Loudspeakers & Power Ratings Part III: The Test Results
In parts one and two, I made a case for the idea that real power handling was intimately tied to the nature of the signal, as well as the electromechanical parameters of the speaker under test. We reviewed basic electrical concepts of power, relying on the universal concepts of Current, Voltage and Impedance in determining what the actual power delivered to the speaker load was, instead of the much more common method of calculating it by a knowledge of the voltage and nominal impedance only.
In Power Handling part three, I perform an experiment to determine if the underlying theory has merit, or if I am simply full of hot air. (Hot air is not good for either loudspeakers or authors.)
Today we will examine test results of three (3) samples of the same 6.5” midbass speaker provided with the courtesy and generosity of Axiom Audio. All three samples are tested to make sure they are normal in regards to their DCR (to be sure there are no shorted turns or large Impedance variations). All three drivers are subject to a sweep from 20 Hz to 5000 Hz to be certain that none of them are buzzing or rubbing so that we can ensure the samples are not defective prior to power testing them. Having decided I had 3 good samples, I hooked these small (but valiant) drivers to a QSC PLX3002 power amplifier, capable of delivering 550 watts per channel (stereo mode) into an 8 ohm load. I start at the modest level of 28.3 volts, or 100 watts into a nominal 8 ohm load, with one channel using white and the other pink noise. The signal is generated by Sound Tech Spectra RTA software (Version 1.32.16a) running under Windows 2000 on a Digital Audio Labs Cardelux Sound card. The balanced analog outputs of the Sound card are connected directly to the input of the QSC PLX3002 amplifier.
The speakers were wired with short lengths of 16 gauge stranded copper wire, and the voltage was measured at the amplifier terminals with a FLUKE model 45 five digit multimeter. This meter is one of the few that can actually give us true RMS readings without the caveat that the signal must be a 1000 Hz sine wave (Like most “True RMS” meters sold do).
6.5 Inch Woofer with 1
inch Diameter Voice Coil
When the speakers were subjected to the signal, I could tell immediately from the tonal characteristics, which was driven with pink versus white noise. Even deaf, if one looks at the drivers under test, the motion of the Pink noise driven speaker is considerably larger than the white noise driver under test. What is also obvious from this is the cooling of the VC (Voice Coil) caused by air passing across its surface while it pumps back and forth was far better on the speaker with pink noise than it was on the speaker with white noise.
WHITE NOISE OSCILLOGRAM
PINK NOISE OSCILLOGRAM
Both of these drivers were powered simultaneously and placed without an enclosure into the bottom of an empty speaker box with the front of the box facing up so the sound (and heat) emanate out of the box. The drivers are 6 inches apart so they are in a similar ambient temperature, in an air conditioned lab set to 72 degrees F. Although the voltage across both speakers was the same, if my proposed theory is correct, based on the impedance variations of a normal loudspeaker, and this one in particular (see below), the actual power delivered to the load by the pink noise should be much higher.
IMPEDANCE VARIATION VERSUS FREQUENCY OF SPEAKER UNDER TEST
If that is in fact the case, the pink noise speaker should perish first. As review, we can look again at the spectrum of white noise (equal energy per unit hertz) versus pink noise (equal energy per unit octave) below.
POWER SPECTRUM OF WHITE VERSUS PINK NOISE
Since white noise is weighted heavily towards high frequencies where the Impedance (AC resistance to flow of current) is very high in a typical woofer owing to voice coil inductance, then the actual power delivered will be much less than what is delivered with a signal having more low frequencies. Therefore with the same voltage, the Pink Noise speaker is actually receiving MORE power, and will therefore burn out sooner than the white noise speaker. (If my theory is correct).
After about ½ hour, the tonal characteristics of the speakers began to change. There was no smell of burnt VC, so I assumed this was a result of a large change in the DCR of the voice coils with its subsequent changes to the speakers frequency response. What was also clear from watching the speakers is how much more the white noise speaker was moving just a ½ hour after the beginning of the test compared to the beginning of the test. (Compliance increase due to break-in I theorize.)
After 45 minutes of cooking (I mean testing) the speakers at 28.3 volts without any evidence they were going to fail, I raised the signal input by 1db to 125 Watts to further stress them. I continued this 1db incremental rise in drive until one or both of the speakers failed. After 10 minutes at the 125 Watt RMS level, I notice the smell of VC was getting strong so I began checking in every 5 minutes to see the health of the patients. About 15 minutes into the 125 watt portion of the test, I started to hear a slight clicking noise. Free air testing with Pink Noise is a particularly brutal way to destroy a loudspeaker. In most instances (especially in the high-fidelity end of the speaker business) most speaker/box combinations are far more stiff than are the speakers alone in free air. The result is signals that will destroy the speakers in free air do not do so when the same exact speaker is fitted to a properly designed box. The box, especially in a sealed or closed box design, adds a great deal of stiffness to the suspension that is not available from the spider and surround alone. That stiffness results in far less excursion of the woofer at low frequencies and therefore less chance of self destruction from application of very low frequency energy.
After ½ hour at 125 watts, and no destruction, I raised the level of the signal by another 1 db to 160 watts. At this level, I was smelling burnt VC within the first 30 seconds and my belief was that we have reached the end of the line for at least one of the test participants. 15 minutes later, to my surprise, the burnt VC smell was still powerful, but more amazing is that both speakers were still going strong, and neither has of yet bit the dust. (Nor ignited the box stuffing!) (An examination of the failures after the fact reveals why this small loudspeaker would endure so much thermal energy, so please gentle reader, read on).
See also:
Recent Forum Posts:
fmw;462359
Wouldn't work with this test. You need a constant source so that your measurements will have meaning.
I agree, I was being sarcastic when I made that statement. Basically I am not a big believer of the claim that more speakers are damaged by underpowered amps. I am sure it happens, but probably not that often. I damaged mine once by overpowering it, period. I can't imagine who can withstand distorted mid high frequencies at home in their private listening environment. People do turn boom box up way too loud in semi public or public places, but those things can blast all day long fully distorted with no problem. I guess they don't have real tweeters to worry about.
Paul_Apollonio;464768
I hope that helps to clarify. If not, let me know.
Sincerely,
Paul Apollonio
Actually I did understand your article as I understand Fourier analysis quite well. In fact that was why I made the point that people tend to exaggerate how speakers are more likely damaged by being underpowered. As you explained so clearly, it is the tweeter that gets damaged due to the high frequency contents of the clipped signal. When a solid state amp clipped, it makes the speakers sound terrible. So I would bet that normal people would turn the volume down long before the tweeters melt down, because normal people are sensitive to the distorted mid/high frequency signals. In doing so they will save their tweeters every time!
Regarding my suggestion of testing with real music, again I think you and fmw might have misunderstood my point. I was just trying to say that while you can damage the tweeter with tones easily, if you use real music, a normal person would not like the distorted mid/high frequency signal and would turn the volume down right away before the tweeter gets too hot. In my house even if the music sounds loud but not distorted, someone will still grab the remote and turn it down if I am not quick enough to do it myself.
In fact, I would like to argue the opposite that a high power amp has a greater chance to kill a tweeter because a normal person can withstand much louder level of SPL if the music is loud but is free of distortion. That's how I fried a tweeter as I mentioned in my last post. In fact at the time I had no excuse as I wasn't even listening to classical music that has quite passages. I simply got it too loud (no audible distortion) for a little too long, when I first added an amp to the front channels.
Anyway, thanks for clarifying your point. As I said I understood it the first time, but you did an excellent job explaining it the second time. On the contrary I completely failed to make myself clear in my post clear.:o
I have an M&K MX-100 subwoofer in storage at the moment (because it is broken). I have had it fixed twice. I had it for several months (after I initially bought it) before I had a single problem with it. It sounded great, had awesome low frequency extention, as well as high output. One day I played the movie Resident Evil: Apocolypse and during one of the scenes there was some inaudible low frequencies present. I was watching this movie at near reference level in a small to medium room. My other relevant equippment was a pair of PSB Centuri 600i floor standing speakers and a Carver AV-505 multichannel amplifier.
The Carver and the subwoofer shut down during this scene. The Carver came back on, the M&K did not. If I remember correctly the mains fuse on the subwoofer blew, and I know for certain the two internal fuses blew as well. When this occured the woofers (there are two) slammed against the magnet structure making a loud clap sound (I assume this might have been the bleed off from the capacitors). I am unsure of what caused it do that. I took the subwoofer's amp plate off, I observed the power transformer was hot enough to burn my hand.
I took the subwoofer to a local repair shop and had it tested. The shop claimed the IC output device was damaged and in need of replacement, apparently the power transformer tested good. No other parts, other than the fuses, where replaced. When they repaired it the where unsure of what fuse size to use for the internal fuses (the mains fuse is clearly stated what it should be). I don't recal if the fuses where too low, or too high of a rating. I got the subwoofer home and tried it out. It worked for a short while (probably less than 30 minutes) and stopped working again.
The repair shop has a warranty, which they honored. They replaced the IC again, and put the correct fuses in it this time. I am unsure if they retested the entire subwoofer. It's conceivable that another part could be damaged and in need of replacement due to the neglagence to find the correct fuse value the first time because it blew again after a few short days, and I don't believe I was being very hard on the subwoofer. M&K made very good subwoofers designed for high output, and low bass. I would expect a $1000 subwoofer to perform as such.
Both woofers seem to be fine, there is no rubbing sound when you press on the cones and they both work when connected to an amplifier.
Have I just been horribly unlucky and am I just getting all these low frequencies imposing taxing loads on the subwoofer's amplifier, or is there something massively incorrect with the subwoofer's amplifier? I am aware these factors aren't likely enough to figure it out, but what's the likelyhood of it being one way or the other.
Thank you in advance for any advise or insight,
Seth
PENG;462350
I hope this article can make a dent into another popular misconception (I mean the other one being how important ACD ratings are) that more speakers are damaged by being underpowered than overpowered. I always find it hard to believe even some loudspeaker manufacturer’s sites are perpetuating such nonsense. The thing is people with low power amps will only tolerate audible distortions to a point before they reluctantly turned the volume down, hence before they blow the tweeters. On the other hand, people with high power amps (happened to me once) may get carried away and end up turning the volume up too high especially when they could not hear distortion during the quiet passage of a symphony.
Surely clipped signals are not good to speakers and will damage them, but there is no need to exaggerate it to the nth degree. I know for a fact you can use a 5WX2 (claimed) to power a pair of 20W (claimed) rated speakers all day long at levels you can hear major distortion and that boom box will not blow the speakers. On the other hand I guarantee you will blow those same speakers in no time with a 50W amp sending unclipped signals.
Now, it would be nice if Audioholics would do more tests using real music instead of test tones.
Actually the article speaks directly to this point. Perhaps it was not clear. Regardless, it is worth reiterating. Here goes. CLIPPED SIGNALS are dangerous to High Frequency devices, NOT woofers. It is because the percentage of power in the spectrum shifts from low to high frequencies that this is so. As the article clearly shows, high frequency energy is not absorbed as easily as low frequency energy by inductive woofers (anything with a conventional voice coil will be inductive). Therefore, the clipped signal is (with the RMS value held constant) LESS dangerous to the woofer, and more dangerous to a systems tweeter. If unclipped music has 10% of the power above 2khz, a clipped version of the same signal (test or music) will have considerably more. Therefore, the tweeter sees a much higher percentage of the signal when it is clipped than not. The woofer sees a lower percentage of the signal since it has shifted up. Also remember an amp that can deliver 500 watts RMS power (at low distortion) will put out 1000 watts (or close to it) when clipped (turning your sine wave into a square wave). Of course, the distortion (Which manifests itself as a series of harmonics) goes WAY UP. If this point is not clear, you might have to look at the impedance curves to see that the woofer has a higher impedance (resistance to current flow) at high frequencies than low frequencies. Since most of the signal power of music is in the range of 100 to 500 hz, when those musical signals are clipped, the harmonics which result go into the range of the high or mid frequency devices. Again, dangerous for the tweeter, not the woofer.
As for using music as a test signal, we would love to. This would make the tests far more entertaining. The problem is that music does not have a controlled crest factor or spectrum and is therefore not a reliable way to measure speakers. The best we can do is to use a repeatable signal with the same spectral distribution of energy and crest factor. The standards committees are lagging behind your intuition on this point with regard to adopting such a realistic standard.
I hope that helps to clarify. If not, let me know.
Sincerely,
Paul Apollonio
Djizasse;461696
It can't quite understand everything. Had to digest the "crest factor" definition and read the article twice. Thanks for destructing stuff in the name of Science
It is not difficult to understand, so perhaps I did not explain it well. Let me try again. Crest factor is the ratio of instantaneous Peak to long term RMS power in the signal.
Thats it.
Sincerely,
Paul Apollonio
