How long does it take to open a relay?

Return to the Industrial Control and Automation Index

In a previous article we explored relay properties by determining the length of time it takes to close a relay. In this article, we will answer the complementary question by exploring how long it takes to open a relay. You will find a great deal of similarities between the two questions along with a few unexpected surprises.

Please review the previous article if you have not already done so. Pay attention to the material regarding time constants and the meaning of a L/nR designation.

Test Circuit

The test procedure is nearly identical to that used in the previous post. The picture is repeated here as Figure 1. It includes a representative industrial relay plus socket, the relay driver as shown in Figure 2, and an Arduino Nano Every to toggle the driver. A Digilent Analog Discovery 3, probe BNC adapter. The 10 X probes allow the Analog Discovery to measure voltages as large as +/- 250 VDC.

There are minor modifications to the high-side driver and placement of R4. The original MPSA56 is replaced by a higher voltage 2N5401. This is necessary as we will encounter higher voltages as the relay is deactivated. The R4 resistor is moved so that it is in series with the flyback diode D1.

**Tech Tip: You may object to using a 1N4001 diode in this high voltage situation. After all, the inductive kickback of relay K1 will develop nearly 100 volts. Yet the 1N4001 diode is not stressed in this situation as it conducts one diode drop of approximately 0.7 VDC while the relay deactivates. In the forward direction it will encounter a the 24 VDC. The anticipated voltage and current are well within the design maximum of the 1N4001 diode.

Inductive Kick and the Flyback Diode

Energy is stored in the magnetic field of the inductor. When we turn off transistor Q2, the magnetic field collapses causing a voltage spike across the coil of K1. If we were to anthropomorphize the relay, or more appropriately, the inductor contained within the relay, we could say that the inductor attempts to keep the current constant both before and after transistor Q2 is turned off.

The “constant current” action associated with an inductor will give rise to a voltage. Without containment this voltage will rise to several hundred or even over a thousand volts if necessary to maintain the current. This excessive voltage will destroy transistor Q2 if it is not clamped.

Recall that we are using a high-side driver (Q2) for the relay. Please take a moment to observe the polarity of this voltage spike. Many readers will assume a positive spike based on their previous experiments with a relay driven by a low-side NPN transistor. Not so in this example. Instead, the voltage as measured at the collector of Q2 will immediately swing from the 24 VDC to a negative voltage when Q2 is turned off. The amplitude of this spike is limited only by the resistance of R4 and the forward conduction of diode D4. Please review the Figure 2 schematic to convince yourself that Diode D4 is forward biased when the collector of Q2 becomes negative.

Figure 1: Test setup to measure relay activation time

Figure 2: Schematic featuring a high-side PNP relay driver (Q2) and a current shunt (R5).

Results with R4 Shorted

Most systems are implemented without the addition of resistor R4. Instead, the flyback diode is placed directly across the relay’s coil. The configuration is so common that industrial relays such as the one used in this experiment include an optional diode module such as the one shown in Figure 3.

Figure 3: Flyback diode and LED indicator module for the Finder brand relay used in the experiment.

This parallel diode is effective and relatively simple to use. Unfortunately, it results in a slow to open relay. This is relayed back to the inductive time constant mentioned in the previous article:

\tau = \dfrac{L}{R}

Where L is the relay coil’s inductance and R is the internal relay resistance. When compared to the original driving voltage (24 VDC in this example) the diode is effectively a short circuit.

Think back to your schooling where you were introduced to the capacitive discharge circuits. Recall the initial charge problems where the energy was dissipated across a resistor. This is no different. The energy is stored in the inductor’s magnetic field. When disconnect from the source the terminals are shorted. All the energy is burned in the inductor’s internal resistance with a tiny amount in the diode.

The results are shown in Figure 4. There are three panels:

  • Upper: The orange trace (CH 1) is the relay activation voltage as measured at the collector of Q2. The blue trace (CH 2) is the relay current as measured across the R5 shunt resistor.

  • Middle: The blue trace (CH 2) is the voltage as measured on the relay’s normally closed (N.C.) contact. This N.C. contact is transitioning back to closed.

  • Lower: The blue trace (CH 2) is the voltage as measured on the relays normally open (N.O.) contact. Recall that we are deenergizing the relay. The N.O. contact is transitioning back to open.

Figure 4: Deactivation waveforms for the relay including the coil’s current, normally closed, and normally open contacts.

Based on the Figure 4 data we observe:

  • the armature movement is first observed at 8 ms when the N.O. contact toggles.

  • there is a 1.5 ms time of flight from 8 to 9.5 ms. In this “flight time” neither the N.0. nor the N.C. contacts are connected to the circuit.

  • first contact with the N.C. contact occur at 9.5 ms.

  • the contacts bounce for 4.5 ms starting at 9.5 ms extending to 14 ms.

The wiggle in the in the current mentioned in the previous article is still present. This change during the flight time is attributed to a change in inductance as the armature’s metal plate departs from the inductor’s central iron core.

When compared to the previous article, we see that the relay with the parallel flyback diode opens rather slowly. This relay in a L/R system activates in 8.8 ms here defined as t_0 to final bounce. It closes in 14 ms.

Increasing Opening Speed

We can increase the speed of the relay by changing the time constant. One way to accomplish this task is to add resistance in series with the flyback diode. To understand how this operates we need to consider the relay activation circuit as independent from the relay turn off circuit. This piecewise operation involves the switching action performed by the high side driver Q4 and the flyback diode D1. Note that they are independent of each other and do not operate at the same time. The flyback diode does not activate when the relay is powered by Q4. Likewise, Q4 does not conduct while D1 is forward biased by the collapsing magnetic field.

Recall that the relay coil has a resistance of approximately 1 kΩ. To construct a L/2R system we will place a matching 1 kΩ resistor in series with the D1. The results are shown in Figure 5:

Figure 5: Deactivation waveforms for the relay when operated in a L/2R environment.

Figure 5 shows that the first sign of relay deactivation occurs at 5.2 ms when the N.O. contacts open. This is in the ballpark of a 2 X speed increase. We could speculate about this apparent discrepancy between the L/2R difference between closing and opening times. Perhaps it has something to do with the added magnetic material and consequently difficulty of changing inductance. Perhaps the unburdened (less material) closing core can build a magnetic field faster than the burdened coil with the magnetically attached ferrous armature can collapse a field.

It’s a hypothesis to be explored another day.

Thoughts?

As for other items of interest, there is a slight decrease in flight time with a minor increase in bounce time. Note that the voltage on the collector Q4 jumps to approximately -20 VDC when the relay is turned off. This is approximated as a 50 VDC voltage drop across the transistor. This is well within the V_{CE} rating of the chosen 2N5401 transistor.

We can take this process further by adding additional series resistance to the flyback diode path. Figure 6 shows the results when resistor R4 is changed to 6.6 kΩ. The time to first observable change is now 2.4 ms. The cost is a significant increase in the Q2 V_{CE} voltage. The transistor is now feeling the difference between the -100 volt spike and the 24 VDC rail. This is getting close to the transistor’s V_{CE} design maximum voltage. Once again, there is a small reduction in flight time with no appreciable change in contact bounce.

Figure 6: Deactivation waveforms for the relay when operated in a L/8R environment.

Lesson Learned

Have we backed ourselves into a corner?

The collective wisdom of engineers tells us that faster relay deactivation is a desirable goal. The working theory is that coil energy dissipation will result in the faster moving contacts. This should result in longer relay life as faster moving contacts are better able to extinguish the arc that naturally forms between contacts opening under load.

Unfortunately, this arguable limited experiment does not support that theory. Instead, it shows that relay flight time is loosely associated with the L/nR time constant. Recall that we defined flight time in terms of a double-pole contact. This is the time when the armature is in motion with no connection to the N.C. or the N.O. contact.

This contradictory thinking is also supported by contact bounce time and the signature of that bounce. Like a basketball, wouldn’t a faster moving contact have more kinetic energy causing it to bounce higher with continued bouncing for an extended period? That does not appear to happen.

This has been an interesting exercise. What should we do next to better understand these “simple” relays?

Your comments and suggestions are most welcome.

Best Wishes,

APDahlen

P.S. There is a way out of this velocity corner as described in this article: The MOSFET Active Clamp: The Case Against a Relay’s Parallel Flyback Diode

Return to the Industrial Control and Automation Index

New Information

The results of the previous experiment were troubling, as they didn’t support conventional wisdom regarding fast contact opening. To gain more knowledge we will try the experiment again. This time using the large 3-phase contactor as shown in this picture.

Note that it was necessary to remove the bidirectional diode assembly from the contactor’s body. This diode can be seen on the box next to the contactor. Also, it was necessary to add an auxiliary contact module to the top of the contactor. This provided access to the N.O. and N.C contacts.

The results are stunning as shown in the next figure. With the bidirectional diode removed and a 1N4001 connected across the contactor’s coil, it took about 105 ms for the relay to open. The “flight time” defined as the time it takes for the N.O. contact to open and the N.C. contact to close was about 10 ms.

Improved Response

Rather than follow the L/nR method, I used the active clamp as introduced in a follow up article involving an Active Clamp MOSFET river. The results are shown here.

Observe that the horizontal time frame is not the same between these two images.

When compared to the flyback diode, the active clamp opens the contactor in 18 ms. That is nearly a six-time speed increase. The flight time is reduced by a factor of 2 to about 4.5 ms.

Conclusion

Myth confirmed!

The opening time for a contactor is very much dependent on quick demagnetizing of the coil. The velocity appears to be faster which should result in faster arc extinguishing for contacts that are under load. Finally, as shown in these arguably limited experiments, the physics of the mechanical assembly matters. It would appear the effect is greater for the large contactor.

Best Wishes,

APDahlen