I was asked by a fellow forum member to look into designing a loudspeaker protection system. Basically something that will help a speaker survive an amplifier fault condition.
Amplifier faults often result in the speaker output being connected to one of the power rails because when a transistor fails, Murphy's law states that it will always fail in the "fully on" state. Because the power transistors are usually connected to the power rail at one end, and the speaker output on the other end, you can see how this is a disaster in the making.
Now: The standard way of doing loudspeaker protection is to have a little sub-circuit which monitors the speaker output for DC. When a significant amount of DC is detected, it disconnects a relay which sits between the amplifier output and the speaker terminal, thereby disconnecting the speaker from the faulty amp and *hopefully* saving the speaker.
The problem statement is this: It becomes increasingly difficult to disconnect relay contacts which are carrying high levels of DC current. What happens is that the high levels of DC cause an arc to form between the contacts as they start to pull away from each other. This arc is often so severe that it literally welds the contacts together, so now you have a situation where the relay has been powered down, but the contacts are fused together, thereby not saving the speaker at the other end.
In 95% of amplifiers out there, this isn't too difficult because there isn't a huge amount of total power (Joules) sitting in the power rails. The extent to which we have a huge amount of power to contend with is dependent on the power supply of the amplifier obviously, but more specifically to the total amount of storage capacitance we have sitting there. While your transformer may only be a couple of hundred VA with the capacity to deliver a couple of Amps per winding, large storage capacitance can deliver massive current in a very short timeframe. I believe it is this massive burst of current that more often than not is the cause of the relay contacts welding together.
We know that most DIY amplifier builders tend to go a bit extreme on the storage capacitance (myself included) so this is a very real issue for us DIY guys.
Very few relay datasheets will even list the maximum DC breaking capacity of a relay. It has nothing to do with the DC "making" ability either so it makes things a bit tricky:
If you have one of the normal types of relay rated for 20 Amps AC at 250V, I wouldn't think it would be able to make and sustain a DC connection of more than about 10 Amps at 100V DC. From there, I wouldn't think it would be able to reliably break a DC connection of more than about 5 amps at 50 V DC. In reality DC breaking figures may very well be much lower still. You start to see the problem unfold...
So what can we do?
There are two standard options which come to mind:
1. Some manufacturers make relays specifically for high current DC switching. The downside is that they are extremely expensive - starting around R 2.2k each.
2. Many designs implement a triac system which in effect shorts out the power stage or the driver stage. The theory then is that the amplifier will blow fuses.
What then happens in the case where we don't have fuses between the power supply and the amplifier? A number of people connect their amplifiers in this way for various reasons. It means there is no fuse which can blow when a triac or other shorting out mechanism is used, and the amplifier will most likely either melt down completely (Fukushima) or hopefully at some point a circuit breaker in the mains DB will trip. Hopefully there is a fuse connected to the primary of the transformer which will at some point go open circuit, but even that isn't always a given.
One way or another, it means catastriphic failure of the amplifier and / or power supply.
What then are we to do here if we don't want to spend R 2.2k on a high capacity DC relay?
I have a rough idea in my mind about how to temporarily limit the current going through the speaker relay so that it can be disconnected without welding the terminals. After this, the system must release all of the current limiting circuitry and hopefully also allow the amplifier to die peacefully (and remain fixable) rather than have a full scale nuclear core meltdown which is cleaned up with earth-moving equipment and chucked into concrete lined landfill.
Obviously, all of these mechanisms must be "non invasive" - we don't want all kinds of additional bits and pieces in-line with our speaker output because it tends to degrade the sound quality...
Please feel free to share your thoughts or simply join the discussion.
Kind Regards,
Ian.
Amplifier faults often result in the speaker output being connected to one of the power rails because when a transistor fails, Murphy's law states that it will always fail in the "fully on" state. Because the power transistors are usually connected to the power rail at one end, and the speaker output on the other end, you can see how this is a disaster in the making.
Now: The standard way of doing loudspeaker protection is to have a little sub-circuit which monitors the speaker output for DC. When a significant amount of DC is detected, it disconnects a relay which sits between the amplifier output and the speaker terminal, thereby disconnecting the speaker from the faulty amp and *hopefully* saving the speaker.
The problem statement is this: It becomes increasingly difficult to disconnect relay contacts which are carrying high levels of DC current. What happens is that the high levels of DC cause an arc to form between the contacts as they start to pull away from each other. This arc is often so severe that it literally welds the contacts together, so now you have a situation where the relay has been powered down, but the contacts are fused together, thereby not saving the speaker at the other end.
In 95% of amplifiers out there, this isn't too difficult because there isn't a huge amount of total power (Joules) sitting in the power rails. The extent to which we have a huge amount of power to contend with is dependent on the power supply of the amplifier obviously, but more specifically to the total amount of storage capacitance we have sitting there. While your transformer may only be a couple of hundred VA with the capacity to deliver a couple of Amps per winding, large storage capacitance can deliver massive current in a very short timeframe. I believe it is this massive burst of current that more often than not is the cause of the relay contacts welding together.
We know that most DIY amplifier builders tend to go a bit extreme on the storage capacitance (myself included) so this is a very real issue for us DIY guys.
Very few relay datasheets will even list the maximum DC breaking capacity of a relay. It has nothing to do with the DC "making" ability either so it makes things a bit tricky:
If you have one of the normal types of relay rated for 20 Amps AC at 250V, I wouldn't think it would be able to make and sustain a DC connection of more than about 10 Amps at 100V DC. From there, I wouldn't think it would be able to reliably break a DC connection of more than about 5 amps at 50 V DC. In reality DC breaking figures may very well be much lower still. You start to see the problem unfold...
So what can we do?
There are two standard options which come to mind:
1. Some manufacturers make relays specifically for high current DC switching. The downside is that they are extremely expensive - starting around R 2.2k each.
2. Many designs implement a triac system which in effect shorts out the power stage or the driver stage. The theory then is that the amplifier will blow fuses.
What then happens in the case where we don't have fuses between the power supply and the amplifier? A number of people connect their amplifiers in this way for various reasons. It means there is no fuse which can blow when a triac or other shorting out mechanism is used, and the amplifier will most likely either melt down completely (Fukushima) or hopefully at some point a circuit breaker in the mains DB will trip. Hopefully there is a fuse connected to the primary of the transformer which will at some point go open circuit, but even that isn't always a given.
One way or another, it means catastriphic failure of the amplifier and / or power supply.
What then are we to do here if we don't want to spend R 2.2k on a high capacity DC relay?
I have a rough idea in my mind about how to temporarily limit the current going through the speaker relay so that it can be disconnected without welding the terminals. After this, the system must release all of the current limiting circuitry and hopefully also allow the amplifier to die peacefully (and remain fixable) rather than have a full scale nuclear core meltdown which is cleaned up with earth-moving equipment and chucked into concrete lined landfill.
Obviously, all of these mechanisms must be "non invasive" - we don't want all kinds of additional bits and pieces in-line with our speaker output because it tends to degrade the sound quality...
Please feel free to share your thoughts or simply join the discussion.
Kind Regards,
Ian.
