More Imaging Improvements, need to get both level and delay correct

Previously I determined I needed about a 0.1 ms delay added to the right channel to bring the center image into the exact center position.  It's funny because the speakers look to be exactly equidistant, theoretically I could achieve the same result by moving the right speaker about 1.5 inches back, but it's already as far back as it can go because of the subwoofer and other equipment being in the way.  And it's strangely hard to dial in the correct position acoustically by moving the speakers which is very difficult to do especially considering how very important the angling is too.

It was also very hard to come up with the 0.1ms delay number itself, if I just play with the delay setting with the remote control I almost invariably came up with larger delays, which then were later found to introduce other issues.  The 0.1 ms delay is the minimum which causes the image center to move at all, and that's what the correct setting seems to be as it doesn't introduce other issues, and then somehow it seems right afterwards if not during the adjustment process itself where the ear keeps saying "more, more" until everything goes to hell.

The testing involved listening to mono'd sources and trying to get everything in the phantom center.

Getting the image exactly centered is surprisingly important in creating a holographic image.  Just a little bit off and and you have ping pong stereo with no depth.  Of course you cannot with multiple listeners siting side by side...that is one flaw in acoustical stereo.  You can get correct centering for multiple listeners only if they sit in front and back of each other, but that leads to other issues.

I don't like center channels either.  While that is the obvious solution if you MUST have multiple listeners side-by-side with at least some semblance of image centering for all, center channels always sound unnatural to me when I am at the sweet spot myself.  When you have a correctly centered phantom center, you have depth, width, and everything, with a center channel you just have left to middle to right, end of story.

But when I did this testing some weeks ago, and posted about it here, I did not at all try to equalize the loudness of the two channels, which has to be done with a different kinds of test.  Getting the loudness equal (and not influenced by delay) requires playing the same signal in one channel, then the other channel, and adjusting until they have the same loudness and general sound.

I did this playing pink noise only in one channel, then the other channel.  I'm shocked that there is no test CD with this kind of test.  I had to make one by taking a pink noise recording (from Stereophile Test Disc 2 IIRC) and then using Audacity to mute one channel or the other and making a new audio file from that.  Sadly I cannot distribute this test because it uses Stereophile's original "content."  I will someday have to regenerate it using the pink noise feature of Audacity itself, and that will also involve determining if that is as good as the pink noise from Stereophile, so I keep putting that off.

I suppose most audiophiles use an analog preamp with a balance control you can swing from far left to far right so they don't need such test signals.  But not only do most digital preamps lack such a feature, many purist preamps don't have balance controls either, and some that do don't allow you to swing from far left to far right.

The pink noise in each channel can be measured objectively with an SPL meter with "C" weighting.  You do not want to use the more well known "A" weighting, because that is for noise which is the stuff you don't want to hear.  The "C" weighting is for Content, the stuff you are trying to hear.  A flat contour gets it wrong too.  I use an SPL app on my iPhone to do the measurements nowadays, which gives me high accuracy and digital readout.  I use the slowest available response on my app, which is Slow and then do a mental averaging of the number.

I was shocked to find the right speaker was about 1 dB louder.  I wondered if it was because I'm using two vintages of Acoustat 2+2's on either channel.  But actually, I had previously attenuated the left speaker by 2.3 dB in the Levels menu of my Tact 2 Digital Preamp.  I had long forgotten about that.  I think I was trying to compensate for the delay difference by only using level adjustment, which is impossible.

I got the best result dialing back the attenuation to 1.0 dB.  Then the speakers measured the same loudness.  And when that was done, I also noticed the speakers sounded much more similar.

I then tried re-adjusting the delay in the left channel.  But the best image centering still seemed to occur with 0.1ms delay.  It seems like delay is the more powerful localization mechanism, so even if the levels don't quite match (within one dB anyway) the center image position is almost entirely controlled by the relative delay.

But getting the levels matched better than one dB, and with the correct delays too, it greatly improves the depth and realism of the image, I determined in subsequent listening.

Current settings:

Delay: 0.00 ms Left, 0.10 ms Right

Level: -1.0dB Left, 0dB Right

This was in center back seating position of the room, I think it should work in the closer up position too.  Sometime in the past few months I had eliminated ALL EQ's from the Acoustat feed because I figured they were not well optimized for the back of the room (and maybe not for the more frontal listening position either).

The overall response seems fairly flat until you get to midbass frequencies 200 kHz and below where there are some dips, those are problematic to adjust because I'd need to add "boost" which I try to avoid as it is headroom lowering.  The subwoofer feed has several EQ's to deal with room modes, but the subs are cut off with a steep FIR linear filter above 120 Hz  (FIR linear filters are digital magic which introduce no phase shift) so the subs can't help with dips above 120 Hz caused by back of room reflection and dipole bass loss.

Though I've done this in the past, I generally do not like leaking some sub base above 120 Hz to "help" the panels.

The left and right channels measure about the same from 300 Hz on up.  They sounded very different before the level adjustment, after the level adjustment they still sound a bit different.

I need to dial in EQ's better AND measure and adjust bass and panels separately, and then go back and check levels again.

After any kind of EQ adjustment where channels are equalized differently, you need to re-measure channel loudness as I just did, using single channel pink noise and a C weighted SPL meter, and not rely on image centering which is more controlled by delay when the channel levels are already fairly close.