Speaker positioning is generally the most important adjustment that can be made to a hifi system. Listening position is important too, actually both are important together, but there are more aspects to the speaker positioning. One of those aspects is the speaker angle, and this is super important when you have electrostatic speakers, because their sound is enormously affected by the speaker angle
At my first ever listening party for the River City Audio Society, I got only one negative comment: the system sounding harsh. I admitted there was some truth to that, and in fact it was in mind when I switched to the Krell. The harshness went away. Usually. Not always. It has also always seemed to be very listening position dependent. I aimed the direct beam of the speakers to converge just behind my head. If someone sits too far back (as the commenting person was) they sit in the direct beam, which sound harsh or peaky or something. This sometimes happens to me also when I don't check the chair before sitting down. It seems to migrate slowly away from the speakers over days of time, and often need to be put back in place.
I've been thinking about other things also, such as the piles of electronic equipment behind each speaker. I decided that had to go back about eight months ago, and planned to buy a new rack in August. Right then, other things came up needing to be replaced. Now I'm putting new rack (to be positioned on the side of the room behind the equipment) back on the calendar and I hope to buy it in January. There's also the not-so-little issue of making room for it by moving stuff around in the living room that needs to be done first, and I'm thinking it might be good also to replace the windows, because once the rack is set up that's going to be a monumental task. That could push the new rack beyond January. And so it goes.
But meanwhile there's no excuse not to re-examine the speaker angle, which is just what I started doing last weekend. Just on a whim, actually (that's often how great things start), I rotated the speakers as shown above from a looking down view. I rotated the beam of the speaker further away from the listening position so that now convergence only occurs about 7 feet further back, near the end of the room.
This changed the highs dramatically and for the better I thought. Just before this change I was noting that the singing on Supertramp's Crime of the Century was more than a bit twisted, it was all scrunched up. With the new angle, the singers sounded much more like real, if unshaven, people. Overall this was the finest audition of this recording I've ever had, and I've been listening to it on many fine systems including a friends Quad ESL system in 1979.
But how can this be, I wondered. On typical box speakers, the beam of the speaker generally provides the flattest response, and many box speakers begin to have serious tonal irregularities off the beam, such as collapsing upper midrange, not to mention loss of highs which is almost universal. And then, compared to box speakers, electrostatics are said to be even more beamy.
But this simplifies the actual situation greatly. And there is even a never stated truth for electrostatics: straight on the beam they sound horrible. This is puzzling for a speakers whose very design seems to eliminate distortion everywhere on the membrane.
But there it is. Something is definitely wrong on the beam, I determined the next day. I played the Stereophile pink noise track on repeat, attenuated the left channel by 99.9dB on the Tact "Level" adjustment, and used various iPhone apps, the cheaper one called Analyzer providing an RTA and generator only seemed to give the best pictures.
On the beam, I saw rising response above 4k and very jagged and generally rising up to my super tweeter at 18k, which had a large peak I tamped down by 5 dB right then...though there's still a peak it's not out of line with the 20-20k peak and on this app even shows tiny rolloff just at 20k.
I tried many different angles, ultimately putting down yardsticks to get some idea of the angles involved. Right about 1' from the center of the front base, and 5 inches over, the response is much better. It's at this point it might have maximal flat extension, but still jagged, and still a rise from 4k on up. That's about the angle I used to be on. At 7 inches over, the rise at 4k is pretty well tamped down, and further over it becomes a broad dip instead, with the highs generally rolled off. Curiously some people might like the sound as far off the beam as 50 degrees best of all. From about 60 degrees on around, you have a very depressed but perfectly smooth highs. There's sound even on the edge of the speaker just like that.
My thinking was that 30 degrees nows looks just right, I'm currently at 22 or so--which was a huge improvement over the 10 degrees or less I was at originally. No wonder I had issues with harshness! 6k-10k, the so-called harshness region of hearing, was strongly elevated. I'm not thinking 30 degrees looks best of all, and coincidentally that will give me "zero toe in" relative to the room itself, with the speakers being parallel to the front and back walls.
Which makes me wonder if electrostatic speakers don't have some serious HF resonance issues, but helpfully those resonances are very directional, making it possible to escape from them by being off the beam. I had heard the need to be off the beam from a long time Quad owner, so I figure this is not an Acoustat only issue.
Here's an article on measuring speaker polar response.
Sigfried Linkwitz has a different idea. He says that the on-axis response of loudspeakers should not be flat. This is because they are normally oriented about 30 degrees from our head axis, where we are too sensitive to sounds around 4kHz. He recommends shelving the response (though in this blurb, the amount he recommends is unclear). His Orion speaker is designed to be dipolar like my electrostats.
Here's a critical discussion of Linkwitz' shelving idea. Critics maintain his shelving only compensates for something, either an inherent boost due to speaker width, or it compensates for other changes in the design which added a boost. But it also reveals (as Linkwitz himself doesn't, at least at the link above, unless you buy the speakers) exactly what the shelving needs to be. It is said to be a -3.2dB shelf at about 4kHz. There must be less than 0.5dB of loss at 2kHz for it not to sound too soft.
It could very well be that the peaking I measure above 4kHz is a reflection or beaming type issue. And it could be that the best solution might combine off-axis speaker orientation (relative to listener) and a bit of electronic EQ.
It's worthwhile taking a look at the measured quasi-anechoic frequency response of the Quad ESL-63, a speaker I continue to admire (and have often wished I had) as measured by John Atkinson in 1989. Scroll down to figure 7. Notice that there is a general tendency toward rolloff starting just at 4kHz (call Sigfried!) which heavily notches out 5kHz and just barely returns to baseline around 6kHz. By 9kHz another steep dip is starting, reaching minimum just below 10khz. Then there is peaking in the 11-14kHz region. It almost looks as though the design is intended to suppress a resonance around 12khz with general rolloff.
I did not do quasi-anechoic measurement, but my near field measurement doesn't show as much rolloff at 16 Khz (highest pure tone I can still hear) until I'm about 50 degrees off axis.
All this suggests to me that it's OK to be pretty far off the beam with Acoustat 1+1's, unlike Quads.