|Elliott Sound Products||Project 208|
Loudspeaker DC protection is always something of a mixed bag. Units such as Project 33 are well behaved and will offer a high level of protection for the speaker. Should the amplifier fail, the most common failure mode is for an output device to short-circuit, causing the output to swing to one supply rail or the other. This is going to cause damage to the speaker, and a voicecoil subjected to (say) 70V DC won't survive unless the excursion is very brief.
The task is harder with high-frequency drivers, because they have much smaller voicecoils with very little thermal mass, so damage can be almost instantaneous. However, in a system with a passive crossover, no DC can get to the HF driver(s) because there's a capacitor in series. For low-frequency drivers, we may set an arbitrary limit of perhaps 50ms, which allows full power at 20Hz, but will disconnect the speaker if the signal remains at full voltage for any longer. Unfortunately, it's not quite so straightforward, and there are many other factors that need to be addressed.
BTL (bridge tied load) amplifiers pose additional problems, as it's theoretically possible for one amplifier to 'go DC' while the other keeps working. While the amp will not last for very long (the 'working' amplifier will fail sooner rather than later), it may survive for long enough to destroy the loudspeaker. Project 175 (Single Supply BTL Amplifier Speaker Protection) is the solution for this, but it's designed to be installed inside the amp chassis - it would be difficult to make it function as an external unit because it requires the ground and DC supply rail for its references. The designs shown here might protect the loudspeaker when used with such an amplifier, but it's far from guaranteed.
With well designed circuits and internal DC fault protection built into most high-quality amplifiers, there are countless examples of poor design that almost guarantees failure at some point in the amplifier's life. Underestimating the peak dissipation of output devices is uncommon in most commercial offerings, but there are still plenty of examples of amps that have not been thought through. They may not be on the market for very long, and some will fail. Once repaired, the owner may well decide that it's not worth keeping, especially if it managed to destroy several hundred dollars worth of speakers when it died. The new owner will be unaware of this, and the process could easily be repeated.
There have been a number of patents granted over the years for an 'amplifier powered' protection system, but some are seriously flawed [ 1 ] while others are best considered a concept at best. There are many dependencies, and there is no 'one size fits all' solution. The circuits described here are probably about as simple as it's possible to make them, consistent with being able to do the job required. However, this does not mean that it will protect any driver from any amplifier, as that requires a power supply that provides a known fixed voltage that powers the circuit.
If we attempt a circuit that doesn't rely on anything other than the output of the power amplifier, things become complex. Without its own power supply, the circuit must rely on whatever is delivered by the amplifier. As long as it is an AC signal (whether a sinewave or music), the protection must not react in any way. Should the amplifier fail and output DC (a not-too-common but very destructive failure mechanism), the speaker has to be disconnected. This needs to happen as quickly as possible, but the circuit has to ignore everything that is not a fault. This is surprisingly difficult, especially if very low frequency signals are allowed to get through the amplifier. For this reason, a high-pass filter should always be used, limiting the amplitude of anything below the lowest frequency of interest.
The system needs to be designed to handle the maximum likely input voltage (AC and DC), but must still work if the protected speaker is used with a smaller amplifier. Because the voltage is lower, it will take longer to react, but the two should balance out reasonably well to prevent voicecoil failure. Ultimately, it's a balancing act - detection speed vs. allowable voltage and low frequency limit. While it's easy to say that all high power PA/ sound reinforcement systems should use a high-pass filter, many don't and it's up to the 'sound guy' to keep everything within permissible limits.
A common failing with many of the protection circuits published is that the relay is wired incorrectly. With DC voltages above 30V, it's inevitable that the contacts will arc when opened, and if the arc persists (which it will with a 70V DC supply), then there's no protection at all. This is described in some detail in the Project 33 article, and that also shows the correct wiring for the relay. It's imperative that the relay contacts short the speaker when activated, as this allows for an almost complete 'meltdown' of the relay (due to arcing) while still protecting the load. Care is needed to ensure that the contacts don't short-circuit the amplifier, as that will only cause more damage. Yes, the amp has already failed, but there's no reason to cause even more damage if it can be avoided.
As noted above, one thing that should be included in every sound reinforcement system (but is usually missing) is a high-pass filter. There are very few systems that can handle frequencies below 30Hz or so, and a steep filter that removes everything below 25Hz is a worthwhile investment. An example is Project 99, a 36dB/ octave filter that's designed specifically to remove 'subsonic' signals. These stress drivers, and (if present) use valuable power reserves in amplifiers and cause unwanted cone excursions that don't contribute anything to the overall sound. Using such a filter is somewhere between 'highly recommended' and 'mandatory' if either of the circuits described here is used.
The circuits described are not guaranteed to protect loudspeaker drivers from a failed amplifier under all possible fault conditions. While every care has been taken to ensure that the circuits themselves perform as described, there may be some circumstances that cause false triggering (excessive very low frequency power for example). The system should always include high pass filters to ensure that frequencies below 20-30Hz are rapidly attenuated.
The relay remains the 'weakest link', and with very high powered amplifiers it may be unable to completely prevent DC from damaging the speaker(s). This is particularly true if the relay fails internally due to an arc. Arc suppression is especially difficult when the supply voltage is greater than 30V DC (the typical maximum quoted for most common relays). Consider using two sets of contacts in series for high supply voltages (anything above ±35V). Do not use a 'standard' miniature relay, but aim for one with heavy-duty contacts and a generous contact clearance (0.8mm is the suggested minimum).
It's unrealistic to expect any protection system to protect the connected drivers from damage under all foreseeable (or unforeseeable) situations. The range of amplifiers is vast, some include protection systems internally, but many do not. The range of power supply voltages is also vast, ranging from basic BTL (bridge tied load) amps with ±35V supplies (roughly 200W into 8Ω), up to Class-D (switchmode) amplifiers with supply voltages up to ±100V. If (when) a BTL amplifier fails, the most common failure will be one amplifier only, and while it's theoretically possible for the other amp to continue to work 'normally', that's unlikely in the long term (more than a few seconds or so).
Neither of the circuits shown can handle a situation where a high level AC signal voltage and a DC offset are present simultaneously. The Figure 1 circuit will activate if the AC component of the combined signal is at a frequency well above the filter's natural rolloff and/ or is at a comparatively low amplitude. For example, a test with a 30Hz, 25V peak AC signal together with a 35V DC offset shows that it will work, but it's not something I'd rely on! The simplified version shown in Figure 2 will not activate under the same conditions.
A DC detector that's built into the power amplifier will (or should) always perform reliably, regardless of the applied signal, because each channel (of a BTL amplifier) can be monitored independently. This isn't possible with an external circuit that has no fixed ground reference, and must rely on the signal from the amplifier to be able to work.
Both of the designs shown here are designed to operate with a minimum frequency of 20Hz. Operation of any high-power sound reinforcement or hi-fi amplifier is not required below that, and if there is significant energy at very low frequencies (<20Hz) the circuits may false trigger. This will create very nasty transients, with more than sufficient amplitude to damage tweeters or compression drivers. Ideally, neither circuit should be used in any system that uses passive crossovers.
There's not a lot to this circuit, but its operation is more complex than it appears at first. Even a small mistake with relay wiring could be fatal for the amplifier, so it must be tested thoroughly before use. There are basically three separate sections, the DC detection circuit, power supply (derived from the fault voltage), and the trip circuit which drives a relay. The relay disconnects the speaker from the amplifier, and shorts the speaker terminals. To detect DC but ignore the audio signal requires a filter, and this is responsible for most of the delay between a fault appearing and disconnection of the load. If it were instantaneous, the relay would constantly open and close the contacts in sympathy with the applied audio.
The power supply is comparatively easy, but with a very high power amplifier, the voltage will be much higher than is desirable. It could be regulated to something more sensible, but that would involve even more parts. For example, if the amplifier is capable of 50V RMS output (310W into 8Ω, 620W into 4Ω), the detector's power supply will have a peak DC voltage of up to 70V if the amp is pushed into clipping, and the same if the output stage fails - the full supply voltage is normally presented to the loudspeaker. Of course, the voltage under drive or fault conditions may be positive or negative. The bridge rectifier ensures the correct polarity, regardless of the amp's output voltage.
The current required when the circuit is in 'standby' is very low (i.e. with signal at varying levels but no amplifier fault). As shown, even with a 70V peak signal (50V RMS), the current is less than 10mA. Providing that small current from an amplifier with a typical low output impedance (generally no more than 0.1Ω) will not cause audible distortion in any sensibly wired system. The power supply voltage and current will vary of course, and with no signal it will be zero.
Note that the power supply bridge rectifier must use high-speed diodes. In a full range system, the frequency will be up to around 15kHz with most material (some may extend to 20kHz), and 'ordinary' mains diodes will fail, because they can't turn off quickly enough. This causes significant reverse current that will cause the diodes to run hot (or very hot), and they will not survive. UF4004 diodes (the 'UF' means ultra-fast) will be quite sufficient in this role, as will any similar device. High current is not required, so heavy-duty fast diodes are not necessary.
Figure 1 - Amplifier Powered Speaker Protection Circuit (#1)
The component values shown are designed for a power amplifier having supply rails between ±35 and ±100V. For lower or higher supply rails, a few changes may be needed. The circuit has been simulated with the equivalent of a 1,200W power amplifier (I don't have one to test it with), and as low as 100W (both are 4Ω ratings). At low supply voltages it takes longer to activate if there's a DC fault, but of course the speaker power dissipation is also greatly reduced, and the two tend to balance out.
Operation is straightforward, but not necessarily intuitive. One section that's very easy to work out is the power supply - it's simply a bridge rectifier followed by a capacitor, which appears to be far too small to be useful. However, the circuit's purpose is to detect DC, and to ignore the normal (audio) output from the amplifier. The DC supply will become 'solid' if there's an amplifier fault that causes the amp's output to become DC (by far the most common failure mode that causes speaker failure). With a 70V DC fault, the relay is activated in about 33ms. Lower voltages cause a corresponding increase in relay activation time (about 65ms at 35V).
The DC detector uses an optocoupler (4N28 or similar), which follows a filter that removes the AC component. The optocoupler's output will be activated only if there is DC (or an unrealistically low frequency) present at the input. There are countless other optocouplers that will work equally well, and I used an LTV817 for testing. The fault output may be positive or negative, so a low voltage bridge rectifier is used to ensure that the optocoupler will work with a fault voltage of either polarity. The output provides gate current to the MOSFET (Q1), which turns on the protection relay. If it's found that you get false triggering at low frequencies, you can reduce the value of R6 (nominally 100k). Don't reduce it too far, and make sure that you test the circuit with a DC input to ensure that it works reliably!
The MOSFET specified (IRF630) is overkill, but it's rated for 200V, and they are under AU$2.00 each. You can use any number of others - it's not at all critical. However, you must ensure that the one used is not designed for logic, as the gate threshold voltage is too low and it may operate (intermittently) with normal signal. A heatsink is required for the MOSFET, especially with systems expected to be driven by high-powered amplifiers. The heatsink should not be less than 10°C/ W, which will cause the MOSFET to run at around 25°C above ambient (the temperature inside the enclosure!). Mechanical support is needed, because the system will be subjected to intense vibration in many installations.
The MOSFET uses a current limiter (Q2), which is designed to provide roughly 120% of the normal operating voltage to the relay, to give it the best possible chance to activate, even if the NC (normally closed) relay contacts have slight welding. D12 should be 1N4004 or similar. The value of R7 is determined from the following ...
IRelay = VRelay / RRelay × 1.2
R7 = 0.6 / IRelay
For example, if a 24V relay has a 576Ω coil, the current is nominally 42mA (50mA at 120%), so R7 will be 12Ω. The relay must be considered sacrificial - if an amp fails, the relay may be destroyed if the contacts arc. A bit of additional coil current is not likely to be an issue in practice. This is one relay specification where we can take liberties - we don't want the coil to burn out, but if it overheats we don't care much because it should be replaced after an amp failure anyway. The same comments apply to the Figure 2 circuit shown below.
The fuse shown is optional but recommended. Without it, the arc drawn across the relay contacts may result in a complete relay meltdown, but the fuse itself is fairly critical. I suggest an HRC fuse (high rupturing capacity), and it needs to support the current drawn during normal audio at the maximum suggested power for the system. For very high power systems (> 1kW) that means at least a 15A fuse. For example, a 620W amp (4Ω) will deliver around 7A RMS into the speaker(s), allowing for a rather minimal 5dB peak to average ratio (highly compressed material, with the amp just clipping transients). This may sound unrealistic, but it's not. A 1kW amp will provide around 14A under the same conditions. In the interests of maximum reliability, you'd probably use a 20A fuse to prevent 'nuisance' fuse failures. It's worth reading the Fusing - How to Apply Circuit Protective Devices to see the characteristics of fuses - like so many other areas of electronics, they are not as simple as they seem.
The following circuit is a simplified version, but it will work almost as well. As long as there's audio, Q1 turns on with each half-cycle and keeps the voltage across C3 below the threshold voltage for the MOSFET. This happens even at the lowest frequency of interest (20Hz) and at any amplitude up to 70V RMS. Like the Figure 1 circuit, it uses a power supply that cannot maintain a steady voltage, but if the amplifier 'goes DC' due to a fault, it will have a solid power supply to activate the relay. The same current source circuit is used to drive the relay, but DC detection is not as fast as the Figure 1 design. Should the amplifier fail with DC output of 70V, it takes 60ms before the relay is activated, and this is extended to about 95ms with a 35V fault voltage. The time can be reduced by reducing the value of C3, but you'll have to verify that it doesn't trigger with 'normal' programme material.
Figure 2 - Amplifier Powered Speaker Protection Circuit (#2)
The AC detector ensures that transient voltages (or sustained high power) cannot turn on the MOSFET. As long as AC is present, C3 remains discharged as Q1 turns on twice for each complete input cycle, and keeps the voltage across C3 below the MOSFET's turn-on voltage. This is particularly important if the amp is driven to heavy clipping, as that could maintain more than enough supply voltage to turn on the relay without the discharge circuit. Once the AC signal is replaced by DC (an amp fault) there is no drive signal to Q1, so C3 charges until the MOSFET turns on, thus operating the relay. The same comments apply to the MOSFET used as for the one described for the first circuit. The threshold voltage is more critical in this version though, so be careful with substitutes.
Despite the simplifications, this circuit will activate the relay within 100ms after the AC (signal) is replaced by DC (fault), assuming a supply voltage of 70V. The relay activation time adds perhaps 10ms, but this depends on the relay. Most are fairly fast, and the small extra time delay is not usually a problem. A limitation is that the Figure 2 circuit cannot separate AC from DC effectively, so if the amplifier develops a fault where there is significant DC but the AC signal is still present, the relay will not activate. While such faults are very rare, it remains a possibility (albeit a remote one) with some designs. I doubt that it's a major concern, but if you want the most reliable DC detection then use the Figure 1 circuit.
The relay wiring is identical to that used in the Figure 1 circuit, and relay requirements are shown in the next section.
Note the way the relay is wired in the drawings, and make sure that you see Figure 6 below! The speaker is normally powered via the 'NC' (normally closed) contacts, and when the relay operates, the speaker is shorted (but not the amplifier's output!). This helps to prevent arc current from passing to the loudspeaker. Never simply use a normally open contact alone to 'protect' a speaker, because it usually won't. The relay(s) required are known as '1 Form C' - aka SPDT (single-pole, double-throw) or changeover (normally open [NO] and normally closed [NC] contacts). The alternative is 2 Form C - DPDT (double-pole, double-throw). These are dual changeover types so the contacts can be wired in series.
I strongly suggest that anyone contemplating building the designs shown here reads the Relays (Parts 1 and 2) articles, to gain a full understanding of the strengths and weaknesses of electromagnetic relays (EMRs). MOSFET relays cannot be used because they are normally off, and an internal battery would be needed for normal operation. I doubt that anyone would consider that to be a good idea. An EMR is the only sensible choice, and how it's wired (and the use of capacitors) determine if it will protect your loudspeaker.
One additional limitation that you'll come across is that many relays have a lower current rating for their NC contacts than for the NO contacts. This is largely due to the fact that more contact pressure is available when NO contacts are closed by the coil. All relays use a spring to restore the armature after operation, and that spring must be weaker than the available magnetic force or the relay won't activate at all. As the armature gap closes, more electromagnetic force becomes available, allowing higher contact pressure for the NO contacts. In the application described here, there is no choice - only the NC contacts can be used, because there's no DC supply available until there's an amplifier fault.
The relay is critical, and the vast majority of those available are rated for only 30V DC. While the current rating is also a limitation, it's not quite so serious. The current rating is (usually) the average, and it can be exceeded by higher peaks in normal use without too much concern. However, no commonly available relays are capable of breaking 70V DC (or more) at a current of around 20A. As the contacts open, an arc is drawn that will maintain current flow, and it will also cause a great deal of heat that can (and does!) melt the internal contact structure. Some automotive relays claim to be able to break 75V, but they have a very high coil current (typically around 250mA, 12V coil). This makes the relay switching MOSFET dissipate a great deal more power, thus requiring a heatsink. I'm rather wary of such claims, especially where high current is involved, but is is an option worth investigating.
With very high-powered amplifiers, there is a risk that the contacts may weld themselves closed if their average rating is exceeded. The 10A relay I used for contact resistance tests was subjected to 50A for a few seconds, and the contacts did just that - I had to apply 24V to the 12V coil before the relay had enough 'grunt' to separate the contacts. I know this is a pretty severe overload, but everything is important if we are providing the last line of defence (and that's exactly what this is).
The capacitor across the relay contacts is intended to suppress the arc, but it's very much a compromise. The cheapest is a bipolar electrolytic, as used in budget crossover networks. These are fairly cheap, and during normal operation they don't pass any current. The cap will absorb the initial voltage across the contacts, but to be effective at high voltages this may not be sufficient. A larger capacitor can be used, but it may be unrealistic to expect complete arc quenching.
I have tested the capacitive arc 'quencher' circuit, and was able to suppress the arc with 70V DC across an 8Ω load completely, with only 2µF (10µF is still recommended). However, the relay still has to be considered 'sacrificial' - if the amplifier fails, so too may the relay. However, it's far cheaper than the loudspeakers. The very high peak current may also kill the capacitor, so the whole system needs to be checked thoroughly after it has operated due to an amp failure.
Another relay type that may be worth trying is automotive relays. These are readily available and generally inexpensive, and are designed for very high current. 24V versions have coil resistances from 250Ω to 330Ω (96mA to 72mA respectively). This will place a far heavier load on the switching MOSFET, and it will need a more substantial heatsink. With a 70V DC fault voltage and almost 150mA for a pair of relays, MOSFET dissipation will be nearly 7W. The heatsink thermal resistance needs to be no greater than 5°C/ watt or the MOSFET will overheat and (probably) die. Note that the value of R7 will need to be reduced to allow the MOSFET to supply the required current (around 6.2Ω for 96mA).
Figure 3 - Automotive Relay Internals
The relay shown above was sold as a 40A type, but that's highly optimistic. It does have a wider than 'normal' contact gap, measured at around 0.85mm. At 10A, the contact resistance for the NC contacts measured 4.7mΩ and the NO contacts measured 3.8mΩ (when closed, naturally ). These are fairly cost-effective (typically around AU$4.00 each including sockets), but the coil does draw much more current than most other relays. Normally, I wouldn't suggest automotive relays at all, because their insulation between coil and contacts isn't good enough, but in this application it doesn't matter. However, the relay shown cannot break 70V DC at 20A or more - it will be destroyed! So will your speaker(s) which will still get high current DC via the arc. A parallel capacitor was tried with this relay and it seems to break the arc reliably at 60V DC and up to 10A.
Another general style or relay that will (hopefully) survive is one of the TE Connectivity 30A relays shown in the T9A Series Datasheet, but if it ever has to activate with a high-voltage supply, it will almost certainly be destroyed. However, it's a great deal cheaper to replace the relay than the loudspeaker driver(s). Unfortunately, this style of relay is not available with more than one contact set. The Omron LY2-0-DC24 relay is DPDT, and rated for 10A. Without capacitors it will not survive breaking 70V DC, even with the contacts in series, but with them installed it should be possible to break the arc. Predictably, it's neither possible nor practical for me to try to test every relay available.
Note that the '1 Form C' relay is rated for 20A (NO contacts) but only 10A for the NC contacts. This arrangement will handle up to 400W average power, but the peak current may be well in excess of the rated capacity. Peak current with 70V supplies into 4Ω will be around 17A, and you may need to select a heavier duty relay. The T9A series is a suggestion, but you have to be prepared to run your own tests. The selection of relays that can handle more than 20A is quite limited.
Lest you think that I'm exaggerating and that it can't possibly be as bad as I claim, cast your eyes upon the following photo. What you see is all that remains of the upper contact set after a sustained arc. The relay shown is a heavy-duty type, and internally it's almost identical to one that I used for some testing (but not to destruction). This class of relay typically has 0.7mm contact clearance, where common 'miniature' types only have 0.4mm contact clearance. Despite the increased contact spacing, the arc completely destroyed the contact set.
Figure 4 - Relay Destroyed By Arc
To obtain a higher voltage rating, you can use two relays with the normally closed contacts wired in series. This arrangement reduces the voltage across each set of contacts and thus might be sufficient to prevent arcing. In the datasheet linked above, the initial contact resistance is quoted as 75mΩ, but this is rather pessimistic and would mean a contact dissipation of 67W at 30A (which is clearly not possible). I tested a 10A relay with 50Hz at 10A RMS, and measured 60mV (6mΩ), and even at 20A I was only able to measure 132mV across the contacts (6.6mΩ), including the internal connections. At 20A, this represents a loss of 2.64W - almost negligible compared to lead losses, but it's a lot of heat in the small area of a pair of contacts. At rated current (10A) dissipation was only 600mW. Relays in an audio circuit never have to deal with maximum current continuously, so a 30A relay is only needed to keep contact dissipation low, and as an attempt to break the arc. The higher rated current helps to protect against the contacts welding themselves together in normal use.
Figure 5 - Arc Voltage, 60V DC, 8Ω Load
The above is a direct capture, measured across the relay terminals. The power supply was set for 60V DC, and no suppression caps were used across the contacts. The period of the arc starts as soon as the contacts open, and continues to supply 30V DC to the load until it eventually extinguishes. This test was done with a heavy-duty industrial relay, and indicates that the arc impedance is low enough to supply considerable current to the load - in this case, around 3.7A. The arc is noisy, both electrically and acoustically. The arc sounds like white noise, and the frequency spectrum extends well into the radio frequency (RF) band.
Despite the shortcomings, there is some comfort to be had in the electromechanical approach. Relays are used in their billions, in all manner of applications from consumer products to heavy duty industrial systems. They remain popular because they are so reliable, and are far cheaper than electronic 'equivalents'. Their one failing is the inability to break high DC voltages reliably, which is (unfortunately) the very task asked of them for speaker protection. In general (and provided the relay is wired as described), relay protection is reliable and effective, and has always been the most cost effective approach. Using parallel capacitors is a 'brute-force' arc suppression technique that can work surprisingly well.
Figure 6 - Series Contact Relay Wiring Details
The relay wiring shown above uses two relays, with the coils either in series or parallel and the contacts in series. The T9A Series relays have 144Ω coils for 12V, or 576Ω for the 24V version. Relay resistance is therefore 288Ω for two 12V coils in series (83mA), and two 24V coils in parallel gives the same total resistance and current. The 12Ω resistor (R7) shown in both circuits above has to be reduced to 6.8Ω if you use two relays. While the relay coil is driven to well above the maker's recommendations, the over-voltage condition probably won't last long as the loss of audio from the enclosure will alert the user/ operator that there's a fault.
Both sets of contacts interrupt the DC fault current, the second relay shorts the load, and the capacitors help quench the arc by absorbing the initial energy as the contacts open. High values of capacitance are more effective, but there's a cost (and size) penalty. You can add resistors in series with the caps to prevent a high discharge current from welding the contacts, but that reduces the effectiveness of the arc quenching action of the caps. For a circuit that may never function, it's not realistic to have a large and expensive system that will never activate unless an amplifier fails. It's very important to ensure that the circuit remains functional, despite perhaps years of inaction. Relay failure (due to a sustained arc) is a far cheaper option than replacing expensive high-power speaker drivers!
I've tested a relay with 0.8mm contact separation with 70V DC into a 4Ω load, and without capacitors it will arc every time (as shown above). As little as 2µF is enough to prevent the arc from forming at all, so the 10µF suggested should be more than sufficient. Note that the cap(s) need to be as close to the relay as possible, because any additional resistance or inductance reduces their effectiveness. Despite this, if the expected fault voltage is greater than 50V or so, I strongly recommend using two relays, with two sets of contacts in series.
Everything must be constructed to a very high standard, with no possibility of failure even when subjected to heavy vibration inside the enclosure. These criteria are not trivial. Should a fault develop in the circuitry, you will be unaware that there's anything wrong until an amplifier fails and sets the speakers on fire. As noted earlier, the dedicated system operator will periodically apply around 30V DC to the enclosure input terminals to verify that the relay(s) operate normally under fault conditions. The speaker will make a fairly loud noise as the DC is applied then disconnected by the protection relay. Note that the power supply has to be able to deliver at least ½ amp (assuming that the 'Test' switch is included), thus ensuring that the speaker will not be damaged, but providing enough current to ensure reliable operation. This test simulates an amplifier failure, and is not without some risk!
I recommend that you include the 'Test' switch, as that minimises the current needed (as shown above). Remember that the switch must handle the full amplifier current during normal operation, so it must be a heavy-duty type. This lets you test the system with minimal current (about 500mA) and with greatly reduced risk of speaker damage (other than tweeters/ compression drivers - they won't like it, but they should not fail).
You actually can get relays that are rated for up to 125V DC with a 15A contact rating. Information is scant, but RS Components sells one made by TE Connectivity (Part # V23009A 7A 52). The cost is around AU$475 (yes, really!) and it's highly unlikely that anyone will pay that much. We have to make do with what we can get, preferably costing less than the loudspeaker it's meant to protect. Ultimately, the relay is the constructor's responsibility, as those available are dependent on your local suppliers - there are far too many of both compenents and suppliers for me to make an absolute recommendation (something I usually avoid for just this reason).
Having decided on the version you wish to use, my suggestion is that it be housed in a diecast aluminium enclosure, with the case acting as the heatsink for the MOSFET. The input/ output connectors should be Speakon types, as they are designed to handle the current from high-powered amplifiers. The completed protection circuit can be external, with input and output clearly marked, and the LED should be visible. For a less cluttered stage setup, the box can be mounted inside the loudspeaker enclosure, with internal screw terminals for input and output. Ideally, it will be removable without having to remove speakers or rear panels, and a flat mounting plate is suggested. This also means that it is always in circuit, making it tamper proof for equipment that's used by others.
Before the circuit (either Figure 1 or 2 version) is installed, it must be tested in conditions that are equivalent to those in the 'real world'. This means connecting the input to the speaker line, but without the relay wired in. The system should be run normally (or abnormally if that's the way it's used), and the LED monitored. Under all operating conditions and at full power, the LED should remain off at all times. If it flashes, that means that the circuit has activated, probably due to excess low frequency energy. It is not necessary to connect speakers if the test can be run in the workshop or wherever the system is normally stored when not in use.
Should the LED come on, the timing/ filter circuits need to be slowed down. That means that C1 and C2 (Figure 1 circuit) or C3 (Figure 2 circuit) need to be made larger. Doing so will delay the operation of the relay, and reduces protection. The constructor may also find that these caps can be reduced, depending upon the programme material. Making them a lower value increases the detection speed, so provides better protection.
The requirement for thorough testing is not optional. This is a circuit that will normally remain dormant for most of its operating life. It can (and will) only operate if there's an amplifier fault, or if a frequency well below the detection threshold is applied. The threshold has been designed (very deliberately) to have a -3dB frequency of less than 0.5Hz, as this is necessary to accommodate high amplitudes at 20Hz. The values of the filter/ timing capacitors are designed to handle 70V RMS at 20Hz without triggering.
70V RMS is (theoretically) obtained from a power amplifier with ±100V supply rails, but in reality the supply will be higher than that. Such an amplifier will be able to deliver 1.2kW into a 4Ω load. For amps that can provide more (and they exist, but I'm unsure why), the values of C1/ C2 (Fig. 1) or C3 (Fig. 2) will need to be increased. This is why testing is so important!
It would be ideal if the circuit would latch when a transient causes false triggering, but that's not possible because the only power source comes from the amplifier. That's why it's so important that the circuit is tested thoroughly before it's put to use. False triggering with a speaker having passive crossovers will almost certainly destroy compression drivers, so I do not recommend that either circuit is used with full range enclosures with passive crossovers.
It's essential that you make sure that the circuit never false triggers in normal use. While this increases the time before the speaker is disconnected, it also means that the circuit is inaudible when it's being used. There are many compromises needed for this type of circuit, and it's up to the user to ensure that it works as intended, and only disconnects the speaker if there's an amplifier fault.
The two designs are set up for use with amplifiers having a supply voltage from ±35V up to ±100V, and using a 24V relay having a nominal coil resistance of around 570Ω (about 42mA coil current). This is set by the source resistor (R7) for the switching MOSFET (nominally 12Ω), which limits the current to 54mA. The extra current helps to ensure that the relay operates, despite perhaps years of remaining inactive. The LED resistor (2.2k) will allow a current of 10mA, ensuring that the LED is bright enough to be seen, as it indicates an amplifier fault.
This will cover the majority of cases, but particularly high powered amplifiers having a supply voltage greater than ±100V may need some changes. The MOSFET is rated for 200V, far greater than the supply voltage used in any known amplifier, but with a higher supply voltage it will need a larger heatsink. For example, with a 100V supply, the MOSFET will dissipate 4W, and without a decent heatsink it may run very hot. You will need to change the 12Ω resistor (R7) if the relay you select draws more (or less) current than designed for. There is some leeway, but the MOSFET current must exceed the relay's rated coil current by at least 10%.
By far the biggest problem with very high supply voltages is interrupting the DC fault current. Consider an amplifier with 100V rails (70V RMS output, 1.2kW output). The fault current with a nominal 4Ω load will be over 25A, and attempting to break that without a purpose-designed relay (which will be hard to find and very expensive) will lead to a complete melt-down inside the relay. If such a system is being used, then I'm afraid that you are pretty much on your own. The arc suppression capacitors may (or may not) be sufficient to prevent an arc. Any very high-powered amplifier should include DC fault protection internally. If it doesn't, buy something else!
The most brutal protection scheme of all is the so-called 'crowbar', which typically uses a high-power TRIAC or back-to-back SCRs to short-circuit the amplifier's output. The result will almost certainly be a totally destroyed amplifier unless it has good fuse protection (note that some amplifiers have no DC fuses at all). Finding a TRIAC that can handle the massive instantaneous current from a kilowatt amplifier is a challenge, but SCRs are available that can handle the current with ease. For example, 50A SCRs (over 500A for 10ms) are available for around AU$10.00 (but up to ~AU$35.00) each.
The circuits shown above can be adapted easily for a crowbar circuit, but it's not something I'd recommend. Although it's given that the amplifier has failed if it presents DC to the speaker, risking further (and possibly catastrophic) damage isn't recommended. If you were to try this technique, inclusion of a fuse is a must, but that adds complications. For example, do you rate the fuse for the maximum power the speaker system can handle, or something less? For a cabinet rated for a maximum power of 1,200W, you need a 20A fuse, assuming 4Ω impedance. You shouldn't use any old 20A fuse though - it needs to be an HRC (high rupturing capacity) type, because the peak current may be over 100A.
If the same system is used with a lower powered amplifier, it may not be able to supply enough current to blow a 20A fuse quickly enough (or perhaps not at all) to prevent further damage. This could easily destroy the amplifier. Then there's the ever-present risk of something fairly trivial (such as a subsonic signal occurring briefly at power-on or power-off) that 'false-triggers' the circuit, and blows up an otherwise perfectly good amplifier. This is a very real chance, and it's not one I'd be willing to take.
This is the reason I don't recommend (and nor will I describe any further) crowbar circuits. They are brutal, and totally unforgiving.
The circuitry described is (as far as I'm aware) unique. There doesn't appear to be anything like it available in the market, although there are some commercial enclosures that claim to have inbuilt protection circuits. It's unknown (at least to me) if these circuits work as intended or not, as no details were found on-line. There are a couple of examples of systems that could be adapted (one is shown in the references section), but the design is flawed as shown in the patent drawings, and cannot be recommended for anything.
Speaker protection isn't trivial, and the vast majority of circuits shown elsewhere won't work with supply voltages above 30V. As soon as high power amplifiers are included into the equation, everything gets harder. Ideally, all high-power amplifiers would have DC protection built-in, but regrettably this is not always the case. Providing DC detection and disconnection using only the amplifier's output makes everything that much harder.
There is no doubt whatsoever that speaker systems need to be protected from amplifier faults. It only takes a few seconds for a 70V DC supply to burn a voicecoil, as it's pushed out of the gap and held stationary by the magnetic field. 70V DC across a 4Ω voicecoil is 1.225kW continuous, and no loudspeaker ever made can handle that without failure. An amplifier with ±100V supplies will try to push that to 2.5kW (25A DC!) so survival is limited to (maybe) 100ms or so. While it's guaranteed that the amplifier's power supply will reduce that voltage somewhat (no-one designs for that much continuous DC output), unless the amplifier has a fail-safe protection scheme installed, your speaker(s) will be toast (literally).
Many commercial power amps include DC protection as standard, but equally, many do not. In some cases it's advertised, but the relay used cannot break an arc if there is a fault (some speaker relays are used only to prevent turn-on/ off noises from the amplifier). To save yourself the (not inconsiderable) bother of building external protection circuits, verify that whatever you plan to buy has internal protection that works. If it doesn't, I suggest that you avoid it completely, regardless of any other claims made. The simple fact is that competent sound reinforcement amps will sound the same (especially at 100dB SPL or more), and it's well worthwhile to spend a little more to get inbuilt protection rather than to hope that the amp doesn't fail.
While you may not be aware, Class-D (switching) amplifiers are not immune from DC failures. All that needs to happen is for one of the output MOSFETs to fail, and like most semiconductors, they fail short-circuit. So, it doesn't matter if the amp is Class-B, Class-G or Class-D, MOSFET or bipolar transistors. Failure in the output stage nearly always results in DC at the output, and it's nearly always the full supply voltage. Failure that gives a large DC offset but still provides (at least some) audio is very uncommon, but it can happen. It's more likely with an amplifier that's DC-coupled throughout, as DC from a mixer or preamp will be amplified along with the audio. DC-coupled amplifiers have no place in any audio system IMO, due to the risk of an external failure in peripheral equipment causing DC at the output. A good high-pass filter is your friend, and one should always be used.
Please note that the circuits shown here have been tested and verified to work, but the relay is a different matter. I've run tests that show that the capacitive arc suppressor works (and it works well), but you need a relay that has the widest possible contact gap. It's up to the constructor to find a relay (or relays) from a known reliable supplier, and be prepared to test one or more to destruction. Anyone who has ever used an electric welder (or has seen one being used) will be well aware of the awesome power of an arc - it's an excellent way to move molten metal from one place to another!
|Copyright Notice. This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is © 2020. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro- mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference while constructing the project. Commercial use is prohibited without express written authorisation from Rod Elliott.|