Speaker Sensitivity: Advertised vs Measured

[taisho]

Considering all of the options, I like the bar + area best. Not the most elegant looking, but it shows everything in the data at first glance.

True, I also like a version where the speakers are sorted by measured sensitivity descending. Would make it harder to find a particular model or brand though.

 

[hardisj]

I don't see what is wrong with the original post's presentation method. I work as analyst for a missile system where I purge tens of thousands of run sets to find faults and commonalities ... I cannot tell you the number of variety of plots I generate and see on a daily basis. It's way too easy to get caught up in making something for the sake of it. Most of the time the original method was sufficient and succinct.

 

[MZKM]

I work...for a missile system where I purge tens of thousands of

Thought that was going to take a dark turn.

 

[hardisj]

Thought that was going to take a dark turn.

lol

 

[stunta]

I don't see what is wrong with the original post's presentation method. I work as analyst for a missile system where I purge tens of thousands of run sets to find faults and commonalities ... I cannot tell you the number of variety of plots I generate and see on a daily basis. It's way too easy to get caught up in making something for the sake of it. Most of the time the original method was sufficient and succinct.

Agreed. I work with more than a dozen data scientists and they drown me with graphs all day every day. Original plots look fine. Stacked bar graph is not a good fit for this data. The sensitivities are within a small range so normalizing the variance is not needed either. Thank you @MZKM for this exposé.

 

[Selah Audio]

Try this with Tekton.

 

[wemist01]

The top chart might be better as a bar chart since there is no relationship between x-axis data categories ("we see greater spread in the transition from K to M as the lower alphabet woofer reaches the limit of its extension"). The strange gaps that open between non-aliged data points adjacent to others are pretty much random and have no particular meaning. You might do one bar as the spec, one bar as the measured, and a third bar at the bottom as the difference. And you could overlap them, or at least have the difference overlap the first two.

 

[anmpr1]

I wish I had some good data on the various ESL's I've owned. Acoustats claimed like Soundlabs to be 88 db/watt/meter. In fact I'd say more like 78 or 80 is for real with a heavy lean toward 78 db. Stereophile did measure 101 db/2.83v/meter which was lower than the claimed 105 db for the K-horns, but still awfully high.

The Klipsch discrepancy is especially problematic imo, as I often see them recommended to newbies based on that sensitivity. Go to a random audio forum, ask for advice on what to buy, say you like to listen loud, and that you only have an avr; chances are you're gonna get a lot of Klipsch recommendations.

For their little box speaker, Klipsch figures can be attributed to marketing. Anytime marketing gets involved, things deteriorate. That is not an excuse. It's just the way things are. However, my guess is that any Klipsch speaker is going to be a reasonably easy drive for most amplifiers. With this in mind, the current Klipsch company is not Paul Klipsch's company. It's a too familiar scenario--would Gordon Gow and Frank McIntosh be happy with their progeny?

When discussing sensitivity one has to weigh the lab with real life. I don't know how they measure now, but the old horn based Klipsch were measured in their anechoic chamber. Real-life 'in the living room' figures were always lower. But not by much. Richard Heyser measured 98dB SPL at 3.5 meters, and suggested that 104dB at 1 meter was not an unreasonable figure (that was the company's spec). Heyser measured at 3.5 meters since in his opinion that was as close as you wanted to get to the loudspeaker in practice.

Also, SPL sensitivity is FR dependent. Heyser offered the following K-horn graph:

FWIW I have used La Scala II's with both a 12 watt/ch tube unit (6V6GT push pull) and AHB2 100 watt/ch SS unit. With either it will get too loud too quickly.

Regarding Acoustat. These were relatively difficult loudspeakers to drive regardless of SPL. They did not mate well with many cheap amplifiers of the day, and I went through several trying to find one beefy enough to drive them to decent SPL. I finally settled on their own MOSFET (Jim Strickland's design) 200 watt/ch amplifier. It was a long time ago, and I never had any objective data, but I suspect it was the amp's interaction with the wide impedance load, both reactive and resistive. Also, unlike horns (which just keep getting louder) the electrostatic panels reached a certain SPL and then seemed to just stop huffing. What I wanted was a cross between the Quad and K-horn. Somehow. LOL

These days, with cheap watts, the benefit of high sensitivity is (or should be) lower distortion as the drivers are not asked to do as much at a given wattage, and hence should not distort as much.

However that is, with these little monkey coffin shoe-box speakers, and modern amplifiers, it's mostly a wash. Not much to care about. Unless you are talking something from specialty days, like an LS3/5a with a Futterman OTL amp. Or if your idea of amplification is a SET. Then things must be considered.

 

[MZKM]

Try this with Tekton.

Ha, I did this because I was arguing with someone on another forum about their Tekton (me stating how they are like Klipsch in terms of stating sensitivity). I know the tweeter used in their model, and showed how far its datasheet sensitivity was from Tekton‘s stated sensitivity; then they went and measured using a voltmeter and an SPL meter, me then stating that them doing a corner-loaded, in-room, c-weighted reading doesn’t prove me wrong, then they responded with in-room is what they care about, which wasn’t my point (which was seeing which companies give anechoic and which in-room).

 

[MZKM]

These days, with cheap watts,

Don’t forget max SPL. If a speaker is 85dB sensitive and has a max SPL of 100dB, it doesn’t matter how many watts you throw at it.

 

[exaudio]

The first chart shouldn't be a line chart.

What would you suggest?

Stacked bar chart

I also like a version where the speakers are sorted by measured sensitivity descending. Would make it harder to find a particular model or brand though.

I used @MZKM's data to create a stacked bar chart with the speakers sorted by measured sensitivity. When the measured sensitivity was the same as or higher than the claimed value I didn't bother to display it.

 

[Selah Audio]

Ha, I did this because I was arguing with someone on another forum about their Tekton (me stating how they are like Klipsch in terms of stating sensitivity). I know the tweeter used in their model, and showed how far its datasheet sensitivity was from Tekton‘s stated sensitivity; then they went and measured using a voltmeter and an SPL meter, me then stating that them doing a corner-loaded, in-room, c-weighted reading doesn’t prove me wrong, then they responded with in-room is what they care about, which wasn’t my point (which was seeing which companies give anechoic and which in-room).

Yes, I previously noted one model being rated 3dB higher than the SB Acoustics tweeter rating (one that used just one tweeter). Before that I thought I had seen everything.

 

[MZKM]

I used @MZKM's data to create a stacked bar chart with the speakers sorted by measured sensitivity. When the measured sensitivity was the same as or higher than the claimed value I didn't bother to display it.

View attachment 68949

I'm fine with whatever, though I don't like shading to accommodate colorblindness and I like going alphabetical so brand trends can be seen and specific models are more easily discoverable.

So,

A:

Or
B:

 

[mhardy6647]

As soon as I saw the thread title I thought Klipsch, even before opening.

Someone should forward this to them for comment.

Stereophile made a similar comment in measuring one of their speakers.

You are not alone. I don't this was an issue with Klipsch back in the Colonel's day (i.e., "Klipsch & Associates" of Hope, Arkansas).
Klipsch of that era had plenty of other issues, but (underperforming) sensitivity wasn't one of them

 

[Xyrium]

However, my Tektron 2A3 SET amp with 2.5wpc drove my Kipsch Forte II's in a large, high-ceiling living room cleanly to quite loud levels

Ok, now we're definitely not talking apples to apples with the Forte's! Nice rig!

Nice work, and thank you, @MZKM!

 

[kaka89]

@MZKM How about double bar?

 

[KaiserSoze]

The first chart shouldn't be a line chart.

That's what I was thinking exactly. I was puzzled by it at first. Also, if you want to know how the actual (measured) sensitivity of a given speaker or brand compares to the average for all speakers, the best you can do is sort of eyeball it from the 1st chart.

 

[Koeitje]

Why are we still making bar charts if this tells you everything you need?

 

[kaka89]

Why are we still making bar charts if this tells you everything you need?

Because some people might want to know the absolute sensitivity in SPL, rather than just the difference.

 

[hardisj]

Because some people might want to know the absolute sensitivity in SPL, rather than just the difference.

That’s easy, then.

He can swap the actual SPL sensitivity as a data legend above/below each bar. This can be done in excel by simply referencing a different column that has the SPL values. Squish the y-axis. Then he will have the difference represented by bars and a label at each bar telling what the value is.

Or have both: concatenate the difference and absolute values with a character break (char(10)) and have that at the bar. Put the SPL in parenthesis if desired. Whatever looks best. Either way, you can have all the data needed without having to choose an unintuitive layout.