The physics and limits of relative loudness perception

Oops, I meant moire patterns rather than diffraction gratings in my last comment. 

Tubes are another one of the more common sources of imbalances - so it might seem odd you got the imbalance after going from tubes to SS - unless you had an imbalance in the tubes that was actually offsetting the room

@mulveling Thanks for all your debugging tips -- lots of good info. Luckily in my case there was no underlying physical problem. It was just signal processing with the one channel’s 0.8dB gain. I found it very interesting that going to SS provided more precise image focus so that the problem was unmasked.

I can hear very subtle image shifts with as low as a .2 db change in relative balance, with .5 db shifts being clearly discernable.  This surprised me because I cannot reliably hear a full 1 db change in overall level. 

@larryi Yes, this is what I was really asking about in this post! Clearly, my setup context was way too detailed, but this is very surprising to me, too...

I find it very interesting that volume deltas that we cannot hear on their own become easily discernible when hearing a "downstream consequence", for lack of a better term, like image shifts. I think it is the auditory analog of diffraction gratings.

I can’t seem to find any research on the relative acuity for different kinds of sonic phenomena, like volume vs. image shifting. Anyone know of any? 

It makes me wonder if, for example, measures of image shift could be used to create some quantitative metrics that could describe what are currently qualitative descriptors like imaging and focus. Or perhaps to gauge whether the gain factors of initially-matched tubes are drifting apart over time. You might not notice based on other sonic qualities, but a periodic "image check" could help you make a determination.

When I have my Levinson No. 32 preamp in my system (which allows for very small incremental changes of volume and/or balance by remote control), I can hear very subtle image shifts with as low as a .2 db change in relative balance, with .5 db shifts being clearly discernable.  This surprised me because I cannot reliably hear a full 1 db change in overall level.  I consider balance control an necessity in a system.  My favorite linestage is a tube unit I own which has a balance control, but not with such fine steps.  Still, having some control is WAY better than not having such control.  This is particularly true with tube gear where level differences are much more likely to occur because of slight differences in tube gain and effects of aging.

Left-to-right channel imbalance is very noticeable at 1dB, and can be annoying even at 0.5dB and a bit below. Yes, I’ve had aysmmetric rooms cause it, usually perceived as between 0.5 - 1 dB. But it can also be caused by components. Of course, sometimes it’s not just a flat SPL difference - it could be specific frequencies or phasing / timing issues that enhance our perception of the bias.

Since I’m more analog based and generally don’t use DSP, I’ll swap L/R cables along certain components to allow imbalances to cancel out where possible - I don’t really care if my absolute L/R channels end up flipped; music is not like movies. BIG difference if you let 0.25 + 0.25 + 0.25 = 0.75dB versus leveraging cancelling to net 0.25.

Suspect cable contacts / connections can also *easily* cause imbalances. Keep them clean, and periodically ensure you don’t have a cable with internals going bad. I think one of the more common culprits is those litttle DIP switches often used in phono stages. I HATE those, ugh. Tubes are another one of the more common sources of imbalances - so it might seem odd you got the imbalance after going from tubes to SS - unless you had an imbalance in the tubes that was actually offsetting the room (I’ve had this happen)!! Cartridges too, though not as common as you’d think - I’ve had some 40+ MC cartridges, and precious few with notable imbalances. Speakers too. 

I think the old "1dB is the minimum discernible by humans" is imperfect science. 1dB of absolute level might be negligible to casual listeners, but we’re not that. And 1dB of imbalance is far more discernible than 1dB of absolute level. This is surprisingly effective.

Ooof, channel imbalance has driven me crazy at times. I’ve thought about having dual (mono) stepped attenuators with say 0.25dB steps to allow for balance correction - e.g. from Khozmo or Goldpoint. Some preamps will throw in a balance pot but it’s inevitably one of those awful sounding Alps "Blue Velvet" RK27 things which is a no-go for me. 

@sfgak glad to see you've gotten to the bottom. And no one has golden ear in this space full of people nearing or in retirement age zone who can't hear pray 10kHz effectively for the most part. Hell music itself lacks any hearable content of good volume past 14kHz so don't worry about not detecting ±0.5dB differences 100% of the time. Happy listening

I found this interesting site: https://www.audiocheck.net/blindtests_level.php?lvl=0.5

I can, with 100% reliability, distinguish sound levels of +/-1.0dB. But only about 60% reliability for +/-0.5dB

Clearly no "golden ears" here 😬

@kofibaffour 

Thank you for the detailed response and the links to Erin’s videos. You provided me with a better vocabulary to describe what I experienced. Before I reframe that, the speakers are DALI Epicon 6, with a pair of asymmetrically placed Rythmik F12G subs. They have been carefully set up, including active crossover with the subs, and phase matching of the signals at the crossover points. All this can be viewed on my Virtual System page.

The facts were:

1. With the 0.8dB (unintentional) right channel gain boost and the Primaluna amp, I did not perceive any noticeable change in the location of the center vocals.
2. With the Apollon swapped in (no change to speaker location or angle), and still with the 0.8dB right channel boost, the vocals shifted perceptibly to the right
3. With the Apollon still in, but removing the right channel boost, the vocals shifted back to the center, where they should be.

Based on the new vocab you have given me, what I think was going on is:

1. The Primaluna tube amp has good imaging, but relatively poor focus. Very likely, the 0.8dB did shift the image of the vocals in the Primaluna just like in the Apollon, but with a spread out focal point, some of the location data overlapped with the center phantom image, and so the voices were "close enough" to center that psychoacoustically, the brain puts the vocal image in the center.
2. The Apollon has good imaging too, but better focus than the Primaluna. Thus, with the 0.8dB right channel gain, no part of the image of the vocals overlaps with the center of the soundstage, and thus the perceived shift of the focal area to the right.
3. I suspect that, if I were to swap the amps back, now knowing what to listen for, I would perceive a slight shift in the vocals using the Primaluna, but not to the degree of precision / noticeability of the Apollon. 

I find the psychoacoustics interesting, in that the shift wasn’t really perceptible with the Primaluna, but was very apparent with the Apollon. 

I was pleased with the soundstage and imaging before, but even more pleased now.

@052rc Thanks, but this does not appear to be what I experienced. I used SE connections from pre the Primaluna, and balanced to the Apollon. But both connections were the same type on each channel of the respective amps, and the gain switch was set identically for the two channels. Besides, the SE to balanced switch would result in a gain of 6dB. Here, the difference was 0.8dB, and was (unintentionally) introduced and later (intentionally) removed by software. It was this 0.8dB difference that shifted the image. 

Maybe the issue is with the new amp. If it has balanced connections along with single ended, you may want to start there. When a signal is balanced, the output increases approx. 4-6 db louder. Make sure you have both channels running either balanced or SE. If not, one channel will be louder and it will shift the imaging to the right or left.  

why do these posts always go on and on without mentioning the speakers used? I never understood it. Also, humans can hear 0.5dB and above to varying degrees of correctness provided it is a wide enough Q boost. if it's a narrow Q, nope, you won't hear 0.5dB changes

also for the imaging and whatever question, yes, a very ideally below 1dB matching across speakers, with the right control in directivity patterns, good room treatment, all lead to uncanny phantom centre and precision of images in the sound field

Reason why, the best imaging and soundstage I've actually heard is in a mastering studio of an acquaintance that has been acoustically controlled with  with Ci5160REFM-THX | KEF International as mains and two subwoofers - JTR Captivator RS1 — JTR Speakers

And it had such incredible depth, and the phantom centre was uncanny, and I mean uncanny. I've heard people talk about uncanny but never felt something like that before, had me doubting if they were messing with me and hiding a centre channel somewhere lol.

He really was proud of his system, and he uses that as both his general music listening and mastering setup. HEaring that cured me of any, "oh you need the speakers 4 ft off the wall or else your soundstage vanishes"

Cos most studios encode these soundstage attributes using in wall speakers or soffit mounted speakers.

These two videos are a good to watch if you have the time - https://youtu.be/FHjy0ZV0V7U?feature=shared and https://youtu.be/8qihRNB660M?feature=shared 

Watch the first and follow with the second, right after

Long-winded comment but your post brought back an awesome memory @sfgak