Speaker DC protection with relays
An important feature of an audio amplifier is to provide a system to protect your speakers in case the amplifier breaks down and puts DC on the speaker terminals. No speaker will take kindly to the presence of 50V on it's voice coil, so that has to be prevented. One way of doing that is to include a VI limiter in the amplifier or power supply. However, it's hard to design one that doesn't negativly influence the sound quality when the amplifier is driven near it's maximum (which happens sooner than you might think). Other solutions include polyswitches, fuses, etc. But the best solution, in my opinion, is a relay with DC sensor. Relays have no non-linear components and their resistance is small.
However, you need to pay attention to how you wire your relay and how to protect against arcing between the contacts. Breaking 50V at 12A is quite a hard thing to do, but as the following experiments will show, there are ways to do it.
I made videos of the experiments so that you can see what happened.
I designed and built the following test circuit:
Figure 1: DC circuit tester schematic.
The detector part is copied from Rod Elliott's P33.
The 4A fuse is what the rail fuse in an amplifier would be. The 3.9R resistor and 3A fuse are a simulated speaker. The fuse will tell me if the speaker has received too much punishment.
It's power supply is a 2x18V 8.33A, recitified to 50V. Supply capacitance is 20.000 µF.
Initially, I also wanted to test the effect of D2, which serves to shorten the relay release time. However, preliminary tests indicated it doesn't have much effect (the fuses in my simulated speaker didn't blow any less often), so I forgoed that. I forgot to remove jumper JP2 when testing the Amplimo LRZ relay, so the information sheet at the beginning of each film is in error about that (it is actually bypassed, not in-circuit).
The Amplimo LRZ is a special relay designed specifically for speakers. It has two contact areas, one of tungsten (to take the arc upon making and breaking contact) and a silver one to bypass the tungsten contact (tungsten has a very high melting point, but it is a poor conductor). It looks like this on the inside:
Figure 2: Amplimo LRZ relay.
The Omron G2R-1-E is a normal single pole dual throw relay:
Figure 3: Omron G2R-1-E relay.
I wanted to test the following aspects:
- Does a snubber circuit help to quench an arc? In my tests I used a 100 Ohm resistor + 1 µF capacitor, connected over the relay contacts.
- What happens when you do and don't connect the speaker to ground in fault-mode.
- The effects with a simulated speaker (3.9 Ohm resistor and 3A fast fuse) and a real 8 Ohm driver.
The amplimo is a single throw relay, so point 2 doesn't apply to that.
I planned the following experiments:
- Test 1.1: Amplimo LRZ, with snubber, simulated speaker.
- Test 1.2: Amplimo LRZ, with snubber, real speaker.
- Test 1.3: Amplimo LRZ, no snubber, simulated speaker.
- Test 1.4: Amplimo LRZ, no snubber, real speaker.
- Test 2.1: Omron G2R-1-E, with snubber, NC contact to ground, simulated speaker.
- Test 2.2: Omron G2R-1-E, no snubber, NC contact to ground, simulated speaker.
- Test 2.3: Omron G2R-1-E, with snubber, NC contact not to ground, simulated speaker.
- Test 2.4: Omron G2R-1-E, no snubber, NC contact not to ground, simulated speaker.
- Test 2.5: Omron G2R-1-E, with snubber, NC contact not to ground, real speaker.
- Test 2.6: Omron G2R-1-E, no snubber, NC contact not to ground, real speaker.
The relay, simulated and real speaker all held up fine in all tests. I performed the test dozens of times and there was no arcing. The speaker made a pop but survived, and the simulated speaker's 3A fuse remained intact (although you could see the filament deform).
However, I performed some preliminary experiments with the LRZ before and even though it held up quite well, at some point, an arc did strike and the relay was destroyed:
Figure 4: Amplimo LRZ melted.
The simulated speaker also didn't survive this (the 3A fuse blew).
This relay faired a little less well, even though it's quite a sturdy one (16A at 30Vdc).
I'll start with the experiments where the NC contact was not connected to ground. Even with a snubber circuit, and arc formed. With the simulated speaker, the 3A fuse blew after the arc was sustained for about a second. With a real speaker, it was even worse. The arc was sustained for about five seconds while the speaker was at maximum excursion, reproducing the sounds the plasma arc made. Obviously, the speaker didn't like this... The funny thing is, though, that it survived the abuse. However, it should be clear that you don't want to tempt fate and test if you precious drivers will survive that too.
I didn't even do the experiments without snubber, because if with snubber the relay arcs, it would do so without one too (I confirmed this in preliminary tests).
Here are videos of the experiment:
- Test 2.3: Omron G2R-1-E with simulated speaker, NC not to ground.
- Test 2.5: Omron G2R-1-E with real speaker, NC not to ground.
Then the experiments where the NC contact was connected to ground. It behaved pretty well. An arc formed, but because the arc is to ground now, as opposed through the 8 Ohm speaker, the current exceeds the 4A rail fuse by a lot, so that blew. The arc was momentary and both simulated and real speaker survived. Even the relay was relatively intact. It showed some charring, but not much.
Here is the video of the experiment:
One very important conclusion can be drawn from this: wire the relay so that the speaker is connected to ground through it's NC contact. Even if your relay is underrated, you can get good protection that way. In my tests the contacts didn't even weld together, so the relay could be reused, but my advice is to consider the relay as expendable; when it's been used to break fault current, replace it with a new one when you repair the amplifier. The Omron G2R-1-E has a transparent cover (which I removed for clarity) which is very convenient in this case, because you can see you need to replace it.
Also, you can't asume that any commercial DC protection system is wired up correctly. I know for a fact that the Velleman K4700 does not have the speaker wired to ground with the NC contact, so that circuit is unlikely to protect your speakers. The relays are even lighter than the one I tested, so if you use one of those, you might want to rewire it (which is difficult, because the speaker is not connected to the moving contact, the amplifier is. You'd have to remove the relay from the board).
What could also be imporant, is that your amplifier has fuses per channel, as opposed to for the entire amp. When an arc strikes, it's the rail fuse's job to blow so that the arc is quenched. If one amplifier has a maximum current rating of 4A, two amplifiers would be 8A. An 8A fuse is a lot harder to break than 4A.
Also, make sure you wire the detector so that it doesn't lose power when the rail fuse blows. It should be wired before the amplifier's fuses, with a fuse of it's own. The reason is simple: it needs power to operate. If your amplifier has a negative fault, the relay will turn off, blowing the negative rail fuse. Your amplifier will most likely then have a postive fault (because negative supply has disappeared). Some DC detectors require the presence of both + and - rails to operate and in this scenario, it will no longer be able to detect the postive fault.
- Video of test 2.1: Omron G2R-1-E with simulated speaker, NC to ground.
- Video of test 2.3: Omron G2R-1-E with simulated speaker, NC not to ground.
- Video of test 2.5: Omron G2R-1-E with real speaker, NC not to ground.