Occasionally, a design needs to be updated because it is being placed on a new platform, or because the parts needed to build the equipment are no longer manufactured. Does the equipment need to be requalified, or will replacement parts change the electromagnetic profile, and therefore the original test results? If the original equipment was tested to an older standard, are the new and old standards similar enough to assume compliance? If you have had these questions, know that you are not alone. They are commonly asked.
Companies may try to avoid performing new EMC testing by supplying “Qualification by Similarity” analysis and reports. To properly report similarity to a previous design, all aspects of the electromagnetic profile of the changes made should be addressed. This can be a rather complicated process and difficult to analyze completely. And the smallest changes in components, positions, routings, and the like, can drastically change the electromagnetic profile. And yet full requalification can be expensive and possibly outside the budget or the timeframe allowed.
First, about the use and updates of the standards. In the case of commercial testing of aircraft, the governing standard is DO-160, which has been through many iterations. Yet, the overall method and limits for emission measurement have remained essentially the same. One major change has been the control of measurement sweep speeds and bandwidths used since the introduction of DO-160C. For military equipment and MIL-STD 461, the changes that came with Revision D were more dramatic, including similar bandwidth and sweep speed controls, but also changing 10 µF feedthrough capacitors on the power lines to 50 µH LISNs. Power line lengths for conducted emissions were changed from <1 meter to 2 to 2.5 meters [1], which also changes the line inductance and thus the impedance over the frequency. Current measurements were changed to voltage measurements, among other differences.
Changes to aerospace standards regarding susceptibility testing were significant over time. Radiated test levels were increased significantly. Lightning and ESD tests were added. Bulk current injection was introduced for low-frequency RF conducted susceptibility. So, changing from an older version of DO-160 or MIL-STD 461 might highlight some serious issues. Will the filters and shielding be adequate for increased field and current levels? Is there any transient protection that will need to be added that was not required earlier?
The documents “EMCS Project Number 0178 (April 6, 2000), Results of Detailed Comparisons of Individual EMC Requirements and Test Procedures Delineated In Major National and International Commercial Standards with Military Standard MIL-STD-461E”, and “NUREG/CR-6782, Comparison of U.S. Military and International Electromagnetic Compatibility Guidance,” have both tried to compare MIL-STD 461 to various commercial EMC standards, including DO-160.[2] The SAE published AIR6811, “Equivalence of Equipment Environmental Qualification Standards for Civil and Military Aircraft Equipment,” an excellent document for comparisons of DO-160 with MIL-STD-461, MIL-STD-704, and MIL-STD-810.
Although the equipment remains unchanged, similarity analysis may not be easy in cases where standards are changing. For commercial testing and MIL-STD 461, conducted emissions are measured as a voltage on an LISN. Relating that to DO-160, which measures current, can be very difficult. Yes, you can deduce the currents from the equipment to the 50 µH LISN based on the impedance over frequency of the LISN, but a different impedance curve is used for 5 µH DO-160 LISNs. The shorter DO-160 power line will have less inductance than a military or commercial electronics setup, thus less impedance over the frequency range. The output impedance of the equipment will also play a role in the overall current readings. The analysis becomes rather complex and problematic.
In the case of simple changes of a unit or system, qualification by similarity may be reasonable. For example, a component becomes obsolete, the new component is a drop-in replacement, and the test standard does not change. When no circuit board changes are introduced, and filter components stay the same, only the component change may need to be addressed. Does the new component operate at a different clock speed? Does it create different harmonics due to faster risetimes or higher current demands? Can the filter and shields manage higher amplitudes or higher frequencies generated in these cases?
Ultimately, it is best to consider the efforts and time expended on a justification for Qualification by Similarity. If the task is difficult and time-consuming, it is likely due to many aspects of the design and electromagnetic profile, which may be difficult to determine or predict. In this case, it is often best to perform some testing, such as conducted emissions to verify filters, and at least lower frequency radiated emissions to verify the shielding quality. Comparing conducted emissions of an unmodified (original) unit to the updated unit can show any changes in the emission spectrum and the filter’s performance. Increases in the emissions may need to be addressed, while decreases may be reassuring. Note that emissions and susceptibility have a relationship – if the emissions cannot get out, the susceptibility signal is likely not easily getting in, which is true for both the filter and shielding. These simple tests tend to be inexpensive and quick, and may be able to answer questions raised by the customer.
Having the assurance of these emission results that compare older and newer equipment designs may be coupled with a Qualification by Similarity analysis. The analysis is then backed up with data to strengthen the argument made for similarity.
Endnotes
[1] Ken Javor reminded me that all radiated testing in the MIL-STD 462 test methods used greater than 2 meters. This required the power lines to be extended for all transitions between conducted testing and radiated testing.
[2] Several groups and individuals have similar works, such as NASA/CR–2000–210400, as well as works by Brian Farmer and Vincent Greb.
