• English (United States)

How to Pick a Relay

Relay Questions

The story behind a time delay relay specification or Man Does Not Live By MIL-STD-704 Alone

There's a lot more to specifying a time delay relay than meets the eye. Any manufacturer's data sheet is a good start, but you can't always stop there. You've got to dig beneath the surface of each spec and ferret out the things which really relate to your specific application. Otherwise, you're liable to spend a lot more than you bargained for. And end up with a lot less. Or something worse.


Whether you want to admit it or not, the time delay relay you specify is going to be used out there in the real world. Which means it's got to be done right. Unfortunately, "close" doesn't make it in Relayland.


As a systems designer, you can't be expected to know all the ins and outs of a time delay relay specification. You may want somebody to guide you through the various nuances of meaning which lie hidden beneath the surface of the parameters on the printed page. We've found that there are seven "major considerations" which, if properly understood, can lead you to specify a better, less costly time delay relay. We won't attempt to detail each of these seven parameters in this insert. We will, however, take a quick look at them and share some of the insights we've gathered over many years' experience as a time delay relay manufacturer:


In our last insert (How To Pick A Relay #11, Clearing The Multi-Definitioned Air of Time Delay Relay Terminology), we discussed the different functions which time delays perform. Basically, this covers what you want the time delay to do for your system. Do you want a delay-on-operate? Delay-on-release? Or what? How long a delay do you need to get the job done? And, is this the minimum time only or the maximum time only? There are a lot of different combinations, so it's always best to check with the manufacturer to make sure you're both on the same base.


Specification of the nominal input voltage (28 VDC or 115V, 400 cycle AC) is not a complete answer. Beyond that, the time delay manufacturer must know your power supply specifications with particular regard to voltage limits, transients, power interrupts and so on. (A graphic example of a mis-specified power supply can be found in How to Pick a Relay #10, High Voltage Three-Minute time Delays, Transients and You.) Typically, system voltage supply is specified "per MIL-STD-704." So, let's take another look at MIL-STD-704. It's not a magic number or a cure-all in itself. You've got to be more specific. Is it Category A, B or C? And, does the unit just have to survive the transients that will be imposed upon it by the specification; or does it have to perform its intended function while being subjected to all those bad things? As it turns out, about 80% of what we supply is meant for MIL-STD-704. Nevertheless, it's always a good practice to let the manufacturer in on what you've got cooked up. Especially if your specs are "just a little bit different" than the mil spec.


These are usually pretty easy for you to specify because they're controlling your load. The thing to remember here is that the time delay manufacturer who also manufactures electromechanical relays is a better source than one who doesn't. There are several advantages to such a company. A big one is knowing how to make the time delay and the relay work together. Even a seemingly simple thing like turning a relay off and on can be costly if not handled in the smartest way. (For more on this, see How to Pick a Relay #7, Relay Coil Suppression.)


Specification of the environmental conditions often poses a problem because of the variety of possible conditions found lurking on the manufacturer's data sheet. Just remember that your overall systems specifications provide basic guidelines for environmental conditions. And, when you specify the output relay, you automatically impose the relay specification limits on the entire time delay relay. Obviously, a hermetically sealed output relay gets you around most of the environmental problems you're apt to face. But there may be a few things you have not thought about.

For instance, if you mount a relay on a wing strut, you've got to figure out how much additional vibration you're going to come up against. Finally, it's a good idea to specify temperature two ways; the temperature the relay has to be able to survive at (ie, storage); and the temperature range over which it has to operate. If these are different, you can save money by letting the manufacturer know beforehand.


It all boils down to this: the tighter the tolerance, the more money the time delay relay costs. What you have to decide, then, is what you can get away with and still get the job done. An important thing to remember is that a reputable manufacturer will put some kind of a "pad" into the tolerances he lists. So, if you specify +/- 5% in order to make sure you're going to get +/- 10%, you've overcompensated. And needlessly increased the cost of the time delay. Tolerance, temperature and money are all interrelated. A wide temperature range by itself is easy to make. A narrow tolerance by itself is easy to make. But mix these two things together and it can cost you a lot of extra money. A final tip: always verify with the manufacturer that his tolerance is specified over all temperature and voltage variations.


Here, the manufacturer wants to know if you have any size limitations. Or if you want the unit mounted in some special way. In regard to size, there are a couple of good rules of thumb to follow: First, the volume of the output relay plus 50% is needed to package the output relay. Second, it is desirable to have an additional 1 1/2 cubic inches for the electronic timing function. Obviously, you can specify any size and shape you want. It's only sensible, though, to use the cans, headers and mounting styles the manufacturer is already tooled for and save the nice retooling charges that can be accrued if you want it "just like that only 1/20,000th of an inch less in some dimension." Besides checking manufacturers for their particular size and shape cans, it's also good to look at an established relay mil spec for other package configurations.


Once again, merely saying you'd like "EMI protection as per MIL-STD-461" is not always adequate. You've got to look beyond the surface of the mil spec and define the exact performance requirements under your particular EMI environment. A little extra thought early can save you a lot of money later on. We don't pretend to know everything there is to know about time delay relays. But we've lived through enough mistakes along the line to be able to solve 99.9% of the problems you're most likely to come up against. So, if you're about to specify a time delay relay and you've got any kind of question, just give us a call. Our applications group is waiting to give you some experience-nurtured, realistic answers.

Specmanship and relay reliability or It's not what you've got, it's how you say it

There probably isn't a single engineer around who hasn't been caught in the often-very-wide gap between "specifications" and "requirements" because of specmanship.

Specmanship implies things that just aren't so. You might get what you expected - but only under uniquely controlled conditions. Or for more money. Or under some other very special circumstances. And relay reliability figures are one area in which you can get burned: the same life test data can be used to calculate reliability figures that differ by an order of magnitude. Getting higher reliability ratings is a lot easier than getting higher reliability.

But to be forewarned is to be forearmed, so here's some ammunition.


Figure 1 is a generalized curve relating failures and life for almost any manufactured product, with the life period of the device divided into three major sections.

The "wear out" period begins when the failure rate starts rising because of general wear and tear on the device. This period shouldn't be of much concern to the user, and won't be if it starts after the rated useful lifetime of the equipment. By this time, you've gotten your money's worth and can replace the device. A regularly scheduled maintenance program prevents these potential failures from having any effect on your system performance.

The "early failure period" starts when the product comes off the end of the production line. No matter how good the design or how tight the manufacturing Q.C., the initial failure rate will be higher than normal. These failures are generally conceded to be the result of "production processes," but whatever the cause these incipient early failures must be weeded out.

Screening (burn-in or aging) catches these early failures. The screening must be rigorous enough to fail all the "weak" units, but not stringent enough to affect the operational life of the units that pass. For relays, the burn-in is generally 5% of the rated useful life and at loads that will minimize contact degradation.

There are a lot of definitions for the remaining period, but we'd like to be pragmatic: The useful life starts when the manufacturer ships the relays and ends when they have been operated for their minimum rated lifetime. This is the period of random failures where reliability data applies.


Relay reliability is demonstrated by performing life tests on a representative sample randomly selected from each manufacturing lot. Lot sizes vary to a maximum of about one week's production of a particular configuration. Generally, the testing is carried out at the maximum ambient temperature and at the rated contact loads in a controlled environment. The samples are cycled to the minimum rated life of the relay, and the failures are used to calculate the demonstrated reliability.


Figure 2 shows the results of life tests on a sample of 100 4PDT (four pole, double throw) relays. Testing was terminated at 100,000 cycles (each cycle is the relay coil energized and de-energized once) and five failures were observed.

For the data to be meaningful, testing must be terminated on a failure. If, instead, it stops on a successful operation, accepted practice is to add one failure to the number observed on the assumption that the next operation would have been a failure. The approach is slightly pessimistic, but very realistic. The failure rate is then calculated as:

Failure rate =
(failures + 1)
total operations
x 1,000,000

(in failures per million operations).

This is where specmanship comes into play. By not adding the pessimistic "1" to the number of observed failures, the failure rate is decreased and the reliability is "improved." The reliability is "improved" even more if an "operation" is re-defined to mean "one contact make/break," since each subject relay has eight contacts.


Even if the right definitions are used and the calculations are pessimistic, the reliability figure is meaningless if an inadequate sample size is used. Figure 3 shows the accumulated sample sizes for testing to various reliability levels. The 100 units used in the earlier example were adequate to test to the M Level. Notice that sample size and the number of observed failures do not decrease proportionately. Any reliability data based on "projections" from smaller sample sizes than those specified in the table for the observed failures are totally invalid.

Getting the wrong level of reliability for your application can be expensive, in dollars if you buy too much, in excessive failures and down-time if you buy too little. Specifying the right level is easier if you translate the terms "failure rate" and "confidence level." For example, the failure rate at the P Level is the equivalent of having one failure when using 100 relays (regardless of the number of contacts per relay) for a lifetime of 100,000 operations each. Do you want to pay for this much reliability? Do you need more, or less?

And the "confidence level" is the equivalent of the tolerances on physical measurements. It gives the range of uncertainty of the reliability data and is inherent in the sampling process.


You can short-circuit any chance of specmanship by getting all the facts. Ask the manufacturer to provide you with:

1) Sample size (is it representative of the lot?) and demonstration period (number of lots, months, etc.);
2) Method of computation (pessimistic? Based on coil operations, not contact operations?);
3) Confidence limits (and is this supported by the sample size?);
4) Life test conditions.

This latter point is important because life tests are carried out under specific conditions that probably differ from those in your applications. The failure rate may not be applicable under your load or environmental conditions. If this is the case, the manufacturer can usually provide suggestions that will give you the reliable performance you need.

Another good move towards getting the right performance is to check with Leach before you buy. We have years of experience at delivering high performance, high reliability relays behind us and our applications group is standing by.

Figure 1
This chart is a generalized
picture of the characteristic
failure of relays. The total
life period is measured in
operations since relay failures
are primarily due to functional
operations rather than
accumulated time.

The early failures should be
eliminated by the relay manufacturer's
in-house screening. Then, the
useful life period starts when
the relays are bought and it is
generally 100,000 operations or
more. Reliability data applies
to this constant failure rate
period of random failures. When
this period ends, failures start
increasing as the relays
physically wear out.


Some mechanical aspects of the relay or Did you hear the one about the independent poles?

Our present world is so thoroughly electronic that the mechanical properties of the relay - an electromechanical device - are often overlooked. Sometimes, this is because the relay has been used successfully in one circuit, so it should be all right for a "similar" one. Or, the relay appears to be such a simple device that all you need to do is pick up a catalog and order a specific model. Or, the circuit details are classified and the relay is bought without the supplier knowing the application. Whatever the reason, an all too frequent result is that the relay bought meets specifications but not requirements and another relay "fails."

Take the relays that recently "failed" in some military ground support equipment, while operating completely and entirely within specifications. The relay was a half-crystal can height, four pole, double-throw unit with low level contacts. Since the units were to go into field test equipment, the effort spent on screening parts was considerable. The relays in particular received a great deal of attention because each piece of test gear used forty-eight of them.

The failures occurred in twelve of the forty-eight that operated in a switching mode, one second "on," one second "off," over a thirty-second interval. The output of a photomultiplier tube, a current of about 100 nanoamps maximum, was summed by a integrating capacitor during the "on" time. During the "off" time, the opposite terminals of the photomultiplier tube were connected to the same capacitor, thus effectively subtracting the dark current of the photomultiplier tube, and leaving only a charge proportional to the light energy impinging on the tube. The approach was good, but even with a standard non-varying light source, readings were erratic. One of Leach's ubiquitous field sales engineers was asked for an explanation of the failures, even though the relays had not been manufactured by Leach. No, he didn't solve the problem right there. But he did get one of the failures sent to our factory for analysis.

Thorough testing at the factory revealed that the relay functioned completely within the manufacturer's published specifications. The "failure" was a potential failure mode for any relay, but one which had almost been assured by the unorthodox relay construction. The relay experienced a "make-before-break" condition (two poles making before the other two broke, and vice versa). In the usual application, non-simultaneous operation may be undesirable, but is seldom critical; in this case it was. The manufacturer had almost guaranteed that this would happen by using two 1/2 crystal can size relays on a single header, and with only electrical connections between the two. Without a rigid mechanical inter-connection the two sets of contacts could and did have different operate times.

The integrating capacitor was able to discharge momentarily through the shorted relay contacts during the switching cycle, with the result that the energy stored on the capacitor varied from cycle to cycle. Leach's recommendations was for 100% simultaneity testing of the relays before installation, for a change of relays, or for a circuit design change to blank the operation of these relays during switching. The circuit design change was preferred, and the "failures" stopped when the change was implemented.

This example illustrates the importance of timing when more than one set of contacts or more than one circuit are interdependent. Relay operation is not instantaneous. The relay is a mechanical device and times are associated with the various positions of the contacts. A detailed understanding of the mechanics of relay operation can often save the user a lot of grief.


Figure 1a is a slightly-modified version of a circuit given in MIL-R-6106F (ASG) for measuring the physical performance of relay contacts. It is typical of the circuits used for this purpose. Note that physical displacement is NOT measured by this circuit - an electrical analog in the form of the current through the contacts is the parameter actually being measured. Extra care must be taken to ensure that only the effects of contact motion show up in the results.

It is absolutely necessary that the load on the contacts is below the minimum values that will allow arcing (voltage 6V or less, current 100mA or less). If arcing is permitted, the actual performance of the contacts will be masked. The results will look better, but will not reflect the true situation and will probably lead to another relay "failure." The solid line in Figure 1c indicates the actual performance of a set of double-throw contacts; the dotted line indicates the "results" with arcing of the contacts.

Operation of the circuit is straightforward. The waveform of Figure 1b is applied to the coil, and the results are acquired by monitoring the voltage drop across Ry. In this circuit, arcing is prevented by using 6V, but a higher voltage could be used if the resistor were chosen to limit the current to 100 mA or less. The second resistor, Rx = Ry, is inserted to differentiate between the contact current for the energized and de-energized relay states.

"Operate time" represents the time from the application of the actuating signal to the time when the contacts first meet. But, final contact has not been made yet. The moving contacts have kinetic energy and when they strike the fixed contacts they behave much like a metal ball dropping on a metal plate would - they bounce. The final actuation time is thus longer than the operate time. The final actuation time is usually specified; operate time or initial actuation time seldom are. Thus, it can be seen that it is quite possible for an overlap of operation to occur when more than one pole is operated. This is in fact what happened in the misapplication described earlier.

A similar analysis applies to the de-energization of the relay, as shown in the right half of the waveform in Figure 1. The bounce exists on this portion of the switching cycle largely as a function of the relay construction. Relays that use a spring to return the contacts to their de-energized position suffer in this respect. The return springs are at minimum extension (and therefore at minimum force) in the "off" position and give the least stabilizing force. Since each bounce is the equivalent of another operation, contact life can be greatly reduced by excessive bounce. The problem is further aggravated when the relay must operate in an environment that subjects it to high shock and vibration levels.


Leach's Balanced-Force Relay obviates this type of bounce problem by applying the same amount of force to the armature for both the energized and de-energized states. This is done by using a permanent magnet as shown in Figure 2. (Solid lines indicate the de-energized state; dotted lines show the energized state.) Additionally, the armature is in inertial symmetry to further reduce bounce problems.


We have covered some of the physical aspects of relays that could lead to misapplications. We would like to strongly recommend that you discuss all your applications with the relay manufacturer, and discuss them with your circuit diagram in front of you. Whatever your application, our Applications Group has had years of experience at supplying the right relay for the right job. Let them do the same for you. Please call.

Relay, reliability, part II or There's no such thing as a free lunch

Some relay misapplications we have seen might have been prevented if the nature of the problem had been determined at the outset of circuit design. Solutions often exist that can meet seemingly contradictory requirements.

Recently, a computer manufacturer chose a relay to serve as the interface between his computer logic and a high speed printer in the outside world. One of the prime limitations on this relay was size, so he called out one of our 2-pole subminiature relays with contacts capable of handling currents up to 2 amps resistive.

When the system was turned on, the relay functioned satisfactorily a few times before the contacts welded shut and the printer went on and on, spewing paper all over the room.

The customer quickly called to advise us that our relay had "failed" although his load was drawing only the rated current (he had just rechecked this). Further questioning determined that the printer was indeed drawing only 2 amps, but to a very inductive load.

An inductor will store energy equal to 1/2 LI2, where L is the inductance (henries) and I is the current (amps) through it. When the contacts separate the load from the supply, this energy must be dissipated someplace, and the usual place is in an arc across the relay contacts. The result is melting and vaporization of the contact material with metal transfer from one contact to the other if the applied voltage is DC. The effect is further aggravated if the load circuit has a long time constant (high L/R ratio).

In this case, the load was highly inductive, with an L/R ratio of about 0.06 (60 milliseconds). The contacts in the relay he had were suitable for inductive loads of 1 amp and a time constant of 0.025. Our recommendation was that the customer use one of our relays that was rated for 5 amps resistive and would handle his inductive load.

The customer couldn't accept this suggestion because of his space limitations. His program was so far along that he could not change the physical parameters in his equipment, yet we could not get the 5 amp contacts into the space allotted to us.

The solution was a diode connected as shown in Figure 1. When the contacts break, the reverse voltage on the inductor has a discharge path through the diode that does not pass through the contacts. The energy is dissipated harmlessly.


Where size and weight are critical and overload conditions are transitory (as is the case with non-resistive loads), additional circuitry might be more advantageous to the user than going to larger relays capable of handling the overloads. An added benefit is that even with smaller contacts, operating lifetime is increased.

Arc suppression in AC circuits is less of a problem than in DC circuits because the arc is extinguished whenever the current passes through zero. Arcing still occurs, however, and the life of the contacts can be extended by using either of the networks shown in Figure 2.

The resistor-capacitor network of Figure 2 (a) is designed so the circuit seen by the applied voltage is effectively resistive. The resistor and capacitor are chosen to give a time constant for the suppression network approximately equal to the time constant of the load, or RcC = L/RL. The resistor should be in the order of 100K Ohms, and the diodes should have a peak inverse voltage of about 400 volts. Make arcing is eliminated because of the apparently resistive nature of the load; break arcing is eliminated because of the apparently resistive nature of the load; break arcing is eliminated because a circulating discharge path is now available for the stored energy.

Figure 2(b) is useful for both AC and DC loads. Arcs could be suppressed using either the capacitor (C) or the resistor (R) alone, but this will cause problems. The charging current of the capacitor alone during make would tend to weld the contacts shut, so the added resistor serves as a current limiter. If the resistor is used alone, it draws current the whole time the circuit is activated and wastes energy. When both are used together, the protecting resistor should approximately match the load resistance (RL). This strikes a balance between the need for a high resistance during make as a current limiter and the need for a low resistance in the discharge path during break. When choosing the capacitor, particular attention should be paid to its breakdown voltage. A high voltage spike is generated by the inductance (L) when the current is abruptly interrupted (=-L di/dt) and the capacitor must be able to withstand it.

Another effective method of arc suppression in DC circuits was shown in Figure 1. The diode must be chosen to handle an initial current equal in magnitude to the normal load current and this limits the applicability of this technique.

Another characteristic of this circuit is the low forward resistance of the diode with a consequent long time constant for the discharge path, and thus a longer time for hazardous arcing. This can be eliminated by placing a resistor or varistor in series with the diode in the protecting leg. The time constant now becomes equal to the load inductance divided by the load resistance plus the added resistance [T=L/(RL+R added)].

For extremely inductive loads, the circuit of Figure 3 provides effective protection. The capacitor is charged through the diode and absorbs energy from the inductive load when the circuit breaks.

A major disadvantage of this circuit is that if the capacitor should short out, the load is activated. It is not a "fail safe" circuit.

Contact ratings can also be extended by the use of two contacts in series to control the load without the use of additional circuitry. The amount of improvement in performance is not as great as with the other methods discussed and depends on contact material and spacing, opening speeds, and load inductance. This technique can carry with it the penalty of generating larger voltage spikes because of the possibility of a more rapid current interruption. But the technique is effective if the amount of improvement required is not too great and the extra contacts are available.

All of the foregoing are useful techniques for getting the most out of relays. Limitations exist on the current and voltage ranges over which relays can be used - so enlist the aid of the relay designer.

At Leach, we have an applications group with an unquenchable desire to prevent relay misapplications and the skills to do so. Call us and add our experience to your design team.


Relay coil suppression or How to keep an airline pilot on the straight and narrow

Relays have been around for many decades, yet have survived evolving technologies and still retain a significant position in today's electrical and electronic systems. There have been changes in materials and changes in design, but the relay is still essentially a simple electro-mechanical switch. With today's ever-increasing need for the switching function, the relay offers advantages over other switching techniques. It can be used singly without auxiliary circuits (aside from a power supply); it exhibits very high isolation between controlling and controlled circuits; it can result in a simple, inexpensive circuit fast enough even for today's high speed world; and it can be compatible with semiconductors.

On the other hand, this compatibility is sometimes not achieved because one relay parameter or another is not considered. A relay is a very simple device - how much engineering time should be spent on it? Let's find out.


The cockpit of a commercial airliner witnessed the results of a lack of consideration for relays. The flight was on schedule, at cruising altitude and making good time. On auto pilot and smooth - when all at once the stories of the last layover were interrupted by bells, buzzers and flashing red lights. Not Christmas - autopilot failure. The rest of the flight was on instruments and the seat of the pants. A safe flight, but a busy one.

What happened? Several transistors in the autopilot computer and controller had been burned out, and this was traced to negative, high voltage spikes on the 28-volt DC line. The cause? Transients generated by relay coils in the system. The solution? Transient suppression. The reason for problems? Overlooking the fact that the relay coil is an inductor.

System designers usually take into account the problems associated with the making and breaking of currents by the relay contacts. In this case, arc suppression had been included on the load side of the relay to extend contact life and reduce the RFI generated by arcs. (Contact protection is treated in a later installment.) The designers had, however, neglected the fact that the relay coil, too, is a non-resistive load and is thus capable of generating interference. The energy stored in the coil inductance is seen as a back EMF across the coil when the drive is removed. This voltage is usually greater than 750 volts and can be as large as 3000 volts in a 28-volt circuit. Few components are designed to withstand voltages of this magnitude.


The first step in curing circuit interference is to limit the magnitude of the coil generated spike. Any of the circuits shown will do this.

The diode in Figure 1 is probably the most popular form of voltage suppression used today. A single diode (D1) can be used, but this is frequently burned out by the application of the wrong polarity to the coil. Diode D2 prevents this type of damage. Diode D1 provides a very low resistance re-circulating path for the energy in the coil, and thus offers the highest degree of suppression available. Because of the low resistance, however, the time constant for energy decay is quite high, and the dropout time of the relay with a diode across the coil is increased at least 2 times - and often 10 times - the normal value for the unsuppressed relay. This slows the rate of separation of the relay contacts, and can increase arcing damage to the contacts. This actuation delay can also be very critical when several circuits are operating interdependently.

One method of suppression frequently used by relay manufacturers, the bifilar coil, is shown in Figure 2. This is manufactured by winding two coils in parallel simultaneously, then shorting the secondary coil inside the case. The resistance of this secondary coil determines the effectiveness of the suppression. The maximum back EMF generated is approximately equal to the applied coil voltage times the ratio of the coil resistances (EMF = V x R bifilar/R coil).

This last equation seems to point up the bifilar method of suppression as one that can deliver an extreme amount of back EMF limiting. This is true, but you never get something for nothing. The bifilar coil in a relay has a considerable effect on relay performance. For instance, the smaller the resistance of the bifilar coil, the smaller the back EMF, but also the longer the dropout delay. The transfer time of the relay contacts may increase as much as 5 times.

In addition to this, because a current flows in the secondary coil, a magnetic field is generated by it. As the armature moves, the air gap changes and causes a change in the magnetic flux, which in turn causes an increase in self-induced current in the secondary winding. This increase not only slows down the motion of the armature, but may even reverse its direction. If this is the case, break bounce occurs. This of course can cause arcing, which damages the contacts and shortens contact life.


A better method of suppression is shown in Figure 3. The zener diode can be placed in series with the shunt diode (D1) shown in Figure 1. Or, two zener diodes can be used back-to-back instead. This latter arrangement has the advantage that it is not polarized. The peak back EMF from the coil is now limited to the breakdown voltage of the zener diode. The breakdown voltage obviously should be chosen to be greater than the applied voltage on the coil. The increase in the drop out time when using this technique is negligible. In essence, the zener diode affects the relay performance only when that performance is out of its normal range; the rest of the time the relay behaves as if the zener were not there. (This is not the case with the bifilar coil, which affects the relay's performance at all times.)

The series RC circuit of Figure 4 is effective for voltage limiting, but is usually used only with coil currents under 100mA because of the voltage drop through the series resistors. This voltage drop can be eliminated by placing the resistance and capacitance both in the shunt path.

Figure 5 represents a common relay interface - the transistor driver. Because of the low currents required by relay coils, economical drive circuits can be readily designed. The transistor eliminates the possibility of arcing that exists when a switch activates the coil directly. Even if a switch is used to turn the transistor on, the transistor serves as a buffer amplifier so that a much lower (and hence less troublesome) current is being interrupted. Operate time can be tailored by varying the value of the capacitor. The zener shown prevents transistor burn-out when the coil is de-energized. The ground side of the relay coil is the preferred location for this type of circuit to make it "fail safe."


No component values have been given for any of the suppression circuits because these are dependent on circuit and relay coil parameters and choices will have to be made on that basis. The aim of suppression is voltage spike reduction, but contact life can be reduced if the techniques are improperly applied. If you have a specific application with which you would like some assistance, don't hesitate to contact us. We have an application group with years of experience at applying these techniques to practical, reliable circuits and would be happy to help you.

Non-resistive loads: another common relay misapplication or How to keep 130 people in the dark at 30,000 feet

The electromagnetic relay, though easy to understand, seems to be subject to more misapplication "failures" than other components. Part of the reason for this is its very simplicity. Designers tend to discount the relay and do not spend as much time analyzing its operation in their circuits as they do with more complex devices.

A common area of misunderstanding - and misapplication - relates to the fact that there is no single voltage/current rating that applies to a given set of contacts under all circumstances. To ensure the successful, full rated lifetime operation of the relay, it must be chosen in terms of the specific application.

Recently, a transcontinental airliner at 30,000 feet was the scene of a misapplication of this type. As dusk silently fell, the pilot toggled the switch for the inside lights. A low level signal was generated and proceeded down the control lines to the relay coil. The relay actuated and contact was made. There was a short pause as the lights came on - then died. 130 people sat looking darkly about them for the time it took the auxiliary power system to cut in.

And another "failure" was chalked up to relays because the manufacturer had not been given enough information about the application.


The engineer, when specifying the relay, was aware of the environmental load imposed on equipment by modern aircraft. The relay called out was realistically specified to operate over the temperature range from -65° C to +125° C and to withstand vibrations up to 20 g's at 2000Hz, and shocks to 50g with durations up to 11 milliseconds. He then included a safety factor for the contacts and specified a current carrying capability 50% higher than the steady-state current he expected his lamp load to draw.

Unfortunately, this wasn't nearly adequate to handle the turn-on transients through the filaments, as these might have been as much as 10 or 15 times the steady-state current, and the contacts were burned out shortly after they were energized.


The load-carrying capacity of contacts is normally given as a current value for a resistive load. When a load is nonresistive, energy stored in the circuit or changes in the load characteristics can drastically overload the contacts.

Although there might be momentary arcing between the contacts on make or break, resistive loads draw an almost constant current at all times. As long as the contact ratings are not exceeded, the relay functions reliably. When ratings are exceeded, problems can be expected. If the applied voltage is direct current, excessive arcing results in vaporization of the metal of the contacts and transfer of the material from one contact to the other. This results in what is known as a "pip" and "dimple" - a built-up area on one contact and a crater in the other. Because of the roughness of these areas and because the arcing melts the metal, destruction of the contacts can be expected. If the applied voltage is AC, the arc tends to quench itself as the applied voltage goes through zero and wear on the contacts is less severe than in the DC case. For a given value of current flow, contacts rated for 29 to 32 volts DC can normally be expected to handle 115 VAC at 400Hz.


Lamp filaments are resistive, but change in value by a large factor from their cold state to their operating state. This effect is so great that the inrush current can be expected to be 10 to 15 times greater than the steady-state value as indicated in Figure 1. If this is not taken into account, the consequences described in our example can result, with the contacts either welding shut or being totally destroyed. Normal practice is to derate contacts to 20% of their resistive load capabilities for a lamp load.

Inductors and transformers act as energy storage devices and can cause excessive contact arcing when the relay breaks the circuit. When operated near their maximum capability, the overload added by the arcing will cause deterioration of the contacts. The material of the contacts is vaporized and, if the voltage is DC, is transferred from one contact to the other. The situation is further aggravated by contact bounce and vibration, and a long time constant for the load (measured by the ratio of the load inductance to the resistance of the discharge path: T=L/R). Local heating of the contacts causes either welding or heavy metal transfer. For this type of application, contacts are normally derated to 50% of their resistive load capacity.


Motors are inductive and will cause arcing on the break, but also draw an initial transient on the make. The peak current can be 5 to 10 times the run current. Contacts are typically derated to 40% of their rating for resistive loads when used to control motors.

Normal derating practices for the contacts have been given, but the best practice is for the designer to discuss his application with the relay manufacturer, as other factors enter into the design of the relay. The amount of arcing and consequent damage can be considerably affected by the rate or amount of contact separation or the amount of contact bounce after the relay is actuated. The designer is the best judge of the appropriate relay for the specific application and should be consulted.

We will continue this series with a variety of techniques for protecting contacts, but in the meantime, please call if you have any questions. The Leach applications group is well staffed and well qualified to help you define your relay requirements. The information we have gathered over the years can be used to prevent a misapplication in your circuits.

High Voltage 3-Minute Time Delays, Transients & You or :30, :29, :28, Zap! (Oops!)

A common time delay relay application is the delaying of high voltage to klystrons and magnetrons. And the trick is to avoid applying full voltage for at least 180 seconds or suffer the loss of an expensive tube. Yet, along these lines, even the most noble of intentions can run awry. For example:


Some time back, an engineer came to us and asked for a special three-minute time delay. Special because he wanted it to be very small. As we discussed his particular requirements with him, the question of transient protection was raised. "No problem," said the engineer. "I've got good, solid power, so I don't need any transient protection. I don't need any recycling protection, just make it small." So we made it small. 1" x 1.2" x .5" to be exact. And he was quite happy. His relay worked perfectly during initial system testing. But later, during more extensive testing, things got pretty hot. A technician switched the power off for a very short period, then switched it on again. The cycle was thrown off. Time-out was less than 180 seconds. Bam! A $3,000.00 magnetron went up in smoke. And so did another before the problem was pinpointed.


his particular incident points out a problem which is all too common in time delay relay specification. In this case, it was the engineer's misunderstanding (or lack of complete information) as to the transients he could expect on his supply line. He had thought in terms of normal system operation. He had not explored the peculiar ramifications of what power interruption during check-out and test might mean to him. And so he got burned.


Solving transient problems is only one consideration in the selection of the right time delay. Maximizing equipment availability and "on-the-air" time is another. The following examples show how two particular customers accomplished both these things. First, by thoroughly thinking out their applications. Then, by specifying a relay to meet their exact needs.


This customer wanted a time delay with a built-in immunity to "short" power interrupt. He said: "If I have a power interrupt for up to five seconds, then power comes back on, turn me back on immediately. Don't make me have to start from zero." He had a good point. So we designed a special time delay that did exactly what he wanted.


Then there was the engineer who approached it a different way. He reasoned (rightly) that you heat up on a gradual curve for three minutes. Then you're hot enough to apply full power. When you lose power, you're coming down on the same kind of gradual curve. "Give me a recycle time," he said, "that's proportional to the time off. If I'm off for a minute, make me time for a minute only. If I'm off for thirty seconds, make me time for thirty seconds, and so on." The relay we designed (Figure 1, bottom) gave our customer two important things: It allowed for interrupts of longer durations than the minimum without hurting the tube. It didn't keep him "off-the-air" for three minutes each time he had an interrupt.


We can design the exact time delay relay for your needs. But we've got to know what they are. And that's where you come in. Literally. After you've designed the system that'll get the job done. After you've looked at it carefully to determine the kind of transients you're most likely (or least likely) to encounter during check-out, actual operation and all other conditions. After you know when power's going to be lost, how frequently, for how long, and so on. If you hold up your end, we'll hold up ours. We'll put together the time delay that best meets the sum total of your requirements. A unit that's been 100% tested over the voltage range at temperature extremes. Then you can go back and do good radar, without having to worry about a time delay zapping an expensive tube, or being off the air when you need to be on. Our applications group stands ready, willing and able to help you.

Contacts: where most misapplications begin or How to save $35 and lose a half-million dollar missile


Electromagnetic relays have been making and breaking circuits for more than half a century. The scope of their applicability has been widened by advances in the techniques and technology of manufacture; industry standards have been established that increase their range of usage; and considerable effort has been expended to increase their reliability. In spite of this, operational failures have not been reduced proportionately. Why not? Because a major reason for relay failures remains the misapplication of the relay. Even with all the changes during its history, in operating principle the relay remains a simple device. It looks unprepossessing on a circuit diagram - an electrically-activated switch. Anything that easy to understand can't be hard to use, can it? This apparent simplicity is the major stumbling block encountered by relay users. The designer tends to overlook this "simple component" and does not give it the same amount of consideration given other components in the circuit. The relay starts working with a virtue turned into a disadvantage. It is simple, but not forgiving. Relay contact ratings are the most misunderstood specifications and the most common cause of relay mis use. It is not enough to call out the currents and voltages that the relay will see. Relay contacts are designed, built, rated and must be specified in terms of the specific type of application for the relay. A 5-amp, 115 volt set of contacts in one circuit may or may not be a 5 amp, 115 volt set of contacts in another application. The designer should specify the type of load, duty and environment to which the relay will be exposed. Incomplete specifications from the user often result in misapplications. The consequences can be tremendous for such a "simple" device.


Consider a recent high reliability application for which the design engineer specified a group of 4-pole, double-throw, single phase, 5-amp relays. Several weeks after delivery of the specified units, a frantic call was received from the engineer - system check-out had just been initiated and relays were failing. What was wrong with the relays? We quickly acquired some of the failed units and examined them, discovering the symptoms of a recurring failure mode in relays - phase-to-phase and phase-to-ground arcing. (Phase-to-ground arcing pits the contacts; phase-to-phase arcing leaves the contacts looking 'blown-up' and occurs between adjacent contacts). The most prominent cause of this type of failure is the misuse of a relay with single-phase contacts to switch poly-phase power. A single-phase relay is designed with contact clearances and dielectrics that will readily handle the peak loads specified when these are imposed by single-phase voltages. In polyphase circuits, the phase angle differences between the applied voltages will result, at some time when the relay is operating, in voltages greater than the relay designed for single phase use can handle. Catastrophic failure often follows, with arcing to ground or the merging of two loads by arcing between adjacent contacts and the destruction of the contacts.

When asked, the engineer stated that he was in fact using the relays to switch a 3-phase motor to and from an ac generator. Our recommendation was that he spend an additional $35 to get a 3-phase relay designed for his type of application, but he declined on the grounds that time was short and his budget was spent. The next failures occurred during system operation. The ac power generator stopped, thrust dissipated, and the missile the relay was on fell back to earth. Total loss: one-half million dollars. Another "relay failure"?


Double-throw contacts are a versatile arrangement that can save the user space, weight and money - sometimes! Too often these savings are only temporary, and are more than cancelled by the first failure due to misapplication.

Figure 1 illustrates a common but potentially hazardous use of a relay with double-throw contacts. Rather than switching between two power sources as shown, the relay might be reversing the direction of a 3-phase motor by swapping phases, switching one source between two loads, or just switching a load from a power source to ground. The diagram applies to all of these situations, and the potential danger of relay misapplication is high. If an arc is created as one contact is broken and is not extinguished before contact is made with the second fixed pole, the power source is shorted and heavy damage results.

All contacts have minimum current and voltage thresholds that must be exceeded before a sustained arc can be created. The minimum voltage required, a function of the material used for the contacts, is 12v for silver (typical of most metals), 15v for gold and 17.5v for platinum. A minimum current is also required, again a function of the material. For silver (typical of many metals), this is 400ma; for gold, 400ma; and for platinum, 700ma. The only lower values are those for cadmium and zinc (100ma) and for carbon (10-20ma). Both voltage and current must exceed these threshold values at the same time for hazardous arcing to exist.

If your circuit will not exceed these threshold conditions, it won't sustain an arc, then the configurations of Figure 1 can be used. If they are exceeded, you could use this circuit for selection only, and have the load carried by another set of contacts. Or consider the circuit shown in Figure 2.

The relays are both single-pole relays, each with 3 normally-open (N.O.) and one normally-closed (N.C.) contacts. The normally-closed contacts serve as safety interlocks - neither relay can be energized until the other one is de-energized.


As stated earlier, there is no single voltage/current rating that applies to a given set of contacts under all circumstances. If the specifications seem incomplete, or your requirements are unusual, call the relay manufacturer. At Leach, we have an applications group with years of product experience ready to go to work on your problem. Call and let us help. 

Contact duty: dry circuit to heavy load or How to keep a sub on the up and up

There is no relay contact that can be used for switching all load levels. Each load level requires tailoring of the contacts for that specific application. Contacts rated as capable of handling "dry circuit to 2A" loads will do so. They cannot, however, handle loads that go from 2A to dry circuit levels. Knowing this might have prevented one misapplication that luckily had only embarrassing consequences.


In spite of the early hour, a small audience witnessed the departure of the submarine on its training voyage. Wives and friends along the channel waved their farewells to the crewmen still on deck. At last they were past the breakwater.

A klaxon signaled "dive" and the topside crew vanished. The last hatch cover clanged shut. Then silence except for the throb of the diesels as the sub sailed on. And on. And on. And on.

We won't describe the language that shattered this placidity once it was realized that the sub definitely refused to submerge.


Back at the base, the failure was traced to a DPDT, 1/2 crystal can relay in the dive control circuitry. We received the relay and a 4-page, typed, single- spaced (but friendly) letter shortly after.

Analysis at the factory revealed that the relay, although rated dry circuit to 2A, no longer functioned satisfactorily at low load levels. It was still within its specifications at its maximum rated load. Cutting the casing off the relay, we found that the original gold alloy plating on the contacts had been burned off.

Checking again with the customer, we received a circuit diagram that showed the relay operating as shown in Figure 1. The solution was right there. Many relays are symmetrical physically and electrically, as was this one. At some time, probably during trouble-shooting before the exercise, the relay had been pulled from its socket, then replaced physically reversed. In terms of the poles and coil, it made no difference. In terms of operation, it meant a failure.

Why a failure when the relay was rated from dry circuit to 2A? Because any relay rated to cover the range of loads from dry circuit to heavy loads is a one-way device. Once it has been used for the larger loads, it no longer operates satisfactorily at the low levels. This is a limitation of materials, not design, and no immediate change in this situation is in sight.


Relays can be classified in terms of the duty or loading to which the contacts are subjected. Since this classification is often also the way to get into a fast argument, let's paraphrase the definition in the NARM "Engineers' Relay Handbook:"

Heavy loads: Current through the contacts is great enough that some arcing occurs under normal operation. Typically, the voltage must be greater than 12V and the current must be more than 400mA for arcing to occur with most metals. The arcing actually serves a useful function - it cleans the mating surfaces. Contact materials must be chosen, however, to minimize electrical erosion. A silver-cadmium oxide has been found to work well at these load levels.

Intermediate loads: Current is below the minimum level for even momentary arcing when the contacts are open. Loads of 50-400mA at 26V are typical for this range. Some arcing can occur during the "make" or "break" of the contacts, but extinguishes itself by the end of the contact transfer. This arcing is usually just enough to carbonize any organic vapors present. The carbonized material ends up as a deposit on the contacts and contact resistance goes up, often leading to failure. The difficulty is minimized by the appropriate choice of insulation, potting compounds and cleaning agents to reduce the amount of organic vapors present.

Low level loads: Currents are in the microamp or low milliamp range, with the voltage across the open contacts well below the melting point voltage of the contacts. No arcing occurs, but organic vapors still are a problem. The sliding of the contact surfaces causes polymerization of the organic compounds with the result that deposits with high, unstable resistance are left on the contacts. The problem is particularly bad with metals that are members of the platinum family. The solution is to use gold or gold alloy platings on the contacts.

Dry circuit loads: No current is switched. The contacts carry current only after they are closed or before they are opened. The currents may be high, as long as they are not switched. Since there is no arcing, contact resistance is kept low by using gold plating or gold alloy contacts.


The contact materials used are selected on the basis of the load they will be switching because materials for one load level are generally very unsuitable for the other end of the scale. Silver-cadmium oxide handles high currents easily, but fails at lower levels where there is no arcing. Gold alloys work well at low levels and eliminate the organic contamination problem, but at high levels erode very rapidly.

This explains why multi-load level contacts are one-way devices. The usual method for making them is to plate a heavy-load contact material with the gold alloy. At low levels, contact resistance is low and stays low. At high load levels, the plating is burned off, exposing the material with the high immunity to arcing. But, since the plating is gone, the contacts are no longer suitable for low levels.

Another method of making multi-load level contacts is to build projections on them. These provide a high unit area pressure at the point of contact, ensuring low resistance. These are one-way contacts, too, as the projection is designed to burn off at high loads to allow the mating of larger surfaces (necessary for heat dissipation and current distribution).


Getting back to our example, the reason that the plating was gone from the contacts, even though the steady-state load on the B contact was only 100mA, is because a lamp is a non-linear load. The initial surge when the contacts first mated was probably over 1A. Obviously, the relay had been operated under one circuit condition for a period of time and then mechanically reversed so that the B contacts were later supposed to handle the dry circuit load of the logic circuit.

If you have an application that's a special problem, call our applications group now. They're interested in two-way communications to prevent relay failures. Use their assistance in your designs.

Coil voltage: no performance spec is independent or How an automatic firing system decided to make love not war

Because a relay is a switching device, failures show up as nonperformance of the contacts. This failure mode is often caused by the misapplication of specifications other than those that apply to the contacts.

A case in point occurred in an airborne military fire control system. The customer specified the need for a relay with DPDT 40A contacts and a coil to operate from a 28vdc power source in accordance with Class A (70° C) MIL-R-6106 and vibration of 10G's to 55Hz.

The relays were delivered and installed, followed shortly by a call from an irate customer. None of the relays lasted longer than a few hours in use, with typical time to failure being just over an hour. This was one case where a cessation of firing increased hostilities.

A relay was returned to us and examination of the contacts disclosed metal transfer from one contact to the other, a sure sign of excessive arcing. With this information, the customer re-checked his load and advised us that the contacts actually switched 50A. Although this is an overload of 25% (and not recommended), the contacts could handle this current.

What now? The relays performed well in tests at the factory - even at the overload level - but continued to fail in the field.

A thorough investigation of the field application uncovered several contributory factors that had not been reflected in the original specification: besides the 25% overload on the contacts, ambient temperature was 50° C higher than specified, the coil voltage was not maintained at the proper level, and the relay was being subjected to vibrations far in excess of its rating.

In the field, the relay was located near the engine exhaust and the actual ambient temperature was 120° C after one hour of flight, rather than the specified 70° C. The end result, in this case, could be relay failure due to an open coil. However, more often than not, the coil would not even get a chance to open because a combination of the wrong coil voltage and high vibration would destroy the contacts first.

MIL-R-6106 specified that the voltage in an aircraft shall remain in the range of 18 to 29 volts DC. In this misapplication, the coil voltage was within the required range when the relay was first energized. Contact was made solidly, as shown in Figure 1A.

Normally, the actuating magnetic force applied by the coil is enough to overcome the restoring mechanical force of the springs, and the armature moves to its full operate position. The contacts are held firmly together by the spring tension in the moving contact arm. In this case as soon as the load started to draw current, the input voltage at the coil dropped to 10 volts - a factor not previously specified. The restoring spring drew the armature up until the mechanical force was in equilibrium with the now low magnetic force. In this position, the contacts just lightly touched, with almost no force to hold them together. (Figure 1B). The excessive vibrations (750Hz rather than the specified 55Hz) soon started the contacts chattering, making and breaking the load hundreds of thousands of times per hour. Arcing was almost constant, with consequent metal transfer and the creation of a crater in one contact and a build up on the other ("dimple" and "pip"). When this had gone on long enough, the contacts no longer mated, and another relay failed through misuse. The contacts actually approached their operating lifetime rating, but the rating was applied to them accelerated thousands of times.

The solution was straightforward - a relay designed to cope with the actual environment. Many headaches would have been saved had the specifications actually coincided with the requirements.


Misapplications can result even when the user knows all of the circuit and environmental parameters because of misunderstandings generated by the present profusion and confusion of terminology. Similar relay items are often described by dissimilar terms that relate to the company or industry in which they are used, rather than indicating the parameter being discussed. To cut through this confusion, the National Association of Relay Manufacturers, in the "Engineers' Relay Handbook," lists preferred terms that should be used to ensure that both the user and the supplier understand what requirement is being specified.

If noise is likely to be a problem, the user should call out the "specified non-pickup" point (see Figure 2). Signals below this level will definitely not actuate the relay. He should also call out the "specified pickup" voltage. The relay must and will operate whenever the signal is at or above this value, and the armature will always seat at its full operating position.

Actual tests for these values will yield measurements that fall somewhere in the grey zone of the chart. The measured non-pickup value will be equal to or greater than the specified non-pickup value, while the measured pickup value will be equal to or less than the specified pickup value. In either case, the earlier definitions for both of these points still apply.

Operation is similar for de-energization of the relay. The armature will remain in its full operate position as long as the coil voltage remains at or above the "specified hold" value and will drop out when the voltage drops to or below the "specified dropout" value. Again, measured values for these two parameters will fall in the grey zone of the chart.

An understanding of the operation of relays and the terminology is a good start toward preventing misapplications, but the best step is to call the relay designer and discuss your needs with him. The reliability of the relays used (and your equipment) can be increased by working with the manufacturer on the details of your application.

We at Leach are ready to help you whatever your needs. The years of relay experience represented by our applications group could be working on your circuits today. Call us.

Clearing the multi-definitioned air of time delay relay terminology or How to say what you mean, not just what you say

In our last insert, we presented graphic results of the communications gap between time delay relay terminology and time delay relays. A designer specified a certain time delay relay for his system. He got exactly what he ordered. But it was the wrong relay for the function he wanted it to perform. A $3000 klystron got zapped. And a red-faced designer learned that what you say isn't always what you mean (a dilemma not uncommon these days considering the plethora of time delay phraseology being bandied about).


The obvious solution is to determine the right words to specify the right time delay relay at the right time. Right? Right. But the responsibility goes beyond the designer. If your technician's terminology doesn't match yours, there's going to be trouble in Relay City. A time delay is apt to be tested for the wrong function. At the least, a relay may be destroyed. At the most, an entire system. Either way, time is lost and money is wasted. All because your people couldn't get together on terminology. Then there's the time delay relay manufacturer. He's got to build the exact relay you specify. Nothing more. Nothing less. A relay that will really do what you want it to do.


We would like to clear up the confusion once and for all. And present An Illustrated Glossary of Time Delay Relay Terminology which, if used faithfully by all involved, will result in better, more reliable systems and avoid forever (at least 99.9% of the time) the costly mistakes of Bad Nomenclature:

time delay relay - A switching device in which a time increment is added beyond that normally inherent in the switching mechanism.

time delay on operate - A time delay relay in which the switching or load circuits are energized at a specific period of time after the input circuit is energized.

time delay on release - A time delay relay in which the switching or load circuits are energized when the input control circuit is energized and remain energized for a specific period of time after the control circuit is de-energized.

positive/negative control - That polarity associated with the control lead (3 terminal input device) to which the control terminal is connected. Either polarity may be selected at the design specification stage of development.

interval timer - A time delay relay in which the switching or load circuits are immediately energized - and remain energized - for a specified period of time after the input circuit is energized.

repeat cycle timer - A relay that repeats ON-OFF cycle as specified, as long as the input circuit is energized, i.e., flashers.

adjustable time delay - Any of the above timing functions wherein the capability of varying the timing period is provided. This variation is typically accomplished by varying a resistance which may be an external potentiometer or an external fixed resistor.

recycle time - The minimum time required between operations to assure repeat performance within the specified time tolerance.

recovery time - The minimum time required to assure repeat performance within the specified time tolerance when the input power is removed during the timing interval. This is typically greater than the recycle time for a given unit.

release or reset time - The time required for the switching device to return to the de-energized position upon removal of power.

latching current - That minimum load current necessary to assure that the output device (SCR) will conduct, and remain in a conducting state throughout the worst-case input voltage and temperature variations. Applicable to solid-state output devices only. Frequently overlooked in the receiving inspection operation, this load current magnitude may vary from 0 to 100mA or more, depending upon the particular unit design. Not to be confused with holding current, that minimum load current that must be drawn by the load to assure that the output device (SCR) remains in a conducting state.

The newly released Mil Spec, MIL-R-83726 is the controlling document for electronic time delay relays.

Obviously, no mere definition can suffice in solving all the problems inherent in time delay relay specification. If you have any questions along the line, don't hesitate to call or write. Our time delay relay applications group is ready and willing to lend any assistance you might require.

This site uses cookies to help deliver an engaging user experience.
To learn more about what cookies are and how to manage them visit AboutCookies.org