volume/transparency ?

Subaruguru -

I learned about the relationship between efficiency and thermal compression from a recording engineer who posts over at the Audio Asylum under the moniker "RBP".

The following is taken from RBP's post:

"Over the last 25 years, I have studied a factor that is not getting the attention it deserves. I have written several papers on this. Call it Linear effeciency.

"Take two designs, Low eff. and High eff. Even though "High eff. is still considerably less than even 10%, these results will prove why higher effeciency Loudspeakers are closer to the truth that Low effeciency in terms of dynamics. High effeciency has it's price. Usually more deviation in the frequency response curve, and large size, plus higher low frequency cutoff per cu ft.

"Here is a reprint of one of over 200 effeciency test I have performed over the years. Note that doubling the power does not necessary mean a rise in 3dB!

"From the Archives:
86dB speaker. DCM Time Window.

"1W.... 85.7 (2.4dB)
2W.... 88.1 (2.3dB)
4W.... 90.4 (2.2dB)
8W.... 92.6 (2.2dB)
16W... 94.8 (2.2dB)
32W... 97.0 (2.1dB)
64W... 99.1 (1.8dB) Severe compression here.
128W..100.9 (1.3dB)
256W..102.2
512W..Voice coil failure.

"OK?

"104dB Speaker...Klipschorns.

"1W.....103.2dB (3.4dB)
2W.....106.7dB (3.4dB)
4W.....110.1dB (3.4dB)
8w.....113.4dB (3.3dB)
16w....116.7dB (3.5dB)!
32w....120.2dB (3.4dB)
64w....123.6dB (3.0dB)
128W...126.6dB (2.4dB) Compression begins....
256W...129.0dB No need to go further risk speaker damage.

"White noise burst for 500mS (1/2 second)

"Clearly the difference between 1 and 128 watts on the Klipsch is 20dB. [actually RBP made an arithmetic error - the difference is 23.3 dB].

"The low effeciency speaker ..only 15.2dB"

This is very interesting. I asked RBP about the greater-than-3dB "expansion" figure for the Klipschorns. From his reply:

"I thought it was meter overshoot at first but my measurement system is deadly accurate....I get the same thing for 50mS as I do 500mS or even 5 sec. It suggests that there is a point (it is around 96dB...96.2 to be more precise) where you get an honest 3dB with power increase of double."

I haven't independently verified RBP's findings, but I think he's barking up the right tree. His measurements explain things I've been hearing in a good high-efficiency systems, but I still haven't found a high-efficiency system that doesn't have some little shortcoming that bugs me.

For example, using the numbers given by RBP, suppose we have a hybrid system with 86 dB woofers and 104 dB horns (padded down, of course). When the music signal says "give me a 20 dB peak", the horns would give you 23 dB while the woofers would only give you 15 dB. So, the speaker would sound bright and thin on peaks. In addition, it would have to be voiced to sound right at a particular sound pressure level - any softer and it's too dull; louder and it's too bright.

Note that with electrostats, different compression mechanisms come into play - there is no thermal compression because there is no voice coil to heat up. Compression in electrostats seems to be related to transformer saturation, among other things, as I've heard low-efficiency electrostats that compress audibly less than higher efficiency ones.

Good one, 61.

Steakster, what volume levels are we talking about? Do you have access to a SPL meter? Also what is the size of your listening room and how far are you sitting from your speakers? Measured decibel levels at the speaker and at the listening position will give insight into whether it's a loudness curve issue or not, plus you should measure the background noise level of your listening room. It may be that the at soft listening levels you are knocking at the room's noise floor.

Drubin, those old Quads were good at ALL levels they could take, IMO :)!

I'd like to take this opportunity to say again that I think listening volumes are the underappreciated X-Factor in audio. If I prefer to listen mostly at lower volumes, this will be a very significant factor in how I build my system and how I evaluate components, particularly (but not only) speakers. My impression is that most people--certainly most reviewers-- listen a lot at live or near-live music levels. Their findings are not 100% useful for me, I think.

I lived with Quad 57's for many years and they were absolute champs at delivering the goods at low volume.

Audiokinesis, nice post. But re differing driver efficiencies and compression limits: why are these related?
The higher sensitivity driver (almost always the tweeter) is padded anyway, so compression issues would be unrelated to efficiencies, no? Not relevant here anyway, but just a thought. I would think that speakers that have a "right" volume only are (accidentally?)designed to produce an inroom response that flattens out VIS-A-VIS our fletcher-munsoned ears. It's an interesting point. In addition to symmetrical "droops" with lowered SPLs, I wonder if the best designers also have to factor in off-axis flare smmothness AT DIFFERENT SPLs to get a speaker that sounds balanced in a wide range of volumes? Interesting....

Streakster -

I find a correlation between high efficiency and low-volume liveliness in conventional speakers, and a correlation between transformer quality and low-volume liveliness with electrostatics (although a very thin diaphragm also seems to help). In general, I find electrostats to have more liveliness and inner detail at low volume levels than conventional speakers. The most lively speakers at low volume levels I've heard are full-range electrostats and full-range horns.

Compression is largely a function of voice coil heating with conventional speakers, so a high effiency speaker that undergoes little voice coil heating will compress less. Sometimes in a multi-way speaker the various drivers will have differing efficiencies, and therefore differing compression characteristics. In such a case, the speaker will sound "right" at a particular volume level, but will sound tonally unbalanced at higher or lower volume levels.

Transformer losses are the primary cause of compression in an electrostat, with transformer saturation resulting in severe compression.

To see a set of Fletcher-Munson curves, go to this address:

http://hyperphysics.phy-astr.gsu.edu/hbase/sound/eqloud.html#c1

Steakster, I don't quite understand your question, so I probably haven't answered it. Is it the imaging that collapses at lower volume levels, or the clarity (transparency)? What speakers do you have?

Yep, it is our ears more than the speaker. The basic premise of Fletcher Munson is that our ears are far less sensitive to low frequencies and high frequencies at lower volumes. Our ears are more or less midrange dominant at lower volumes. So, we don't "hear" the full range sound until we start to turn it up closer to the original level it was played at.

Fletcher-Munson curves!!

From my own experience, I've found that most speakers don't "take off" until they are played at medium volume or higher. I suspect this due to the amount of energy needed to overcome the inertia of the driver, and to properly damp the speaker. This is particularly true with dynamic cone speakers.

However, I noted the same phenomenon with my old Acoustat 2's, which I owned in the late 1980's. With them, however, there was an additional problem: low efficiency, due to the speaker's transformer. Many electrostatic speakers tend to have similar problems, and all of the ESL's that I've listened to over the years don't really start to "sing" until they get some juice.