By Chuck Newcombe
It's been almost 45 years now, and the world of electrical testing has changed so much during that time. In this column, I'll revisit the test techniques I used in 1963 at an aerospace company, and compare them to today's tools and methods.
My example is the process for testing the electrical bonds required in grounding systems by the National Electrical Code (NEC.) The use of standard construction practices and visual observation are useful, of course, but how do you know if the electrical connections you've made will serve the intended purpose?
In the case of the aerospace company, after we had seen instrumentation problems in portions of a production building, I was asked to verify that the receptacle ground connections were less than 15 ohms from nearby building steel uprights - the construction electrical bonding specification in a building that was now live, electrically, 24-7. There was no time that I could shut down power to the building to make ordinary ohms measurements, and low levels of both dc and 60 Hz voltages could be observed everywhere, further complicating the situation.
What to do? I needed to design a test method that could work around the limitations and obtain the desired test data. And, in a facility where we had an almost unlimited test equipment pool to draw from, I had all the resources my inventive mind could hatch a need for. The following is what I came up with.
I decided to measure the ac impedance between the selected receptacle ground pins and nearby vertical steel girders in the building. The task required that I assemble a cart load of test equipment that included:
- A function generator to produce low frequency voltage sine waves. I selected a frequency of 83 Hz for my tests. More about that later.
- A wide band audio amplifier with a mult-tap output transformer, which allowed me to select very low output impedances to maximize the current available.
- A tuned narrow band voltmeter, to read only the voltage drops, produced by the function generator/amplifier combination, in the circuit between the girder and the receptacle jack.
- A low-reactance shunt resistor, to be used with the tuned voltmeter to determine the current flowing in the circuit between the steel girder and the receptacle ground jack.
- A slide rule to calculate the measured ac impedance using Ohms Law.
There were, of course, several cables and test leads to round out the test setup, including an extension cord and 4-outlet power receptacle to supply all of the line operated electronics. The whole setup cost over $4,000, in 1963 dollars. And, finally, there was the requirement to remove paint and primer from the test points on the girder in order to make a good electrical connection.
Could I have measured the dc resistance of the path using similar techniques? The answer is yes, but that would require a synchronous demodulating voltmeter to decode the second harmonic data from the ac signals. I could have done it, but I didn't.
After testing at 83 Hz, I then tried 94 Hz. Why the odd values? I didn't want to errors due to possible interactions between the harmonics of the 60 Hz which I knew to be present, and the harmonics of the test frequency used. And, by comparing the data taken at the two different frequencies, I was able to estimate the inductance of the circuit, and thus to roughly calculate the dc resistance.
So, how specifically, does one make such a measurement - where do you connect?
The method I used involved four leads. One to inject current into the girder, and another at the receptacle ground socket to complete the current loop to the source. A third lead was connected to the girder near the current injection point, and the fourth to the pin inserted into the receptacle socket. (Because two of the four connections rely on the pin to socket connection, this is technically known as a three terminal ohms measurement.)
The current level was determined by measuring the voltage drop across the shunt resister placed in series with the current leads, and, using that current and the voltage drop from the girder to the socket, I used Ohms Law to complete the exercise.
How different it is today. Take for example, the Fluke 1625 Geo Earth Ground Tester. At just over $3,200, this tool does everything I was doing in 1963 and more. It actually does measure the dc resistance using a method similar to that I opted not to do in my tests. The 1625 provides several methods to evaluate earth ground systems as well. As a matter of fact, that is the main focus of its design.
What I really like about this tool is the fact that it can be battery operated, meaning I can forget the extension cords required in my test system.
Epilogue
I just ran a 3-pole earth ground test at my service entrance using the 1621. I took three readings and averaged them - the result was 0.53 ohms. The three readings varied from a minimum of 0.47 to a maximum of 0.56.
I then tried the same test with a 289 dmm, using the ground electrode and the probe. I read 0.85 ohms. Not too bad, you say? Well, I then reversed the leads and tried again. This time I read minus 0.73 ohms, that's right - negative ohms - wrong!
No, you can't average these. All you know for sure is that there is dc present that is rendering both readings suspect, and you might not know that if you didn't reverse the leads. I could probably devise a complicated procedure to get close using the dmm, but you don't have the option to do the three lead test, and it's hardly worth the effort. You really have to know what your doing, and how the meter works, to pull it off.
By the way, the whole process, including laying the wire, driving the ground stakes, testing, and then packing up the whole setup took less than 30 minutes - and that included the measurements using the 289.
I then went to my panel and tested the bonds between the grounding conductor going to the grounding electrode, and the neutral bus in the panel. I used the two pole test in this case. I read 0.08 ohms. I read a similar number with the 289 in this case, and did not see the negative reading when I reversed the leads. Since I didn't compensate for meter lead resistance in either case, I'm guessing that much of that reading was test leads.