|
|
| Elliott Sound Products | Arc Mitigation & Prevention |
Main Index
Articles Index
A great deal of what you need to know about arc prevention and/ or mitigation is shown in the second relays article - Relays (Part 2), Contact Protection Schemes. However, there are many other techniques that were only mentioned briefly, largely because they are either little-known or are still covered by patent protection. While this prevents the circuits from being used commercially without infringing, the information is available from the patent documents, so the techniques can still be discussed.
Where circuits are provided, they will show the general scheme, but with only representative component values unless these were also made available in the patent documents. Since most circuits of this nature have to be designed for a particular set of conditions, component values only apply for a limited range of voltages and currents, and there is no 'one size fits all' solution. Arcing contacts have been the bane of industrial systems for as long as they have existed, but today systems run faster than ever before, so contact erosion becomes critical.
Every time a set of contacts arc, material is removed from one contact and re-deposited on the other. With AC, one might imagine that this balances out, but surface erosion causes higher resistance and greater losses. DC systems are particularly hard, because DC creates bigger and better arcs than AC, even at lower voltage and current. A standard 'miniature' style relay can withstand no more than 30V at rated current, and with typical contact clearance of only 0.4mm, higher voltage will cause a sustained arc that can (and does) totally destroy the relay contacts. A photo can be seen in 'Relays - Part 2', linked above. This can happen with only a slight overvoltage - the voltage and current ratings for relays are not described as such, but they represent 'absolute maximum' values to obtain the rated life.
This article is intended to provide information and ideas - it is not intended to be a construction guide. The basic DC arc prevention scheme shown in Figure 4.2 has been bench-tested, and it works exactly as described. Even with an 80V supply and a 4Ω load, there was zero arcing when the relay contacts opened. This demonstrates that relays can be operated safely at well beyond their voltage rating, but it comes with some risk. Electronic parts can fail, and the result may be catastrophic. Considerable testing is necessary to ensure that whatever you choose to do will be safe and reliable, and always include a fuse or other safety device to guard against severe overloads that may cause additional damage or fire.
The basics of a relay are fairly simple, but there are many styles, and countless variations. Multiple contact sets are common, and most are available with different configurations. The contacts are almost always mounted on phosphor-bronze or similar material that has the ability to flex many thousands of times before breaking (mechanical failures are remarkably uncommon. When the coil is energised (AC or DC, depending on the intended usage), the armature is attracted to the pole-piece, and an actuator pushes the moving (common) contact arm to open the normally closed contacts, and close the normally open contacts. Relays with only normally open contacts are common, but it's not often that you'll come across one having only normally closed contacts. They do exist, but changeover contacts are probably the most common of all.
Figure 1.1 - Relay General Principle
While most people won't necessarily come across contactors very often (if at all), they are really just a large relay. The internals are far more robust, and most are designed for 3-phase mains. While single-phase and 2-phase versions also exist, they are less common. The basic internals are shown below. The most notable difference is that most contactors have two contacts in series, and a wider separation. Many relays (particularly miniature types) have a contact separation of between 0.4mm and 0.8mm, where a typical contactor may have a total separation of 5-10mm.
Figure 1.2 - Contactor General Principle
The primary differentiator between relays and contactors is that the contactor is far more rugged, and contacts are generally spring-loaded to ensure very good contact. Most use an AC coil, with a laminated steel core (yoke) and armature. The magnetic pull is designed to be a great deal higher than any relay, and most use two sets of contacts in series. Due to the size and complexity, they are generally far more expensive than most relays. There's a wide variation in styles, but that's usually only the cosmetics - the principles are unchanged. When the coil is powered, the armature pulls in and closes the contacts. To allow for contact wear and erosion, the moving contacts are usually spring-loaded. The fixed contacts are generally rigid, and power connections are generally bolted onto the accessible terminal.
Most control systems use relays for turning equipment on and off. These remain the dominant control switch, as they are low-cost, reliable and are used in the millions. The cousin of the relay is the contactor - it's principle of operation is identical, as described above. The majority (but by no means all) are activated with AC at the nominal mains voltage, although 24V AC is also common as it qualifies as 'SELV' (safety extra-low voltage). Contactors are used for motor control, and are commonly 3-phase, so have three sets of (usually) normally open contacts to switch the power. Some may include auxiliary contacts that may be used to indicate that the contactor is activated or otherwise, or to signal the contactor state to a system controller.
I don't intend to cover contactors further here, as they are in a different league from the relays that most people will use.
Figure 2.1 - Automotive Relay Insides
It's almost universal that people use a diode in parallel with DC relay coils to absorb the potentially damaging back-EMF. It won't damage the relay, but if not suppressed it will usually kill the relay driver transistor or IC. For example, if a relay coil draws 50mA at (say) 12V, when turned off, the magnetically stored charge has to go ... somewhere. If we assume a 1MΩ impedance, the voltage will theoretically rise to 50kV. This never happens in reality, but in excess of 1kV is common. The diode reduces that to a mere 0.7V, but it has an unexpected side effect. Relay activation and release times are usually shown in the datasheet, but the release (drop-out) time is invariably quoted with no protective diode.
Figure 2.2 - Relay Test Circuit
We can use an automotive relay with a 218mA, 12V coil as an example, which has a resistance of 55Ω, operated at 13.5V. Like most relays, the datasheet will say that the drop-out voltage is 10% of the nominal operating voltage (1.2V). At 1.2V, the coil current is only 22mA. With 280mH of coil inductance (not provided in the app. note, but I measured it on a similar relay), it takes 9.6ms for the coil current to fall to 22mA with a diode, and the relay can't even start to release until the current has fallen below that. This does two things. Firstly, it delays the relay drop-out time, so the figure quoted in the datasheet can't be achieved. Secondly (and more importantly), it reduces the armature's speed because there is still some magnetic energy left in the poles.
The answer to this is described in the Relays (Part 1) article, but is repeated here because it's important. If the simple diode is replaced by a diode in series with a 24V zener, the current falls to 22mA in less than 1.9ms. More importantly, the armature can accelerate at close to its maximum, back towards the rest position. This is because the current derived from the back-EMF decays much faster. You can use a higher voltage zener diode to get even faster response, at the expense of a higher back-EMF. The transistor driving the relay must be rated for at least 20% higher voltage than the back-EMF that will be generated. Some examples are shown in the following table, adapted from a TE-Connectivity application note [ 4 ].
| Suppression | Release Time (ms) | Theoretical Back-EMF | Measured Back-EMF |
| Unsuppressed | 1.5 | ∞ | -750 |
| Diode & 24V Zener | 1.9 | -24.8 | -25 |
| 680Ω Resistor | 2.3 | -167 | -120 |
| 470Ω Resistor | 2.8 | -115 | -74 |
| 330Ω Resistor | 3.2 | -81 | -61 |
| 220Ω Resistor | 3.7 | -54 | -41 |
| 100Ω Resistor | 5.5 | -24.6 | -22 |
| 82Ω Resistor | 6.1 | -20.1 | -17 |
| Diode | 9.8 | -0.8 | -0.7 |
A resistor in parallel with the relay coil (with or without the diode) is an old technique that was common in very early systems - before diodes were readily available. I first saw this used with electric clocks, operating from 1.5V. The de-facto standard was to use a resistor with 10 times the resistance of the coil (allowing a 15V back-EMF pulse). Clearly, allowing a higher back-EMF means faster release times, but at the expense of an added zener diode (or a resistor) and a higher voltage requirement for the drive transistor. However, allowing higher back-EMF has distinct advantages, in that the relay releases faster, which reduces arcing at the contact faces. This leads to less contact erosion and longer relay life. It's a trade-off, but in some applications it can be very important. Control systems are often especially vulnerable, because they have a high 'work load', and down time is very expensive.
While the figures shown are from the referenced application note, they are easily measured with an oscilloscope, and can be simulated. The 'unknown' is the relay's inductance, which is almost never published. It's essential for simulations, but it's generally irrelevant. It can't be measured directly, but can be determined by measuring the resonant frequency with a paralleled capacitor. The armature must be closed manually to obtain the inductance value that determines the drop-out time. Measuring the inductance is difficult, because it's a very low Q circuit due to eddy-current losses in the solid core and armature. It's not necessary, but a series circuit (at very low voltage and impedance) gives the best result.
L = 1 / (( 2π × f )² × C )
While the diode by itself is by far the most common approach taken by the DIY (and audio) fraternity), as seen in the table it's far from ideal. Fortunately, it's rarely necessary to ensure the fastest possible release time, but it is useful for DC protection relays. However, the extra few milliseconds doesn't cause any problems - the problems arise from the relay trying to interrupt a high DC voltage at considerable current. This is not a simple task! It can be made marginally easier by ensuring that contacts separate as quickly as possible.
An arc is formed when ionised air particles bridge the gap between the contacts. Once the voltage exceeds the critical potential (which depends on the contact materials and many other factors), the ionised air particles allow conduction, and the air (or other gas) and vapourised contact material turns into plasma (the fourth state of matter). The temperature of the arc can be over 5,000°C, depending on available current. No known contact material can withstand that - even tungsten, which melts at less than 3,500°C.
Snubber circuits are one way to help extinguish an arc, as the initial energy is absorbed by the capacitor, and the stored charge is dissipated by the resistor. This arrangement does not mean that you can exceed the relay's voltage rating, but it does reduce arcing to the point where contact damage is minimised, allowing reasonable (or at least acceptable) contact life. Like so much in electronics, it's a compromise.
Figure 3.1 - Basic Snubber Circuit
The values for resistance and capacitance are not overly critical. The capacitor needs to be large enough to absorb the energy, but not so large that it can allow significant current flow with an AC supply. The resistor needs to be small enough to let the capacitor absorb the initial energy, but not so small that a high current flows when the contacts close. A reasonable starting point is as follows ...
R1 - 0.5 to 1 Ohm per contact volt
C1 - 500nF to 1 µF per contact amp
There are more 'advanced' snubbers, typically including a diode to allow maximum capacitive arc damping, but these are only suitable for DC circuits. AC is less troublesome than DC, because the voltage and current pass through zero every 10ms (50Hz) or 8.33ms (60Hz), although the two may not happen at the same time (due to phase shift with reactive loads). Any arc that forms usually can't last longer than one half-cycle, but if ionised particles are still present the arc may re-strike if the contacts are pushed to their limits. Relay specifications take this into account.
Because the plasma forming an arc is both fluid and conductive, it can be manipulated by, and creates, a magnetic field. If a magnet is positioned where it will interact with the arc, it can be stretched until it extinguishes (at least that's the theory). This technique is used in some industrial contactors, but it requires experimentation to get the magnet position right. While it certainly works, it's not something I'd recommend because most relays are fully sealed. You can't see inside them, so you have no way of knowing if the magnet is in the right place, or is strong enough to extinguish the arc. Magnetic arc 'extinguishing' systems are a fairly specialised field, and while experimentation is always encouraged, don't expect miracles.
Most magnetic systems are a part of a more complex overall solution, that often includes specially fabricated arc chutes or arc splitters. These guide the arc away from the contacts, and divide it into smaller segments that are cooled by the chute until the arc extinguishes (shown in Figure 3.1). The contacts are usually provided with arc 'horns' (aka arc runners) that rely on the fact that an arc will tend to rise. The effect known as a 'Jacob's ladder' also relies on this - the arc moves up a pair of wires due to convection - the air around the arc is (super) heated, so it rises.
None of these are available in common relays, because they are not designed to interrupt fault currents. Relays are used for control, and are not considered to be safety devices. However, arcs will occur every time the relay is activated with any voltage present and interrupting current flow. Often, the arc is not visible, but it's there anyway, even with surprisingly low loads. Every time the contacts arc, a little bit of damage is done to the contact surface, meaning that contact resistance rises as the relay is used.
You will find arc chutes in circuit breakers (CBs), as these are designed to be a safety cutout. Most use a thermal system for prolonged (but minor) over-current (up to 200% of rated current, where the CB should disconnect within 3-20ms), and a magnetic trip circuit to protect the wiring against short circuits. The magnetic section is designed not to operate unless the fault current exceeds a specific value, and most circuit breakers are designed to be able to break a fault current of at least 4.5kA (4,500A). The absolute magnetic trip value is rarely specified in datasheets, but is covered in the relevant standards for the country where the CB will be used.
I tested a 16A thermal-magnetic circuit breaker, and the internal resistance was 23mΩ. At rated current, it will dissipate 5.9W, rising to 9.2W at 20A, and almost 21W at 30A. With 50A, the breaker could be heard to buzz (the magnetic circuit was on the verge of tripping), but with a dissipation of 57W, the thermal cutout operated in less than 1 second. At 100A, it cut out within 10ms, checked over multiple test cycles. Mostly, about 1 half-cycle was enough to trip the magnetic cutout. The passive arc mitigation system used in circuit breakers is complex, and it needs a photo ...
Figure 3.2 - Circuit Breaker Internal Mechanism
Perhaps the most remarkable thing about circuit breakers is the very low cost (the one pictured was under AU$5.00) compared to the number of precision parts involved. The actuator mechanism is quite complex, as it must provide positive contact closure, but can be tripped with very little force. The bi-metallic strip gets hot at high current and bends upwards. If deflected sufficiently, it will touch the trip mechanism, and only very light pressure is needed to release the contacts.
Should normal mains voltage be applied under fault conditions, a large arc will be created. This is stretched by the 'arc horn', and is then split and cooled by the arc chute. The latter is a series of metal plates (9 in the unit shown) that are insulated from each other. The breaker shown has a fault current rating of 6,000A (6kA). That is normally not possible because the mains impedance will usually be somewhere between 0.5Ω and 1Ω (230V mains). This means that the worst-case current will usually be less than 460A. Higher current breakers are used in 120V countries because appliances draw more current for the same power.
It's almost never mentioned, but AC circuit breakers can also be used with DC. I wouldn't exceed 100V or so, but the wide contact separation and arc mitigation elements should be more than capable of breaking a DC arc quite easily. Naturally, if this is something you want to try, you must test it thoroughly before installation. Should testing indicate that the CB cannot break the voltage and current you are using, then you have to use something different.
Figure 3.3 - Circuit Breaker Cutout Curves
The family of curves shown it adapted from an 'Engineering Talks' article [ 5 ], and shows the expected range where the breaker will activate. The 'B-Curve' is not common, and most systems use the 'C-Curve'. 'D-Curve' (delay) breakers are used when high inrush current is expected, and they allow higher peak current without tripping. The CB shown above is a C-Curve type.
The current scale is normalised to unity. For a 16A C-curve breaker, the magnetic cutout will activate at currents between 5.5 × 16A (88A) and 9 × 16A (144A). Based on my tests, 100A provided reliable tripping, although the open-circuit voltage of my test transformer is only 4V. Circuit breakers don't care about the voltage (other than for arc mitigation), and are operated only bt current flow. Below the magnetic cutout current, disconnection is due only to current through the bi-metallic strip. The load should disconnect within 20 seconds at 3 times the rated current (48A), and this was confirmed by testing.
Note: - any experiments you perform are at your own risk entirely. You will be dealing with fairly high DC voltages, and looking at a sustained arc can cause irreparable eye damage due to the intense ultraviolet light emitted. There is also a risk of serious burns and fire. No experiments should be carried out if you have little experience with high voltage, high current and arcs in general. This is a fairly specialised field, and extreme care is required. If you wish to run this kind of test, I suggest Project 207 - High Current AC Source. This allows very high current at a safe low voltage (around 4V RMS open circuit).
Contact arcing is such a problem for industrial systems that countless patents have been lodged for new, no-so-new, exciting and mundane ways to reduce contact damage. Some should never have been granted because they are 'common knowledge', while others are quite innovative. If you want to know about the various systems that have been devised, a patent search will provide many results.
The passive designs you'll come across are not intended to allow the use of any relay contacts beyond their rated voltage and current limits. When contacts arc, damage is done to the contacts, and the idea is to minimise this damage, not to let you operate the relay beyond its rated current limits. However, active arc prevention (and/ or hybrid relays) do allow you to operate a standard relay at a higher DC voltage than recommended. Active circuit failure has to be considered, because any semiconductor can fail for any number of reasons.
Figure 3.4 - Arc Voltage; 60V DC Supply, 8Ω Load
The arc in the screen capture lasts for a little over 350ms, and this test was done with a relay having 0.8mm contact separation. No suppression was used, but the armature's movement was damped by the external supply used. With a smaller gap, this arc would be sustained and would have a lower impedance, thus hastening the demise of the contacts. How it's dealt with depends on the application, and in many cases the recommended solution is to use two sets of contacts in series. This increases the overall separation distance, and also provides more contact area to help cool the arc, which will cause it to extinguish. Operating contacts at above their rated DC voltage is never recommended, which is why there are so many products made that are designed to quench (or prevent) arcs from forming in the first place. The same setup as described was tested with two sets of contacts in series, and while there was an arc, it was small and extinguished quickly.
As noted above, a snubber network can be used in parallel with the contacts. This will not allow operation above the maximum rated voltage or current, but if properly designed for the application, a snubber will reduce arcing. This can help to reduce contact erosion, but it's not as effective as active techniques. However, it's cheap to implement and can extend the life of a relay. Snubbers cannot be relied upon to allow operation at voltages and currents above the rated maxima.
Passive systems can only suppress an arc, but cannot prevent one from forming. Where it's important to ensure that there is no arc at all, an active system is required. These can eliminate the arc completely, by ensuring that the EMR contacts only carry the active current, with the current interruption function handled by semiconductor switches.
Active arc suppression involves semiconductors and other support components. Unlike the suppression system shown for a circuit breaker, there are definite limits to the current that can be interrupted, and they are not intended for use as a safety cutout. Should a major fault develop that trips the CB, there's a good chance that the controller (which can use active systems) may be damaged. Active systems are intended for use where high loading is expected, and/ or rapid cycling which will lead to early contact failure.
One thing that an active systems allows (and this includes hybrid relays), is that the full contact current can be used with either AC or DC! Normally, DC operation is limited to around 30V for most relays, but if the contacts only have to carry current and never break an arc, then the only limitations are those imposed by the relay's insulation and the contact gap. Even 0.4mm will withstand 500V or more under static (no current) conditions, so if an arc can never eventuate, then the relay's only limitation is contact resistance and insulation ratings. Dry air will not allow an arc at a voltage less than ~30kV/ centimetre (3kV/ mm), so even 0.4mm separation can (theoretically) withstand 1.2kV before a spontaneous arc will develop.
The following circuit has been simulated and workbench tested, and it does exactly what's claimed for it in the patent. Although I came across the patent drawings more-or-less by accident, I was partway there with some other experiments I was playing with. It may appear simple, but the component values require optimisation for best performance. Like many other arc interrupter/ suppression techniques (which includes capacitive snubber networks), the circuit does allow some leakage across the contacts when they are open. This can be hazardous if used in an industrial system, and it would breach regulations if used on an emergency stop system, or a safety isolator.
Figure 4.2 - MOSFET Arc Extinguisher (DC Version)
The above circuit is shown primarily to demonstrate the circuitry necessary to ensure that an arc is quenched (or in this case, not started at all). The circuit is based on a patent taken out by International Rectifier (one of the pioneers of MOSFETs). The patent (US7145758) is still current, so I have only included indicative component values, being those I used for my test. A more recent patent uses additional parts to switch off the MOSFETs much faster than the simplified version shown. The MOSFET will conduct for around 300-500µs (depending on component values used), while the 'enhanced' version turns off in 100µs or less. In this (and the next) circuit, the arc is not 'mitigated', it is prevented from happening at all. The MOSFET will conduct when there's around 12V or so across the contacts, so an arc doesn't get a chance to form.
Workshop tests show that it works extremely well. There is a small arc when the contacts close, which is caused by contact bounce. When the contacts are opened, there wasn't even a hint of an arc, even with a test voltage of 80V DC and a nominal 4Ω load. While that suggests 20A DC, in reality it's less because the power supply isn't regulated. It's still a very severe test, and was done with the same relay used to produce Figure 3.3. With the higher voltage and current, the relay would sustain a continuous arc without the MOSFET circuit - I know this because I tested it (and a mighty arc it was, too!). This is a case where reality and simulation were in 100% agreement. Even after a number of switching 'events' in fairly rapid succession, the MOSFET I used didn't even get warm. Instantaneous power would be about 400W, but the duration is very short (less than 1ms).
It's an elegant solution, and the added cost and complexity is such that it will pay for itself fairly quickly, thanks to reduced 'down-time' of critical equipment. Circuits such as that shown are used where contacts are constantly opening and closing under load, so arc suppression means far longer life for the relay/ contactor. This class of circuit is intended for industrial applications, where contact operation is in the hundreds (or thousands) of cycles a day, and failures are very costly. There are quite a few companies whose livelihoods depend on arc suppression technology, either as users or sellers.
DC is by far the worst for contact arcing. Most miniature relays only have a contact separation of around 0.4mm, and a continuous (and destructive) arc can be created remarkably easily. The DC rating for most relays is 30V DC at rated current, but that's the figure provided to obtain rated life. Most will be able to sustain an arc with a voltage of around 40V DC at rated current. By using a circuit such as that shown, there is no arc as the relay opens - none at all!
Figure 4.3 - MOSFET Arc Extinguisher (AC Version)
In the AC version, two identical but inverted circuits are used in parallel with the contacts, because the polarity is unknown. One or the other circuit will conduct, depending on the polarity at the time the contacts open. The total active device dissipation can be very high with either circuit, depending on the voltage and current. However, it has a brief duration (generally less than one millisecond), so unless operated with a short duty-cycle (with many switching events per minute), the average dissipation is low enough that it won't cause problems. However, if you happen to be switching 10A at 230V, the peak dissipation may exceed 2kW, and the MOSFET(s) used must to be able to handle that.
A MOSFET such as the IRFP450 is rated for 500V, 56A pulsed drain current, and a dissipation of 190W (at 25°C). The safe operating area graph indicates that 300V at 10A (3kW) is permissible, provided the duration is less than 200µs. This is not a recommendation, but is an example of a device that may be suitable.
There is a place in audio for the AC version shown - loudspeaker DC protection. A normal relay cannot break the DC output from a failed amplifier if the voltage is much more than ±30V. The Figure 4.3 circuit will break almost any DC voltage of either polarity reliably, something that's simply not possible with any standard relay. Normally, the relay should always be connected so it shorts the speaker (not the amplifier!) when the relay opens due to a fault. In this role, the relay is considered 'sacrificial' - it will almost certainly be destroyed (but your speakers are saved). Don't use any circuit that doesn't short the speaker, as it won't save anything from destruction with more than 30V.
These are the ultimate for the prevention of arcs, including those created by contact bounce. They are covered in detail in the article Hybrid Relays using MOSFETs, TRIACs and SCRs, so will only be discussed briefly here. Because the 'solid state' switch is activated first, the electromechanical relay's contacts never have to switch much current, and the EMR reduces power dissipation to the lowest level possible. When released, the solid state switch remains on until the EMR has released, so there can be no arc.
However, this comes with some complexity, including the requirement for an isolated driver for the electronic switching. This is easy enough with TRIACs and SCRs, but is more difficult with MOSFETs. However, there are solutions for this, and example circuits are shown in the article. An electronic timer is also needed, which can be a simple comparator, a 555 timer, or it can all be controlled by a microprocessor. There are tangible benefits, especially with high current (particularly DC at more then 30V), if the switching cycle is short, or if very precise timing is required.
Magnets can be used with DC relays to achieve rapid and effective arc quenching. An arc (which is plasma) is electrically conductive, and therefore a magnetic field can be used to force it out of the contact gap. The polarity is important, so you can't swap the positive and negative terminals of a magnetically quenched DC relay. The idea is to use the magnetic field to pull the arc away from the contacts, stretching it until it extinguishes. Arc 'horns' may or may not be provided as well, as these help to stretch the arc and move it away from the contacts.
There are commercially available magnetic arc-quench relays available, and it's it's also something you can do yourself. However, you need a relay with a see-through case so that you can observe the arc and ensure it quenches properly. While at least one amplifier manufacturer I'm aware of has used this trick, it's not very effective for speaker protection. This is because the fault may be a positive or negative voltage, and magnetic quenching is polarity-sensitive. This can be solved by using an electro-magnet, but that adds (most likely considerable) cost to the final product.
With the increasing use of high voltages in electric vehicles, the need for safe, reliable arc-quenching will only grow. 'Solid state' switching (e.g. MOSFETs, IGBTs) has an inherent flaw, in that the failure mechanism for most semiconductors is to fail shorted. If used as an emergency cutoff subjected to an overload that kills the semiconductor, there's no way to disconnect the power unless a high-voltage DC relay is incorporated. Given the propensity for DC arcs to be self-maintaining, this is not an easy fix. Most high-voltage, high-current DC relays will use an arc chute and arc 'breaking' plates (see Fig. 3.2). The chute deflects the arc, and the arc breaking grid breaks the arc into small segments and removes the heat needed for an arc to be maintained.
By adding a magnet, the arc propagation speed can be increased, drawing it into the grid faster. Even where an arc chute and grid are not used, the magnetic interaction will pull the arc away from the contacts until it can no longer be sustained.
Figure 6.1 - MOSFET Arc Extinguisher (AC Version)
I've not shown polarities, nor have I attempted to optimise the configuration. It's intended only to show the general principle. The magnetic field of the arc is circular, according to Fleming's Left-Hand-Rule. Feel free to look this up, as it's somewhat outside the general scope of this article. Since magnetic arc-quench circuits are polarity sensitive, they are not suitable for AC. Unlike a MOSFET (or other semiconductor), a well designed magnetic arc-quench relay may be suitable for emergency cutoff systems.
Where the ultimate reliability with very high voltages and/or currents is needed, specialty relays are available that use a vacuum or suitable gas to quench the arc. Sulfur hexafluoride is one that's widely used (see Sulfur hexafluoride circuit breaker on Wikipedia.
It's now possible to arrange a switching system for almost any imaginable load, over a wide range of voltages and currents. Switching DC need not be the problem it's always been, but there is an inevitable increase in complexity. Semiconductors used in conjunction with EMRs provide capabilities that were not possible in the early days of switching systems, but careful design is essential to ensure that the electronic parts run as cool as possible. This often means adding a heatsink.
While heatsinks are a nuisance and add cost and bulk to the end product, operating any electronic parts at high temperatures reduces their allowable dissipation, and failures become more likely. Whenever a hybrid solution is used, it's essential that there's a safety cutout in the system, so that a faulty semiconductor doesn't wreak havoc on a machine or an entire production line (and yes, that can happen easily if you miss something that causes a switching system to become a short circuit).
No system can ever be ideal in all respects, and the art of design is to work through the compromises needed (and compromises are always needed in any design) to arrive at an end result that does what's needed. Amateurs who have a good understanding of the risk/ reward equation will usually over-engineer the solution, since they may only be building a small number of switches. The industrial designer is forced to push everything to its limits to keep costs down. There's not much point having the 'best' system available if it's so expensive that no-one will buy it.
This article is intended to show principles, and is not a construction or design guide. However, it should help if you find yourself with a seemingly intractable problem where arc mitigation or prevention is required. It's unlikely that many DIY builders will need more than 'moderate' power - up to perhaps 1kW or so, and the parts needed are not especially expensive. For those who simply want to experiment with ideas, this should give you a head-start.
Main Index
Articles Index