Baffled about computer power

[Phelonious Ponk]

It is fairly typical in hardware world to not disclose model/brands of components as a way to protect design. Often finding the right part takes considerable work and companies try to hide their identity going as far as erasing the numbers of components, burying them in glue, etc.

Thanks, Amir.

I went to the EA site and found what I was looking for. I must have asked wrong...

Some will claim that inserting yet another component in the signal chain is a negative thing, but in this case they are dead wrong. The transformers used in the Final Drive pair are simply the best on the planet. These use the Finemet core and winding technology from Japan. These are some of the most expensive signal transformers available which have been customized specifically for this application. The wiring inside the Final Drive uses cotton-insulated continuous-cast silver from Japan. Only the best ELMA and C&K switches with the optimum contact plating are used for the selectors. Neutrik XLR and Vampirewire solid-copper direct-gold plated RCAs are used. Only the very best and the sound quality improvement reflects this.

But The sentence in bold is exactly what I would have thought. This is, essentially, a switcher inserted in the signal chain, and not a simple one. Conventional wisdom would say at its best it can only be audibly neutral, with a transformer in there it's very likely to be additive, regardless of the quality of materials. The previous section offers some explanation (sorry for jumping around):

The benefits of #7 should be obvious. The ground-loop created by a stereo amp or monoblocks and the DAC or preamp driving them has to be the most problematic in most systems. If it does not create hum, there is a lot of RF pickup of the big loop antennas of the ground-loops

This must be over my head. I would think there would be simpler ways to break a ground loop, without introducing additional transformers to the system, and that if the breaking of that loop eliminated enough RF pickup to audibly drop the noise floor, that would be easily measurable and demonstrable, that such measurement would have been an often-repeated, essential part of this product's development, and that the best of those results would be the center of the case for the product's efficacy. Yet there are no measurements or relevant specs on EA's site.

http://www.empiricalaudio.com/products/final-drive

Tim

 

[Empirical]

The measure of how well any implementation does provide true differential input is the Common Mode Rejection Ratio - I am curious as to what do you consider a good enough CMRR (as not even transformers get a totally unlimited ratio)?

It is never good enough, and the problem is that particularly on USB interfaces even the published numbers are somehow compromised. These USB chips usually have decent specs, but don't seem to meet them because measures to reduce CM noise like filters are very effective.

Amir: Digital sample timing. A lot can vary these from the source, to cable to receiver. There are also mitigation techniques in the DAC for upstream issues. What makes this situation the most difficult, is the DAC attempting to guess at the sample timing on every input pulse. And using a transport like S/PDIF that embeds such clock in the data stream itself.

I have I2S, HDMI I2S, S/PDIF and AES interfaces on my products. I have found that S/PDIF, if the chips are carefully chosen and the design executed well, can perform virtually identical to I2S. I think the problem with most designs is that they either use the wrong chip or the implementation is poor. Sometimes the cable is at fault too.

When you come down to Oregon, I will demonstrate this for you.

High-performance implementations (in DACs) can deal with this very effectively. In my WSR magazine article I showed for example how even over the "dirty HDMI" connection where every AVR did poorly, the Mark Levinson 503 processor passed this test nearly with flying colors. I say nearly because one could do even better with good async USB but we do know how to do extract the clock well. It is just that it takes a lot of engineering skill and material cost. It involves what is called "mix-signal design" meaning you need to know both analog and digital design which many engineers are not trained or experienced in.

The easy way out of the above, with a computer server or proprietary connection to a DAC, is to reverse the roles. Put the DAC in charge of the clock, have it generate a super stable one and then tell the upstream source to track it. An async USB interface accomplishes this by generating a local clock and having the PC (the USB source) track it. What is left then is the accuracy of local clock in the async interface. In the case of external async USB, you still have a bit of consideration left with respect to the S/PDIF output. With internal ones this can be avoided.

Async USB is certainly an improvement over Adaptive because it establishes a new Master Clock, however it did not turn out to be the panacea that all of us manufacturers had hoped for. We were are frankly shocked when we realized that the USB cable still made a difference and different computers and playback software still sounded different.

Common-mode noise is certainly a large part of this and the poor CMRR of the USB receiver chips is at fault. I was the first manufacturer to put forward this hypothesis and create a hardware solution for it, the Short-Block. I received untold abuse on Audio Asylum for proposing this BTW.

What we are learning now I think is that there are other more devious things at work in the computer that still need to be addressed.

Here is some more anecdotal evidence that I have been hearing over and over on the forums (and some I have verified myself):

1) Differences in FLAC and .wav file SQ is only audible when using the audio stack. IF the music is generated over Ethernet or WiFi, these differences disappear. I speculate that the CODECs for FLAC don't work correctly when used on-the-fly to play music.

2) With Galvanically isolated USB interfaces, the USB cable does not matter, but the power supply for the computer still does matter.

3) Different players sound different even though they are supposedly bit-perfect.

4) Different rippers result in different sounding tracks, even though they are also bit-perfect and maybe even using Accurate-Rip.

Even though rippers like iTunes produce bit-perfect tracks, I believe they muck with the offset. Offset is the number of leading nulls in the track before the music data starts. I have a number of test tracks of the same exact music data, but with different offsets. These were generated by an engineer that I converse with on Audio Asylum. Each sounds different. At first I thought this effect was due to the feedback nature of the Sigma-Delta converter, but I have customers with NOS DACs that also report hearing these differences. BTW, the guy who generated them cannot hear the differences - his system is evidently not capable.

I even hear differences in AIFF versus .wav. This says that even Apple cannot create their own non-compressed format without screwing-up the SQ.

Also baffling

Steve N.

 

[Empirical]

This must be over my head. I would think there would be simpler ways to break a ground loop, without introducing additional transformers to the system, and that if the breaking of that loop eliminated enough RF pickup to audibly drop the noise floor, that would be easily measurable and demonstrable, that such measurement would have been an often-repeated, essential part of this product's development, and that the best of those results would be the center of the case for the product's efficacy. Yet there are no measurements or relevant specs on EA's site.

I wish there were a simpler way. If the signal is transmitted from one component to the next using hardwired connections and all of the components have AC safety grounds, then there will be ground-loops. There are several ways to break these, but unless you change the signaling method to optical or transformer-coupled, you still have ground-loops. If each component had its own AC isolation transformer and a copper ground-rod for safety ground, then you would have some isolation, but the currents can still flow through the earth grounds.

Ground-loops are entirely system-dependent and AC service dependent, so making measurements does not make sense unless it is on a specific system and this measurement must be made using differential probe and floating ground or the test equipment will create yet another ground-loop. Some systems will be worse than others. If you use a single dedicated circuit and outlet for all of your components and keep the cords short, this will minimize the effect of the ground-loops. Even better to twist all of the power cords together to minimize loop area. Trying to make such measurements on one component with test equipment does not tell you much, except for maybe verifying the CMRR of the component, which is already speced in the datasheet for the input device.

Steve N.

 

[puroagave]

^
Steve, you had great sounding room at the Newport show. one of my favorites and definately one of the best all-digital demos for sure.

 

[Empirical]

^
Steve, you had great sounding room at the Newport show. one of my favorites and definately one of the best all-digital demos for sure.

Thanks. I hope I played some music you liked. I want those Vapor Audio Nimbus speakers BTW. Never heard bass like that before.

Steve N.

 

[puroagave]

the music was great, lots of boogie factor definately showed off the systems strengths. yes kudos to the vapor nimbus speakers, im sure its not last we'll hear of them. the thought of what they would sound like in my system crossed my mind.

 

[Julf]

I wish there were a simpler way. If the signal is transmitted from one component to the next using hardwired connections and all of the components have AC safety grounds, then there will be ground-loops.

Unless safety earth and signal ground are kept completely separate. Safety earth only needs to be connected to chassis, and doesn't need to be connected to signal ground (and if you use fully balanced / differential connections, signal ground isn't needed either).

 

[Empirical]

Unless safety earth and signal ground are kept completely separate. Safety earth only needs to be connected to chassis, and doesn't need to be connected to signal ground (and if you use fully balanced / differential connections, signal ground isn't needed either).

Have you ever tried this? Even with differential signaling, you need ground between components.

Steve N.

 

[Julf]

Have you ever tried this?

Even with differential signaling, you need ground between components.

Are we talking analog interconnects or USB? The best one is of course optical toslink - 100% guaranteed not to cause ground loops.

 

[Phelonious Ponk]

I wish there were a simpler way. If the signal is transmitted from one component to the next using hardwired connections and all of the components have AC safety grounds, then there will be ground-loops. There are several ways to break these, but unless you change the signaling method to optical or transformer-coupled, you still have ground-loops. If each component had its own AC isolation transformer and a copper ground-rod for safety ground, then you would have some isolation, but the currents can still flow through the earth grounds.

Ground-loops are entirely system-dependent and AC service dependent, so making measurements does not make sense unless it is on a specific system and this measurement must be made using differential probe and floating ground or the test equipment will create yet another ground-loop. Some systems will be worse than others. If you use a single dedicated circuit and outlet for all of your components and keep the cords short, this will minimize the effect of the ground-loops. Even better to twist all of the power cords together to minimize loop area. Trying to make such measurements on one component with test equipment does not tell you much, except for maybe verifying the CMRR of the component, which is already speced in the datasheet for the input device.

Steve N.

i think you named the simpler way - optical. And of course measurements would make sense, even if they vary from system to system you could demonstrate that you had actually lowered noise while adding a transformer Quite a breakthrough,I'd think

Tim

 

[Empirical]

Are we talking analog interconnects or USB? The best one is of course optical toslink - 100% guaranteed not to cause ground loops.

USB. This is differential.

Steve N.

 

[Empirical]

while a trasformer can reduce common mode noise, and while a trasformer made from better materials can have less "self noise", it aint gonna eliminate regular old noise other than what all transformers do, and thats have a passband that they pass, and thereby, can reduce noise or anything else out of their passband.

While we can hear noise, in a decent audio system with speakers, at the normal listening position, it has to be pretty bad to be audible. Noise seems to be the new can to be kicked around in digital now though.

I think the noise that we are talking about is audible because it is riding on the audio signal, not very audible as background noise when there is no music playing or between sounds. Ground-loops make very good loop antennas and pick up HF noise. This noise is not generally audible from the listening position until you get some music with HF energy in it. Then the noise becomes more obvious by masking the detail. Once the noise is eliminated, it gives the illusion of a blackness in the background, much more detail and better focus.

Steve N.

 

[Julf]

USB. This is differential.

OK, USB is a bit of a problem because while the signalling is mainly differential, it isn't completely differential. There is ground-referenced voltages that matter too.

With completely differential, fully balanced signals you do avoid ground loops, with "partially differential" like USB it is trickier.

 

[1audio]

while a trasformer can reduce common mode noise, and while a trasformer made from better materials can have less "self noise", it aint gonna eliminate regular old noise other than what all transformers do, and thats have a passband that they pass, and thereby, can reduce noise or anything else out of their passband.

While we can hear noise, in a decent audio system with speakers, at the normal listening position, it has to be pretty bad to be audible. Noise seems to be the new can to be kicked around in digital now though.

Its not the noise that you hear, its the noise that modulates the signal you are listening to. There are a number of mechanisms that allow noise to essentially modulate the signal (audio) but have little or no presence when there is no audio to modulate. Jitter is one of these phenomena.

Also a transformer can work very well to remove out of passband noise. But even with transformer coupling there are still parasitic capacitances that will pass noise currents. At 24 MHz (typical DAC master clock) 10 pF (a typical interwinding capacitance of an SPDIF transformer, some are better and some are much worse) is 660 Ohms, enough to pass some current around. With the 140 dB signal to noise potential of 24 bit audio it doesn't take much to degrade the performance. 140 dB below 2V is something like 200 nV. And to put that into perspective if your chassis has 100 mOhms impedance between the connector grounds. It only take 2 microamps to get 200 nV of ground coupled noise to degrade the potential SNR. That is a very small amount of current. The 600 Ohms with 5V of master clock at 24 MHz will pass 8 milliamps if given a chance, 60 dB more than the noise floor example above. Its actually a testament to the care and tenacity of the designers that they can meet the exceptional performance we can get.

 

[mauidan]

Demian,

What improvements can one expect to hear by upgrading the SMPS on your PK90-USB and PK-100 units with the optional linear PSU?

 

[Phelonious Ponk]

I think the noise that we are talking about is audible because it is riding on the audio signal, not very audible as background noise when there is no music playing or between sounds. Ground-loops make very good loop antennas and pick up HF noise. This noise is not generally audible from the listening position until you get some music with HF energy in it. Then the noise becomes more obvious by masking the detail. Once the noise is eliminated, it gives the illusion of a blackness in the background, much more detail and better focus.

Steve N.

The electrical noise from ground loops modulates with the audio signal?

Tim

 

[Joshua_G]

Hi Amir,

…
You say 9 out of 10 suites at shows don't sound. Since those companies are all by definition "high-end" would you like me to conclude that you don't consider high-end, high-end?
…

Isn't 'Hi-End' defined by the price tag?
Is the price tag carry any guaranty for SQ?

Here is the other reason we are crap listeners: we don't agree on the same thing!!! Why is it that people buy so many different amps, CD players, cables, etc? Definition of a good hearing would be to hear the truth.
…

Nope, audiophilia, or caring about the SQ of sound setup, is driven by personal tastes and preferences, it has nothing to do with 'truth'. Many 'audiophiles' don't even care about fidelity (which, in my dictionary, is the aspiration to come a little close to live music).

 

[Joshua_G]

Hi Steve,

IMO, there are so many different views simply because:

1) not all audiophiles have trained their ears the same

2) not all audiophiles systems are at the same level of refinement

Yes, though it's only part of the picture.
There are 'audiophiles' who's main concern is the way the setup 'sounds', while there are those who's main concern is the degree to which the reproduced music is engaging and involving.
People from those 2 major 'camps' will seldom agree on SQ of any piece of gear.

Other than that, many things you wrote here echo in me (if what I wrote is proper English).

 

[Groucho]

...while there are those who's main concern is the degree to which the reproduced music is engaging and involving.

Not necessarily. As was discussed earlier, it may not be the reproduced music as such, which the user finds engaging; it could be many other factors, like valves that glow blue. People acknowledge the placebo effect when talking about it in isolation, and are even prepared to accept that of course their impression of an audio system is swayed by many things other than the reproduction of music. But in the next post have forgotten all about it when relating how their latest tweak "sounds".

 

[Joshua_G]

Not necessarily. …

As I said, there are various 'camps'.