• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Phase Noise!

Purité Audio

Master Contributor
Industry Insider
Barrowmaster
Forum Donor
Joined
Feb 29, 2016
Messages
9,051
Likes
12,150
Location
London

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
191
Location
Australia
Makes absolute sense to me - phase noise is an easy way to measure actual, rather than theoretical performance, and from day 1 using digital playback I have always heard the very poor performance when cold, which steadily improves over a long time frame, which can be days long for some devices. Nice to see that someone has devised a simple, repeatable test for this ...

The subjective impression when cold is that the playback has poor resolution, much detail is completely absent, very lacklustre treble; overall, a very flat and non-involving sound; then, apparently correlating to this drop in jitter noise, with ongoing conditioning all these negatives steadily diminish and completely vanish in the best cases; it's a steady and slow process of continual change.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
How long did they let their "Symetricon phase noise test device" warm up for?

(I think the company is, in fact, "Symmetricom").
 
Last edited:

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
Looks likely to be true. Also likely such noise at its worst when cold would indeed be inaudible.

Also why would it be higher at 1 hour versus any other time?
 

dallasjustice

Major Contributor
Joined
Feb 28, 2016
Messages
1,270
Likes
907
Location
Dallas, Texas
I guess this paper is intended to help repair audioquest's long pseudoscience history? Whether one lets a DAC warm up or not, I bet nobody could hear the difference without knowing whether the oscillator had become stable.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
Also why would it be higher at 1 hour versus any other time?
I meant what I asked in my comment above: how long was the test gear powered on for? In the instruction manuals for various phase noise measuring test sets the performance specifications apply after a "30 minute warm up period" (for example). Presumably the test set has a similar time-related (or is that more likely to be temperature related?) characteristic to the DAC because it, too, will be based on a crystal-controlled frequency reference. If the test set was powered on at the start of the test we might be seeing its characteristics combined with the DUT. If it was powered on from cold intermittently for the measurements then that would affect the results differently.
 

Speedskater

Major Contributor
Joined
Mar 5, 2016
Messages
1,610
Likes
1,333
Location
Cleveland, Ohio USA
A "30 minute warm up period" is often part of the 'boiler plate' in the test equipment manual. That is, it's part of the standard text that is coped from one model and generation to the next.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
A "30 minute warm up period" is often part of the 'boiler plate' in the test equipment manual. That is, it's part of the standard text that is coped from one model and generation to the next.
No doubt that's the case. But it doesn't mean it isn't a relevant instruction in this particular measurement. 'A crystal oscillator' is a boiler plate component in test equipment that is copied from one generation to the next, too.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,498
Location
Monument, CO
Measuring phase noise accurately can be a challenge, and after a very quick skim IMO there are far too many variables in this to say for certainty that phase noise is worse when cold for all DACs. I noticed it seemed to spike up at 1 hour, what's up with that? Let alone how much it matters to the listener. I will note that RF systems and test equipment routinely uses temperature-controlled oscillators, often with built-in heaters, to ensure frequency stability. Oscillator topology is a big player. But phase noise is also heavily dependent upon the circuit design, particularly the oscillator's amplifier and biasing circuit. For my present work we measure it well enough in the open lab, but for truly accurate measurements I was in a screen room, or at least a well-constructed screen box, to isolate the device under test (DUT) from external noise sources. Of course measuring phase noise in the -140 to -160 dBm/rtHz range is a bit more challenging than in the -60 to -100 range...

I have used test instruments that specify from none to one hour to one day of warmup before use, the worst (longest) being a highly-accurate microwave instrument that specified 48 hours before reaching rated specs. Did it really take that long? Almost certainly not, but we had to follow the manufacturer's guidance to be able to certify our devices. These days an hour suffices for most equipment (better designs, better parts, better compensation methods).
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,529
Location
Seattle Area
I can't figure out how they measured the output of a DAC this way. Are they instrumenting the oscillator in dragonfly? It seems to be the case.

They talk about <1 picosecond phase noise after 24 hours. Here are the measurements of the Dragonfly by stereophile: http://www.stereophile.com/content/...da-converter-measurements#uKdkaMK4RyxO7krk.97

1012AQDfig11.jpg


1 picosecond would correspond to -153 db in above graph. Clearly the device generates a lot more noise otherwise. The static noise level is at -135 db which corresponds to 10 picoseconds. Low frequency jitter is evident in how the main signal has been broadened. That seems to extend to -120 dbFS which corresponds to more than 40 picoseconds.

In addition to those, there are peaks that are correlated to internal circuits that rise up well above the noise floor. These are more problematic than any random noise (although here they are too low to be of audible concern).

It is also odd that they don't show the standard output from Microsemi's phase measurement device. They are far more informative and correlate well with stereophile measurements:

upload_2016-10-6_10-10-54.png


Notice the important gray shaded graph that shows the instrument's own phase noise.

BTW, I did search for warm up requirements for the 5120a/5125a shown above and there is no warm up spec. Nor was there a reference in the manual. This is a clever device which measures its own distortions and subtracts them from input so perhaps it is more immune. The device has two inputs however, one of which is the reference clock. That clock will have warm up specifications. THere is no reference to what clock was used by Audioquest so we can't determine if we were seeing its characteristics instead of DAC's oscillator.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
Look at the input frequencies. They were measuring a 22.58 mhz crystal directly. Might not even be an audio DAC. Could be an instrumentation piece of gear for high frequency radio. You see such numbers with those sometimes.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,529
Location
Seattle Area
Look at the input frequencies. They were measuring a 22.58 mhz crystal directly. Might not even be an audio DAC. Could be an instrumentation piece of gear for high frequency radio. You see such numbers with those sometimes.
Ah, I didn't notice the frequency column there. So that explains that they were indeed measuring a clock oscillator and not the DAC analog/audio output. They do say this is from Dragonfly DAC though:

"For the purposes of providing valid measurements, a DragonFly DAC was used on a Symetricon Phase Noise Test Device."
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
Ah, I didn't notice the frequency column there. So that explains that they were indeed measuring a clock oscillator and not the DAC analog/audio output. They do say this is from Dragonfly DAC though:

"For the purposes of providing valid measurements, a DragonFly DAC was used on a Symetricon Phase Noise Test Device."

The next question is why didn't they show phase noise with and without a Jitterbug in use. ;)
 

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
191
Location
Australia
What's amusing is that we may have a measurement technique here which correlates to when people say the sound is not up to scratch - yet the rush is on to insist that this can't possibly be audible :rolleyes: ! What's the saying ... "You can lead a horse to water ... "
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,529
Location
Seattle Area
What's amusing is that we may have a measurement technique here which correlates to when people say the sound is not up to scratch - yet the rush is on to insist that this can't possibly be audible :rolleyes: ! What's the saying ... "You can lead a horse to water ... "
Can we lead you to water Frank? Read my post on translating their phase noise error to db. Then read this article on limits of our hearing: http://www.audiosciencereview.com/f...erception-of-sound-loudness-dynamic-range.24/

The maximum dynamic range of our hearing is 116 db SPL.

Now take this data back and explain why you think their measurements relate to audibility, never mind that what they show is their clock oscillator performance, not anything coming out of the DAC.
 

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
191
Location
Australia
Can we lead you to water Frank? Read my post on translating their phase noise error to db. Then read this article on limits of our hearing: http://www.audiosciencereview.com/f...erception-of-sound-loudness-dynamic-range.24/

The maximum dynamic range of our hearing is 116 db SPL.

Now take this data back and explain why you think their measurements relate to audibility, never mind that what they show is their clock oscillator performance, not anything coming out of the DAC.
Amir, I appreciate where you're coming from and what you're trying to convey. However, the huge flaw, to me, is that you're trying to treat the significant problems in audio quality as something that is in itself very precise. That is, you can nail a specific number, of a distortion or dB reading, as a go/no-go measurement of subjective audio quality. But everything I've explored in audio quality issues says that it's a nightmare world of string, sticky tape, fencing wire, and chewing gum which holds together all the artifacts that cause less than optimal sound; that is, it falls apart or completely alters in nature at the slightest provocation. The only way out of this madness is to assert that there is an optimal state of a system, any system, and to just be able to measure in some fashion when it falls below this optimum - and the measurement mentioned in the OP may just be one of those. The numbers read off doing this in themselves do not mean anything that can be strung together as an "explanation" for why this is so - it's a measure of confidence, and no more than that.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,986
Location
Riverview FL
upload_2016-10-6_20-56-10.png
 

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
191
Location
Australia
Amir, you're trying to convey that if you precisely measure phase noise in a certain test situation, using a certain method, that it's possible to translate that into a particular level of distortion, which by the book is inaudible - and that is perfectly OK: years ago I went through the exercise of working out precisely how much error it effectively created in amplitude if the DAC "reading" moment occurred a certain time interval away from the correct point - and amazingly, :D, it worked out just like the spec's said. Which said that the obsession with jitter back then was a nonsense, and I have never worried about it as a problem in itself. But that didn't stop me worrying about the sound not being as good, sometimes, as it should have been. And I have always been interested in some measurement regime that shows a change of behaviour in the DUT that correlates well with audible, subjective, performance characteristics. The fact that the numbers don't compute is irrelevant - what matters is that the numbers change in a way that corresponds to what one hears.
 
Top Bottom