Contrary to popular belief, the act of troubleshooting is not always about your tools; it's often about your mind. Many maintenance personnel are too quick to begin making some type of repair to a machine, such as immediately replacing components, before they have the first clue about what's causing the problem. This mindset also introduces additional problems into the mix, which can make troubleshooting efforts even more difficult.

If used properly, one method that's been around for years can save you valuable troubleshooting time, and that's the half-splitting method. The logic behind this approach is simple. With every step in the process, the troubleshooter eliminates half the circuit as the cause of the problem. Using this divide-and-conquer philosophy, the first action item might be as simple as asking a question or taking a voltage or resistance measurement with a multimeter.

Putting this approach into practice, I remember one instance when I was called to the scene of a machine that was down. Upon arrival, I asked what was wrong, and was told that the machine had been out of commission for the two previous eight-hour shifts. At this point, virtually everything had been changed, including the programmable logic controller, power supplies, and a motor drive. The perplexed repairmen from the previous shift told me that no matter what they did, the machine continued to blow a small control fuse.

Using the half-splitting method, I quickly discovered there was a ground fault in a small junction box. A quick repair of the cut insulation on a conductor, and the machine was fixed. Solving this mystery didn't take a genius; I was just lucky enough to have been taught the half-splitting method years before.

Pure probability. Let's review an old card trick to help demonstrate this method. Ask someone to pick a card from a 52-card deck of standard playing cards. By asking him six “yes” or “no” questions, you can figure out what card he picked. The first question would be, “Is your card red?” Let's say he says “no.” Now we know the card is black, having already eliminated 26 cards from the deck — not too bad with one question. The next question, “Is your card a club?” He says “yes,” eliminating 13 more cards. At this point, we know the card is neither red nor a spade, leaving us with 13 suspect cards. You ask, “With the ace being high, is your card a 7 or below.” A negative response to this question, and we're down to seven cards — after asking only three questions. We now know the card is a club (8 or higher). Half-split again by asking if the card is a jack or higher. A “yes” takes us down to four cards. Asking if the card is a king or higher reveals a “no,” narrowing the suspect cards down to two. The next question solves the problem. “Is your card a queen?” Another “no,” and we know the card is the jack of clubs. Asking six questions eliminated 51 of the 52 cards.

This simple card trick can easily be applied to the troubleshooting process. Simply look at a circuit as a group of individual components. You know one of them is not functioning properly. Therefore, what questions can you ask that will begin to eliminate the largest group of components?

Sample situations. Let's take a look at some electrically focused hypothetical examples to further explain this method.

Example 1. Say you're working with a piece of equipment with 20 doors. For safety reasons, all of these doors must be closed in order for the machine to operate. Each door has a normally open limit switch that is held closed when the doors are closed.

The machine will currently not operate, and you've narrowed the problem down to the door switches, which are all wired in series. Rather than checking each switch, check 10 of them first. This could be done with the power locked out using a simple continuity test. If the first 10 switches show continuity, they are not the problem. You have 10 left, so half-split those by checking five of them. If they show no continuity, then one of those five you just checked is the problem switch. Half-split those by checking three, then two, then one — and you're there.

Example 2. Trouble is reported on a machine that will not start. A “circuit enable” light on the control cabinet tells you that the start or circuit enable relay is not energized.

This piece of equipment has 10 safety switches that have to have their contacts in the closed position before the machine can operate. These switches are all wired in series and might consist of pressure switches, flow switches, limit switches, and temperature switches.

When all of these contacts are closed, the coil of the start relay will energize. Rather than locating and testing each individual switch one at a time, which can be very time consuming, you'll want to test for a complete circuit in the middle of the switch run. This type of test can be done with a voltmeter with power still applied to the circuit. Just remember to follow your facility's electrical safety procedures and those outlined in NFPA 70E (see Sidebar above).

A good first step is to measure voltage across the coil. If no voltage is present, but we know that the circuit control voltage is functioning for other parts of the same control circuit, this would lead us to believe one of our 10 switches has a problem. Leave one of your voltmeter leads on the relay coil opposite the switches. Put the other lead in the middle of the switches. Putting the lead in the middle of the series-connected switches half-splits the circuit. If the circuit control voltage is still present, then the five switches between that point and the source are closed, leading you to believe the problem is in the five switches between your voltmeter leads.

With this one measurement, you've proven the functionality of the other five switches. Your next step is to half-split once again. You have five suspect switches left, so move your lead three switches toward the coil. If voltage is still present, you have proven that those three switches are functioning properly. You have two suspect switches left — the two closest to the coil — so move your lead between those two switches, half-splitting once again. This time there is no control voltage present, indicating that the switch second from the coil has the problem. This could be a switch out of adjustment, a broken or loose wire, a safety door open, or pressure too high in the system. Although you still have to find out what is wrong with this switch and fix it, at least you've identified the culprit.

Applying this approach. It may not always be parts of a circuit you're trying to eliminate. Instead, it may be entire parts of a machine or process. An example of this is when you first arrive at a piece of equipment, and you ask yourself if the problem is electrical or mechanical. This obvious first step is often overlooked. An example might be a conveyor belt that will not move when started. If the conveyor belt will not move but the drive motor is turning, this could indicate a broken belt or problem with a pulley.

Half-splitting can be used to find many electrical problems aside from open switches. Removing parts of the circuit and testing to see if the fault is still present can also be used to find ground faults. Half-splitting can also be used in all types of circuits, not just series-wired switches. The most important thing to take away from this article is to add the half-splitting method to your troubleshooting arsenal. Trust me when I say — it will save you hours of time spent in vain.

Lewellyn is president of Lewellyn Technology, Inc., Linton, Ind.


Sidebar: Safety First

Whenever you're troubleshooting, it is extremely important that you follow all safety precautions and procedures. If your company does not have a documented electrical safety program as required by NFPA 70E, you should encourage its creation. Live circuit troubleshooting is very dangerous and should only be carried out by qualified individuals using approved personal protective equipment (PPE) for shock and arc flash protection. Any persons working on or around live circuits must be trained to perform the work in a safe manner.