EMC Tests using High Voltage Spark Gaps
In industrial environments, spark flashover often present the most common, toughest and widest ranging interference sources. It is therefore obvious that in addition to the standard test methods (e.g., IEC/EN 61000-4) a spark test should also be carried out. The interference impulses from IEC/EN 61000-4-4 more or less simulate the spectrum of from a spark. However, the fields from IEC/EN 61000-4-3 (10 V/m, CW) are not in any way similar to spark generated fields.
Sparks result primarily from switching systems (e.g., relays, switches) and in electric motors (collector) but can also result from electrostatic and atmospheric discharges, ignition systems (combustion motors), welding equipment, etc. Interference generated in this way is extremely broad-band (up to 1 GHz). For a breakdown at atmospheric conditions, a field strength of approx. 3 million V/m is necessary. In the immediate vicinity of a spark the resulting field strength is of the order of several hundred thousand V/m.
Tests on a Measurement Transducer
The test object is a standard Pt-100 measurement transducer, type RTM 70-C, with a measurement range of 0-100°C = 0-10 V. The 0-10 V output of the module is connected to a seven-segment display which shows the ambient temperature (measured by a Pt-100 sensor) in °C (resolution 1/10°C). The high voltage sparks is applied to the sensor cables during normal operation. During the tests, the frequency of the sparks is varied between 1 and 100 Hz. No change in the display (1/10°C resolution) could be detected during any of the tests. The entire spark current (with frequency components or up to approx. 1 GHz) has to flow through the module before returning to the high voltage generator via the earth cable.
In such tests various interference mechanisms come into play:
Field related interference: The spark gap is approx. 1 cm. Separation from module, approx. 5 cm. Measurement transducers can tolerate field strengths of several ten thousand to several hundred thousand V/m. In spite of the fact that the measurement transducer is not shielded (plastic casing), it is unaffected by the interference. Variation of the order of a few mV can be detected.
Conducted interference: The Pt-100 sensor is connected to the measurement transducer via two-wire technology. The positive wire leads the measurement current through the sensor (approx. 1 mA), and at the same time the drop in voltage is measured at the module. The negative wire leads the measurement current back and serves as the zero point for the voltage measurement. The wire is connected to the module¢s analog ground. The spark is applied at the negative wire. Coupled into the neighbouring positive wire are high-voltages interference signals (direct galvanic and via the electromagnetic fields) which are completely eliminated by the over-voltage arrester at the input and by the various filter stages. At the output and the power supply connections, larger interference voltages will certainly occur, but will not affect the signal. The display (also available from SOCLAIR ELECTRONIC) has over-voltage arresters and various filters at the signal input and power supply which renders it completely immune to interference voltages. None of the wires are screened. The entire HF spark current flows through the module's ground system without causing interference of any kind.
A spark applied to the positive cable must, of course, have a noticeable effect: due to direct galvanic coupling, an interference signal which contains a DC element is superimposed on the measurement signal. The DC element cannot be eliminated by the filters. However, of critical importance is the fact that with this type of coupling, the over-voltage arrester and the filters protect the module from high voltage (approx. 30,000 V).
How much does good EMC protection cost?
It goes without saying that such extraordinary high interference security is not easy to achieve. However, assuming the relevant know-how, the cost of the product scarcely increases. Of greatest influence is the layout (the geometry of the circuit, and of the ground in particular), the grounding concept, the circuit technology and the selection of EMC-resistant components. The additional circuit elements (over-voltage arrester, filters) add no more than a few percent to the total price, assuming optimum design, and the other measures cost practically nothing. If the measurement transducer is designed properly then additions such as metal casing, screened cabling and other expensive screening elements become unnecessary.