Default standard for speaker "sensitivity" measurement listings? Anechoic? In-room? Other?

There is no standard. The measurements are often more from the sales department than the engineering department. Just use as a rough guide. 

Speakers which play nice with tube amps

Tube power is more expensive. That is why horns were so common in the 1950s when tubes were King. So you’re looking for a speaker with sensitivity in the low 90s at the very least, higher if the room is bigger.

As tube amps get larger, often they do not sound as good. OTLs are an exception simply because they lack an output transformer. So a more efficient speaker is helpful.

Avoid 4 Ohm speakers, and speakers with dual woofers IF the woofer array is thus 4 Ohms. Tube amps are less efficient driving 4 Ohms; you can lose an octave of low end response from the output transformer simply by using the 4 Ohm tap (loaded at 4 Ohms). Bass response is often a casualty of higher powered tube amps! So really pay attention to this.

Its possible to put similar woofers in series, thus creating a 16 Ohm load without affecting how the woofers act in the cabinet otherwise. A 16 Ohm load in the bass is great for any tube amp- it will be lower distortion and will have more impact. So if you can find a 16 Ohm speaker so much the better; they are pretty rare.

If the speaker is higher impedance you’ll have far less issues with speaker cables. The lower the impedance of the speaker the more critical the speaker cables become, to the point that monoblocks are needed so as to keep the amps as close to the speaker as possible so shorter speaker cables can be used.

@atmasphere Thanks so much. Very helpful post!

It is my expectation that all measurements are anechoic unless otherwise stated.

For consumers A) They usually exaggerate and B) They are only useful as general, relative guidelines.

It's my understanding fewer are anechoic than in-room measurements. And, when revisions are made in subsequent versions of a speaker, one needs to ask if they truly went back and retested both anechoic and in-room measurements again.  

Sensitivity listed on a line item ad seems to be more of a general guideline at a selected frequency range noted on basic tech spec sheets. Seems to vary from one manufacturer to the next on what frequency they are sharing with you. Gotta look at the actual graphs across the entire frequency range to get a better handle on what's actually taking place with a speaker; where the dips and peaks are. 

I would say only 1% of speaker makers have access to an anechoic chamber. And extremely few of those have access to a chamber large enough to work below 100Hz. Even the few that do have chambers use them mostly for research and product development testing.

Most numbers from experienced manufacturers are calculated in half space, speaker on its back pointed up at the sky, no boundaries close by. That’s not easy to do either, as some just use a parking lot or roof top. Avoiding boundaries nearby is critical to avoid any gain being added by reflections of sound off boundaries.

This is the reason many manufacturers don’t bother, its not a level paying field as many products have deceptive measurements. Some more consumer oriented just put a spec down that sells speakers.

Brad

Jim Salk told me his measurements were anechoic because in room measurements are meaningless. I do not know how he does those measurements. Not sure how he would access such a facility. If he calculated in "half space" as lonemountain indicated, I would expect he would have used that phrase. Puzzling.

When I look at the measurements on the Salk site, none of them show anything below 200 hz. He may well be using a measurement method something like lonemountain describes. That is to take nothing away from his speakers, because I have found them to sound very good every time I have heard them.

Jim Salk told me his measurements were anechoic because in room measurements are meaningless.

Except that is where we all end up listening to their products.

That's why in room measurements are meaningless to a manufacturer unless every room was like the one they were measured in. 

Except that is where we all end up listening to their products.

I totally get Jim Salk's point; it's valid to insist on a metric which can be compared to other identical metrics. The question I would raise is the "half space" measurement equivalent, technically, to the anechoic one? Otherwise, instead of the apples-to-oranges problem (of anechoic to in-room) we have an apple-to-pears problems (of anechoic to half space).

Jim Salk might have built his own "anechoic room".  Some passionate engineers do exactly that, build one in their office or their warehouse or factory.  They are not necessarily true anechoic, as that is very expensive, but it can give you an idea.  This can yield good info but with limitations.  One of the most critical things in a true anechoic environment is noise floor.  If there is any noise leaking in, say HVAC, a truck driving by, wind around the building, it will screw up your measurements.  So if you built a chamber in your office, you'd turn everything off when you measure.  The fact that he states measurements only down to 200Hz is admirable though, as 200Hz would represent 5 or 6 feet dimension- whihc sounds like it clould have been a custom built chamber.  The other possible reason for the 200Hz is the measurement microphone being used to measure, which may not have had reliable flat response below 200.   

Here again,  MUCH we dont know about how it was measured, which makes the measurement relaible only in comparing his speakers to each other.  His measurements are not useful in comparing them to other brands, which would have been measured differently unless it was done in HIS chamber using HIS system.

This is the problem with specs: many many ways to affect results that make them  not comparable to anyone elses measurements.   Its like saying your table is  "brown", its not an absolute and is not translatable - there are 1,000's of different browns.  

Brad

I have no idea how Salk measures all their speakers but some are measured with Praxis. You can see it on the measurement at the top so could be quasi anechoic. 

This is the problem with specs: many many ways to affect results that make them  not comparable to anyone else's measurements.   It's like saying your table is  "brown", it's not an absolute and is not translatable - there are 1,000's of different browns.  

Great point. At least with "brown" we can safely assume it's not being lit by a red or blue or green lamp. It is not-black, not-grey, and not-red -- at least.

Look closely at the graph, box#1 (83.3db) at 500hz, and box#2 (83.7db) at 1kHz.

@russ69 "Except that is where we all end up listening to their products."

Yep, the bottom line is how do they sound in your room, on your system.

HIlde45, yes you get it. the challenge is many want to assign "brown" a specific value. Will we all know for certain what color your table is if you call it a "brown"? Saying a speaker is 92dB efficient means zero. Saying its 91dB SPL 1w/1m at 1K is a proper spec. If someone else says on an spec sheet "91dB SPL" , that is not the same.

Ive seen companies post specs of a speaker as "20 to 20K". No plus or minus variance (so we dont know how much it varies in between 20 and 20K), no distance or power spec. Inverse square law says if you stuck a microphone at 1 meter or at half that distance, those specs would be 3dB apart. So a 92dB (no power or distance) spec could really be 89dB 1 w/1m @ 1K- or worse. For someone searchng for a highly efficient speaker to mate up with his low pwer tube amp, this ommission of info could be costly as it would mean the speaker actually needs twice the power you thought it did.

Another example is your speaker spec sheet says "20-20K". Is this comparable to a spec sheet that says 100-15K +/- 2dB? Which one is better?

Its impossibel to say from this little information but I’m gonna bet the 100-15K +/- 2dB spec is a better speaker because the "20 to 20K" is deliberately deceptive. It could be + or - 20dB, which would be a crap loudspeaker. You have no idea if the spec doesnt tell you how much the Frequency Response varies between the two points. . 20-20K +/- 10dB means the Frequency Response varies as much as 20dB up and down, which is huge! A variation of +/- 2dB is the normal variation we like to see, so the real spec of this 20-20K speaker could be 200-8K +/-2dB @1K ! If it was really 20-20K +/- 2dB, you can bet they would print that +/- 2dB because they would sell more! Everyone in the engineering and manufacturing business knows what a proper spec looks like so this isnt some innocent ommission. This 20-20K is something I often see with "no name" speakers, or bluetooth speakers, or something that is of low quality. So for some companies, sales today is the single most important thing; for others its the integrity of the brand and building up a reputation for quality.

Brad