@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
The physics and limits of relative loudness perception
I have recently been experimenting with the combination of tube pre / Class D power amplification; previously I was using all tubes. An interesting thing happened.
I switched from a Primaluna with EL34s to an Apollon Audio implementation of Purifi Eigentakt, to see what it would sound like. There was the immediate "this sounds kinda clinical" reaction, where the warmth of the power tubes was replaced by the more analytical sound of {solid state, Class D}. This was neither good nor bad itself, but a difference in presentation that took some time to adjust to. I also experimented with my HQPlayer settings (sent to a Pontus II OG), and arrived at a very pleasing tonality [comment: changing HQP filters and dither can alter the sound significantly, and was a lot simpler than auditioning 13 new DACs]. I found that, among other changes, the soundstage became more precise than with the Primaluna.
But... there was one very annoying thing: Singers who previously were front and center were now slightly off to the right. This was directionally consistent across multiple albums / singers / genres, although the exact degree of shift varied somewhat. I was racking my brain trying to figure out why. Turns out, when I was in my previous house, the system was set up in a non-rectangular room, and a previous set of adjustments using REW and HQP had resulted in the right speaker getting a gain boost of 0.8dB (just gain, no phase adjustment). It was on a different page of the HQP interface than the corrections for bass resonances, and I had missed it.
For a year in the new, rectangular, room, with the Primaluna, this did not appear to bother me. But it was glaringly obvious with the Apollon in the system. Once I found this vestigial setting and removed it, the singers swung back to dead center. Supposedly, our brains are not good at hearing under 1dB of sound difference, yet at 0.8dB this was very obvious. This leads to some questions, on which I hope those more knowledgable that I might help shed some light:
- Have there been any scientific studies done on how sensitive human hearing is to loudness in the determination of object placement in the soundstage? I.e., that you might not hear small relative volume differences, but these differences can be easily perceived via the imaging of the soundstage?
- Is it loudness, timing (i.e., phase), or a combination of the two that creates the sense of image?
- If so, is it a reasonable generalization that, at a similar price point / quality level, most (not all) solid state power amps will have better gain control or timing control vs. frequency than most (not all) tube power amps at that level? Or can different implementations of either topology have the better control of loudness/timing?
In any event, like several others in the forum, I am enjoying the combination of tube pre with Class D power. The HQP experiments with "warmth" have led me to order an LTA Aero, which is arriving in a few weeks. I think overall, the combo of new power amp / new DAC will be a nice improvement for my system. And it has been a great lesson in the importance of overall system synergy.
Any observations, references, or experiences would be much appreciated.
- ...
- 10 posts total
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. |
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. |
@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.
@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.
|
- 10 posts total