New DS measurements in latest Stereophile


#1

And guess what? They show the improvements that we’ve all suspected and were expecting. More proof that Ted and the whole PS Audio team aren’t just shaking bits around to see what comes out. Great job!


#2

When you say ‘they show the improvements’ do you mean the Stereophile follow-up articles?

I tried to find updated measurements online based on your thread heading, but with no success…(?)

Are there any in the paper version?


#3

The paper version (March 2015) has a short followup by JA with A/Bs of “spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with 24-bit data”, “spectrum with noise and spuriae of dithered 1kHz tone at –120dBFS with 24-bit data” and “spectrum of 50Hz sinewave, DC–1kHz, at 0dBFS into 100k ohms”. He also shows the “HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms” of 1.2.1 with some comments.


#4

Thx, I just downloaded the digital version.

How does the PP compare with 1.2.1 on the measurement side of it?


#5

I didn’t do any measurements of the kind that JA uses on PP, I don’t think PS Audio did any either. I checked the obvious things like channel alignment, absolute polarity, Left vs Right channel correct with all possible input rates and formats. Levels consistent with each different input rate and format… Apparently I didn’t check the left / right balance control. When we get that bug fixed I can probably spend the time to do some of the JA tests.


#6

If I understood correctly, a new revision of the OS doesn’t necessarily build on the penultimate one (for good or bad).

Is this correct, because in this case things may take a step for the worse on certain areas (e.g. some measurements)?

Does PP build on 1.2.1?


#7

No, I wouldn’t say that. The average SQ differences in a set of compiles are small compared to the (measurable) ones we got in 1.2.1 and now in PP. There seem to be some stinkers and some better than average compiles in each release so there may be some overlap of quality from release to release but again, the average sound quality in each set of compiles has been getting better when we make things technically better.

I can take any set of FPGA code and constrain the router to make it suck. And I can similarly constrain the router so most of the compiles sound the same. But in general I like to let it run wild and see if we find a new best sound quality. But the key differences I make are explainable and purposeful (tho sometimes bigger or smaller than I expect :) )

If it were simply a random walk from release to release then we’d “merely” compile up a couple of hundred versions and find a beaut for all time.

The differences that 1.2.1 made were quite measurable and audible - I don’t claim all of the audible differences were measured, but, for example we all know that a quieter background makes a noticeable difference and we all (I assume?) heard that in 1.2.1.

My ears tell me that PP has yet a quieter background and I expect that I’ll see that in the measurements.

Anyway PP has all of the enhancements from 1.2.1 and in general the things we’re doing in the FPGA should make the sound quality better. As I fix bugs or add features I also take the opportunity to tweak the code for better sound quality based on my previous experience. The differences in PP probably wouldn’t be as obvious if I backed out the 1.2.1 changes.

If I have some time this weekend I’ll try to do some measurements, but it often takes the bigger part of a day to rearrange everything around the bench to do these measurement.


#8

Ted/Paul;

What are your impressions of the last Stereophile measurements - do you think those satifies the ‘measurement-mafia’ from a marketing standpoint?


#9

I’m not sure - I think JA’s tone has changed a little, but that could just be me. Still there are a certain set of people that will always poke at an architecture they don’t like philosophically. And some others who are big fans of what they see instead of what they hear. I also have some friends who have always loved CDs over analog or SACD, often they are very critical of the rising noise floor in SACD (tho there’s nothing there at all in CDs, so I’m not sure they are being rational.) They don’t understand what 1/3 octave sliding filter is telling them anyway. I’ll let Paul talk about the marketing angle.


#10

Measurements are Internet fodder and the source of argument among enthusiasts who listen with their eyes.

My expectation is also that those who were critical of the DS based on the previous measurements will continue to be irrationally critical; it is of their nature.

PS Audio publishes few specs. I doubt PSA will ever attempt to use detailed measurements for marketing, preferring to let their equipment do the singing.


#11
Elk said

…PS Audio publishes few specs. I doubt PSA will ever attempt to use detailed measurements for marketing, preferring to let their equipment do the singing.

If they have a good set of measurement numbers they should be using them in their marketing, good numbers do appeal to a portion of the market. Actually that is the first thing I look for when considering a new purchase. The performance numbers, and the price numbers. Equipment singing is useless to me when deciding what to buy. Bad numbers and I certainly wouldn't buy it.

#12
brodricj said Equipment singing is useless to me when deciding what to buy. Bad numbers and I certainly wouldn't buy it.
You do not audition equipment before a purchase? You rely on numbers instead?

With respect to a DAC, what numbers would you like to see?

What manufacturers are providing these numbers?

Which numbers truly correlate with good sound so you can judge by the numbers?

I find the numbers fairly useless. All DACs have flat frequency response, fairly low output impedance, good broadband S/N, low distortion. There are differences to be sure, but I certainly cannot listen with my eyes and tell you how each will sound based on specifications.