I often receive data files recorded from power quality monitors to analyze the reason a piece of equipment experienced an event. This analysis is followed by determining the cause and who is to blame. Others want to know if their supply quality is acceptable or “normal” (whatever that means) for their unspecified equipment or if they should call their utility to improve it.
These are all interesting challenges that can often be overcome but only if the data presented is valid. Despite advances in PQ monitors that try to make the connection, configuration and setup of the instruments less prone to human error, there are still many times when the data cannot be used to achieve a certain goal. Just a few simple checks can take some of the human error out of the equation.
First, does the monitor have the capability to do what you want it to and capture the type of data you are interested in? Reading the specifications on the instrument brochures can be a daunting task. Try looking for these few things. Is the voltage and current range compatible with your electrical system and equipment? Is the sampling rate at least twice the highest harmonic frequency you are interested in? Is the storage capacity large enough for the time that you want to monitor and the amount of data you think you will collect?
That last one is tricky because it is difficult to predict all variables. Since less-experienced users tend to save a lot of waveforms and nearly every parameter on the menu, storage capacities in gigabytes is probably a safe bet.
Next, connect the instrument’s voltage leads and the proper rated current probe on the correct conductor pointing in the right direction. Despite claims to “fix incorrectly connected instrument data,” that is generally only possible if every voltage and current waveform is saved over the entire time you are interested in seeing what is happening. You can’t post correct a delta connection on a wye circuit with phase A probe on phase B conductor pointing toward the source instead of the load, and so on. It is unlikely that you would have captured a neutral-to-ground voltage swell, since the line-to-line voltage of the delta connections wouldn’t see that. Determining power parameters and the directivity of sags and swells (upstream or downstream) requires the proper phase relationships, which can’t be recreated without waveforms. In addition, you have to know which is the correct phase relationship after you have left the site.
Going back to the current probes or current transformers (CTs), are they properly sized for the current levels expected, and are the range selection switches (if applicable) and scale factors set properly? Putting a 30-ampere (A) CT on a circuit with fault currents over 100A likely will result in clipped waveforms and incorrect maximum values. Conversely, a 1,000A CT looking for less than 1A of fifth harmonic current out of 5A rms total current is likely going to have a lot of noise in the data. Most CTs have a 3x overcurrent range and can read down to 10 percent of full scale, but check the specs to be sure.
Also, if you suspect there might be a DC component in the current (or voltage), you should have a probe that can pass DC (such as a Hall-effect probe for current), and ensure the instrument’s measuring inputs aren’t AC-coupled.
If the problem occurs frequently and you plan on monitoring only for a day or two, you can set the periodic capture or journal rate to one minute or less, provided the storage memory is adequate. When doing a month-long benchmark survey, there is rarely a reason for setting that rate to less than 10 minutes, especially when the instrument records the minimum/maximum/average values over that interval.
Similarly, setting the waveform capture more than 30 cycles pre- and post-event is usually not necessary, except when doing inrush measurements of large motors, transformer energizations, or the startup or switching of generators.
Finally, if you are a new user or an infrequent user of the instrument, it’s a good idea to leave the threshold settings for triggering event capture in the default or preprogrammed setups from the manufacturer. These values are often based on typical limits used by many customers and referenced to standards, where +/–10 percent from nominal is the sag/swell limits. Customers have set the instantaneous peak triggers to 170V, not realizing that a 120V rms undistorted signal has a peak value of that, which can result in a continuous recording of nothing useful while the memory fills up. Another favorite memory filler is to set current limits within normal operating ranges of the loads.
Of course, there is the fall back of reading the user’s guide or operator’s manual. But short of that, follow the above steps each time, and the data you record (and I will analyze) will likely have information that you (or I) need to resolve the problem.