Should amplifier output be restricted to the max power rating of the speaker?

[Dave TR3A]

Since you are using a Flex 8, have you considered/tried using the compressor?

Compressor - miniDSP Flex Eight / Eight-N User Manual

docs.minidsp.com

@NTK That's a fantastic idea.

I'm unfamiliar with how compressors/limiters work and wasn't able to figure out what to do with just the info in the manual. I contacted MiniDSP support but they weren't helpful.

I'd welcome any tips about how to use that functionality.

 

[peng]

Correct me if I’m wrong:
It is my understanding that an underpowered amp is more detrimental than one with more power.
In other words, it is better to have more power with more headroom.
Less power risks clipping and hence more chance of damage.

Not always true, it depends on the specific use case. For example, but just one example, if have a speaker such as the KEF R11, the specs simply say:
Amplifier Power
15-300W (recommended)

So, in that case if you have a 20 W rated amp, you might say it is underpowered, except for near field use and you don't listen very loud, but compare to an amp rated 500 W, that 20 W will likely not blow the speaker even if you turn the volume to maximum, and the little amp would be clipping badly, but the 500 W can definitely damage the R11 if you do the same with the volume dial, even though the amp may not be clipping at all. It is a silly extreme example though, just to show that it depends.

So there is nothing to correct you as you are neither wrong or right.

 

[Everett T]

What I've learned from this thread is:
the driver (or speaker) specs can't be trusted to be useful for that purpose
most of the time, the power being used is so small that the concern is more hypothetical than real

The T/S parameters are valuable, especially for modeling the box. However when you go active you need something like the Dayton Audio (or need to design a new passive crossover) DATs 3, it will give you the full suit of measurements you need and don't have to worry about variations from the T/S parameters or a bad deiver.

They have a more automated system, but @ $699, I'd stick to the manual DATS over the DATS LA. Erin still uses his DATs 3 for individual driver measurements, along with a majority of speaker designers in DIY and smaller audio companies.

 

[DVDdoug]

Compression won't help.

Compression and limiting "push down" the peaks and loud parts, allowing everything to go louder (i.e. more average power) without clipping.

 

[Pareto Pragmatic]

What prompted this thread was curiosity about whether or not such a restriction was a prudent precaution or unnecessary.

It's a prudent precaution against a low probability but high cost event.

 

[Dave TR3A]

Do you connect the woofer polarity 180 degrees from that of the midrange or tweeter because of the capacitors? Are all the amplifiers input fed full range or do you apply any active crossover as such or pass filtering before being amplified?

@Jose Jimenez

The DSP sends different signals to the different drivers. Here are the crossovers and filters I've configured:
Tweeter to midrange

• 4500 Hz
• Midrange has has a L-R 48 dB/octave low pass filter
• Tweeter has a L-R 48 dB/octave high pass filter
• No HPF on the tweeter

Midrange to woofer

• 400 Hz
• Woofer has a L-R 48 dB/octave low pass filter.
• Midrange has a Buttterworth 6 dB/octave high pass filter

Woofer to subwoofer

• 80 Hz
• Woofer has no filter applied; it just runs off
• Subwoofer has a Butterworth 48 dB/octave high pass filter
• No LPF on the subwoofer

Why do I have such an odd mix of filter types and slopes? Through a process of trial and error they gave me the flattest SPL/frequency.

I worked with REW to find the best polarity and delay settings.

• I didn't do anything with respect to polarity on account of the capacitors.
• The tweeters and midranges are delayed to time-align with the woofers.
• The woofer and tweeter are normal (i.e., 0 degree) polarity; the midrange polarity is inverted (i.e., 180 degrees).

 

[somebodyelse]

Depends if you are an ESL-57 owner

For those that aren't aware this is because electrostatics have a fairly specific maximum voltage, and exceeding it will cause arcing which will cause local damage to the diaphragm. IIRC early versions relied on Quad's amps having limited output voltage below the 57's maximum, while later versions has an internal limiter board to protect the speaker, possibly at the expense of the amp. More recent electrostatics often have other sorts of arc protection, like insulation on the stators. This is a different situation to conventional cone and coil drivers where you're trying to avoid either driving the coil into hard excursion limits (slamming the coil into the magnet or other hard end-stops) or avoiding over-heating the coil to the point where permanent damage occurs. The latter is very much a dynamic process with typical music signals, and in modern active speakers there is often a model-based protection that involves characterising the driver to allow short peaks while limiting longer term power to avoid overheating.

 

[DVDdoug]

The DSP sends different signals to the different drivers. Here are the crossovers and filters I've configured:

IMO - It's a good idea to add a capacitor in series with the midrange and tweeter as a high-pass filter in case the "unexpected" happens. The cutoff frequency should be lower than your DSP crossover frequencies so it doesn't affect the normal DSP operation.

 

[NTK]

Compression won't help.

Compression and limiting "push down" the peaks and loud parts, allowing everything to go louder (i.e. more average power) without clipping.

I think it should work. Even in miniDSP's manual protecting speaker driver is given as the first example application.

If the limiter with a high ratio value (hard limiting), it should be the same as having a lower power amplifier that clips clean and hard. And we know from Alex Voishvillo's research that hard clipping is not particularly unpleasant (contrary to popular belief). If a lower value of ratio is used, it will become soft clip.

 

[bmc0]

If the limiter with a high ratio value (hard limiting), it should be the same as having a lower power amplifier that clips clean and hard.

The rate at which the gain can change in a compressor/limiter is limited by the attack and release time parameters. It doesn't clip the waveform like a clipping amp.

 

[Dave TR3A]

I think it should work. Even in miniDSP's manual protecting speaker driver is given as the first example application.

If the limiter with a high ratio value (hard limiting), it should be the same as having a lower power amplifier that clips clean and hard. And we know from Alex Voishvillo's research that hard clipping is not particularly unpleasant (contrary to popular belief). If a lower value of ratio is used, it will become soft clip.

View attachment 520824

@NTK @bmc0 I had hoped to use the Compressors in my MiniDSP Flex 8 DSP to restrict the amplitude of the DSP's output signal amplitude from 2V RMS (at 0 db FS) to match the input sensitivity of my amps (0.775V RMS, according to the manufacturer).

I could figure out that going from 2V to .775V was -8.3 dB (i.e., -8.3 = 20log(.775/2.000)) but I had no idea what one would use for the required "Ratio", "Attack Time", and "Release Time" settings. Since I couldn't make sense of the info in the User Manual, I wrote to MiniDSP's support group asking what I should use.

The support group responded quickly and told me to use:

• Threshold = -9 dB FS (since only whole numbers are supported)
• Ratio = 50
• Attack time = 15 ms
• Release time = 15 ms

I didn't (and still don't) understand how the values for Ratio, Attack time, and Release time were determined.

Although that would allow me to address one problem (amplitude of DSP output signals being much higher than the input sensitivity of the amps), it still left me needing a solution to keep from blowing the drivers by having the amp deliver more current and/or higher voltage than the drivers were rated for. That led me to create this thread, asking "Should amplifier output be restricted to the max power rating of the speaker?".

Among the learnings that I've had from this thread are:

• the driver (or speaker) max power ratings specs can't be trusted/aren’t useful in determining a safe/unsafe boundary
• most of the time, the power being used is so small that the concern should be more hypothetical than real

I'd still like to be able to protect the drivers so that they can't be overdriven, but it seems that may not be feasible.

 

[Everett T]

@NTK @bmc0 I had hoped to use the Compressors in my MiniDSP Flex 8 DSP to restrict the amplitude of the DSP's output signal amplitude from 2V RMS (at 0 db FS) to match the input sensitivity of my amps (0.775V RMS, according to the manufacturer).

I could figure out that going from 2V to .775V was -8.3 dB (i.e., -8.3 = 20log(.775/2.000)) but I had no idea what one would use for the required "Ratio", "Attack Time", and "Release Time" settings. Since I couldn't make sense of the info in the User Manual, I wrote to MiniDSP's support group asking what I should use.

The support group responded quickly and told me to use:

• Threshold = -9 dB FS (since only whole numbers are supported)
• Ratio = 50
• Attack time = 15 ms
• Release time = 15 ms

I didn't (and still don't) understand how the values for Ratio, Attack time, and Release time were determined.

Although that would allow me to address one problem (amplitude of DSP output signals being much higher than the input sensitivity of the amps), it still left me needing a solution to keep from blowing the drivers by having the amp deliver more current and/or higher voltage than the drivers were rated for. That led to me creating this thread, asking "Should amplifier output be restricted to the max power rating of the speaker?".

Among the learnings that I've had from this thread are:

• the driver (or speaker) specs can't be trusted to be useful in determining a safe/unsafe boundary
• most of the time, the power being used is so small that the concern should be more hypothetical than real

I'd still like to be able to protect the drivers so that they can't be overdriven, but it seems that may not be feasible.

What about adding a resistor to the circuit, that would give you a solution for this issue?

 

[NTK]

The attack time determines how quickly the overshoot (see figure below) decays to the compressed level. The release time determines how quickly recovery is. In the class notes I found through a web search below, it suggests a attack time of 40 ms and release time of 135 ms to match the human auditory reflex. But a faster attack will provide better protection.

I'd suggest you experiment with it. Start with a low threshold level (so the limiter activates very frequently), and see how it sounds and how loud it gets. Progressively increase the threshold until you find the setting that it will limit the volume to not exceed your desired maximum, and minimizes degradation to sound quality at normal listening levels.
https://ccrma.stanford.edu/courses/192b/192b_Lecture_1_18.pdf