Yes, theoretically that would be the best (because we could deliver infinite power to the speaker), but not from the perspective of the (real) amplifier.
Last edited:
Some distributors still use this older and less accurate spec label. Sensitivity and Efficiency are different by way of label and reference. Sensitivity to voltage and efficiency to wattage. Impedance is not ignored at the wattage spec as you will have to factor in the impedance to get the correct sensitivity level to adjust your drivers accurately. Wattage is really only useful for mechanical or thermal power handling in models for boxes or meltdown potential, or maybe the maximum loudness by calculation of impedance at output level.
Most of the manufacturers that I've used over the last 5-6 years have either switched or already have had their drivers rated at dB/2.83V/1m.
2.83v into a 4 ohm load is 2 watts so by using that as a standard you are comparing 1 watt to 2 watts. A 4ohm speaker should also be driven with one watt or 2v into a 4 ohm load. Amps are rated in watts not volts. The 4 and 8 ohm versions of a driver usually produce the same spl when driven by the same power in watts.
We don't equate drivers by watts. We use 2.83V. If you are using watts you will have to comp for impedance in comparing the 2, whereas voltage you won't.
That is far from universal. Only about half of manufacturers use that standard, half use watts. I just went through a bunch at the Madisound site to verify this as fact.
It should be universal, but it is not quite yet. I meant what I said that most manufacturers I've used drivers from in the last 5 to 6 years have either been at voltage/m, or switched to it. I didnt say ALL of them use it.
Manufacturers that gives the sensitivity data as dB/W/1m are gives us just a calculated number or measured avarage SPL but the W is still just calculated and the measurement starts with that formula: 2.83V to 8Ohm is 1W.
For example, if a manufacturer wants to give use a SPL/W/1m curve for a nominal 8Ohm driver, they set the output Voltage of the amplifier to 2.83V, and measure the SPL curve at 1m distance.
Yes, that would be 1W to the driver, but only if the impedance of the driver is constant 8Ohm, but that's very rarely the case. An impedance of a 8Ohm driver can be 100Ohm at one point and 6.2Ohm at another point and this is not at all extreme.
So in reality, they just measured the 2.83V sensitivity of the 8Ohm driver and called it 1W, but this is misleading, because the impedance of the driver is probably not constant 8Ohm, so what came out of the amplifier (and delivered to the speaker) is not 1W, at least where the impedance are differs from 8Ohm (it probably differs almost in the whole spectrum, we can see it).
For example, if a manufacturer wants to give use a SPL/W/1m curve for a nominal 8Ohm driver, they set the output Voltage of the amplifier to 2.83V, and measure the SPL curve at 1m distance.
Yes, that would be 1W to the driver, but only if the impedance of the driver is constant 8Ohm, but that's very rarely the case. An impedance of a 8Ohm driver can be 100Ohm at one point and 6.2Ohm at another point and this is not at all extreme.
So in reality, they just measured the 2.83V sensitivity of the 8Ohm driver and called it 1W, but this is misleading, because the impedance of the driver is probably not constant 8Ohm, so what came out of the amplifier (and delivered to the speaker) is not 1W, at least where the impedance are differs from 8Ohm (it probably differs almost in the whole spectrum, we can see it).
Last edited:
Agree that impedance varies, A LOT with frequency, but the proper-honest way is to rate it at nominal impedance.
Which is fine because over most of the curve, impedance is HIGHER than nominal,except in a band typically around 250Hz
But chosen test voltage depends on nominal impedance too, so for 8 ohm it´s 2.83V ; for 4 ohm it´s 2V and for 16 ohm it´s 4V
Now on purpose using 2.83V on speakers known to be 4 or even 2 ohm nominal, dipping to about 60/70% at some frequencies, is cheating/sleazy/misleading, pick one or more.
Just a few days ago I read a post (was it here?) about a Klipsch bookshelf speaker claiming impossibly high 94-96dB , but of course, Ad was based on "everybody knows Klipsch speakers are high efficiency"
But it was measured by a "serious" Magazine; tester wrote, politely, "we could not fully confirm rated sensitivity", he found impedance dipped almost to TWO ohm (on a supposed 8 ohm speaker) , that is was pulling (or tried to) way more power than expected and suggested "choose your amp carefully, not all can drive it"
The understatement of the year 😉
Which is fine because over most of the curve, impedance is HIGHER than nominal,except in a band typically around 250Hz
But chosen test voltage depends on nominal impedance too, so for 8 ohm it´s 2.83V ; for 4 ohm it´s 2V and for 16 ohm it´s 4V
Now on purpose using 2.83V on speakers known to be 4 or even 2 ohm nominal, dipping to about 60/70% at some frequencies, is cheating/sleazy/misleading, pick one or more.
Just a few days ago I read a post (was it here?) about a Klipsch bookshelf speaker claiming impossibly high 94-96dB , but of course, Ad was based on "everybody knows Klipsch speakers are high efficiency"
But it was measured by a "serious" Magazine; tester wrote, politely, "we could not fully confirm rated sensitivity", he found impedance dipped almost to TWO ohm (on a supposed 8 ohm speaker) , that is was pulling (or tried to) way more power than expected and suggested "choose your amp carefully, not all can drive it"
The understatement of the year 😉
It's true that, at the spectrum ends, the impedance is usually higher than nominal, but in the middle of the spectrum, it's usually lower than nominal.
That's also true, but the 1W realizing only at the frequency, where the impedance is exactly the same as the nominal impedance of the driver, so at a very few frequencies, or a very narrow band usually.
Therefore, it is not so consistent to talk about dB/W in terms of sensitivity because we are sending constant Voltage to the speaker rather than constant Wattage.
Last edited: