 508-921-4557, e-mail: [email protected]

Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!