The Devil in the Details

This month we step outside the confines of Ask the Experts to address topics covered in previous power quality articles. John DeDad, editorial director of EC&M, clarifies some points made by fellow Ask the Experts veteran, Mark McGranaghan. And Dan Carnovale, Eaton/Cutler-Hammer's power quality solution manager, responds to questions concerning his article published in June.


An article in the August issue of EC&M, “Controlling Harmonics Locally in Commercial Facilities,” lists benefits and disadvantages of various filters. I'm particularly concerned with the section on neutral blocking filters, as I have installed a number of these filters in my facility over a period of four years. I've had no problems with any of my mixed loads. However, the article says that these filters may compromise the fault protection of my system and should not be used with mixed loads because of the potential for voltage distortion. Should I worry?

DeDad's answer: NEC experts at the International Association of Electrical Inspectors (IAEI) have thoroughly investigated the neutral blocking filter and its application. In fact, an article published in the January/February 2003 issue of the IAEI News, “The 3rd-Harmonic Blocking Filter, A Well Established Approach to Harmonic Current Mitigation,” outlines all aspects of the Code requirements that apply to the filter and pays particular attention to grounding issues. The filter is shown to meet all Code requirements and won't affect the fault protection integrity of your system.

Your concern may have been prompted by the article's somewhat unclear discussion of the differences between systems serving phase-to-neutral, single-phase loads and those serving 3-phase loads. The 5th harmonic voltage distortion is quite damaging to 3-phase motors. This phenomenon is discussed in an article published in the November 2002 issue of EC&M, “Harmonic Current Distortion and Voltage Distortion.” Further research shows that this valid concern about 5th harmonic voltage distortion and 3-phase motors shouldn't translate into concern about 3rd harmonic voltage distortion for mixed single-phase loads.

The correct application of neutral blocking filters results in increased 3rd harmonic voltage distortion, which appears on the voltage waveform as “flat-topping.” This distortion has existed in 120/208V distribution systems — even without neutral blocking filters — for several decades with no documented evidence of damage to any type of load. Nor is there any evidence of damage in systems where these filters have been in use for many years.

Since 3-phase motors don't use the neutral, they're not affected by a filter that's inserted in the neutral of the distribution system.

Further, a manufacturer-commissioned modeling study noted in an August 2001 research report titled “Neutralizer Harmonic Blocking Device Study,” by Alexander Emanuel, Worcester Polytechnic Institute, showed that the effect of 3rd harmonic voltage on small single-phase motors is negligible.

In contrast to these concerns, the effect of 3rd harmonic voltages on computer power supplies was shown to be beneficial as the efficiency of the supplies is increased in PQTN Brief 20, “Energy Use of Personal Computers,” prepared by EPRI-PEAC in May 1994.


In the article “Power Factor Correction and Harmonic Resonance: A Volatile Mix” (June 2003), the author describes a de-tuned harmonic filter as having a “parallel resonance.” Most capacitor-inductor circuits designed for power factor correction and harmonic filtering will be connected as a single tuned “series-resonant” circuit. The series filter is then connected to the network in parallel (shunt). Also, the author's statement that “harmonic filters provide the same 60-Hz reactive compensation as capacitors” is an approximation and not precise. The reactive compensation at the fundamental (60 Hz) is related to the inverse of the natural frequency, squared, of the filter. The increased compensation is due to the increased voltage impressed across the capacitor by the filter's inductor. The difference is substantial for low-order filters like the 3rd and 2nd harmonic. Can the author respond to my comments?

Contributing Author Dan Carnovale's answer: Regarding your first comment, you're correct in that the filter is connected in series (capacitor and inductor) and therefore has a series resonant point, which is the tuning frequency we select (4.7th harmonic for a 5th harmonic filter, for example). However, my point in the article was that every capacitor, whether or not it's part of a harmonic filter or just a straight cap, has a parallel resonant point with the power system impedance in front of it. Therefore, every harmonic filter has a series and a parallel resonant point.

Harmonic currents injected into the power system flow from a drive or other harmonic source and can flow through the filter or toward the utility source. Those two impedances are clearly in parallel, which is why we use the term “parallel resonance.”

As a general statement, you can say that the parallel resonant point for a harmonic filter is about one harmonic order below the “tuned” or series resonant point of the filter and is therefore typically below any significant source of harmonics, assuming that the filter was designed properly for the system of interest. Fig. 1 depicts a sample system in which a filter could be installed. Fig. 2 shows the series and parallel resonant point for a 4.7th (5th) harmonic filter. Notice that the parallel resonant point is at the 3.97th harmonic for a 4.7th harmonic filter. Changing the source impedance components, such as transformer impedance or system impedance, on a distribution system will alter the results and detune the system.

This is why people are often concerned when they apply a “fixed” filter on a power system. The fear is that it will eventually become “detuned” with the system. However, this really isn't a big deal unless the utility changes configurations or the main service transformer is changed out, thus altering the system impedance.

Regarding your second comment, the point I was trying to make was that harmonic filters also provide capacitive compensation. This was merely meant to show a general concept. My statement was intended to address the general reader who might not understand the intricate details of harmonic filter design. A reader who doesn't fully understand the details of harmonic filters wouldn't likely follow the explanation that voltage rise on the capacitor plus some kVAR reduction from the reactor would yield a net effect of an increase in kVAR for the filter. You're correct that given the increased voltage rise on the capacitor in a harmonic filter, you'll see an increased kVAR value out of the filter.

Note that as a manufacturer of harmonic filters and capacitors, our company — as have other low-voltage manufacturers — has taken the approach that if somebody wants a 5th harmonic filter to provide 400kVAR (a very specific catalog number) of capacitive compensation, the filter is designed to match that 400kVAR requirement. Customers don't need to know that the manufacturer has supplied higher voltage rated capacitors (550V) or that the kVAR on the capacitors doesn't match the output, just that they're getting 400kVAR if the filter is applied on a 480V system. Medium-voltage filters are typically individually designed and therefore, all of these details are well known and typically understood by the end-user.


Regarding the article “Power Factor Correction and Harmonic Resonance: A Volatile Mix,” is it correct that the parallel resonant point will always be below the series resonant point?

Carnovale's answer: Yes, the parallel resonant point is always below the series resonant point for a harmonic filter. The tricky part is when you have multiple-stage filters like 5th and 7th. Then you have multiple parallel resonant points. If you switch off the 5th filter and leave the 7th filter on, you'll likely hit parallel resonance near the 5th harmonic. For that reason, you should always turn on the lowest order filter first and turn off the highest order filter first.

For arcing loads, like arc furnaces, you'll typically see multi-stage 2nd, 3rd, and 4th filters to address all of the low-order frequencies on the system. If you aren't careful, you'll run into one of these parallel resonant points at one of these frequencies if you simply apply a single, higher-order filter. In fact, we worked with a steel mill where we corrected a problem like this. The ideal correction would have been to install multiple-staged filters or a static VAR compensator, but as always, the cost of the solution was the deciding factor and the least expensive fix prevailed.