ever, some equipment generates appreciable power at higher frequencies. It is for this reason that the measurement is specified over the range of 20 Hz to 1 MHz. For equipment with significant high frequency energy, capacitive coupling to the user can present a hazard, and the measuring equipment must have adequate bandwidth and sensitivity to perform the measurement. A wide bandwidth TRMS meter such as the hp 3400 is one example of a suitable instrument.

There is a commercially available instrument, the labsMate Model LMT series Touch Current Tester/Leakage Current Tester. This unit incorporates all three of the networks-burn, perception/reaction, and let-go. It isnĄŻt clear from the operating instructions what its bandwidth or high frequency power dissipation limitations are. The provided instructions specify measurement accuracy up to 1 kHz. However, we checked the bandwidth of this instrument in our lab for the perception/reaction setting at modest currents and found adequate performance to at least 100 kHz.

Component Selection and Calibration

The issues in component selection are primarily those of voltage rating and dissipation. For the perception/reaction network, where the only currents expected are low-level leakage, the voltage and dissipation ratings are not critical. However, if either of these two conditions occur, voltage and power ratings will be important:

1. Significant high frequency currents are expected, and compliance depends on weighting, or

2. an accidental connection to mains power occurs during testing

In the case of high frequency energy, significant currents can be passed while still meeting the requirements because of the low-pass weighting of the signal. In this case, RB would dissipate significant power.

In the case of an accidental connection across the mains, current flows through RS and RB, and significant voltage appears across CS and C1. Because the power dissipated in the resistors will be large, the most likely effect will be the overload and opening of RS and/or RB. Consequently, the most logical design choice is to rate C1 and CS for the likely expected voltages (usually 250 VAC) in the event of a mis-connection, and to perform a mini-calibration before use in the form of a resistance check for continuity of RS, RB, and R1, combined with a spot check of the transfer function at select frequencies.

A more detailed calibration would involve the following steps:

1. Verification of the resistance between terminals A and B at 2.00 kOhms

2. Veriication of the resistance between terminals A and U2 (at the R1C1 junction) as 11.5 kOhms (or equivalently from point B to U2 as 10.5 kOhms).

3. Frequency response of the transfer function (U2/VAB) over the entire frequency range. It should be adequate to check the response at 100 Hz, 1 KHz, 5 KHz, 10 kHz, 100 kHz, and 1 MHz, as well as to verify that the transfer function varies smoothly in the interval between those points.
LabsMate INC.
address: 728 W. Stottler Pl. Gilbert AZ 85233 U.S.A.
www.labsmate.com
info@labsmate.com
occurs.

Frequency Sensitivity and Measuring Network

IEC 60990 gives three measurement networks for standards writers to select. These are the unweighted, the perception/reaction, and let-go networks, respectively. These differ in the manner in which currents of different frequencies are weighted. The burn network is essentially unweighted, while the other two give higher frequency currents less weight for purposes of comparison to a given limit. They do this via the addition of a low pass filter section which attenuates higher frequencies. (See figures 1, 2, and 3).

 

0103reflectionsfig1.gif (18078 bytes)

Figure 1: Unweighted touch current measuring network, used for assessing potential for
electric burn in some standards

0103reflectionsfig2.gif (19427 bytes)

Figure 2: Measurement network with weighting appropriate for perception and reaction. This is used in
IEC 60950. Leakage is U2/500. Calibration can be made by performing a DC check and measuring the
frequency response U2/VAB.

0103reflectionsfig3.gif (16836 bytes)

Figure 3: Measurement network with weighting appropriate for assessing let-go currents. This is
less commonly used than the perception/reaction network.

 

The perception/reaction network is perhaps the most commonly used. It is the one specified in IEC 60950, and the one weĄŻll focus on. The schematic for this network is shown in Figure 2. It implements the frequency sensitivity weighting shown in Figure 4 and in Table 1. The net current reading, for purposes of comparison to a limit (e.g, IEC 60950, accessible metal to ground) is the r.m.s. value of U2 divided by 500. At low frequencies, we see that this is the voltage across RB due to the current flow. At higher frequencies, this voltage is further attenuated by the single pole filter R1-C1.

 

0103reflectionsfig4.gif (20265 bytes)

Figure 4: Weighting function implemented by the perception/reaction network. An increasing
amount of current is required as frequency increases to produce the same physical effect as a
low frequency current.

0103reflectionstable1.gif (65559 bytes)

Table 1: Transfer function of perception/reaction network. The second column indicates the transfer
ffunction U2/VAB, and is the easiest to use for calibration (see text). The third column is the inverse
of this ratio (i. e., VAB/U2). The last column shows how great an input voltage would actually be
required as a function of frequency to the network to get a weighted indication of 1 mA (i. e., U2=0.5 volts).

 

Note that the weighting curve, Figure 4, goes up with frequency while the transfer function of the measuring network, U2/VAB declines with frequency. There is no contradiction. The weighting curve graphs the factor by which a high frequency current would have to exceed a low-frequency current for equivalent physical effect. The lowpass measuring network implements this function by de-emphasizing high frequencies.

For most equipment the measured currents are predominantly at the power frequencies of 50/60 Hz. How