They seem to forget the PDFs from TI and others describing the jitter requirements of a data converter and what that looked like at audio frequencies.
I also wonder if you can preserve the level of additive phase noise performance that TI and Linear get in their top ADCs in an audio grade converter that’s got integrated DSP functionality. If only picogates and GHz clock buffers with fs PN are good enough for the jitter fetishists here, what do the buffers inside an AK4499 look like in terms of performance and what is it like at the destination?
Last edited:
Understood, FFT measurements, THD, TID and so on.
Enough useless to understand the effects of phase noise (or jitter in time domain) at the output of a DAC.
maybe I'm wrong, so I'm curious to know what you understand from the measurements you are referring to.
Please don't make things up. I chide John on his support for stupid nonsense like Bybee, etc. I have never criticized his designs. In fact I published here an all FET pre-amp that was open-loop with no feedback that I still use. For your information no devices are inherently linear.
It looks like the picogates are good enough.
Indeed they degrade the noise floor but not the close in phase noise.
Attachments
Instead you have not understood, I believe in measurements, indeed I have got an expensive gear to measure the phase noise and I have published several plots.
But there is something that the measurements fail to explain and in this case I rely on hearing.
I am always waiting to know which measurements you are pointing out, since I have limited scientific education.
This has nothing to do with proofing by measurements that amp A with 0.01 % always sounds better than amp B with 0.001%.
Or where does one draw the line?
Just the other day here, I read, iirc, Toole stated that a speaker brand A would be preferred over the JBL M2 because of it's added early reflections, mostly because the on average bad recording quality of studio's would make that so. Had better recordings been used, the JBL would've been picked the favorite.
What's the point?
This point was new to me and after all the decimal points of distortion in amps, directivity and house curves, f-response of speakers etc etc this was yet another factor thrown in that made listeners prefer one speaker over the other.
In other words. Absolute numbers are meaningless and one can't draw conclusions about it's meaning, certainly not when measured in isolation.
Yet here we are again, talking decimal points of distortion in amplifiers (or oscillators for that matter), yet claims about perceived better sq must be proven by measurements.
I've heard the clocks. They're definitely a help in getting soundstage, bass, depth and rhythm to a higher level. I also heard them in equipment and some things change, but on average it isn't worth the time and money. To what level should phase noise be absent I have no clue whatsoever. Point is it available, can be measured, just like the single approach amplifier, and is readily available.
We can't take the rest of the audio set out of the equation, just as we can't rule out subjectivity in assessing sound quality. Expectation bias goes both ways, we are not in total control.
Scott, I'm sorry I misunderstood.
I know that no device is inherently linear, but I think you will agree that a vacuum tube is much more linear than a transistor and a mosfet.
It depends on the application and input and output signals. SET triode amps have plenty of distortion within their rated power range. My JFET pre-amp is -90dB at full line level out from a MM cartridge even thought it has no feedback.
Okay, but you are talking about jfets and low level signal devices.
When the required power level increases you cannot use jfets.
Then even a good 2SC2922 is far from the linearity of a 45, a RE604, a 211 or an 845.
That might be true but you have to admit it's ironic talking about 45's in the same context as -140dB jitter on a DAC.
BTW I have a friend who's grandfather worked for Simplex Cable here in MA (they were doing the trans-Atlantic cable at the time) that had a system that consisted of 3 stages of 45's in p-p with inter-stage transformers driving a single speaker mounted into a hole in a closet door. He said it was the best sound he ever heard.
BTW I have a friend who's grandfather worked for Simplex Cable here in MA (they were doing the trans-Atlantic cable at the time) that had a system that consisted of 3 stages of 45's in p-p with inter-stage transformers driving a single speaker mounted into a hole in a closet door. He said it was the best sound he ever heard.
Last edited:
Well, I admit it might be nonsense, but realistically speaking, I think the quest to improve audio power devices has failed miserably.
Maybe because there wasn't a real commercial interest.
The powerful non-linear device with tons of negative feedback seemed like the panacea for all ills.
But later it turned out that the problems this approach added far outweighed the benefits.
To be honest I have never listened to a truly realistic audio system, each approach has its advantages and its disadvantages.
And since there is no perfect system, I prefer vacuum tube distortion over solid state coolness, as I prefer to give up demolition bass and listen to an open baffle speaker.
In the end it probably remains ironic, but it's not really a nonsense.
Speaking of drivers lately I'm listening to the old Coral holey basket, it's just wonderful.
Last edited:
Unlike some other "distortions" testing for audibility of jitter doesn't need to be so difficult. Most commercial high performance OXCO's have a voltage trim. Its not that hard to add and I don't believe it brings much of a burden on phase noise (I have not seen any references). Simply introduce noise on the VC input. The phase noise increase will not be a lot from my past experiments but from this plus your ability to measure how much you have added you could then explore what level of noise actually degrades the audio. What I have seen is an upper band limit for the VCO input around 1 KHz. That should still enable meaningful measurements. And the ability to see if any is measurable on the audio output. All you should need is a good sound card and a clean single tone test track to see the added jitter as noise. You can also easily see stationary noise (hum) with that.
The story about noise modulation not being seen with an FFT may have some basis but its not hard to conjure a test that will see something like that. However noise by definition keeps changing so attribution of a change in noise is not simple.
The story about noise modulation not being seen with an FFT may have some basis but its not hard to conjure a test that will see something like that. However noise by definition keeps changing so attribution of a change in noise is not simple.
No, it doesn't seem like the -150dB has been tested yet .
The only claim was that a -97dB @10Hz XO was clearly not good enough in listening tests.
If THD sets the limit then wouldn't the speaker THD always bottleneck (almost) everything before it? Surely, if you've ever done listening tests, you've heard differences between 'bad' and good measuring DACs/amps, even when speaker/headphone distortion would dominate. Is there an explanation for this?
The only claim was that a -97dB @10Hz XO was clearly not good enough in listening tests.
If THD sets the limit then wouldn't the speaker THD always bottleneck (almost) everything before it? Surely, if you've ever done listening tests, you've heard differences between 'bad' and good measuring DACs/amps, even when speaker/headphone distortion would dominate. Is there an explanation for this?
You certainly did not follow this thread and the links within the posts.
Regarding speakers distortions masking other distortion sources, I would rather not open another can of worms at this time.