Acoustat and other electrostatic speakers

I've come across an interesting discussion oriented around building replacement or improved versions of the Acoustat panels:

http://www.diyaudio.com/forums/planars-exotics/164430-how-construct-cube-louver-acoustat-9.html

Here's a short description of the Acoustat design: wire stators with milspec PVC coated wire (this was determined to be better than more resistive insulators such as Teflon, read Jim Strickland's white paper), 1/2 inch styrene louver frame, HS65 mylar membrane coated with a somewhat less-resistive-than-usual coating, felt damping in back

The HS65 material is actually heat shrink, shrunk into correct tension.  It tends to expand around tension points over time, but can be re-shrunk with blow dryier.

HS65 is far thicker than the usual membrane material, 0.00065 inch or 0.65 mil or 16.5 micrometer
Martin Logan uses 0.5 mil or 12 micrometer membranes (a bit thinner)
Quad uses 0.13 mil or 3 micrometer.  (Many electrostats are in the 3-6 micrometer range.)

Thus the Quad membrane is less than 1/5 the thickness.  Many people think that thinner membrane yields greater transparency.

Using Kurzweil as Sine Wave Generator

I've tried this before, so I thought I'd try it again.  I start with the Default Program 199 on my Kurzweil K2661 synthesizer and modify it to use the Sine keymap and remove touch sensitivity.  I program the mod wheel to control volume.  That makes for a nice keyboard sinewave generator which sounds far cleaner than my B&K generator.  Today I sent digital from optical out through C02 converter to coax, then into coax digital input 2 on my Tact, thus it was unnecessary to run Masterlink as I usually do (and for some reason Masterlink was not locking onto output from C02, perhaps bad short cable).

Playing around, it was easy to hear all the way down to C0 (16.35 Hz), or at least that last note can be unambigously felt.  Various rattles between the upper 20's and up in the walls, fireplace, and stuff; I need to make a map of all the rattles and fix them.  There clearly is a broad depression in the area between A2 (110) and A4 (440).  That was why I had previously run subwoofer up to 165, but that was not an effective strategy.  Still, I wouldn't say that A2 or A3 is inaudible, just noticably lower than A1 or A5.

Measurements of Acoustat 1+1

I found this graph at Izzy Wizzy's Acoustat site

The response looks remarkably flat, however you may notice a declining average response (with peaks) below 500 Hz (my speaker does that too) and roll-off starting almost precisely at 14kHz (as I have found).

Acoustat specified a "18kHz" upper frequency.  That seems close enough to 20k for most people.  By normal standards, that would mean -3dB at 18kHz.  That pretty much fits the above graph, though it could be more like -6dB depending on where you put the 0dB point.

But what we really might have wanted was flat response to 20kHz, with rolloff to -3dB at, say, 30Khz.

I've seen an interesting argument that we need flat response to wherever (say, 20kHz) even if the recordings and ears roll-off before that.  One reason is that the speaker rolloff adds to the recording rolloff (and every other rolloff) making it sound audibly deficient, even if every one of these rolloffs is inaudible by itself.  Well this is pretty obviously true, but we often don't think of this when we are presented frequency response specifications, or think that we don't need a speaker with response above 16kHz because we can't hear 16kHz tones.

Discussion of Angular Separation of Loudspeakers

From Peter W. Mitchell and published in Stereophile Magazine.

http://www.stereophile.com/asweseeit/36/index.html

Just as I said in previous post, the major advantage of wider separation between speakers is in improved ability to resolve instruments close to each other.  Well, that's sort of like what I said, and I was certainly thinking along those lines.  And he says that adjustments of this angle are among the most important thing that can be done for playback realism.

Reveals that the technically correct angle for recordings made with a coincident microphone is 90 degrees.  However, that does not work well with recordings made with spaced omnidirectional microphones; a "hole in the middle" develops.  For those recordings, an angle 60 degrees might work better.  And notes that many audiophiles and even dealers in USA use angle lower than 60 degrees in the range 45-60 degrees.  60 degrees is a commonly accepted quasi-standard "compromise" listening angle recommended by many books and articles.  That is simply when the speakers are as far from the listener as they are from each other.

So concludes that moving listening chair forwards and back depending on recording is not a bad idea. Peter himself simply leans forward and back in his recliner, covering a distance of 18 inches, depending on recording.

Sonic Paradise Returns

I played some FM radio last night and this morning, KPAC Classical Music.  It was beautifully spacious sounding and I was relieved of the chore of choosing music.  It was below freezing cold outside, but much like paradise inside with the living room freshly cleaned after a party on Sunday night.  Heated and humidified with Windtricity (TM, wind farm electricity purchased through municipal power company) as that is the only kind of electricity I buy (helping to make Texas #1 in both windpower and sonic excellence).

The fun is back in living room system now.  I think the fun had gone by December of last year, when accumulated adjustments had crushed the deep bass and neutered the supertweeters.  Now I've restored the bass, mainly by cutting the subwoofer crossover point way back down to 85 Hz, allowing the level to be raised significantly.  And I've restored the supertweeter levels to a very high level, attempting to use a lower  augmentation frequency, but now almost back to where it was before.  And I've moved the listening spot 6 inches closer to the speakers, almost exactly giving me a "perfect" 60 degree listening angle.

You don't appreciate how a wider stereo image sounds just by thinking of hearing sound at the periphery.  The main virtue is being able to hear more clear separation near the middle.

I noticed little if any sound coming from the supertweeters (15.5kHz crossover point, but you can hear lower frequencies if they are loud enough) on the FM, probably mainly thanks to the modified XDR-F1HD tuner being so incredibly quiet.  It occurs to me it would be nice to have remote control switch to control supertweeters (on/off).  That would require either hooking a serial controller to the Behringer crossover, or using something to switch the line or amplifier level signals, such as my ABX test box.  I have a wide range of line level remote control gizmos, but unsure if I want to put any old piece on the high rez supertweeter line.

Reasons why omni ribbon is hard to integrate

Last night read review of latest Magnepan 1.7 in The Absolute Sound.  There was much discussion of how pure aluminum ribbon tweeters don't integrate well with planar magnetic panels.  The new Magnepan 1.7 is said to be an integration champ because all of its drivers are quasi-ribbon planar magnetics (like ribbons, except the aluminum is bonded to plastic diaphram).

Now I'm having integration difficulty with omni ribbon tweeter and electrostatic speaker with characteristics similar to the Maggies.  Electrostats in general are considered to mate better with ribbons than almost any other kind of driver except for other ribbons.  But it's still not perfect.  Why not?

The first thing another audiophile might look to is the omnidirectional radiation pattern of the ribbon tweeters that I am using.  However, IMO, the omnidirectional pattern fits reasonably well with the figure 8 response of the electrostats, and may even be helpful in filling in the gaps that would otherwise be perceived as making the speaker sound beamy.

What I think about a lot is the possibility that the ribbon has much greater headroom than the electrostatic driver.  It can handle up to 500 watts, just in the range 10kHz-35kHz.  As a result of having much greater headroom, it sounds more open and adds an open "effortless" quality to the sound it augments.  However, as a result of having more headroom, it has less compression.

I'm beginning to wonder if part of the problem is that the ribbon tweeter is *too good*, and it doesn't fall back behind the panels sonically because it doesn't compress like they do.  The integration difficulty stems from the fact that the panels just aren't as good.

This almost suggests that using the supertweeter in "augmentation mode" (where there is no crossover on the main speaker) may not be the best way.  Perhaps the panels need to be crossover over for best integration.  (I am still not planning to do that.)

Rolling off the treble for accurate reproduction

For some reason, this appears to have become nearly ubiquitous.  Nearly all loudspeakers intended for music listening roll off the high frequencies somewhere above 5Khz.  By design.  Loudspeakers which are intended for professional uses such as monitoring recordings or mixing sessions are called "monitors".  That designation is supposed to indicate (or at least suggest) more extended treble response which was not rolled off by design for more pleasant casual music listening.

I also came across this when setting up my TacT 2.0 RCS Room Correction System.  While the Tact does permit you to set up your own custom room curve (a feature I always use) it also comes with a number of pre-configured room curves.  These "room curves" represent deviations from flat measured response that are considered to be more listenable than flat response.  Nearly all, if not all, of the preconfigured room curves have treble roll off starting well below 20kHz.  I have always chosen to extend frequency response flat to 20kHz with a custom curve.



Here is the "optimum" room curve as suggested by Bruel & Kjaer a few decades ago:

Note that it shows a very gradual roll-off that accumulates to 9dB reduction from the midbass to 20kHz.

One rationale given for this is that small rooms reflect high frequencies more than the larger rooms recordings were made in.  Attempt to replay music recorded in a larger venue will result in brighter-than-original sound.

A second rationale sometimes given is that recordings are mixed with speakers having typical roll-off.  (However, quite likely recordings are mixed with "monitor" speakers not having typical roll-off used in home loudspeakers.)

Anyway, this suggests one explanation for why even though I have set my Elac 4PI II supertweeters at a measurably flat (with midrange) level, the sound may be too bright.

The second explanation is that my overall system response, with or without the supertweeter, is tilted toward the extremes, below 50Hz and above 10Khz.  I know of no better way to fix that than with the Tact RCS, which I hope to get to re-calibrating this week.