speaker sensitivity dividing lines

What do you consider to be the dividing lines between low sensitivity, medium sensitivity and high sensitivity speakers ? Here are some thoughts on the subject and please keep in mind that i'm basing these spec's on the speaker being fed 1 watt @ 1 meter with the results averaged over a wide frequency bandwidth ( 100 Hz - 10 KHz). While this "somewhat" takes the impedance variance out of the equation due to using 1 watt rather than 2.83 volts, a speaker that is more sensitive may not be "easier" to drive due to high levels of reactance and / or impedance swings. As such, the lines between a "resistive 92 dB speaker" ( medium sensitivity ) and a "reactive 95 dB speaker" ( high sensitivity ) could be blurred in terms of why a big amp can't drive a more sensitive speaker but a smaller ( yet "beefier" ) amp can. Then again, that is a whole 'nother can of worms for another thread.

The reason that i bring this up is that we may all have slightly different ideas as to what is high / low sensitivity. In order to make conversations a little more easily understood and get to a point where we are all on the same page, coming to some type of mutual understanding as to what we are using as reference points might make things easier. I think that this would come in handy for such things as an "Audiogon FAQ's" type of section that will probably pop up sooner or later.

As such, these are the basic guidelines that i tend to follow when looking at speaker sensitivity with the above criteria taken into account. I'd like to hear from others as to what their "dividing lines" are and how we could come up with an "Audiogon reference" when discussing speakers & efficiency ratings.

83 db's and below = ultra low sensitivity

84 - 87 dB's = low sensitivity

88 - 92 dB's = medium sensitivity

93 - 97 dB's = high sensitivity

98 dB's and above = ultra high sensitivity

Obviously, these figures are somewhat random but you have to draw the line somewhere as far as "spec's" or "performance on paper" goes. Any and all comments / suggestions welcome. Sean
That looks 'spot on' to me Sean. I think that's a rather good tool for beginners. That would put me right smack in the middle of medium @ 90 db.

Gee, on paper it looks well enough. Some how it's hard for me to imagine 87 db's as low. With out doing any research, I would have guessed more speakers are 86-87 db's than not.
My Vmps's are 91db, not to hard to drive, but still take a fair amount of power. I would say they are in the medium sensitivity camp. I think your pretty much right on with your chart keeping speakers with wild impedance swings out of the picture as you stated.
Unsound: In the days of yesteryear where sealed designs walked the Earth in great numbers, there were TONS of speakers in the 85 - 88 dB range. Since driver technology has increased over the years, vents of one kind or another are FAR more common and many designs "share the load" with multiple drivers covering the same pass band, efficiency has crept up a bit over the old school "one driver per passband" sealed box of yesteryear. From what i've seen, I think that most speakers are in the 88 - 91 dB range nowadays. I could be wrong though and that's why i asked for other points of view. Sean
I would factor in amp/preamp gain also. I have a 26 db gain preamp and a 4 db gain 2A3 SET and 86 db @2.83V/4ohm speakers. This front end has about the same effect on these speakers as a 100W amp with a typical 12db gain preamp (but at different volume settings). So, to a low gain front end, they would be "low" sensitivity. But to a high gain front end, they act as a higher sensitivty speaker. Maybe that's one reason, along with the level of our recorded source material, why we have different opinions as to what we call what. With that aside, I would call anything over 90db/1w/1m "high" sensitivity, anything below 85 "low" and in between "typical" since that is what I think most speakers are - 86 to 89.
I wonder if the resurgence in SET amps and Horn loaded speakers has rearranged the dividing lines. Panels speakers seem to have become more effecient, especially since many more, now have dynamic cones for the lower frequencies.
Gs: Correct me if i'm wrong, but you can only put so much signal into any gain stage before it begins to distort and overload. On the same hand, an output stage can only put out so much power before it starts to distort and overload. As such, driving the piss out of ANY part of the system is going to result in distortion being fed to the speakers and your ears. Obviously, it might sound a lot louder ( distortion of any type typically increases "apparent volume" ) but you are still limited to X amount of signal at a "relatively clean" level. I do agree that tubes overload more gracefully ( with less "nasty" distortion ), which allows one to push them harder with less listening fatigue though.

Having said that, I don't think that ones' electronics has anything to do with the sensitivity rating of a pair of speakers. Whether or not said electronics will drive a set of speakers to "good" or "comparitively loud" levels is another story. That, as far as i can tell, is food for another thread. Sean
Too many manufacturers spec a bit optimistically, I've noticed. How many 88dB/w claims end up a dB or two short over at JA's bench, for example. Sensitivity averages at 1k are usually predicated on midrange sensitivity, which hovers around 88dB unless squashed down to build bass in a small sealed 2-way, for example. Higher sensitvity numbers also abound with MTM (D'Apollito) arrangements, as the tweeter can now run pretty straight out, yielding some 90-91 designs.
An equally important corollary is room liveliness. My 88dB Spendors in my lively HT large toom are a LOT louder with a watt than my Parsifal Encore 88dB in my very damped smaller living room. I don't doubt that the difference is at least 3dB, or 1/2 amp power.
A good example of the design decision exists within the Spendor range. The small 3/5 comes in only at 84-85 dB/w.
In the larger SC3 center channel Spendor simply adds a second mid/woofer to make a horizontal MTM. The tweeter padding is now loosened, and a GREAT 88dB speaker is born.
In some ways I prefer its sound to the previous method they used to raise efficiency: the 6.5" two-way 3/1p I have flanking the center. Same efficiency, slightly different tonality. So it may be that the popularity of D'Appolito designs has raised the average efficiency across the market.
Then there's the curve-bump down at SET-land, with those 95 dB midrangers sucking 2-10watts, eh?
Ernie: That is a good point about "optimistic" specs from manufacturers. I think that Klipsch did just that with their RB-5's. From what i can remember, they claimed something like 95 - 96 dB's and they actually measured more like 91 - 92 dB's ( give or take ). While 91 - 92 dB's for a small bookshelf type speaker is a pretty respectable level of output, NO amount of break-in or "test procedure differences" could account for the 3 - 5 dB difference in rated specs vs measured performance in my book.

Then again, many people do not realize that speakers can and do play louder after the drivers are fully broken in. The stiff suspension due to a driver being new reduces both excursion capabilities and therefore reduces output. As such, a driver should really be used under normal conditions for a good period of time before actually trying to "spec" it. New "out of the box" figures will rarely jive with what one ends up with after a good amount of actual use on a pair of speakers. I have seen JA state similar things and even mention that he didn't think that the speaker was fully "broken in" when taking measurements, but he did test and publish the results just the same.

As far as your comments go about comparing speakers with "equivalent" sensitivity ratings, tonal balance / frequency response can drastically affect what we perceive to be "apparent volume". That is why i said that the sensitivity should be rated and averaged over a BROAD frequency spectrum, as this approach would tend to minimize differences in readings where one speaker was very peaky in specific areas and another was pretty even across the band.

Too bad "audio standards" really aren't "standardized". Sean
Yes these specs can be useful when pairing componets...but the bottom line...speaker sensitivity has little bearing on how a speaker SOUNDS....it does have some relation on how a speaker will BEHAVE with sufficient or insufficient power and at certain volume levels...if you like a speaker(which is a selection process in itself)...then finding suitable power is not really a mystery....i know there are those that believe this is a hi-end art...and all things being equal..the amp has the least influence in a system vs. source and speakers....like cables....there are many who get and ego trip "hung up" on speaker specs...
Phasecorrect: Have you ever actually watched / measured how an amplifier loads into a speaker and how the speaker can modulate the amplifier ? In some cases, you can literally watch the input waveform from the preamp or source change as the amp has to deal with the reflected EMF of a speaker. This can vary as signal amplitude changes i.e. the harder that you drive the speaker, the more reflected EMF that you have to deal with. Obviously, some amps will deal with the reflected signal a LOT better / differently than other amps. This is not to mention the amount of crosstalk that is generated between channels in such a situation. If you or anyone else think i'm nuts, try talking to ANY electrical engineer worth a salt and see what they have to say. As to whether or not this is audible, you betcha. The fact that many people may not have ever really noticed this probably has more to do with the fact that most speakers present relatively "benign" or "easy" loads and the amplifiers aren't being worked to death in such situations. That does not mean that such occurances don't take place on a regular basis though for some of us with reactive speakers.

As far as my thoughts regarding sensitivity of a speaker go, i never implied that one should delete "low" efficiency speakers from their list or only consider speakers of "reasonable" sensitivity. My thoughts in trying to work with this is that many newcomers don't understand that there is such a thing as "matching" when it comes to building a quality component system. All things being equal, starting off with a more efficient speaker will typically make amplifier selection a little easier and open up more available options. If they knew that a speaker was considered "less sensitive" as compared to others, they might be able to make a more informed decision as to the suitability of said speakers and amplification choices. I was NOT saying that this is the most important spec, only that it is one of several factors to look at.

For the record, i have speakers that are 2-3 ohms that are appr 82 dB's output. I also have 3-4 ohm speakers that are about 85 - 86 dB's output. If i had only used speakers that were 8 ohms nominal and 90+ dB output, i would be in a world of surprise in terms of wondering why my amp sounded "bad" ( clipping with lack of control ) and the speakers wouldn't play very loud. If i had some type of a chart to compare sensitivity / output levels of various speakers, i would know that if i went noticeably down on efficiency, i would have to go up on power output to achieve the same spl's and level of control. However, if i stayed within a dB or two from old to new speakers, i might not notice a big difference. Granted, there are other variables here that could come into play, but i'm talking "simple" stuff here for those that aren't that "audio educated", not trying to teach the world physics in one thread or post.

Try to keep things in perspective. If we could cram all we needed to know about each subject into one thread, there would be no need for these or any other forums. I'm still learning and i hope others are too. Sean
Using a spectrally-weighted specification for sensitivity sounds divine, but probably isn't necessary. I don't think any manufacturer would leave a midrange bump at 1kHz just to get a higher sens spec! At least I hope not.
The problem lies in the bandwith, of course. Since we perceive average loudness in the midrange mostly, then maybe an "average sensitivity" from 100Hz to 10kHz would be sufficient, and not have errors related to bass response averaging, nor tweeter roll-off.
Make sense? And is the spec to be on-axis, or a power-response? Shouldn't make too much difference. But what do you do with bi and di-polar radiation?
In general it is nice to see a trend toward less-thirsty cross-overs and slightly bigger boxes that use volume and bottom -end driver piston area to boost efficiency a bit.
Nothing like being irked by an 84dB baby monitor that sucks an amp down and still gets congested as hell....
Sean...you obviously know more about this than I do...over the years I have had 4,6, and 8ohm monitor speakers with varying sensitivity....all being powered by a 40 or 60w intergrated amp that can handle these more "difficult" loads...and even at moderate to loud volumes...never felt my system was underpowered...however my listening room is fairly small...which comes into play as well...at any rate...I have never felt compelled to really add any additional power...as we all know...even doubling your power has very little advantages...often under 3db...and low to moderate listening levels actually require very little power...often 10w or lower....these examples might be "oversimplified" ones...but they used to illustrate a point (I hope!)...cheers
Sean makes an extremely important point - EMF is a crucial factor in amplifier-speaker interaction and it is often ignored.

When it comes down to it sensitivity is really not a very useful specification at all. Impedance, capacitance, and back EMF tell you much more. Impedance and capacitance are easily measured and graphed. Back EMF is not as straightforward but it's pretty easy to tell if a speaker is pumping it out in large amounts or not from listening tests. Because Back EMF feeds on itself, and the louder you play, the more you are generating, it's a really important factor in loudspeaker-amplifier matching.

For more info see an article we wrote: