The "Bits Are Bits" Fallacy and Noise In Mixed Signal Systems

"Bits are bits". We've all heard this too many times when talking about digital audio. Some people feel that's really all they need to know or say, "bits are bits". I admit, it does have a pleasant ring to it and if you silently repeat it over and over in your head it becomes mantra-like. Bits are bits. Unfortunately when talking about mixed signal systems, "bits are bits" holds the same relevance as "om".

"In this trifling particular, then, I appear to be wiser than he, because I do not fancy I know what I do not know." Socrates

I am not claiming any expertise in the field of noise in mixed signal systems but I know enough to know that we cannot ignore it when talking about what goes on in a digital to analog converter. In computer audio, or more appropriately digital audio, we are necessarily talking about mixed signal systems since we mix a digital and analog signal in the same box. A DAC. (Captain Obvious is even shaking his head in disbelief that we have to say this to refute claims made by people who believe all you need to say is "bits are bits" and you can magically make real issues disappear)

Let's take a look at some basic electronics. The following quote is from The Electronics Handbook, Second Edition edited by Jerry C. Whitaker:

Noise In Mixed Signal Systems
It is also common, however, as stated earlier, for electronic systems to begin and end as analog circuits, but in between have digital logic subsections. Such a system is called a mixed signal system, and noise is a concern both at the input to the system and at the output. Noise at the input is converted into errors in the digital logic circuit; the data then picks up further errors as it is processed within the digital logic circuits, and when the conversion from the digital to the analog domain is finally made, these accumulated errors become noise once again at the digital logic circuit’s output. There are some mixed signal systems that combine digital and analog circuits on a single semiconductor strip substrate, and these can produce crosstalk, which originates within the digital logic circuits and creates noise within the analog circuit section. The most common of these are power supply noise and ground bounce.
...
But digital-to-analog converters used in this process are subject to all of the noise problems of analog circuits since they themselves are analog devices, and rely on resistive components, which have thermal noise limitations; thus, the digital information necessarily enters the digital system with noise present.
That's pretty straight forward stuff. Of course this is only the tip of this iceberg, but it gets us to a position where we can very comfortably say—saying "bits are bits" misses the most relevant bits when dealing with mixed signal systems.

On—"What does this have to do my hi-fi?"—here's an excerpt from UpTone Audio on their REGEN product (see my review for more from UpTone on the REGEN):

The REGEN is at its core a single-port USB 2.0 hub. All hubs actually contain two USB interfaces and a full-blown USB protocol engine. It is not just working at the analog level, it is actually receiving the data from the DAC, putting it in a buffer and retransmitting (and the other way for the packets from the DAC).

It uses a selected USB hub chip to create a new USB stream to deliver a very high signal integrity to the DAC's USB PHY, thus decreasing the PHY’s contribution to packet noise. It is called “REGEN” since it completely REGENerates the data signals that cables are messing up—it’s not just a re-clocking. Because it uses clean power and a low jitter clock, the output of the hub has low noise and low jitter. To be most effective, and to maintain best signal integrity and ideal impedance matching it is best positioned right at the input to the DAC, thus its small size, low weight, and included male>male USB ‘A’>’B’ adaptor.

The result is that the PHY in the DAC doesn't have to use any of its pre-processing circuit arsenal so the packet noise is as low as it is going to get.

If we take the information from The Electronics Handbook and use it to fact check what UpTone Audio is saying, we come up clean; reducing packet noise inside the DAC addresses a known problem in mixed signal systems. Here's a hint for any of my "bits are bits" believing friends; if you want to make a rational argument as to the efficacy of devices like the USB REGEN, talk about noise.

Of course this isn't anything new nor is dealing with noise restricted to the recently reviewed REGEN, Wyrd, or JitterBug (it's not even restricted to audio applications).

Here's Antipodes Audio founder Mark Jenkins from my review of their DX music server (see my review for more info):

"...All chipsets generate electronic noise that will interfere in some way with the signal carrying the digital data, and the level and frequency of the noise has an audible effect on the analog output of any DAC. It is easily heard – it just does not fit with the simplistic accepted digital theory of how these things work."
Simplistic. That's very polite.

It's also worth noting that "bits are bits" doesn't even apply in the digital world [footnote 1]. If you were responsible for hiring a company to design and install a network to support a hedge fund and a vendor came in and said “Bits are bits so we’ll just buy the cheapest stuff”, you’d throw them out on their ear. The fact of the matter is, whether or not a bit gets flipped is inconsequential *if you are only concerned with the integrity of the data*.

You can have an Ethernet cable that does not meet spec and injects errors into your network but the data is successfully passed. If you install 100+ shitty error producing Ethernet patch cables in your trading floor network, the combined error rate can slow down network performance. That's bad. If you hire someone to diagnose your network performance problems and all they do is check to see if data is being passed from point A to point B and they declare "The bits are the bits! You're good to go!", you've hired yourself someone not qualified to do the job.

My job is to listen and report. The Schiit Wyrd, UpTone Audio REGEN, and the AudioQuest Jitterbug each improve the sound of my hi-fi. Thousands of these devices have been sold and many owners are reporting hearing improvements as well. We also have other reviews of the JitterBug by people like John Atkinson of Stereophile where he also reports hearing "...a significant improvement to sound quality." with the JitterBug in his system. If anyone thinks they can refute all of this by saying "bits are bits", you know you're talking to someone who thinks he knows more than he knows.

I was talking to my wife about this topic, yea I know I feel sorry for her too, and she suggested coming up with an analogy in order to try to make my point more clearly. Here's what I cooked up.

Let’s imagine I just reviewed a pizza and I liked it better than the last pizza I reviewed. You read my reviews and discover both pizzas use the same mozzarella and conclude my reviews are nonsense because those pizzas must taste the same since they use the same cheese.

I point out that there’s more to a pizza than cheese to which you respond, “cheese is cheese”.


Recommended Reading
If you'd like to learn more about this subject, here are a few articles on AudioStream that go into more depth.

There's no such thing as digital: A conversation with Charles Hansen, Gordon Rankin, and Steve Silberman

A Q&A with John Swenson Part 1, Part 2, and Part 3

Cables, Bits, and Noise: How Cables Can Make A Sound Difference

Here's a good outside reference for USB, "At 480 Mbits/s, Signal Integrity Becomes An Issue In USB 2.0 Designs"

For additional outside references, just Google "Noise in mixed signal systems"


Footnote 1. It's worth noting that Ethernet cables do not transport "1s and 0s".

Here’s Kurt Denke of Blue Jeans Cable:

"…the relevant Ethernet specs do not use simple one-zero binary encoding. They’re multilevel — the most extreme example being 10GBaseT over Cat 6A, which uses PAM-16, a 16-level encoding scheme which results in something that looks a whole lot more like an analog waveform than the simple square-wave pattern that people imagine. In other words, instead of 'all ones and zeros' the signal has sixteen different voltage levels, rapid transitions between which need to be correctly read."
And here’s John Swenson from a comment on Computer Audiophile talking about signaling rates on gigabit Ethernet:
"So each pair is seeing 125 million analog level changes per second, each level change represents 2 bits of original data, for a total of 8 bits of data 125 million times per second, which is the gigabit."

COMMENTS
TheAudio-VisualJunkie's picture

I think people battling on both sides of this topic are simply wasting their breath and time. It's like trying to convince people weather or not god exists. You won't succeed to convince them of something they don't want to be convinced of – one way or the other.

Bits are bits, right? Water is water, right? H2O – simple chemistry science. But what about all the pollutants in the world? Is the water same everywhere? You can apply similar logic to cables and draw your own conclusions…

Didn't we hear enough of this never ending argument?

Michael Lavorgna's picture
...I think it's important for me to try to keep the dialog in the realm of the rational.
PeterV's picture

I am not a believer of the 'bits are bits' church. But what does annoy me is that this discussion is ignoring the simple question if this digital noise influences are being stored as well as soon as they arrive a medium like a NAS harddrive or CD optical medium or a simple USB or SD card medium. In case these bits are continously influenced by transport media like 'inferior' ethernet- or USB cables, than there should be an audible difference also during the final storage proces. If this would indeed be the case, than all digital signals are truncted multiple times during their transport via the internet.. But we all know that this is not the case, and 'we' audiophiles seem to be focussed only to the last meters of bit transport effects inside our audio system.. It isstrange that somehow this important point of the 'bits are bits' discussion is overlooked or ignored. I have the impression that during transport of bits, indeed noise will have influence on the timing and DA conversion. We know that since it is audible and even measurable. But if it is indeed the case that as soon as bits are stored, thet will not contain any of the transport noise anymore and indeeed have become again simple and pure bits, why is this fact not intelligently used in a much more dense (external) transport - (internal) storage - (I2s to I2S ) transport on IC direct to DAC..? I hope you understand what I am trying to make clear.

Michael Lavorgna's picture
Have you ever heard of or used a water filter?

The important point to keep in mind is we are talking about what happens inside a DAC.

kirkmc's picture

Wow, there's so much wrong in what you're saying....

If what you say was the case, then every text document on your computer would gradually change to a long series of random characters.

Nothing personal, but a lot of people who postulate about things like this really need a primer to understand how computers work. In short, much of what Michael wrote above is simply wrong, at least as far as the "bits are bits" argument is concerned. He's right about noise that is added after the bits are converted to analog, and this is a well-known and not uncommon problem. But saying "The 'Bits are Bits" fallacy..." in his title is misguided, and simply incorrect.

Michael Lavorgna's picture
.
streamfan's picture

I understand a 'bit' where some people are coming from in the 'Bits are Bits' world. Computer error checking ensures data integrity, (Otherwise everyones documents would slowly dissolve)and yes, you get what you pay for in terms of how much bandwidth is used up and lost via retries, so quality of pipeline matters. It is like surfing the web when at that edge of your wireless routers range. You may get the page, but also a lot silent resends were made behind the scenes.
What I'm wondering, since I don't know exactly how DACs work, is what is actually error checked? Since that might result in an audible distortion similar to a TV that produces a blank square when it doesn't know what the data is for that pixel. Does any component in the audio chain pump data the same way that a BluRay seems to pump data to the TV. Lost bits from a light scratch on the BluRay produce a visible pixel problem. Sound from a BluRay usually stutters with introduced silence on errors.

I'm not sure what I'd hear as a lost bit error or as lost signal from the noise floor.

Michael Lavorgna's picture
Please re-read the included quotes from The Electronics Handbook and UpTone Audio. We are talking about noise not data integrity.

If you want to know how this manifests itself in terms of listening to music on a hi-fi, read my reviews of the Schiit Wyrd, UpTone Audio REGEN, and the AudioQuest JitterBug.

After all the theorizing, actually before the theorizing, we just need to listen.

streamfan's picture

Currently quite happy with all the sounds my system produces.

Müstrkraaft's picture

I'm growing sick of this debate. As audiophiles, and consequently a minority sect of this society, we need to become experts in this field. It is the very scarcity of our kind that demands mastery of our craft. Having just scraped the surface on the way signals, both visual and audio, work, I must say that no one who has not delved into this subject can contribute much to the conversation. It is vastly complex, and those deeply involved have little-to-no interest in the discussion. I believe they are irrevocably committed to the furtherance of our field. This is a deep conceptual rabbithole. It would be a miracle to see someone as deeply involved, who could adequately explain the science behind the conversion process.

Shp's picture

I think the challenge for the "bits is bits" crowd is that so many manufacturers don't provide results for how their products improve quality. And nearly everything in the cited literature is measurable.

Audiostream has written that they don't have the resources to do all the measuring, and I accept that.

But when the technology guarantees that it's the same bits at both ends of the wire, it's important that manufacturers show which of the valid technical problems they solve. The CAT6 standard makes clear how it's better than CAT5. Can't fancy digital cable A show measurable improvements in EM/RF rejection or data throughput rates over a cheap Amazon Basics' wire?

Then, with the aid of Audiostream's fine editorial team, my ears can decide if these technical improvements sounded different enough to me to justify their price.

Wouldn't an experiment like the following help quantify the differences between cables (or regens or whatever product you wanted)...

After matching for volume level, etc., digitally record 10 samples of Sample A and 10 samples of Sample B. Average each set to minimize the effect of small differences in the environment/recording/digital conversion. If the difference between A-average and B-average is bigger than the deviation within the A or B set, then you have quantified the difference in sound. With the right software, you could actually play the "delta," sort of like facing two speakers toward each other and reversing one speaker's wires would tell you the "stereo" of the two channels.

Archimago's picture

That's the Audio DiffMaker test you're describing in a nutshell.

Shp's picture

I didnt know there was a product for it. I just visited their site and that seems like it.

I think every bits is bits article/thread should concede that this is all totally measurable at the micro level (interference, error rate, throughput) and macro level (audio diffmaker).

Audiostream should put a few products through a macro comparison.

kirkmc's picture

Bits are bits. What you're talking about is what happens when those bits are converted from digital information to analog sounds. And that's where the problem is; not in the digital cables themselves. If error correction is managed correctly at the end - say, at the DAC level - then any noise that's added comes from the DAC itself, not from anything like gravity's influence on cables. The quote from the book makes that very clear: the noise comes from the analog, not the digital. If there is crosstalk, it's not changing the bits, but the analog sound.

Archimago's picture

My reading of the quote as well. The concept of digital is an abstraction of 1's and 0's which is subject to redundancy and error correction. Within this domain of maintaining data accuracy, it is "bit perfect". Conversion to and from in the analogue domain subjects the signal to noise just as any analogue signal has a noise floor at some level.

Even if the digital processing is "noisy", the bits are generally not corrupted (unless truly horrible noise which interferes with signal integrity beyond the threshold of what is considered '1' or '0'; if this happens the product should appear obviously broken - computer crashing, data corrupted, macroblock errors in video, dropouts, spurious beeps and bops...). But this noise generated in the digital processing can be transmitted to the analogue output which is more susceptible.

Practically in the "real world" it is as you say, noise is generally the function of the DAC unless the transport is so horrible that it transmits a ton of noise down a non-galvanically isolated interface and the DAC does nothing to attenuate the noise. One can "hear" this at times in severe circumstances putting the DAC near an electrically noisy computer or placing a cell phone over the DAC. Reputable DACs should not be so sensitive to noise and reasonable transports should not be so severe either.

Michael Lavorgna's picture
Such a system is called a mixed signal system, and noise is a concern both at the input to the system and at the output. Noise at the input is converted into errors in the digital logic circuit; the data then picks up further errors as it is processed within the digital logic circuits, and when the conversion from the digital to the analog domain is finally made, these accumulated errors become noise once again at the digital logic circuit’s output.
I read this as saying noise is a concern at the input of a DAC since that's what it says. Of course we're talking about what happens inside a DAC but the fact that the data is passed in tact misses the issue of noise in a mixed signal system...like a DAC. But if you want to say "bits are bits" to describe the issue of noise in mixed signal systems, that's fine by me. I wouldn't.
Clever Dean's picture

With Ethernet errors don't accumulate. Every time packets hit a switch or router they get regenerated. That is why if you have a 600 foot run of CAT5 you drop a switch in the middle of it. The spec calls for 328ft. As long as the cables meet spec for construction and installation data is going to get passed error free when it is all said and done.

There are tons of different digital transmission systems, some more forgiving and better at error correction than others.

When you quoted John Swenson I hope you realize that he stated he signalling rate was 125Mhz. When SNR is spoken about on GB Ethernet we are talking about noise at the 125Mhz level.

The reference clock or GBE is 25Mhz BTW and clock multiplier used to get it to 125.

What's the ceiling of human hearing?

Michael Lavorgna's picture
I'm not talking about an Ethernet cable connecting to a DAC. I referred to Ethernet to show how saying "bits are bits" does not tell us anything important about network performance.
Clever Dean's picture

Of course. Protocol analyzers tell us this along with OS. I just collected 3.2 GB of packet data off my local AMD E350 (a rather dated CPU) playing back music for ~4 hours over 150N wireless.

0 packet loss and 1 resend. I used WireShark.

Clever Dean's picture

We were talking about AUDIBLE performance? If you take a 2.5GB high resolution album you only need, in real-time (if such a thing existed on Ethernet) 694 Kilobit / second. So 70Kbyte a second. Even 10 base Ethernet clears this by 30%.

What kind of possible network performance issues could we be talking about where you can pick up a brand new Windows 8 computer with SSD and GB wired Ethernet for $159 delivered to your door?

Never mind the fact that on the final leg from the switch all the traffic, the signal etc has been 100% recreated/regenerated.

I've had the AQ Cinnamon in for 10 days and neither my ears, WireShark, RightMark, or my Omni mic show me any difference. My through put of a copying over a 8GB .vhd file hasn't changed. The hash I generated is the same. It doesn't matter if I use my Tera cable or the AQ cable. My network performance is identical.

Out side of that it's a nice looking well constructed, and ultimately over priced cable for me. Now if you like the aesthetic then it's worth it.

Michael Lavorgna's picture
...I am not talking Ethernet cables and noise in this post. If you'd like to talk about Ethernet cables, comment under an appropriate post.
Clever Dean's picture

what would you like to say in that regard then? The point I'm trying to get across is that in the section of network that you control (in your home) what performance issues are we talking about where the cabling passes spec? If you are running a 10/100 network and CAT5e and it's all with in spec what issue is being solved?

Michael Lavorgna's picture
"You can have an Ethernet cable that does not meet spec and injects errors into your network but the data is successfully passed. If you install 100+ shitty error producing Ethernet patch cables in your trading floor network, the combined error rate can slow down network performance. That's bad. If you hire someone to diagnose your network performance problems and all they do is check to see if data is being passed from point A to point B and they declare "The bits are the bits! You're good to go!", you've hired yourself someone not qualified to do the job."
AlexMetalFi's picture

...this is not some rampant problem the industry is somehow grappling with on a day to day basis. Let's not over sell it. In fact, the cable is the LAST place you usually look for issues with respect to network performance problems.

Also latency requirements for high frequency traders are a *BIT* different than trying to stream a FLAC file over a home network.

Michael Lavorgna's picture
It's also worth noting that "bits are bits" doesn't even apply in the digital world [footnote 1]. If you were responsible for hiring a company to design and install a network to support a hedge fund and a vendor came in and said “Bits are bits so we’ll just buy the cheapest stuff”, you’d throw them out on their ear. The fact of the matter is, whether or not a bit gets flipped is inconsequential *if you are only concerned with the integrity of the data*.
...
If you hire someone to diagnose your network performance problems and all they do is check to see if data is being passed from point A to point B and they declare "The bits are the bits! You're good to go!", you've hired yourself someone not qualified to do the job.
This is what I said re. Ethernet. Do you disagree?
AlexMetalFi's picture

...if all they do is walk into your DC and go "bits are bits" and call it a day. Yeah, you hired the wrong person.

But Michael, no one is saying "bits are bits" and leaving at that though, which you seem to be constantly drumming on about.

Michael Lavorgna's picture
...and leaving it at that. Do a search on the REGEN or JitterBug and you'll have a few hours-worth of reading people saying "bits are bits" so these devices cannot possibly do anything. I've had an industry professional say I'm promoting "snake oil" by reviewing the REGEN, JitterBug, and Wyrd because "bits are bits".

Since I've had all of these devices in my systems for weeks, I've heard how they improve the sound of my systems. Debate over. At least for me.

AlexMetalFi's picture

How does the Ethernet cable contribute to any noise inside a DAC though?

The packets arrive in computer memory as 1's and 0's FIRST before being whisked away again for digital to audio conversion. That's all in the digital domain.

qsysopr's picture

Alex I've tried several times to talk to people on information processing and the used protocols. It is a Sisyphos work. They mix up all and everything. They do not understand that once the electrical signal is interpretated as a '0' or a '1' everything else relies on digital information technology. Most of them do not even understand the different protocols and error correction mechanisms that enable them to use their mobile and connect to the right person, when the right number is dialed.
I asked a person who wanted to sell verrry expensive stuff for RFI rejection, why he does not use fiber optics, which are immune to RFI. Got the answer that fiber optics sound worse - period. How to discuss with these people?

kirkmc's picture

Wait, you're the one who started writing about "Bits is bits." Now you're saying that's not what you mean; that it's the noise inside a DAC. That noise has nothing to do with the bits, and it's outside the digital domain.

Michael Lavorgna's picture
I'm talking about more than one issue here. You are stuck on "bits are bits" and data integrity when that's clearly not what I'm talking about.
kirkmc's picture

Then why does your article title begin with "The 'Bits Are Bits' Fallacy? If you're saying that's not what you're talking about, change the title, and change much of the text. Bits are bits, whether you like it or not. Your point about noise being introduced after the D/A conversion is not only valid, but very important, but you decided to bring up the bits are bits issue, which clearly is problematic to you.

Michael Lavorgna's picture
...to suggest that devices like the REGEN, Wyrd, and JitterBug are "snake oil". That's why I said this in first paragraph, "Unfortunately when talking about mixed signal systems, "bits are bits" holds the same relevance as "om"."

The only problem here is you have not understood what I've written and are stuck on data integrity.

bit01's picture

+1
I must agree with the words by kirkmc

kirkmc's picture

Also, you say this:

It's worth noting that Ethernet cables do not transport "1s and 0s".

You're simply wrong. It doesn't matter how many channels the data flows through. Ethernet protocol frames have a set format, and their data is bits. It may be in hexadecimal, which is just a different way of encoding binary, but it's still just bits making up each frame.

As for it being accurately read, that's why there is error correction and checksums built into the Ethernet protocol.

The frame ends with a frame check sequence (FCS), which is a 32-bit cyclic redundancy check used to detect any in-transit corruption of data.

https://en.wikipedia.org/wiki/Ethernet_frame

If the redundancy check doesn't match, then the frame is requested again. And, yes, this all happens very, very fast.

AlexMetalFi's picture

All the bits that an Ethernet cable carry over the wire ALWAYS land in computer MEMORY first before it hits any kind of DAC.

And all of these bits are by definition going to be "bit perfect" else the entire Internet would be in big trouble.

The only real issue is when the hardware can't keep up with the incoming data rate. Packets are then dropped on the floor, i.e. you will hear drop outs in the audio. It's an all or nothing proposition. Btw, audio streaming applications aren't nearly that bandwidth intensive else Spotify and Tidal wouldn't be in business.

So the idea that an Ethernet cable will do anything to the sound is well...

Michael Lavorgna's picture
...carry electrical signals of varying voltages which represent the 1s and 0s. If you want to say those voltage levels are 1s and 0s, that's fine by me. I wouldn't. As far as error correction and checksums and data integrity, I agree and said so. My point with a network is data integrity doesn't tell you about network performance.
kirkmc's picture

It's always electricity, all the way down. That electricity is converted from bits - pretty simple, since it's just on/off to represent 0s and 1s - and at the end of the path, it's converted back. A bit is merely a representation of data, and it could be sent as electricity, light, heat, or any other form of information that can be later decoded. You're straying from the issue here.

Michael Lavorgna's picture
I responded to your comment that I was wrong in saying Ethernet cables do not carry 1s and 0s because they actually carry an electrical signal. You agree yet you say I'm straying from the issue.

Hmm.

kirkmc's picture

I would strongly recommend anyone research this topic to get this book:

Principles of Digital Audio, Sixth Edition by Ken C. Pohlmann.

It covers all the basics of digital audio from recording to reproduction, and the chapter on error correction, while quite complex, shows exactly how bits are bits.

The book is five years old, so isn't fully up to date regarding things like high-res audio, DSD, etc., as well as music streaming technologies. But all the fundamentals are there.

Clever Dean's picture

Without having the reference book in front of me I believe your quoted section is referring to 'Johnson Noise'.

This is where discrete components are only thermally good for about 21bits of resolution. This applies across manufactures as it's roots are based in thermodynamics.

Also, I have to point out some very gross missunderstanding:

Ethernet does indeed transport 1's and 0's. Computers are strictly binary. Higher up it's abstracted for us humans.

Clever Dean's picture

I wonder how many USB DAC manufactures have like functionality already in built?

deckeda's picture

(Hey, somebody had to say it.)

Our lexicon has failed us. We were brought up to believe digital = noise-free and so, how can bits be bits + noise? Wait, wut?

And yet it's true, and that is the crux of "there is no digital" because, there is no digital [by itself, in any practical sense.]

I also would like to know if the various "regen" products wouldn't simply (and even more cheaply) rather just not be built into DACs for example. Or a "powered" USB cable. Surely those kinds of manufacturers know as much about why they could help as anyone else.

kirkmc's picture

Digital is noise free. Bits and noise means that the bits themselves are just fine, noise free, happy and clean, but the device that converts them adds the noise. It's not the fault of the digital domain, but poor design and manufacture that adds the noise.

Michael Lavorgna's picture
...in practice, digital is not noise free.

Clever Dean's picture

Let me start out by saying that I'm not letting the data portion of this off the hook.

He could be hearing wall wart power supply. Not only is he turning off the switch, he's turning off the switch power supply. I would be curious as to the switching frequency of the supply.

Michael Lavorgna's picture
...is well known. After all, the protocol was originally designed with error correction to prevent noise from corrupting the data. It's also why higher CAT levels use additional shielding and higher twist rate tolerances - to prevent noise from corrupting the data.

Of course Ethernet works in the sense that "bits are bits", data is passed without issue in a properly functioning network. What I'm saying is that noise is not something we can make disappear from Ethernet by saying "bits are bits". We know there's noise present.

That said, this post was intended to address noise in a USB DAC. The fact that noise in a DAC is an issue from input to output means that saying "bits are bits" cannot make this noise issue go away.

Of greater importance, imo, is that I heard clear and easily recognized improvements in the sound of my music when using the REGEN, Wyrd, and JitterBug. So have others. We can argue endlessly over data integrity but I find it much simpler and more to the point to focus on my enjoyment of listening to music. I enjoy this experience more by employing these devices.

Clever Dean's picture

In the context of the video you provided I just curious if it was the power supply. That's all.

Again Archimago did some analysis using some USB hubs that certainly showed differences. I wish manufactures took the same care.

Michael Lavorgna's picture
...devices did not "take care" in their design because they choose not to share the measurement results used during design and development is unreasonable, imo.
Shp's picture

Michael,

Recently Michael Fremer posted on Analog Planet recordings of two different phono step-up devices, polling readers on which they preferred.

Perhaps you could do the same for the Regen or any of these other components. A file, B file, and an AudioDiff-style diff file.

Assuming you have available an ADC or other way to record and digitize the signal, it would probably take less time to run the test and post the files than to keep debating "bits is bits but not." (If only comedians cared about this stuff, I could see a Who's on First? routine coming out of this.)

Digital software could also graph differences in the files to help quantify how different the two signals really are. You could have some real fun downsampling the files to see how what bit-depth the differences are audible.

I am posting this in all seriousness and am completely open-minded to the listening test.

I hope you consider it.

Scott

2_channel_ears's picture

'nuff said.

VK's picture

...we would not have a functional computer, and internet would not work (neither our smartphones, etc, etc). But i agree with the noisy problem of the analog side of some DAC. So in this case, i have to blame the engineers behind the DAC project. They are messing up the analog part of it, because, bits ARE bits from the Spotify/Tidal/Netflix servers in God-knows-where in the world all the way down to your computer until they reach the "front door" of a DAC. After that we have problems. But until then, bits WERE bits.

Best regards!

kirkmc's picture

Yep, just as pizza is made of cheese, and dough, and tomato sauce, and anchovies, influenced by heat and time.

Clever Dean's picture

Is there any measurement data for the Regen product?

Archimago did a series using several intermediary USB hubs from Computer to TEAC to XPS-1 and showed 8Khz reduction (sometimes as much as 16dB).

Michael Lavorgna's picture
.
Clever Dean's picture

If I had a USB DAC (I'm running PCIe) I would certainly consider a discrete device like this. It's real grace is the fact that foot print is so small.

AQ is doing a killer business with their DAC's. I hope the jitterbug circuitry makes it into an upcoming DAC of theirs.

Michael Lavorgna's picture
AQ determined the most effective way to deal with the issues they felt needed addressing was to design a device that sits at the source, a computer in many cases, not at the DAC. So I'm not sure how they could build this into a DAC.
Clever Dean's picture

Can plug right into the USB port so the jitter bug circuitry would go first(?).

Michael Lavorgna's picture
I wasn't thinking about the DragonFly.
Clever Dean's picture

.

Michael Lavorgna's picture
;-)
Michael Lavorgna's picture
...of the REGEN performed by John Westlake over on the Pink Fish Media forum.

And here are a few relevant comments from John re. the results:

Now with the ReGen USB Data, notice not only the much cleaner waveform but far more importantly that there is very little "Runt" data - the background is Black. The USB Hub IC in the Regen has for the most part cleanly repackaged the data.
...
For sure any signal with unwanted modulation is not a good thing to have "flying around" a DAC PCB - these modulation effects will be very visible with an RF spectrum analyser looking at the Audio outputs - how the Audio system copes with this is anyone guess, and this is not the only mechanism this unwanted modulation can cross into the Analogue domain.
marce's picture

Firstly bits and noise are two different considerations, though interlinked. But the way this is worded and often worded on audiophile justification sites/threads/adverts just obfuscate what the real issues are.
Bits are bits, sorry but that is how it is, we spend hours with SIV (signal integrity verification) tools fine tuning layouts to ensure that the bits are bits, that's how we can discuss these things on the internet, because those bits are getting through.
Noise, or EMC is always a problem be it mixed signal or just plain digital (EMC and signal integrity are actually two faces of the same coin, refer to the likes of Henry Ott, Ralph Morrison for more in depth learning, I would recommend it). But mixed signal design is done everyday in all fields of electronics not just audio and the problems are solved and overcome without all this hand waving, speculation and in some cases dodgy components, why cant this be done in audio.

A prime example is the Jitterbug, no measurements just marketing prose, yet all other noise reduction devices I am aware of and would use publish specs and the noise spectrum they target... though adding common mode USB chokes to the USB lines is done internally to most USB hosts to avoid EMC problems when cables are plugged in....
As said one you accept that bits are bits and the whole point od SIV is to get them bits from a to b with the data being read without error is what it is all about...

ChipotleCoyote's picture

...which is kind of the crux of the whole thing, isn't it? While this argument keeps being phrased as "bits are/are not just bits," we're surely talking about noise considerations. As for these particular devices, the Schiit Wyrd says it "isolates your USB DAC from the noisy USB power coming from your computer"; Audioquest describes the JitterBug as a "USB Data & Power Noise Filter; the UpTone REGEN's description spends a great deal of time talking about ground noise issues, packet noise modulation, the computer's "noisy" power bus, etc.

Honestly, I'm perfectly happy saying that "bits are bits" -- I've been a computer programmer of one kind or another for thirty years -- and also saying "circuits have noise issues." The problem these devices are trying to solve is ultimately in the analogue realm. I haven't heard any of them, but this doesn't sound nearly as banana crazypants as people keep trying to make it out to be.

PeterV's picture

Hi Michael, yes your 'waterfilter' reflection is exactly the point.A regen or a jitterbug or a high-end USB cable all act as such 'waterfilters' and they do work, since differences are clearly audible. But why not connect a NAS or SSD or SD card point-to-point as close as possible to the DAC in order to minimise signal path and conversions and RFI and other interferences? Best cable is no cable at all..isn't this true? Such architecture might bring the discussion on audible differences between cables and DACs to another level and less or not depending on filters or other devices to correct errors during or after transport of bits to the DAC. Point is that in this worldwide discussion regarding 'bits behaviour' and introduction of digital noise during, much technology, costs and 'waterfilters' are required and developped by the audio industry for the last meters of data transport, while we all know and seem to accept that potential RFI and other digital artefacts and influences on the same data will occur during the thousands of kilometers of transport before they arrive in our home... In dense areas the same bits travel through non audiophile ethernet cable and fibre optics and wireless connections. So what is required to cancel this noise in our house, if the noise outside seems to be not so important..? This is very unlogical to me. Does it mean that as soon bits are stored bitperfect, all analog noise and jitter is intrisically deleted? If so, why is'n this fact used in audio properly? How come that we are able to download Highres files coming from a server on the other side of the world is NOT truncted or influenced by jitter or any other form of degradation, while this does happen inside our audiochain at home...? Ignoring this fact simply makes it all too easy for the 'bits are bits' fallacists.. If bitperfect transfer is possible via the world-wide web, what is the problem to achieve the same in our audio gear..

Michael Lavorgna's picture
If you read the title of this post you'll see I'm talking about "noise in mixed signal systems". If you read the post itself, you'll see that the issue we're addressing is noise in a DAC. Your focus on "the bits" misses this key point.
PeterV's picture

Hi Michael, thanks for your feedback.

Your article and those of the electrical engineers in the footnotes made it clear that transfer of bits towards a DAC can become polluted by jitter and analog noise.

The only point what I am trying to address is that all these methods of using better cables, purifiers, regenerators etc. still leave their personal sound signature on the final analog sound after the bits (including the impurifications intruduced during the transport), have been processed by the DAC..

To my understanding, the DAC is nothing more than a computer, which will work best if the datastream (indeed in form of bits) have not been infected by artefacts like jitter or other errors.

Why not shortcut all this and freeze the bits properly and pure closely to the DAC and in this way assure the digital content will arrive much more accurate, less distorted by any form of analog noise and jitter? It simply annoys me to understand that the data are available in pure form on my NAS or harddrive or USB stick, but that the last part of transporting them to the DAC has audible influence on the final result..

Does this require a completely different hardware and software appproach, or is this method actually allready being used for years in ipods and the new high-end versions of it, like Astell & Kern and Audio Research players?

I am probably biased ever since I own the NAD C390DD DAC/Amplifier. From the beginning I was hoping to achieve the shortest signal path for .WAV files directly from a chip via I2S-I2S into the DAC. The purest 'less-is-more' approach which would not require cables or conversions.

But maybe you are right and am I one of those bits 'fallacists' and not understanding the issues you are addressing here or in other topics :-)

Thanks for keeping this interesting discussion on the agenda!

Michael Lavorgna's picture
Listen, enjoy.
marce's picture

A simple question, how do you know you have noise present, have you measured?

Michael Lavorgna's picture
I clearly heard a difference with these products.
marce's picture

No you perceived a difference, with no data to back it up (especially when we are talking the high frequency noise from a computer or other digital system)it is pure perception and nothing else... or maybe Cognitive biases. Of course the manufacturers of these products know that the average Audiophile wont have the tools to measure whether there really is a problem so we have a whole host of USB add on products that address a problem that probably is not affecting the sound quality, but it sells more products. Of course adding all those connector interfaces is going to have a detrimental effect further negating any possible improvements.
Regarding the jitterbug looking at the layout it has too much capacitive coupling to actually filter the noise effectively, again there are no published measurements with that product.
:)

Michael Lavorgna's picture
If you do, that's fine by me. And I don't buy the "Cognitive bias" argument one bit ;-)
PeterV's picture

Hi Marce, what makes you so convinced that what someone hears is nothing more than a perception? I am sorry to say, but I find it pretty naive to assume that measurements and measurement instrumentation is able to capture all our ears and brain are able to hear and process. It is a very simplistic point-of-view by claiming this!

marce's picture

Sorry but you are wrong, we can measure changes down to points of a dB... Are visual and hearing system is easily fooled as well as the fact that we perceive things, our brain plays about with the data entering our heads.
So if you follow the scientific method you have a hypothesis (you hear a change) you perform experimentation and measurements and if the results are valid you have a theory, nothing simplistic in that it the proper way to do engineering.

marce's picture

Well we will have to disagree, but you cannot design products without measurements to back up what you think you hear, that is definite, especially when we are talking about the possible effect of digital noise on sound quality, as it is often in the MHz range of frequencies.
I wont discuss this any more because I know from discussion from others that have a total belief in their hearing that your view (though mistaken) is like many of a similar mind is belief based so hard to discuss in an open manner.
cheers Marc

Michael Lavorgna's picture
If you read what the manufacturers of these products write, they all perform measurements. If you know anything about the designers of these products, you'll know that they know what they're talking about.

The idea that "marce" knows more than these designers do about their own products, that you "know" they do not offer an audible change in a system without trying them, and that you "know" I must be hearing things, makes you a believer of the most obstinate type, imo of course.

If you want to discuss the design of these products, talk to the designers. Advice: Try being polite, i.e. not dismissive, if you want to have a discussion. If you want to talk about what these products did in my systems, feel free to ask me a question, same advice applies.

Cheers.

marce's picture

Wow you do get tetchy with someone who questions what he is doing and does not blindly go with the flow.
I did not say I know more than the designers, that is your comment, I made an observation, regarding layout for EMC and filter circuits, especially where HF noise is involved. EMC is a big thing in electronics and critical, also you will find most PCs and similar will filter any output that is likely to have a cable attached to avoid EMC problems (Radiated emissions mainly)with the cables acting as antennas.
Just an observation, don't read anymore into it than that.
Au revoir
Marc

Michael Lavorgna's picture
Here, let me remind you:
"No you perceived a difference, with no data to back it up (especially when we are talking the high frequency noise from a computer or other digital system)it is pure perception and nothing else... or maybe Cognitive biases. Of course the manufacturers of these products know that the average Audiophile wont have the tools to measure whether there really is a problem so we have a whole host of USB add on products that address a problem that probably is not affecting the sound quality, but it sells more products. Of course adding all those connector interfaces is going to have a detrimental effect further negating any possible improvements. Regarding the jitterbug looking at the layout it has too much capacitive coupling to actually filter the noise effectively, again there are no published measurements with that product."
Au revoir, pal ;-)
marce's picture

I commented on the layout, something that I am qualified to do as I do sensitive layout and consult companies on good layout practices for EMC, the companies I consult and do layouts are military/medical/aerospace mainly.
I still stand by my view that I would not pass the jitterbug design if it was presented to my for approval.
It is a shame that I cannot post a view here without being put down or having basically non sensical replies to my comments.
Many comment that they don't have an understanding of bit transport and system noise, yet when someone who has 30+ years in this area comments you jump on him because his views do not correspond to your own.

Michael Lavorgna's picture
: good-bye —often used interjectionally
Dr. AIX's picture

Due to the generosity of a fellow audiophile, I will doing a rigorous test of a REGEN box next weekend. The provider and several other members of the local audiophile society will be hear to listen and compare as well as validate the comparison of the digital data presented to the DAC through the REGEN and through a straight USB cable. It should prove interesting.

[self promotion deleted, Ed.]

Michael Lavorgna's picture
...than the hundreds of comments that already exist re. the REGEN. Not to mention the reviews. Besides I, and many others, already know what the REGEN does.

Have fun!

In case you missed it, here's my review.

Edit: I keep thinking, Mark - why would you think I'd care that you're going to listen to the REGEN?

X