PhilipBurgert

Everybody's Converting

April 9, 2009
Analog-to-Digital Conversion Facilitates Data Use
By Philip Burgert

Moving machine signals and other data from machine measurement and sensing devices to control systems and across networks means that most if not all analog signals generated on machines eventually are converted to digital.

For industrial systems and applications, it’s practical to convert analog signals to digital for amplification and computation, multi-variable and bidirectional communication, reducing media cost and complexity or improving system accuracy and speed.

This conversion is necessary when analog signals are to be transmitted using an industrial network, says Helge Hornis, manager of the intelligent systems group with Pepperl+Fuchs. Programmable logic controllers also operate on a digital version of the original analog signal, but since the analog signal is connected to a card in the PLC rack, this fact is less obvious, he notes.

With distributed control systems, programmable logic controllers and remote input/output cards, analog signals commonly are converted to digital via onboard analog-to-digital converters that typically are read and controlled by a local processor. “For industrial networks, a field-mounted analog input block digitizes the analog signal into a series of bits that are then transmitted via the network with analog values typically represented as 12-, 14- or 16-bit values,” says Hornis.

Since the A/D conversion takes place in a field module, cabling from the device to the module is important. Voltage drops in the cable create inaccuracies, and the same is true when the analog device is connected to an analog card in the PLC, adds Hornis. Because the conversion takes place at an earlier point if a field-mounted analog bus module is used, wiring is less critical simply because the cables can be shorter.

Shorter cables also are preferred when considering environmental electromagnetic noise, he adds, noting that, before converting the signals to their digital representations, noise can influence signal levels significantly. Once converted to a digital format, the much stronger protection methods offered by the digital network protect the data, says Hornis. 

Connector selection plays varying roles, depending on the type of signal measured, says Don Lupo, director of marketing and sales for process products at Acromag. “In cases when signals are very low-level—µV or below—the connector materials, thermodynamics of the system and any contact resistance will have a greater impact,” he states. “In these cases every connection becomes a small thermocouple junction due to the dissimilar metals and temperature gradients in the system. The addition of static and dynamically changing leakage currents can also impact the signal measurement across these junctions. When analog signals are greater than 10 Vdc and specifically more than 100 Vdc, the insulation and spacing of the materials play a greater role in isolating the signal from the other channels, earth ground and people.”

The combination of capacitance and resistance of the connectors impact the speed at which analog signals can be passed, and lower connector capacitance, or communications link capacitance, generally means higher bandwidths can be obtained, he says.

Maintaining data integrity in analog-to-digital conversions involves choosing a good communication protocol with the needed reliability, suggests Lupo. In Ethernet transmissions, the use of TCP/IP ensures that data will make it through to the end location, he says.

Analog-to-digital conversion data integrity also can be protected by increased sampling and retries, says Lupo. “It’s always possible that one single piece of input data can get corrupted,” he says. “It’s good practice to include additional sampling and retry algorithms to verify the data before acting on it.”

It also is a good practice to loop output signals back to inputs for data integrity and validation purposes, says Lupo. “Using feedback techniques is the only real way to know if analog or digital outputs are maintaining their integrity,” he says.

Another important test is to determine if there is a lock-up condition on the A/D that causes the data to appear to be in a valid range while it actually is stuck in a fault condition. One strategy for this is to periodically sample a known good reference source to verify the circuit is operating properly. This is generally referred to as a heartbeat or self-test approach, notes Lupo.

Another way to maintain data integrity is to save it to multiple locations and develop a voting scheme before using the data, says Lupo. “If three pieces of data are saved off to three different memory locations, for example, then the data can be considered valid if two of the three memory locations agree,” he says, adding that complete redundancy also helps.

Signal conditioners are important to ensure data integrity and protect data from noisy signals, says Bob Smith, vice president of marketing at Dataforth. “It is advisable to put signal conditioning between your sensor and A-to-D system,” he states. “You should at least filter out the noise of any systems that are not isolated.”

Smith notes that in a data acquisition system the connectors are there to provide an interface between the field wiring and the data acquisition system. “Most connectors today are very, very good for the signal levels that we are looking at,” he says. “Most connectors are nickel-plated or tin-plated for the less expensive ones. They’re all adequate, as long as the installer is careful to get the wire stripped and installed properly with the right amount of pressure, like on a screw terminal clamp.”

The right size wire into the screw terminal clamp that is provided is important, says Smith, noting that a too-small wire in a clamp made for 14-gauge wire might not be grabbed very securely.

The number of bits is another element that Smith also cautions users to consider. “A thermocouple is not very accurate—plus or minus a few degrees,” he says. “The only reason why you’d want a 16-bit converter on there is to actually look at some deviation.” He notes that 16-bit A/D converters are increasingly inexpensive.

Philip Burgert is a freelance writer, specializing in the technical trade media.