I used to think the DAC was 90% of the sound. This sub-$300 DIY project proved me wrong

Tinkering
333

Yes, there is. The fact that the physical layer works with analog voltages is not relevant once the bits are recovered and go into buffers.

Infinites are more of a mathematician’s than a physicist’s thing :slight_smile: Not that it matters, as digital signaling doesn’t rely on infinites. Moreover, a square wave is but one way to transmit a digital signal. There are many other ways, many of which rely on sine carriers (e.g. phase- or amplitude-shift keying, QAM).

Again, that’s just one way to resolve bits. Other ways rely on wave fronts, so the actual shape or amplitude of the wave is largely irrelevant. And when based on threshold, the voltage bands corresponding to ones and zeroes are quite large and rarely, if ever, become ambiguous. If detection is based on wave fronts, then the shape matters even less. If digital transmissions were as fragile as you think, we would not be able to download anything from the web or use undersea cables thousands of miles long.

Which disappears without a trace once the bits are stored in a buffer and are picked up one by one by the DAC’s - not the transport’s - clock.

What happened when you unplugged the network cable with RAAT? The music continued for seconds. It seems plenty clear to me that the exact time the bits were resolved off of the physical transport is completely irrelevant.

4 Likes
334

Tell you what, gents, this is properly interesting :nerd_face:

Zero sarcasm, it’s something I don’t really know about so it’s good to follow the debate

2 Likes
335

Thanks David. I really appreciate your openness and willingness to accept the findings.

This reminds me of the episode about Johannes Kepler in Carl Sagan’s Cosmos. Kepler had been looking for supernatural explanations of the movements of planets, but when the observations of Tycho Brahe, which were the most accurate at the time, proved all his theories wrong, he had to throw everything away and start from scratch. The result was something way less exciting: ellipses. He gave up the occult and embraced science. No parallels to actual living persons…

My understanding is that this is also Diretta’s theory, at least in concept, so I believe this invalidates the reasons for its existence.

I’m not sure I agree with that. When people compare different setups and decide which one is “better” - whatever that means - they probably focus more intently on the novel solution than they used to with their current setup. This heightened focus may, by itself, explain why people hear more details than before. And that’s why I think it’s important not to forget to go back to the previous setup with the same level of focus, before jumping to conclusions.

Take for example the classical experiment with the same wine in different bottles with different price tags. You can say people appreciate the “more expensive” wine because they’re snobs. But you could also say they genuinely enjoy the “more expensive” wine because it commands more of their attention and thus they focus more on its qualities. When you do blind tests, not only do you remove potential bias, you also level the playing field in terms of focus. Bottom line, I think focus is key; it’s called critical listening for a reason.

Regardless, I may have had to persuade my sons to do it (and still need to do it to complete the last phases), but once they started, they did it at their own leisure and did their own switching, without me being around, so I don’t think there was any pressure.

I can understand the engagement, as I explained above. Now that there is some objective data in the mix, some of the listeners may be less inclined to take Diretta’s claims at face value and more open to the possibility that’s it’s just the excitement of a new toy.

I’ve looked at the power rail noise across the audible spectrum. The oscilloscope may not be that accurate in terms of FFT. I think I can do a more thorough analysis of the frequency components using the ADC. I’ll post the results back in the measurements thread when I have them.

If that’s the case, wouldn’t a well engineered DAC be the natural solution to whatever the problem Diretta is trying to solve? The D70 is not an expensive DAC, so it’s possible to do it at affordable prices. Hi-fi is a commodity.

2 Likes
336

Sorry but …

  1. you previously agreed in this very same thread that
    a) jitter on the USB audio connection is irrelevant with an async DAC, and
    b) excess noise on the USB connection could be eliminated with a good USB isolator (or an extremely low noise Diretta target)
    c) and thus that your theorized sources of Dac interference were either coming back in via the mains conduit from the target or via RFI

  2. Now that objective tests didn’t provide any evidence/support for your theory for rails/mains noise, you are back on a theory that noise is propagating on the USB audio connection. Again, if that is really the source of noise in the Dac why not just ensure that the USB connection is well isolated ?? eg. with a device that has been objectively measured to isolate/remove any excess noise (audible or not). That would be a heck of a lot easier than your current setup no?

3 Likes
337

Well, the evidence supports that assessment. The existence of this sector is by no means proof of anything in itself. It’s a self-fulfilling prophecy: the industry produces “solutions” because people believe in the “problems”, and they believe in them because of the existence of the “solutions”.

That’s not how differential signaling works. The signal is simply the voltage difference between D+ and D-. There’s no requirement that D+ and D- be exact opposites relative to some reference; the necessary and sufficient condition for a balanced connection to reject common mode is that D+ and D- have the same impedance relative to the ground.

2 Likes
338

Agreed.

David is correct in that the USB standard does require a ground signal. That is not just in order to provide power (alongside what was the 5V line on traditional USB) which is an integral part of the USB standard. It is also used in conjunction with the D+ and D- lines to signal certain conditions in, for example, target detection and line rate negotiation.

The ground signal in USB is not, however, used by the receiving end to interpret the digital data being carried on the bus. I read somewhere someone referring to this as ‘not true differential signalling’. However, I think it is more fair to say that USB as a whole ‘doesn’t use only differential signalling’. Differential and ground relative signalling are not mutually exclusive provided the domains in which they are used are separated in some way.

3 Likes
339

Hi folks,

I think we have reached the point of diminishing returns on the theoretical debate here.

To @nquery and others pointing out the contradiction regarding jitter: You are right to call that out. In my attempt to explain the fragility of digital-as-analog, I used “jitter” as a catch-all term for timing uncertainty, which muddied the waters. To be clear: I agree that digital audio jitter is a solved problem in modern Asynchronous USB. My hypothesis remains focused on System/Ground Plane Noise modulating the operating environment of the DAC, but I accept that my previous post conflated those concepts. Thank you for the correction.

The State of Play

We now have two very different worldviews operating in this thread:

  1. The Objectivist View: If standard measurements (SINAD, Voltage Ripple) don’t show a change, the audible difference does not exist or is placebo.

  2. The Tinkerer View: If multiple experienced listeners consistently hear a durable improvement that correlates with architectural changes, there is likely a physical mechanism at play that we haven’t found the right way to measure yet.

Neither side is going to convince the other today. And frankly, endless debating of theoretical physics and USB specs is drifting far away from the purpose of this thread.

So, let’s do this:

  • If you want to explore measurements, null tests, and the validity of subjective listening: Please head over to @Marian’s excellent Diretta Measurements thread. That is the perfect venue for that rigorous analysis, and I genuinely look forward to what is discovered there.
  • If you are building this project, stuck on a configuration step, or want to share your listening notes: You are in the right place!

I created this thread to help people build a sub-$300 transport that makes Roon fun again. I’m going to get back to doing exactly that.

To the builders: How are your Diretta kit builds coming along? There’s a new 1 GB RPi5 board that will be available soon in the US for $45. I think it would make a great Target and may even be adequate for the Host now that we have a way to disable Energy Efficient Ethernet (EEE).

3 Likes
340

Whilst I have been reading this thread from the beginning, I have been loath to get involved until recently for just this reason.

The problem, as I see it, is that in the course of this thread you have not only described the ‘how to do it’ and the impressions that it left you with - neither of which I have any argument with. You have also put forward a couple of different hypothesis regarding the physical mechanisms that might be leading to the improvement/difference that you experience.

To me, whilst no-one can or should challenge your (or anyone else’s) experience (or the enjoyment obtained from the process), the validity of any hypothesis put forward to explain that experience in an apparently rationalist manner should be challenged if it does not hold up to scrutiny. Leaving a flawed hypothesis unchallenged lends it a credibility that it does not deserve.

If you wanted this thread to be a thread for reporting ‘experience’ and ‘impressions’, maybe you should not have put forward those hypothesis in the first place.

2 Likes
341

Hi @Wade_Oram,

You make a fair point. I did open that door, and the community has responded by rigorously challenging those hypotheses.

I think we can agree that the “scrutiny” phase has been thorough. The counter-arguments are now clearly on the record for any future readers of this thread.

With that context established and the critique noted, I am returning the focus of this thread to the practical build.

Thanks.

1 Like
342

Arriving back from Thanksgiving I found the kit that David put together for me, and went to work installing it. Some glitches that related to my dCS Upsampler were sorted out with David’s assistance. Got down to some serious comparisons as follows:

  1. dCS Upsampler as Roon Ready endpoint.

a) The Roon Server running on Mac Mini M1 wired to Uptone Audio etherREGEN switch;
b) The dCS Upsampler also wired into the switch;
c) Music coming from Synology NAS into the same switch;
d) Upsampler feeds a dCS DAC;
e) Output to Spectral preamp into a pair of Spectral mono blocks then feeding Wilson Audio Alexa V speakers.

  1. Diretta Host/Target.
    Host connected upstream to the same switch, and downstream to Target feeding the USB input on the Upsampler.

  2. Grimm MU1 streamer. Wired to the same switch, and sending its output via AES/EBU into the Upsampler.

I was able to switch inputs on the Upsampler via dCS Mosaic app, and play via Roon for each zone. All levels set as fixed out of Roon. And only used 16/44 source material.

Results: the dCS Upsampler as endpoint yielded a fine baseline. The Diretta was better, providing what I can best describe as a more etched and seemingly more dramatic play with clearer separation / wider soundstage. However, the Grimm was best as it provided a more refined sonic picture. More natural and capable of nailing the vocal and instrumental nuances. It added a tad more depth and texture, and was more relaxed. And it was better able to retrieve more from the sources and present it in a very natural way. This was especially true of live performance.

But considering it’s a far pricier endpoint than the Diretta, I’d have been shocked if the Grimm had failed to yield a better result. And I cannot discount the dCS components don’t play a significant role in all of this. And there may be some benefits in upgrading the power supplies, cabling, and clocking of the USB that might yield a closer result.

It was certainly worth the expense, time and trouble to test this out. I’m pretty sure I’ll find a home for this rig once I get it into the hands of others in our Bay Area group.

5 Likes
343

You should check out the work by @Marian (above) on the objective measurements of Diretta vs. RAAT. Since they are both bit perfect and there is no obvious transfer of jitter or noise to the Topping D70 DAC, it looks unlikely that there are the differences you think you hear.

But there is always a chance! So, for a fraction of the price of your kit, you can easily get a Cosmos ADC and run similar tests to validate your observations. What an incredible opportunity!

2 Likes
344

Why bother? It changes nothing. Most of us don’t need measured validation for our findings in hifi. I have seen dozens of ”perfect” measurements of gear which, when auditioned, sounds totally different from each other and have unique sound signatures. DACs are the best example of this but few preamplifiers come to mind also. These newest measurements of Diretta by Marian go to the same category. While measurements are interesting to a certain point, there’s no amount of measurements that could make the audible (or imagined, if so you like) differences disappear. Either we measure wrong things or we somehow imagine these constant and steady differences, which usually stay the same even if you audition for months before making the final decision. Even if it was all placebo, which I don’t believe it is, you can’t get rid of it since you always listen to your system knowing the whole equipment chain. Placebo is always there! Nobody listens blind at home, except blind people.

My advice to everyone is to borrow/try the interesting gear (like this Diretta configuration), audition it for a proper amount of time (two weeks minimum) and if it’s good, keep it. If not, return it/stick with what you already have and move on. It’s really as simple as that.

To me, Diretta makes a small but worthwile difference. Slightly more air and space around instruments, compared to streaming over HQP NAA. In a big picture it costs next to nothing, which is a nice bonus.

7 Likes
345

I’m just intellectually curious about how stuff works! I want to know what modifications to signals and DAC performance result in some of these hearing experiences. Measurements provide depth to these ideas as opposed to just claiming something (there’s a Hitchens’ Razor in here, of course).

Investigating the performance of a system can help other folks to make informed buying decisions, as well. I wouldn’t buy a car based only on an enthusiast claiming that it made the road seem more spatial. I want 0-60, 1/4 miles, Nurburgring, EPA, trunk capacity, wheel size, range… I can’t buy cars and try them for a few months to get a gestalt feeling.

1 Like
346

Understood. Having been at the audio system pursuit for more decades than I’d like to admit, I’m with you. However, we all find ourselves in the position of needing to “test drive” anything we add/subtract, since we rarely get to otherwise experience much of “our” system outside of our own listening room. If we can, we obtain a trial piece of gear. And our systems are nearly always bespoke, with a vey unique chain of components and cables, etc. It’s never close to a wholesale drive off the lot test. Even reading trusted reviews in magazines cannot reliably inform us how a product might benefit our own systems. And with the ongoing loss of brick and mortar stores to even hear and possibly borrow gear it’s only gotten worse. So, it’s often, as in this case of Diretta, a roll of the dice.

347

All of this value structure makes me want more and better measurements and evaluation before I get started with buying or evaluating myself! If a manufacturer claims a cable can improve sound quality in any system, it should be demonstrated before I try it in my system. Then I can feel confident about my listening observations. I know there is a potential value-add.

Moreover, I know what to look for. If a speaker has a boost to mid-bass as measured by Klippel, I can feel confident about my sense of flabby bass. In the case of Diretta, there is no impact to the response of a high-quality DAC. I don’t know what to look for other than hopes and extended hypotheses about alternative paths for noise transfer?

1 Like
348

About the car analogy, you borrow a car and take it for a spin before buying it right? At least in Finland, you can borrow the car overnight and drive two full days if you like, even longer I think. I’ve never needed longer than that but I did test drive one same car three times and ended up not buying it. When I bought my current car, I took six different cars for a test drive, few of them two or three times. Obviously I chose the cars to test by reviews, specs and price. But the test drive is the most important thing, just like in hifi. No, you can’t buy cars and test them for months but yes, you can borrow hifi gear and test it for weeks, at least here in Finland.

1 Like
349

I’ve actually bought cars several times without driving them. They are mostly interchangeable when reviews and specs are considered.

350

The car analogy only goes so far because the specifications (resulting from the measurements that the manufacturer wants to present) do not cover all aspects of what makes a car suitable for any given person. Some measurements are not published by the manufacturer because the end users would be unable to interpret them. Some things can’t be measured because they are personal.

For myself, when I want to buy a new car, I don’t go out and test drive all cars on the market. I apply filters to that set of cars based on cost, size, performance, seating and boot capacity and maybe a couple of other parameters that are usually published and then I select a further subset of these cars to test drive. The first filtering is effectively filtering on ‘measurements’. The second is more subjective and personal (do I like the style, how far would I have to take it to get it serviced, is the extra cost over another model worth it to me etc). Finally, I make my selection from that subset of cars that I test drove based in part on subjective criteria. The important point is that any of the cars that I put on my shortlist would likely be good enough on performance grounds because I looked at the specs (measurements).

I choose audio equipment in a similar way. I select a subset of devices that I am interested in based on specs (measurements, cost) and then I further refine that subset based on subjective aspects (aesthetics, value for money and availability for example). This gives me a set of devices to look at further and only then, do I choose the device amongst those remaining that pleases me most (not necessarily differentiated on audio quality).

2 Likes
351

Don’t neglect reviews and third-party testing that provides an evaluation of the car from seasoned reviewers.

1 Like
352

Actually, with regard to car reviews, just as with audio equipment reviews, I pick and choose which parts I pay attention to. If the car review comments on reliability, that will influence me. If it comments on accessibility (e.g. difficult to get into and out of for tall people), that will influence me. And so on. These aspects tend to fall into the measurements category. However, if the review says ‘beautiful styling’, ‘involving driving experience’ or some such, I pay no attention.

Similarly with audio equipment reviews. I pay attention to comments on usability (do the knobs and buttons interfere with each other, is the display clear) and build quality but I do not pay any attention to comments on audio quality based on subjective listening tests.

2 Likes