Signal chain overload effects

Think it might be appropriate to look at what these clipped signals do at the 'end' of the signal chain as a whole.

Firstly there is no Limiter built which can re-construct the original waveform from a clipped one.

Now with squarewaves you immediately have 1.4 times the RMS Voltage of the same peak Voltage sine wave or 3dB increase (twice) in Power from your Amplifier.
Remember Power is proportional to V², so 1.4 x 1.4 = 2 which is where the doubling of the Power output of the Amplifier is calculated.

Next we have very much the equivalent of a continuous signal so the 'Speaker rating which applies should be the Continuous Power value - not the Program Power value which assumes a much greater crest factor for the signal.

Clipping has removed most of the crest factor.

So back to Limiting to try to reduce the potential for 'Speaker damage occurring.
Set the Amplifier input level to full (tamper-proofing) and then monitor it's output to match your 'Speaker's Continuous Power rating less 3dB.
Set your LMS or other Limiter to 'brick-wall' at this level.....

Might preserve your 'Drivers for another session, but be aware of the increase in HF content as shown earlier in this thread

Hope this explains things reasonably!
Mik

Re: "In both venues thy now have NO fire alarm interface"

seem to remember this being a condition of Licenses?
Mik

Test results for a Formula Sound FSM 600 Mixer.

In the first two pics, Line input is overloaded by providing a large three tone input signal, the same as for earlier tests:

This pic shows the Mixer set to have one red LED showing on the Output metering,
note the position of the Fader on Channel 2 which is used as the test input being very low.

Next the Input and Output signals, upper trace showing the output with clipping:


Now by selecting PFL for the input channel, this is clearly overloaded as the metering shows:


At least with this Mixer there is a very good chance of finding overloaded inputs and being able to do something about it!
Mik

If anyone has other samples of Mixers to test, it is fairly easy to set up a repeat of the tests used earlier in this thread..... a DJM800 anyone?

Other parts of signal paths can also be analysed further; should members have questions about this.

Mik

Scott
Firstly: with mixers, they usually have a nominal output level.
This unfortunately varies between manufacturers and is monitored with some form of output metering.
Now to fully detect peaks, special circuitry is used, the 'standard for this' is the Peak Program Meter, which can display either using bar-graph Leds or a special meter movement.
The rise-time of these PPMs are specified by the BBC, but they still will miss very fast transient peaks.
VU type metering is more common. Specifications for these comes from Bell Laboratories. These are much worse for showing short duration transient peaks
There are various other types which don't follow any particular standard so we don't know their accuracy when reading signal peaks.

The actual mixer output signal level then is taken from a continuous signal level indicated by an 0dB indication on the metering.

The headroom above this nominal output level to maximum output, is the 'reserve before clipping' for the signal when peaks occur above the nominal level.

Secondly:
You are generally correct on this. The important thing when clipping occurs is that the Power of the signal increases whilst the peak Voltage remains the same. Also there is an increase in High Frequency content from the distortion components. Have a look at the spectrum displays for 'normal' and clipped' signals in post 2 of this thread - each spike in the spectrum display is one frequency detected.

Finally:
Most of this is already covered but with a signal of increased Power and very low crest factor, as with a clipped signal, plus added harmonic content isn't good for Drivers.
A squarewave signal of the same amplitude as a sinewave one has a 3dB increase in Power or twice!
See page 2 posts 1 and 2 for sample results of an LMS output for the same signal, at normal level and overloading an input.

hope this answers you!
Mik

scott_fury wrote: That's great Mik. Thanks for taking the time to reply. Any chance you could explain about what happens when the power of the signal increases but you say the voltage remains the same? I'm assuming that the impedance doesn't change, so it would be the current? Also, when setting up my system I have usually put the amp volume up full, the mixer on 0db and bring the limiters on the LMS down to stop clipping on the amps (before connecting the drivers I may add!). How do these limiters differ from the clipping of a mixer for example?

Scott
voltage of a signal is measured as the RMS value of Voltage for Audio low level.
Peak Voltage then is 1.414 times this value.
When a sinewave is clipped the RMS Voltage approaches the Peak level of the sinewave such that the Power increases up to a factor of 2 when a full squarewave is reached.
Thus the Power into a fixed load Impedance can double - current also increases pro-rata.
Equally applies to program signals in the same way such that the delivered Power increases but the peak Voltage remains the same when clipping occurs.
Another explanation relates to a somewhat complex maths approach!
Also see Post 1 in this thread?

You are using a good methodology for your setup.
However Limiters shouldn't themselves clip a signal unless you overload their inputs or overdrive their outputs.
Limiters use 'Gain Cells'* which are controlled by signal level to reduce their output depending upon how they are set. Hopefully they don't add any appreciable distortion.
Clipping within a Mixer is simply that a signal amplifier stage reaches it's limit - usually just below the power supply voltage(s) - so has 'run out' of headroom

Mik
*different method but same result with DSP