Quick EMC screening

In our method, we separate active and passive components and then use a two-tiered approach to evaluate the components for their effects on EMC compliance. At tier 1, we review a component's parametric values and its function in a circuit. From this analysis, we can show that certain electrical parameters have a significant effect on compliance, while others have little or no effect.

Figure 1 shows our evaluation process. If we determine that a part needs evaluation, we test it on an Agilent Technologies 4395A network analyzer. We measure impedance magnitude (|Z|), capacitance (C), phase (è), dissipation factor (D), inductance (L), and quality factor (Q) as functions of frequency.

From a |Z| measurement, we can find a component's resonant frequency. If a low-cost replacement component such as a capacitor has a lower resonant frequency than an existing component, it won't suppress noise well enough for our product to meet compliance specifications, and we reject it. If the component's performance parameters are equal to or better than the original component in its intended function, then we conclude that the risk of using the new component is acceptably low.

Figure 1.  Using this evaluation procedure, we determine whether a component meets all review criteria for replacing an existing component.

To illustrate our method, we used our tier-1 process to evaluate two possible replacements for a ferrite. When correctly applied in a circuit, ferrites can reduce emissions—the critical parameter is impedance versus frequency. Figure 2 shows the impedance of the two low-cost replacements compared to that of the original one. The plot shows that low-cost ferrite A (blue trace) produces an impedance below the minimum required by this circuit. Ferrite B (red trace) from another supplier, however, has an impedance somewhat greater than the existing component at high frequencies for noise suppression. Ferrite B will meet the requirements for the circuit in question. Ferrite A won't.

Figure 2.  Low-cost ferrite A (blue) failed to perform as well as an existing component, but ferrite B (red) performed better than the original at high frequencies.

Compare that to the 10 min we need to perform the parametric measurement with the component analyzer. We save 1 hr, 50 min. Multiply this over 10 components per week, and the labor savings exceed one person-day each week. Our new method has proven invaluable in rapidly separating the acceptable replacement components from components that will likely cause problems.

In cases where a component evaluation is insufficient to tell us whether a low-cost component will create compliance issues, we turn to tier-2, a modification of our diagnostic immunity testing procedure. Tier-2 testing isn't a replacement for full compliance certification, but it quickly identifies susceptibility issues that result from a component substitution.

The tier-2 procedure compares test results from a modified product to an existing product. To make the comparison, we need an immunity model of the existing product, which we developed during the product's full compliance certification testing. We store each model's compliance test data in a technical information file, a derivative of the technical construction file formerly used for EU competent-body certification.

The tier-2 method modifies some of the immunity tests to minimize testing time. We currently use the tier-2 method for radiated immunity, conducted immunity, electrical fast transient, and surge tests.

We base our modified radiated and conducted immunity tests on a circuit's ability to respond to impinging RF energy. The response may be difficult to identify because of software averaging, but given repeated exposure, the response will become detectable. EMC standards EN 61000-4-3 (radiated immunity) and EN 61000-4-6 (conducted immunity) for CE compliance require that a product be exposed to RF energy for at least one complete product-operation cycle for each frequency step, called the dwell time.

Unfortunately, dwell time can take up to 20 min for some of our products because of long operational cycles. To complete a cycle, some analytical instruments must wait for a chemical reaction in a sample to take place, which can take up to 20 min. We can, however, reduce the product's dwell time for compliance testing by using pre-reacted samples and special test firmware in the instrument under test. Even so, some products can still take as long as 15 min to complete an operation cycle.

Even if a product's cycle time is short, say 20 s, a radiated immunity test can still take hours because we must subject the product to RF energy at many frequencies. The current radiated immunity test standard requires testing from 80 MHz to 1000 MHz at 1% frequency steps on all four sides of the equipment under test (EUT) at both horizontal and vertical polarities. If we add all the frequency steps, sides, and polarities, we must complete 2040 steps (255 frequencies x 4 sides x 2 polarities). If each step takes 20 s, then 2040 steps will take 40,800 s, or 11.3 hrs of test time.

For conducted immunity tests, the frequency test range is 0.150 MHz to 80 MHz in 1% frequency steps, or 633 frequency steps. At 20 s/step, that adds up to 12,660 s, or 3.5 hrs. Thus, we need 14.8 hrs—nearly two full work days—to complete both the radiated and conducted immunity tests for a full compliance test.

Our tier-2 process shortens this time considerably. Instead of waiting for the product to complete an operational cycle for each frequency step, we sweep through the entire frequency range in 5 s to 20 s. We make up to 100 frequency sweeps for each operation cycle of the EUT.

Obviously, this method would not be acceptable for full compliance testing, but it is fine for evaluating a product's immunity. If we don't detect a response from the sweeps, then we conclude that switching to a low-cost component won't cause the product to fail a full compliance test. If we do detect a response, then we subject the product to full compliance testing to determine the exact response frequency.

To understand the time savings, assume a mean of 50 frequency sweeps at 5 s per sweep. We would thus need 250 s (4.16 min) to test each of a product's four sides at both horizontal and vertical polarities—a total of 33.3 min per product.

For the conducted immunity test, we need just one frequency sweep test at 4.16 min. Thus, we need a total of 37.5 min (0.62 hr) to perform both radiated and conducted immunity tests—a savings of 14.18 hrs (1.78 work days) over the 14.8 hrs needed for a full compliance test.

To perform a frequency sweep, we use an RF signal generator to drive a power amplifier and an antenna. We measure the field strength with an isotropic field probe. The probe's controller has an analog output that we feed into the amplifier's automatic gain control input.

We can achieve similar time savings for electrical fast-transient and surge testing by subjecting the product to the most severe interference levels only. With our products, we have data to indicate that if the product passes under the most stressful conditions, it will pass under lower stress levels. If we detect a failure, we can analyze it by performing a full immunity test. At this level of evaluation, we're looking for a simple yes/no result. That is, does the product containing the low-cost component respond differently from the original instrument?

Figure 3.  A replacement DC-DC converter failed a swept-frequency immunity test. Unlike the original part, the replacement part shut down during a test.

Figure 3 shows the results of a swept test of an existing and a proposed alternate DC-DC converter. We use the converter to power an instrument controller's isolated 4–20-mA output. The original component showed no meaningful response when subjected to a sweep. The proposed replacement however, not only lost regulation, it shut down halfway through the test. The converter regained regulation after testing. Clearly, the substitute component is not a suitable replacement.

If you want to try a method similar to the one we use, you first need a compliance history of your product. That is, make sure you have a full emissions and susceptibility profile to compare the test results against. Without a known set of behavior, you can overlook a minor response. With a known product profile, you can carefully watch for minor anomalies during rapid testing. If in doubt, perform a full compliance test.