Finding a Non-Linear Ground Fault

By Douglas Krantz | Maintenance

One of my stories about finding a ground fault after the end of the day was at a large chain-store distribution center. Until that night, our fire alarm company didn't have anything to do with that store chain. Our dispatcher, though, asked if I could stop there before going home and fix a ground fault.

A few years earlier, another company had installed a new fire alarm system, so they were the company that did the service (until then, at least).

It was a four-loop addressable system, but the building itself was a several-block-long warehouse with around 80 truck loading docks. The problem was there was an intermittent ground fault. When anyone was there to find it, the ground fault was not showing up on the panel. After several weeks of failing to find the problem, the other company quit even showing up.

The problem with that ground fault was that it wasn't a straight-line resistive ground fault, it had non-linear resistance.

Non-Linear Resistance

A common resistor has the same resistance, no matter what the voltage; the current through the resistor exactly tracks the voltage. When the voltage doubles, the current doubles. That's a linear relationship.

Water does not have the same resistance, no matter what the voltage. Neither does a surge suppressor. When the voltage on either is low enough, the resistance is very high. There isn't any measurable current at all through water or through a surge suppressor, at least as long as the voltage stays low.

With water, there is a voltage threshold, or knee, of 5 volts to 8 volts; with a working 25-volt surge suppressor, there is a voltage threshold, or knee, of about 25 volts.

Doubling the voltage across water from 5 volts to 10 volts will often increase the current by 10 to 20 times. Doubling the voltage on a working 25-volt surge suppressor from 15 volts to 30 volts will increase the current by at least 50 times. When the voltage doubles across the knee on either water or a surge suppressor, the current increases many times what the voltage has increased. That is a non-linear relationship.

Keep in mind that the 24-volt (nominal) battery voltage is split up by the ground fault detection circuitry in the fire alarm panel. Often, the real ground fault detection for a fire alarm panel is closer to 14 volts.

Back to the Story

I arrived on site, and, you guessed it, the panel was normal. Checking history, I found which loop the ground fault was on. I knew that this was an "intermittent" ground fault, and the panel wasn't detecting it, so I whipped out my "Ground Fault Detector" and confirmed that the ground fault was truly existing.

Fire Alarm Specifications: Check the maintenance manual for most fire alarm systems. In there they specify the resistance or current level for the ground fault detection threshold.

My ohmmeter, having 3-volts for its internal battery, isn't capable of detecting current when the knee on non-linear resistance is 5 volts or higher. However, my "Ground Fault Detector" is a regular, cheap ohmmeter that has 36 volts (four 9-volt batteries) in series with one of the leads. It also has a resistor in series to keep the meter reading full scale when the leads are shorted together.

The technical name for this device is "Insulation Tester", but at the time, the only insulation testers on the market used a voltage that was too high for fire alarm circuit safety, so I had to make it myself.

Having never seen the building before, and no access to anything like an as-built diagram, I used the divide-and-conquer" technique of disconnecting parts of the loop, and looking for the troubles on the panel. I combined that troubleshooting of Divide-and-Conquer Technique with my "Ground Fault Detector", and found a surge suppressor that had decreased its protection voltage from 25 volts to around 15 volts. Because the Signaling Line Circuit (SLC) went to an out-building electric fire pump about a block away, the surge suppressor was on the SLC.

Even though the panel didn't show it, I had found and fixed the ground fault in about three and a half hours.

The other service company had used their expensive ohmmeters, with internal 9-volt batteries, and couldn't find the problem. Using my homemade "Ground Fault Meter", I fixed it. We got a new service contract.

Note: You can find out more about hard-to-find ground faults at - How Does One Find a Soft Ground Fault?