In the late 1990s, I worked for a company that produced rail subsystems. Our products are custom designed and delivered to automakers, but the end customers are the transit authorities.

One day one of our customers complained that the power relay of our HVAC control box was jamming and draining the car’s batteries. This is bad. The fact that our HVAC doesn’t work is bad enough, but immobilizing a car makes it worse.

Subway cars have batteries that work with most low-power electronic systems. When the subway car is parked in a garage and turned off, there should be no draining of the batteries. To ignite the car, the batteries power the main contactor coil, which sends high voltage DC to the on-board systems, including the charger. The charger also supplies all low-voltage electronics. The rated voltage of the battery is between 36 V and 72 V, with a capacity usually over 150 Ah. A discharged battery means you need a “portable” battery to power your car. The battery is large and heavy and must be rolled around the store to defective cars. This relay failure becomes a big problem for maintenance people.

Do you have an unforgettable experience in solving an engineering problem at work or in your free time? Tell us your story.

We looked for a few faulty relays and checked them. The relays were sticking and the contacts were visibly torn.

The following information about adhesive relays is extracted from various websites and printed catalogs. The correct term is gluing the relay and is different from contact welding. The sticking is confirmed by lightly touching the relay and noting that the contacts are open. Higher current increases the risk of sticking. Relays that switch DC are more susceptible to jamming. Bonding is more common if the power lines are long or inductive, if the load has a high starting current, if the contact surface is too small, and if the metal on the contact surface is of the wrong composition. Finally, cleaning the relays with ultrasound can cause sticking. In my case, I had many of the aggravating conditions: the relay was operating at the high end of the allowable rated current, with a high starting DC current coming from a long car line, and the contact metal composition is standard.

To better understand what was happening, we had to analyze on a microscopic level. The relay contacts are illustrated as perfectly flat parallel blocks without any resistance. In fact, the two contact surfaces are neither flat nor perfectly parallel to each other. Each surface is made up of microscopic hills and valleys. When the contacts come together, some hills will make contact first. The current density is high at the point of contact and since the metal is resistive, localized heating will occur. As the electrodes approach, the contact will be broken. When the contact is broken, the energy stored in each inductor, such as long wires, will generate an electric arc. The rainbow evaporates some metal. The electrons move from the negative electrode to the positive electrode, and the metal atoms move in the other direction and are deposited on the electrode. Each time the coil is activated or deactivated, some more metal is transferred. Over time, a hill reaches through the gap between the resting position and some welding is performed. If the weld is in a small area, it may not be strong enough to withstand a small mechanical impact and we get sticking.

What can I do? I can’t reduce the length of the line: the battery, the train lines and the location of the equipment are fixed. I can’t switch from DC to AC – I’ll need a constant power supply for the HVAC controller and I’ll have to reconfigure the car. I cannot change the load current unless I redesign the controller and select other contactors. I can’t change the size of the relay because it is in a socket in the board and this is the plug relay with the highest rated current and there is no other contact metal for this relay model. I can’t change the starting current because the bypass capacitor on the controller box is needed for the electronic circuit to work properly.

I see two solutions: I can add a plug that connects the relay contact to a larger relay mounted outside the board, and hope for the best, or I can reduce the starting current.

To reduce the inrush current, I can use a simple device designed to do just that: a negative temperature resistor or NTC.

To investigate, I decided to create a test system using a programmable logic controller (PLC). A PLC with its software and accessories costs less than $ 400. I set up a test circuit with the same capacitor and relay, but added a resistor to discharge the capacitor. This setting does not include the linear inductance, but will give me close to the real situation. The sequence of the test is simple: turn on the relay and wait; capacitor voltage measurement; if good, turn off the relay and wait; voltage measurement; if OK, increase the counter and then repeat. If any of the voltage readings are incorrect, the program stops. After about a day the relay remained at about 200, but I accidentally stirred the setting and the relay came off. I changed the setting and let it work. It stopped again with about 400 cycles. This meant that in six months to a year this relay would have to be replaced in all subway cars. Our HVAC failure not only immobilizes the car, its reliability is in the pits.

During the tests, I looked at the NTC datasheets, did some calculations, and ordered a few test pieces. When the tests were completed on the original circuit, I installed the NTC in series with the relay and performed the test. In one week, the count passed over 5000 cycles and the relay showed no signs of cutting or sticking. This means that the relays will last for many years.

I looked for a place to install NTC in the control box. I’m lucky: the box had a few free spaces on the screw terminal strip that connects to the car lines. I can move one wire and put NTC between the corresponding screws.

I informed the project manager about my decision. We ordered NTC and sent them along with a modification procedure to the field office to upgrade all control boxes. Feedback from the transit authority confirmed that they did not have discharged batteries with the modified units. The case is closed.

The lessons learned are: relays are complex devices, and if you want to use them properly, you need to understand all their limitations to avoid sticking to the outlet. Life testing should be done on circuits to check the performance of the design and to confirm failure rates. PLC or Arduino can be useful for quick test settings. Negative resistors can be an old technology, but they are cheap, easy to use and good voltage limiters. NTC manufacturers provide application notes and formulas to calculate the most appropriate model for your needs.

Daniel Dufresne is a retired engineer and has worked in the fields of telecommunications, public transport, consumer products, high-power electronics and custom appliance design. He was also a professor at Cegep de Saint-Laurent and taught courses at the Ecole Polytechnique de Montreal. Daniel has published articles in Audio, EDN, Electronic Design and other publications. He lives in Montreal, Canada and is still working on electronic projects and modifications and repairs of test equipment.

Related content

Previous articleThe Marines’ Force Design 2030 update focuses again on intelligence
Next articleElon Musk will meet today with EU Industry Chief Thierry Breton to discuss freedom of speech, global supply chain issues