Circuit design for space-time sensitive ears
©2016 Loren Amelang
In 1978 I was owner of a stereo shop in Berkeley, CA which became the "last resort" source of repairs on equipment the other shops wouldn't touch. I had the opportunity to study and hear circuits that no mainstream manufacturer of "quality" equipment would have considered. Given my obsession with audible as well as visual space, I was listening for things the mainstream has as yet ignored. Briefly, the path I followed involved more DC coupling, more class A, faster semi's, less feedback, and simplicity. Finally I built myself a system which ended my love/hate addiction to stereo equipment, and moved off to be a computer engineer.
A properly "phased" phono cartridge
(comparable to the way one aligns a tape head stack so that the signals from multiple channels are not only "in phase" as opposed to 180 degrees out, but "in phase" as opposed to 2 or 3 degrees out.)
Requires a verified test disc and an oscilloscope, small screwdriver, and great patience to get the cartridge exactly square with the arm.
Continuously variable input impedance
on the phono preamp.
Roughly set to the spec of the cartridge, fine-tuned "by ear" for spatial clarity.
DC coupling the phono cartridge
to the first preamp device.
I used semiconductor op-amps which avoided voltage offsets.
Actually, DC coupling through the entire system, all the way to the speakers.
Only two capacitors per channel in the entire system - tiny monolithics in the RIAA loop.
Class A amplification through the entire system.
I find this provides the clarity people are seeking with single-tube circuits, while allowing my DC coupled symmetry.
ZERO amplification stage feedback, no negative or positive feedback.
The measured power bandwidth of my system is DC to roughly 2 megahertz...
All circuits powered directly by separate local batteries, with no capacitors across them.
Bias voltages never provided by RC networks or even semiconductor regulators.
Except for the final output stages, individual + and - batteries per channel.
Separate batteries for each stage, four for the phono preamp, four for power amp bias, and the two 75 AH Optima D31T SLA batteries for output power.
Basically, no opportunity for signals from a later stage to feed back into an earlier stage. (I suppose separate output power batteries per channel might be even better, but have never had money or space in my living room for four car-sized batteries.)
Sufficient conductor size for each power source and level.
At the output stage, AWG10 wire less than 24" from batteries to transistors, with a high power relay for power switching.
No stages self-balancing.
Phono preamp and power amp separately hand-tuned for zero balance.
The power amp looks more like a servo driver than an audio device. (It took me a long time to believe this could make a difference, but it does. Taking the frequency response down to zero adds realism even if none of the signal sources have information anywhere near zero.
The exact zero point does not seem to matter, I set it by eye using LEDs flashing with the output power. I do not notice small deviations from zero affecting sound or sound space.)
Obviously no "tone" or "loudness" controls, or any other non-linear stages.
Remote input impedance and separation controls for the power amplifier.
While individual inputs have their own basic input impedance adjustments, I've found it critical to be able to "fine tune" the input impedance remotely while I'm out in the sound field.
A remote separation control (actually just "blend", I have not created a matrix that can exaggerate stereo separation at the line level) seems to interact somewhat with the impedance tuning.
It is eerie how getting this "just right" creates a sound field you can realistically walk around in. And that is my goal - to avoid the "sweet spot" limitation, to get up from the designated chair and move around in solid sound space.
Included with these remote controls is the matrix adjustment for the rear two speakers, which can take them from parallel with the front speakers to being a single "difference" output. This is not essential to my concept, but is a very nice addition when properly adjusted. It is more sensitive to program material than the rest of the adjustments.
With my current single subwoofer and tiny "satellite" speakers, I do not notice the "cone" effect, the mid/high frequency sources are too small and flat. But with previous speakers where the midrange originated from an 8"+ cone, I found the space was clearer when I mounted the speakers "inside out" - with the backs of the cones exposed, and pointing up, 90 degrees from my listening area. I then mounted the tiny tweeters on the tops of the exposed woofer magnets, so the highest frequencies originated from (roughly) the top point of the woofer cones. Sort of a poor imitation of the old Lincoln Walsh "Ohm" design.
(Had to Google to remember that name... Wow - looks like "inverted cone" speakers are now a "thing"! I don't pay much attention to the audio market.
A good example of my almost no-budget equipment... My current speakers used to be a Paramax P6 system, something made for the "white van" ripoff market.
The guys in the (black) van in the grocery store parking lot said it was worth $4K. I ended up giving them $300, which was just about the eBay price plus shipping. The advantage is I had no hesitation about hacking the system to serve my preferences. First modification was removing the crossover components, second removing the perforated metal grilles from the satellite boxes. Even cloth grilles wreck the sound space...
Which illustrates an important factor about what I've described here - these principles regarding sound space are independent of what is normally thought of as "high fidelity" accurate reproduction of the frequency domain. Somehow my brain works in the time domain, and is willing to tolerate less-than-ideal frequency response. I've found many "audiophiles" completely lose interest in my ideas as soon as they discover what limited equipment I own. It might be fun to have more expensive toys to hack, but I'm too content to bother.
I've also found that most "audiophiles" do not notice any advantage to my system. Which makes sense to me because while I was developing it, I found that my testing was one-directional. If I tested an improvement in clarity first thing in the morning, before I had heard any other artificially reproduced sound, it was very obvious when I switched to a less clear signal path. But having heard the less trustworthy sound space, my brain stopped evaluating the now degraded factor, and refused to trust it even if it was fully restored. It took me a long struggle to accept that traditional A/B testing could not be used at the level I was studying. It also made progress very slow, because listening tests were impossible once actual work on ordinary equipment had begun.
I tried several of the past "quad" systems, and except for rendering a few special effects, I found them hopelessly inferior to a simple matrix. I'm not sure the rear speakers add that much to the illusion generated by the rest of my tweaks, but when properly adjusted they add a nice touch. I found the two-channel Dolby formats (even locally calibrated) so gross I've never even tried their 5.1 system. Or any of the more modern multichannel systems.
In reality, most of my listening is now from the internet, via Wasapi or ASIO to USB to S/PDIF to an ART DI/O DAC (which was the favorite of USENET audiophiles back in the day). Somewhat like my designs, they provide a separate (but semiconductor regulated) power supply for each internal stage (I haven't bothered to convert them to individual batteries...)
My conclusion is that resampling (as in Windows KMixer, or Flash SoundManager 2) is the main enemy of a convincing sound space. Similarly, when I had the original Sony DSS (the format that became DirecTV) I found that music recorded from it as a WAV file sounded acceptable but if re-compressed into an MP3 it lost its sound space no matter how high the bitrate. When I can avoid resampling, I find the sound space of 320K streams or even YouTube to be about as convincing as the other aspects of the compressed sound. I have no experience with the sound output of modern Macs...
I once had promo access to WiMP, the Danish/Norwegian streaming service which provides lossless FLAC, and the quality through their dedicated player with ASIO output was delicious. Unfortunately $199 Kr is over $36/mo, which I can't handle, so I've fallen back to Qobuz 320K which is only 5 Euro. Their dedicated app with ASIO output (their Wasapi doesn't work with my hardware) is essential, the same streams through a web browser are miserable. I've not heard the US "Tidal" version of WiMP...
At least at the time I explored the much-hyped Beats Music that Apple bought, they were streaming the exact same tracks (same URLs!) MOG provided, but their player ran them through SoundManager 2 before you could get to them - totally destroying the sound spaces. There was a vocal minority of former MOG fans who noticed, but most people including Apple execs seem to have bought the hype. Guess that's why Dre is a "billionaire" and I'm not...
Revised 19 June 2016