|
Robustness and reliability in long distance data transmission have made RS-485 the industry's interface workhorse, particularly in building and industrial automation. The trend towards increased networking, whether through simple network expansion or upgrading existing nodes with high-performance equipment, pushes the electrical characteristics of conventional transceivers to the limit.
The main culprits are frequency dependent losses of the transmission cable, predominantly increasing at high frequencies. These losses can degrade a signal to the extent that it becomes undetectable by the receiver. Counteracting these losses, high-frequency compensation methods are needed to restore the signal quality. The two most commonly applied compensation techniques are driver pre-emphasis and receiver equalization.
Aimed at helping the design engineer to increase networking performance, this article discusses cable losses and their impact on signal quality. The paper describes two compensation methods, and suggests solutions for typical application scenarios.
Cable Losses
Typical RS-485 cabling is of the 120 terminated, unshielded twisted pair (UTP) type. Because of relatively long propagation times, the cable is modeled as a transmission line (Figure 1a), with its characteristic losses R, L, C and G that yield the familiar line length versus data-rate dependency shown in Figure 1b.
Figure 1: (1a) transmission line model, and (1b) line length vs signaling rate characteristic.
While at low frequencies, the line resistance R determines the maximum cable length. At high frequencies, R and C components dominate, causing the cable to act as an R-C, low-pass filter.
Figure 2 shows the frequency responses of several UTP cables of varying lengths with the reactive losses occurring predominantly at higher frequencies. To ensure sufficient signal amplitude reaches the receiver, cable losses must be compensated through compensation techniques, adding high frequency content to the transmission signal.
Figure 2: Frequency responses for Beldon 3105A cables of various lengths
Cable Loss Compensation
The principle of cable loss compensation is to approximate the inverse of the cable's transfer function as accurately as needed, and to implement it into a functional compensator stage in series to the transmission cable. The summation of the compensator gain and the cable loss ideally results in a net gain response close to unity across the frequency range of interest (Figure 3a).
Two compensation techniques most often applied are pre-emphasis and receiver equalization.
Figure 3: Principle of cable loss compensation (a), via pre-emphasis (b), and receiver equalization (c) Click here for larger image
When the compensation stage is implemented in the driver (Figure 3b), the amplification, or emphasis of high frequency components, occurs before the signal is sent across the data link " hence, the term Pre-Emphasis. If the cable response is equalized at the receiver end of the cable (Figure 3c), it is called receiver equalization.
While both techniques aim for the same result, their functional principals differ, yielding a number of benefits and drawbacks between the two methods.
Pre-emphasis
Pre-emphasis adds high-frequency components to the original signal by boosting the driver output for a fraction of the bit period at each signal transition. The theory behind the signal boost is demonstrated in Figure 4.
Figure 4: Harmonics spectra of a 50 percent duty cycle (a); a 25 percent duty cycle (b); and the combined signal (c)
When transmitting a data stream of alternate zeros and ones, conventional RS-485 drivers transmit signals using a 50 percent duty cycle, whose frequency spectrum purely consists of odd harmonics of the original square wave (Figure 4a). A 10 percent duty cycle pulse, however, shows an almost equal level of even and odd harmonics (Figure 4b). Combining both waveforms to one (Figure 4c), provides a signal enriched with high-frequency content and also with higher signal levels.
Boosting the driver output for a short time interval creates the necessary pre-emphasis pulse that adds odd and even harmonics to the frequency spectrum of the original square wave, thereby counteracting the low-pass filter response of the cable. The results are sharper signal edges at the receiver input and less displacement in time, thus yielding reduced (spell out ISI here) ISI.
Pre-emphasis, however, has several limitations:
1) The pre-emphasis pulse exceeds the nominal signal levels typically by 90 percent. This clearly violates the RS-485 standard, which allows less than 10 percent overshoot during signal transitions.
2) The pre-emphasis pulse generates high-frequency emissions, which does not permit the use of transceivers with pre-emphasis in applications with high restrictions on electromagnetic interferences.
3) Pre-emphasis requires more power per bit due to the square-law relation between signal voltage and signal power. For example, boosting a nominal driver output of 1.5V by 90 percent nearly quadruples the instantaneous power consumption.
|
