Added sweeteners

April 13, 2017
 by Paul McGowan

Every audio company has a design philosophy they ascribe to: ruler flat, textbook perfect, back of the napkin, high feedback, low feedback, no feedback, tubes, solid state, hybrids, rich with harmonic distortion, squeaky clean and lean.

PS Audio has always focused on honest engineering offering musical truth.

Our digital wizard, Ted Smith, is one hell of an advocate for our design philosophy. His latest creation, a new operating system for DirectStream and DirectStream Junior—officially named Huron—will be released in a few weeks. Huron (a free upgrade to all owners) will soar to new levels of sonic truth and musical honesty. But Ted didn’t resort to tricks, or artificial sweeteners to achieve his design objectives.

Here are two scope images showing some of the benefits of Ted’s work on Huron. Note the lower noise floor of Huron relative to Torrey’s (the latest firmware). The first image is Torreys, the second Huron. (the spikes are measurement artifacts and should be ignored).

On the subject of design philosophy, Ted writes:

“There certainly won’t be any artificial sweetening of the sound in any release, all of my experience still indicates that the closer to theoretical/mathematical perfection we get the more life-like and involving the sound is.  Those that wish for some other character of the sound are probably better off with warmer cables, a tube amp or preamp that purposely adds a little THD, etc.  It’s all too easy to assume that any one part of a system inherently exaggerates or modifies the character of the sound, but most of the time those modifications are a result of system specific interactions and therefore shouldn’t be “fixed” in the design of any particular component.  I want the DS to be as faithful and neutral as possible.”

Thanks, Ted. Couldn’t have said it better myself.

Subscribe to Paul's Posts

31 comments on “Added sweeteners”

  1. I take every further improvement as a gift. So far all updates were clear improvements to me.
    In opposite to sweetening, every add. openness in top end (which is already a strength) is very much appreaciated.

  2. The only difference seems to be a reduction in noise at inaudible levels at an inaudible frequency range. The audible spectra look identical. For the not technically gifted, that’s me, please explain why these spectra imply improved audible sound quality (if they do).

    1. They do. But you’re focusing only on a narrow band and there are a couple of things happening. Overall there’s a 3dB reduction in noise – maybe hard to see in these scope photos.

      Also, what you are looking at is the result of a number of factors: lowering jitter, removing power supply interactions, cleaning up problematic code areas. Those actions are what give us the audible improvements, the ultrasonic noise reduction is only a marker – and end result – that by itself perhaps doesn’t make sense, but taken in total, it does.

      Here’s some of what Ted explains as well:

      “Notice that the noise floor rises less overall and (tho it’s harder to see) that it’s about 3dB lower over the audio band. The spikes are artifacts of the measurement setup. The left half of the plots uses much less data so it varies more from measurement to measurement and any differences seen there aren’t as statistically significant.

      The other FPGA code changes are harder to measure, tho they are there mathematically.
      Better low-level linearity, especially for DSD. The audible differences should be on the order of the changes from Yale to Torreys (for those that did hear zipper noise in the Torreys beta). I have no idea how those changes compare to the changes caused by the lower background noise floor.
      More input jitter rejection and less jitter/noise generation in the FPGA. Perhaps we’ll get a feel for the differences here as we listen to the 20 compiles…

      As I was measuring the low-level linearity changes I noticed something I’d missed earlier when measuring the various Torreys betas: The earlier Torreys versions which had zipper noises (including Torreys A, and Torreys B) also had significantly more THD down in the -100 to -120 dB area that was independent of the input level. Because of the DSD noise floor, the differences aren’t always obvious with a steady state measurement, but they show up better with a slowly varying input frequency. That level of THD probably wasn’t audible when listening with the DS volume up above, say, 80 with normal level material, but it might become more significant when listening to very quiet material or using lower DS volumes, say 50 and below with quiet material. Those THD artifacts didn’t exist in releases before Torreys or in Torreys final or Huron, but they might explain part of what people heard between Torreys A, Torreys B and Torreys Final. A little THD does certainly make for a “fuller” sound.

  3. As a registered audiophilistine I can relate to what you say, but what you can’t hear can’t hurt you and it is a free upgrade. I will admit that in my youth, when I was a keen amateur designer and builder of audio gear. I would have been delighted with the improvements shown; relevant or not!

  4. This PS Audio credo of musical truth goes all the way back to the 200C amplifiers, which were uncolored. Combining those amps with a revealing PS Audio preamp, Oracle turntable, Grado cartridge, and Sequerra Pyramids provided as honest a system as you could find in the old days. A little light on bass, but breathtakingly open. In the digital world, that honesty is just as important. Thanks again, Paul.

  5. Amazing work. I can’t wait to see how this works with my BHK preamp and amp. Awesome! Rolling tubes, Firmware updates, and firmware options ( kinda like tube rolling ). Love it. I know of no other company with such a hoist of options! Awesome stuff.

  6. Playing golf (whacking a white pellet around a field, a good way to ruin a nice walk) is starting to seem more rational than consumer audio. For a hobby allegedly rooted in science, the posts so far leave me dumbfounded, but interesting from a psychoacoustic point of view, all the same.

  7. Congratulations, Ted and Paul, for continually improving the reconstruction of music from a AD-process corrupting the smooth analog signal catched by the microphone by digitising it. However aren’t there files in audio where the potential of sound improvement could be much bigger? MQA, XTC, active speakers, digital connections based on glass-fibre cables, better power supplies (ultra-caps based) etc. ???

    1. Well we can’t really give people active speakers, glass interconnects, new power supplies, etc. with a free software update.

      I’d argue that the changes we’ve made in the past with FPGA updates has, at times, been more noticeable than replacing copper with glass, etc. We also know that putting MQA into the FPGA is bad for sound quality (tho we are planning to implement MQA in the Bridge so that base will be covered.)

      Ultra caps in power supplies are a mixed bag. They have much worse energy density than batteries (so can only be used for lower current loads) and they are noisier than capacitors so they fill a narrow band of needs.

      1. Ted, there are a lot of audiophile companies working on improving DACs these days. Is there a commensurate effort on the pro-audio side of the fence to develop better ADCs?

      2. Ted,

        have you really compared copper vs glass?
        I mean a real ST type glassfiber interface not TOSLink plastic shi…
        There are some fine interfaces from HP for instance which in my experience
        have clear advantages, both measurable and audible.


  8. Striving for perfection is commendable but it seems that it will remain a goal to strive for. There are so many imperfections ( distortions) that neutrality in one area reveals the distortions in other areas which were less obvious when things were not as neutral. As long as total neutrality in every aspect of sound reproduction is not achieved balancing the imperfections so that musicality is achieved seems to be the only sensible route. After all music in real life is pleasant and that is what people want from their music systems barring the numerically obsessed. But one in reality can not hear numbers or graphs. Better a musical sounding component with imperfections than an unmusical one with little or no imperfections. After all it is musicality that we are after measurements are only one of the means to an end and not an end in themselves. Just my point of view. Regards.

    1. These changes are based on feedback from previous work doing similar things to the FPGA. People have lauded any lowering of noise we’ve done in the audio band and also have noticed any lowering of noise in the ultrasonic signature. Similarly each time I lower generated noise/jitter with work on the FPGA people hear the expected differences in soundstage, etc. Even JA at Stereophile has noted that tho he can’t measure some of the changes I’ve made, he can hear them.

    2. That’s an interesting post, but makes cleaning out the Augean stables seem like light work compared to perfecting your audio system.

      It does, however, beg the question – what is “musicality”? People go on and on about it, it’s just about the most subjective term it is possible to image. As a rhetorical question, I would suggest that “musicality” might be the ability of an audio system to give a flat response in as wide an arc as possible in the range of say 80-1,500Hz. That covers the vocal range and much of the orchestra (excluding harmonics, obviously). It’s probably most of what we hear.

      Perhaps that it why people like me, who have more modest speakers (in size, but not modest performance) that use a mid/bass driver that goes down to around 40Hz (hence no mid/bass crossover issues or major bass room issues) are more easily satisfied than people with complex speaker systems going down to 15Hz and all sorts of resultant room issues.

      1. I like to judge changes in the code by watching whether other people’s feet tap when they listen… To me it don’t matter how it measures if people aren’t enjoying what they hear.

        1. Commendable suggestion, but people will tap their feet to the most distorting of valve amplifiers and speakers with an Alpine frequency response.

          Try this DSD recording:

          You will have to be a bit more scientific as it is impossible to tap feet to the opening movement of Shost 2. More to the point, it is so quiet (to the point of thinking the download was faulty) that it will test the noise floor of your software.

          1. Hmm, making measurable changes that get closer to a ideal system and then verifying that people like the changes they hear and giving them away for free isn’t scientific? (Further, if someone doesn’t like the changes they can skip the update.) Oh well, you can’t please everyone.

            1. No complaints about free upgrades. They seem to be de rigeur these days. My audio systems run on software from Apple, Auralic, Devialet, QNAP and Audiolab. They are a mix of bug fixes, performance upgrades and features upgrades. For example, the latest Devialet upgrade made the USB socket 2-way. As the unit digitises phono inputs, this allows vinyl to be uploaded in the same way as the PSA phono converter ($2,000 in the UK). Auralic provided a free server app, plug a USB hard drive into any unit and it becomes a NAS drive. Linn also released a free server app that does the same. Devialet are developing a new networking card and operating system, retail $1,500, a RTB upgrade, being given free to people who bought the current system from July 2016. The Linn streamer/DACs seem to have had the most upgradeability, most of which is free. This seems to be the trend, rather than having to pay for upgrades or incur the expense of buying new hardware.

              1. I think free sound quality upgrades are not usual, feature upgrades probably yes.

                Usually the companies don’t let their developers spend months on sound quality upgrades no one would ask for inbetween if they were just sold with the next hardware for 2000$ more a piece.

                With features they have competition.

                Every bit of sound improvement is what the high end community pays for the most. So for me it’s still very special, this is free.

                  1. Surely we all don’t know what they do in detail, but I still read of functional updates. For sure they also make bugfixes, but there’s no indication they do major sound quality updates, too.
                    I guess they’d mention if they did and reviewers would comment it.

                    1. I think quite the opposite. To me, saying the conversion software is improved implies it wasn’t that good in the first place. Improvements are important if people can hear them, but not if they can’t. New features need to be publicised so people know they can use them!

        2. This is what you achieved in first place!
          Even before I first time realized the incredible transparency, ambiance, layered soundstage and open top end, I was shocked that the DS gave the impression of doubling the pace related to my previous two times the price one box player and how much better it shows rhythmic nuances.

          Who wants to experience, listen to Emilie-Claire Barlow/Live in Tokyo, Track “Waters of March”, how she generously decelerates/accelerates towards the rhythm section and how this grooves. Listen to this track, it’s fantastic.

    3. Oliver, an experience I had in using two different $8k phono cartridges showed me the importance of keeping the front end as clean and silent as possible. (I can’t afford either of them, but it was a wonderful lesson to learn.) You can always adjust the system downstream from the source to achieve your particular preference for musicality, but you can’t add resolution and quiet background (or reduce surface noise, in the case of LPs) if it isn’t already present at the front end. In my own experience, I bought speakers that were far superior to my other components, and I’ve spent a decade improving the weaker upstream links (and power supply.) The trick is to have your system be qualitatively in balance from component to component at whatever price level you can afford. And as you’ve said, when you find that musical sweet spot, stop sweating the details and enjoy your music. It sounds like you are already there. Cheers.

    1. The tests were with an Audio Precision which generates (in this case) it’s own test signals. They essentially match my tests with a 16 bit scope and Adobe Audition to generate test signals. Noise floor, THD, frequency response, etc. tests aren’t very sensitive to the source (at least for the DS.)

Leave a Reply

© 2023 PS Audio, Inc.

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram