Controlling communications

Most popular is the 4 to 20mA standard, in which a pair of wires makes a single variable, point to point connection link. Currents in the range 4 to 20mA circulate through the loop, with 4mA defined as 'zero' (as well as the 'loop integrity OK' signal) and 20mA as full scale. A basic telemetry loop consists of a power supply, a transmitter operating as a current source (driven by a sensor), the wire loop and a receiver that measures loop current by sensing voltage across a precision resistor, all connected in series. A control loop has the same components and configuration, but its transmitter is driven by a control signal and its receiver is a current controlled actuator. Assuming perfect integrity, loop accuracy depends on how precisely the sensor or control signal is transformed into a current. This, in turn, depends on the linearity and stability of the transfer function, as measured at the receiver. To maintain system accuracy, the transfer function's sensitivity to factors such as temperature, length and voltage stability must be minimised. The output of a current loop transmitter must also provide very high dynamic impedance – high enough to ensure that drops in loop voltage at the transmitter terminals exert a negligible influence on the output current. The transmitter circuit in figure 1 provides high output impedance (108 to 109 ohms) and operating compliance voltages from 3V to 90V. The lower compliance limit is set by the minimum voltage needed by the op amp, in this case 2.5V. The maximum voltage is set by the maximum allowed power dissipation in the output device – an n channel depletion mosfet. The output constant current source consists of a mosfet depletion transistor in series with a reference quality shunt regulator and current sense resistor (RA). The regulator powers a precision mosfet input op amp, which controls loop current by driving the mosfet gate. The op amp's inverting input connects both to the negative side of the shunt regulator and to the op amp's negative supply terminal. This enables the local feedback loop to keep the drop at RA (and hence loop current) at the same value as the difference between the op amp's non inverting input and the negative common rail (at the other side of RA). The potential at the non inverting input is set by the regulator, the value written to the d/a converter and the values of RB, RC, and RD. The 12bit d/a converter has a differential non linearity of less than 1 LSB, which guarantees a monotonic output. The converter's 3V operating voltage and low operating current (a few µA) enables loop powered operation and a wide compliance range for the digital loop interface. Accuracy of the transfer function from the d/a converter to the output stage is determined by the accuracy of the shunt regulator reference and by the tolerance of the sense resistor (RA), as well as those resistors in the interface network. The influence of all other error components is rendered negligible by the op amp's high gain, input characteristics and ability to operate with inputs down to the negative rail. For higher accuracy, standard trimming techniques can be used. The op amp's ability to operate with inputs at or below the negative rail is essential to the circuit's accuracy, because that enables the amp to force its operating current through the output current sense resistor, where it becomes a part of the output current under local feedback control. The basic loop receiver is a precision sense resistor (RA in figure 2) connected in series with the loop current. An op amp reads the voltage drop across the resistor and calculates the value of loop current. In this case, the resistor voltage is processed and digitised next to the resistor, with circuitry powered by the loop current being measured. The sense resistor is in series with a 3V precision shunt regulator, used to power the amp, as well as the a/d converter and the digital isolator. Thus, the receiver appears to the loop as a 12.5O resistor in series with a fixed voltage drop of 3V that is independent of the loop current. The input amp boosts the drop across the sense resistor with an inverting gain of 10 and shifts the voltage level into the a/d converter's input range. This is possible because the input amp can operate with its inputs at the same level as its negative supply rail. Gain accuracy is guaranteed by the op amp's high open loop gain and by the ratio accuracy of an ultra precision resistive divider. Isolated digital interface Because loop transmitters and receivers must be galvanically isolated, so too must their digital interface and power supply. Transmitter and receiver can be powered by leeching power from the loop operating current and the digital interface can be isolated by either magnetic or optical coupling. The d/a and a/d converters in figures 1 and 2 use a standard three wire interface for reading and writing data. For the d/a converter, three signals flow from the control cpu to the peripheral. For the a/d converter, two signals flow cpu to converter and the third from converter to cpu. These signals are coupled in both directions by a magnetically linked digital coupler that comprises three channels of data. On the sending side, the coupler includes an edge detector that consists of two Schmitt trigger input inverters looking at the same input signal: one at the signal itself; the other at a version delayed by about 20ns. Looking differentially at the inverter outputs, one sees a pulse at each transition of the input signal. The sign of the pulse depends on the sign of the signal edge and the pulse duration is the time delay between inverter inputs. This pulse is applied to the primary of transformer TX, which is connected across the edge detector outputs. At the receiving side, a non inverting buffer is configured as a flip-flop by connecting the secondary of TX between the buffer's input and output. Each pulse from the edge detector sets and resets the flip-flop, reconstructing the edge detector input signal at the output. The propagation delay time is about 15ns and the circuit operates with a minimum pulse width of 40ns down to dc. That action allows data-link speeds up to tens of megahertz. The magnetic coupler imposes an inconvenience: output turn on status is predictable only after the first edge transition at the input. This is not a problem for the receiver digitiser, because the first read can be discarded, but it might be a problem if the transmitter is used to drive an actuator. Optocoupler isolation works just as well, but the data link can operate no faster than a few tens of kilohertz.