Clipping Test Results
________________________________________
Well, Talnlanky and I have finished the test and compiled the results. Our goal with this whole experiment was to get rid of a few possible erroneous preconceptions, learn a little and share the results with all the other readers. We attempted to be as objective as possible with our testing to be sure and give everyone accurate information. We had access to a great supply of accurate test equipment and plenty of samples to work with. I cannot guarantee these results 100% and that they will be the same for everyone, but I believe they are an accurate portrayal of the effects of a clipped signal played into a dynamic loudspeaker.

The basis of this test was to see how the effects of a clipped signal would affect driver failure. Failure in this test was thermal breakdown of the driver's voice coil. Thermal breakdown in the test was when the driver's voice coil either shorted out or opened up due to materials melting down. We scaled a few test drivers with power until we found out how much it would take to have thermal failure within 1 minute.

The test consisted of 16 drivers, 8 small drivers to be tested at 1 KHz and 8 larger drivers to be tested at 170Hz. The two different frequencies were chosen to see if cone movement played a major role in cooling effectiveness.

Two different signals were used in the test, a pure sine wave that measured less than .1 THD at the amplifiers output and a pure square wave that measured at least 34% THD at the amplifiers output.

Two different kinds of power were used during the test, normalized power between the square and sine (differing voltage drive) and non-normalized power between the square and sine (same voltage drive). We did this for two reasons, one; the normalized voltage would represent an amplifier driven into hard clipping, and that is what most people are concerned about. Two; normalized power was used to indicate the difference a square or sine would have on the loudspeakers voice coil at the same power level (like clipping your deck for instance, but still having the output of your amplifier at a relatively low level).

We used two drivers for each set of measurements and gave both results plus an average to help give accurate results. We also matched the pairs used in the test by their voice coil Le measured on a Suncor meter. Of the 16 drivers used, only three gave what appeared to be erroneous results, not to shabby! Power measurements were made using a small signal through line power meter to ensure the square and sine waves were matched relatively well, and then the power increased linearly with the sine to be used as a reference. We were within 10% with our power matching, which is still pretty decent. It is very difficult to measure square wave power into a reactive load. You cannot use an average meter as you are dealing with peak currents in a full duty cycle signal with a crest factor of 1.

The results-

It was very surprising to see how much power these drivers actually handled! The 4" drivers were rated at 5wrms and the 5" drivers at 25wrms. Both drivers took much more than that to be driven to failure as the attachments will show.

Normalized voltage (AKA amp at full clipping). This shouldn't have been much of a surprise to anyone knowing that a square wave with the same Pk-Pk voltage as a sine will have almost double the power. As you can see on the spreadsheet and graphs, power handling severely dropped when the square wave was applied in this test. Time to failure was cut by a factor of 4 for most drivers.

Normalized power was a little more interesting. We could not prove within the scope of our test that a different signals of normalized power would have any different effects on the driver, regardless of the amount of cone excursion. Some people may still argue this, but it seemed pretty clear to us that weather a square or sine was put into a driver at a normalized power level, it still failed about evenly.

An extra test.... Remember when some people say "It doesn't even matter if it is only a few watts, if it's pure distortion, it will ruin your woofers".... Well we took a 10 watt square wave and ran it into a Pioneer 10" subwoofer for a few hours.... Nothing happened to it, of course, the motor didn't even get hot....

This test also only dealt with lower frequency drivers (in the big scheme of things) and is in no way meant to represent tweeters or any other driver that requires a filter network to provide a safe operating environment. That is a whole other ball of wax.

We hope that everyone gets something positive out of this and we had a great time doing the test!

I would like to thank the following people for all their help and knowledge, I know I forgot a few, but here we go anyway-

Talnlanky- Did the whole test, took the pictures and compiled all the data for you guys to see
Victor Tiscareno- Smartest dude I know
Sergey- Helped us realized how tough this would be!
Dan Wiggins- "its real easy Mike, if you have a resistive load....."
Hugh Dean- Designs awesome amps and gave some great pointers
Black00- Always gives us great ideas when destruction is involved
Parts Express- For having 49 cent drivers.....

-Mike Edgar

Graph results

I wouldn'y mind some sort of test to see how/if you can control sound waves directions, it would be tough, but it could help others. I'm sure marshall could help with this.

yes, I've seen the math, but very few people showed open projects about them and it was usually not enough.

I still kills me reading all the different boards online, how people still don't understand about clipping and all the heresay that gets thrown around....

Also reminds me of the myth that cooling is reduced with a clipped signal...  Even though the cone is moving back and forth no less...

Also, people forget that a good majority of voice coils are tested for thermal handling at frequencies around 1Khz, where cooling do to air movement is a mute point...

If you really believed clipping alone killed loudspeakers, everytime you listened to Jimi's guitar, your speakers would fry....

There is only one reason for thermal failure, too much power.  Period.  Regaurdless of input waveform.

You are 100% right, but you are missing the point of this, and the data of the test...


Here is the problem.  There isn't an argument that a square wave contains more power than a sine of equal voltage amplitude (duh), but what people don't understand is that a sqaure or clipped wave doesn't have any more of a detrimental effect than a sine, of the same power magnitude.  This is where the misconception lies.  Many think that a speakers that has a 500watt power handling may be destroyed by a 200 watt amplifier driven in to full clipping, which is never the case due to the fact that an amplifier that can produce a 200 watt sine could only produce a 400 watt square wave (assuming the power supply can do this) and this could never cause thermal failure of a loudspeaker truly rated at 500 watts, mechanical failure aside.  This is where the misunderstanding lies....  

People don't realize that when they burn up a voice coil, it is only due to one thing, too much power, regaurdless of signal.....

A speaker that fails at 100 watts fails at 100 watts.  100 watts of a sine, 100 watts of a square, 100 watts of music averaged.....  This is a fact...

I've got a couple questions on this subject.
When using a square wave, isn't the amplifier sending DC for short periods of time ? And if supplied with DC, won't the coil act like a resistance and possible the amplifier output even more power because of this ?