Audiophilleo

[Elprior]

He there,

 

has anyone heard of these beasts ?

http://www.audiophilleo.com/default.aspx

 

Features seem very cool for the price tag.

Now the comparison table is a controversial exercise (and prone to hot responses).

 

Elp

 

[juanitox]

also interested! two many usb/spdif converter to have a clear idea..

 

[Part-Time Audiophile]

n/t

 

[labjr]

I've never heard of the company or the owner. Competition is good for the consumer though.

 

I wonder if they'll make a DAC next?

 

There must be a large enough market for USB to S/PDIF converters because there seems to be more and more of them coming to the market and Audiphilleo's only two products are converters. M2tech's first product was a converter too.

 

[cfmsp]

http://db.audioasylum.com/cgi/m.mpl?forum=pcaudio&n=73763&highlight=audiophilleo

 

 

ELP,

there was a discussion about Audiophilleo a while back on AA.

 

Back then, I was amused by the quality & sources of info for their comparison chart.

 

A quick review now seems to indicate that they updated it to be somewhat more accurate.

 

Apparently, they've named a feature "BitPerfect", and when first published they didn't even give Weiss credit for having a "bit perfect" indicator. Perhaps someone corrected them.

 

Agreed with your use of "controversial exercize" - these types of comparisons cause way more harm than good (for my opinion of a firm) when info is (even slightly) misrepresented.

 

 

clay

 

PS, interestingly, the products like this I'd consider first (Wavelink and Legato) aren't even on the comparison.

 

 

 

 

[labjr]

"PS, interestingly, the products like this I'd consider first (Wavelink and Legato) aren't even on the comparison."

 

The Wavelink just came out didn't it? Well, can't blame the company for good marketing. Nothing new in this hobby.

 

I guess we'll find out soon enough if it lives up to it's claims.

 

 

 

[barrows]

the Legato is on the comparison chart, the Halide Designs Bridge is missing, I'd let them slide on the Wavelink, as they are brand new at this point. To their credit, the Audiophileo folks encourage the website viewer to let them know about products they may have missed, so that they can update their comparison chart: they seem pretty confident about the performance of their converter. Nice that they offer native driver playback of 24/192 files with Macs, and their jitter specs are pretty impressive if one can believe them. I am not an expert on measuring jitter, but they do give specifics on how they do it. Perhaps Pat (from ART) or Gordon (from Wavelength) can take the time to investigate their jitter claims (in their "spare" time!). It appears that they offer a satisfaction gaurantee...

Who is going to be first to give it a go? Whoever does, please report your findings.

 

[ar-t]

Ahem, it is ar-t.

 

Anyway............

 

I talked to the guy behind that product, when his comparison chart, first came out. Seems to be an ok guy. Might be a tad short on some details, but we can let him by. For a while.

 

Anyway, since that point, I have been asked, by many of you (and folks from other places, as well), how our unit stacks up. So, set out to measure jitter, using the method I believe he uses.

 

Hooked up a unit to our Wavecrest (same as he is using), and measured the jitter on the rising edge. Only I averaged around 1/2 million readings. Got a number, that was less than 4 pSec. Which, was not much different from the (internal) clock's jitter. This should not be surprising, as the clock reclocks the output, directly.

 

(I know some folks don't believe that number, but I am not all that concerned what some forum in Hong Kong thinks. Especially since we have not sent any in their general direction.)

 

OK, so, are these numbers believable, and what do they mean? How do they compare to the jitter numbers we see in Stereophile?

 

Well, as to the latter: not much at all. In a sense, you are comparing apples to apple sauce. Yes, made from the same thing, only not the same.

 

First off, all we have (for damn near all of these measurements) is just a raw number. It does not give us the frequency content, of the jitter. Nor, does it (in most cases) tell us if the jitter is Gaussian or data-correlated. On top of that, a raw jitter number is highly dependent on the carrier frequency.

 

This is important to keep in mind. Most times, jitter is spec'ed at word clock rate (44 kHz, for instance). When you hear someone throw around "Jitter under 2 nSec is inaudible", they are referring to some data that showed that to be true. For Gaussian jitter.

 

If it is data-correlated, that number goes down. How far? "Depends".

 

The other problem is guys like me measure jitter on the component that is the source of the jitter. IOW, the clock. So, if someone uses that 2 nSec mark as the goal, for something operating at 256 times Fs, that number needs to be much lower. Which is why you hear folks like me throw around numbers like "It needs to be down around 10 pSec." (2 nSec divided by 256 is somewhere under 10 pSec.)

 

(Yes, our extensive listening work, on clocks, confirms the clock jitter has to be below 10 pSec.)

 

OK, so what does this mean to you guys?

 

I'll try to put it in terms that make sense.

 

Our clock has a jitter number, of under 4 pSec. This would mean the jitter is inaudible, as it is lower than the magic 2 nSec, at word clock rate. Assuming it is Gaussian. Which it is. That is the strong point, of the Wavecrest. It can break out the jitter components into random and coherent parts.

 

OK.........a bit o/t.....but............

 

A well-known........uh, shall we call it "network controlled audio device", that is very inexpensive, and is a fav with the DIY crowd........

 

If you were to measure its jitter, it would have both components. Not surprising, as the clock and SPDIF driver are in the same chip. I forget what the number is, but it is much higher than 4 pSec, and has a considerable amount of data-correlated jitter.

 

So, does the guy's unit, we are discussing, stack up as well?

 

I dunno. He does not spec what kind of jitter. Come to think of it.........don't think we do, either. (Guess I need to change that.)

 

So, numbers may look good on paper, and may sway some buyers, but they don't always tell the whole story. And we have not yet touched on the frequency content of the jitter. The Wavecrest can not do that. At least not in a manner that is of any use. Random jitter that is very low in frequency is much more detrimental, than random jitter, that is further away from the carrier. This is (largely) a function of the Q of the crystal, used in the clock.

 

Which is one of many reasons why you will see jitter spec'ed at offset frequencies greater than 1 kHz. Numbers are better, at that frequency, and therefore look better on the spec sheet. Also, not as dependent on supply noise!

 

Of course, most SPDIF RX chips will have substantially more jitter, than any of these sub-10 pSec TX units. Then, you have more jitter, at the DAC chip itself. The type of DAC, how it latches the data, and some other things, will affect how much jitter you have at the output.

 

Measured by a totally different method......................

 

There you have it: more than you care to know about jitter.

 

Probably the only thing you really need to know about jitter:

 

You will know there is no jitter, when things sound right. Jitter makes bass sound loose and flabby. Top-end is rough and nasty.

 

Actual comment, from user (some of you might know), that touches on the jitter issue:

 

"How much of all the tweaking crap we went through in the last 20 years was

jitter? -- this sounds (way) better than anything the Theta Jade transport

can get the Gen Va to do (glass, AES, BNC). "

 

(Referring to hooking up one of our units, to his Theta DAC. User is a long-time, dedicated user of Theta products.)

 

Guess that 4 pSec must be real!

 

 

 

 

[RW]

Since the moment I plugged in the Audiophilleo USB/SPDIF converter, I have started listening to music again. In one word, no actually in three words, detail, detail, detail… The Audiophilleo has allowed the music to come through my system as if untouched. The level of detail I have been hearing with the Audiophilleo has me looking for better recordings and not better components. For years I have spent countless hours pretending to be listening music; I have to confess that I was really just listening to my equipment. I cannot say whether the sound stage is deeper, wider or whether the bass is punchier and for once, I no longer care because I trust that the music sounds like it was intended to sound like when played LIVE.

 

That is what I would have written as my conclusion if I were a reviewer. But I am not. I will say that I am not looking for any future upgrades to my Audiophilleo USB/SPDIF converter.

 

 

 

[cfmsp]

 

 

What DAC are you using with the Audiophilleo? And what other USB->S/PDIF devices have you compared it with, i.e. what were you using before?

 

Also, it would be useful to share the rest of your system, as most selections in computer audio seem quite inter-related.

 

cheers,

clay

 

[cfmsp]

 

 

Ar-t,

 

thanks for the sharing details about the Legato.

 

I wonder if we'll see the info you posted here on their comparison chart.

 

They've used less stellar posts from both AA and head-fi as 'references' for the comparison.

 

clay

 

[Part-Time Audiophile]

This is the reason this forum is a great place to hang out. What a great description! I now know more about jitter than I ever wanted to. And now, I feel like I want to know more. ;-) Well done, sir!

 

Now we just need you to crank out a device that allows for high-resolution playback instead of just redbook.

 

[ducks & grins]

 

[RW]

I have been using the MSB Technology DAC III. I was surprised by the degree of difference the Audiophilleo2 made in my system.

 

 

I listen to my system in one of the following configurations:

 

Vista PC-> Audiophilleo2-> DAC III-> Gilmore Audio amp

 

Vista PC-> Audiophilleo2-> DAC III-> Headroom MAX Balanced head amp

 

 

My approach to component selection is simple; I want every component in my system to not have a sound. I don't want components adding or taking away from the recording.

 

 

 

 

 

[Elprior]

He,

 

thank you people for giving this level of feedback so far.

 

I guess I should be pointing Audiophilleo to this thread so that they may be able to comment directly.

 

Elp

 

[firedog]

This product seems similar to the Empirical Audio and M2Tech products for USB transport>SPDIF.

 

Would be interesting to have a comparison of them.

 

Chris, maybe a good idea for a review: USB>SPDIF converter shootout.

 

[barrows]

sorry for bungling your business' name! ar-t! And, thanks so much for taking the time to educate us on some of the aspects of jitter, and measuring it. I really appreciate your input and expertise.

 

[pgruebele]

Thanks Elp for pointing this thread out to me and thanks everyone else for the feedback.

 

I will make every effort to keep the comparison chart up to date and accurate. Unfortunately most manufacturers don't provide published jitter specs, so what little there is came from reviewers or my own measurements. I did get some feedback from users and manufactureres and have made corrections. Due to popular request I also recently tested and added the M2TECH hiFace.

 

Pat, questions to you:

 

1. Specifically how did you measure your SPDIF output with the WaveCrest (since SPDIF will produce various clock periods you must do some filtering)? Is the 4ps measurement for the SPDIF output of the device or the clock inside the device?

 

2. Do you think that the WaveCrest results (RMS period jitter) paint a complete picture of the jitter?

 

 

Regarding jitter in general:

 

Although RMS period jitter is very useful information, it is really integrated phase jitter (or phase noise plots) which gives a more complete view of the jitter that might affect audio quality. This is why we augmented our originally published WaveCrest results with integrated phase jitter specs from 1Hz to 100Khz as published here http://audiophilleo.com/definitions.aspx?jitter

 

The noise floor for our WaveCrest is around 5ps so it is not possible for us to take measurements below that. We have actually obtained values that low in some of our tests. However we spec worst case results for all sample rates (44-192) clocks. Also note that we actually measure the SPDIF output, not the internal clock. Internal clock measurements would show much lower jitter values, but only the SPDIF output is of interest to the user since that is what is fed to the DAC.

 

Cheers

 

Philip Gruebele

Audiophilleo.com

 

 

 

[kamil]

Philip,

I've got the Hiface Evo and am considering taking up the money back guarantee offer on the Audiophileo. Should it be the 1 or 2? The only advantage would be the virtual cable feature, but how necessary is this if the device is connected without the cable and with a long USB cord? Does the quality and length of USB cord matter?

 

[pgruebele]

Kamil,

 

If you don't need any of the features of the audiophilleo1, then I would recommend the audiophilleo2 since it is less expensive and there is no reason to pay for features you don't need. Whether or not the VirtualCable feature would make a difference is difficult to say though.

 

Cheers

 

Philip

 

[kamil]

i guess a shootout could be in the works if budget and time permit.

 

[ar-t]

Hey, bub..............good to see you here.

 

OK, in answer to your points:

 

1.) You can set it up to measure rise (or fall) time, and there are a few options on where on the waveform it will do this. We have the $$$$ s/w (that probably isn't worth the money, but let's stay away from that for now), and it defaults to 20%-80%. Using 10%-90%, the numbers are worse, as there can be more aberrations, due to reflections. We stick a 75-50 minimum loss pad in series, just in case. (Doesn't make much difference with our unit.) At this point, you have several choices for how the data is formatted. Accumulated histogram seems to be the best choice, if all you want is raw number.

 

As I pointed out, the real raw numbers that anyone should care about is how much is correlated. In our case, it is zero. I have measured "competing" units that this is not the case.

 

All the other waveform selections can not yield a worthwhile number, as the period goes all over the map. But, by telling it to just measure the 20%-80% transition, we get around that problem.

 

 

2.) I think it is only a partial answer, at best. All it really shows is that our unit (and probably your unit, as well) have decent clocks, that are not corrupted by either mains noise and/or crud from the SPDIF signal, itself.

 

Now, as for phase noise...................(you may regret leaving me this opening, as I can go on for days about that subject)........

 

For us, that would be the preferred method. It shows not only the amount of "jitter" (us RF types always talk in terms of phase noise, as we think "jitter" is just some number the digit-widgets invented, since they generally have no knowledge of RF, and that is all they can comprehend!), but the spectrum, which is the real issue.

 

But............here is the problem. (Well, actually, there are 2.)

 

First off, as you mentioned the period changing............that changing period makes phase noise plots a mess. In the case of our unit, we would use a carrier of 1.4112 MHz. The plot looks ok, for a while, then you get all kinds on nasty "lobing" sidebands, that make the plot look like doo-doo. So, it is not workable, as method of measuring jitter. Now, if you just take the clock, and assume the output is pretty much the same, you can have a good picture of the output.

 

But, there are ways around this, sort of. We used a slightly different method, than you did. As our unit reclocks the output, we can feed a signal at that rate, into the reclocker. Assuming it cleans up whatever is fed into it, we can get a good estimate, as to what the number is.

 

The other problem, is what phase noise gear, do you use? We have several RF spectrum analyzers, but none have a low enough noise floor, to get really in close, to the carrier. To get one that does, well, they cost around $90K, and there are no deals to be had, on eBay. (There are rare deals on Wavecrest analyzers, which is the only reason we have something that sold for $77K, new.) So, until we can find the steal of the century, the Wavecrest is what we base this on. (We do have some other ways of measuring phase noise, but they are kinda geeky, and only tell another part of the story.)

 

(In theory, the Wavecrest, can measure phase noise, but you don't want to. It uses a stable 100 MHz oscillator, as a time base. Our unit, when used with the proper internal calibration technique, can get numbers that are pretty low, in terms of raw jitter. However, when you take that value, and extrapolate the phase noise, it is not that good. Indeed, the phase noise floor is much worse than any of our spectrum analyzers. So, without valid phase noise numbers to validate it, it is not the final authority on jitter.)

 

OK, now here is where I am going to disagree with you, some. We have been making CD players and digital audio stuff, for over 20 years. We found out, a long time ago, that the key to getting this stuff to sound right was to have a clock, that has very low phase noise, and the close-in noise is much more important. (OK, it also has to have zero data-correlated noise, but how hard is that do, with just a clock? It isn't. But is when SPDIF is involved.) So, we concentrate more on getting the close-in noise down. And assume everything else will fall into place, once you have that solved.

 

The problem with anything designers like us do, or measure, it may have diddly-squat to do with the end result. SPDIF RX chips have horrible phase noise plots. And a lot of it is data-correlated, which is the kiss of death. So, guys who design DAC boxes come up with all sorts of schemes to get rid of that. And as we both have said, different DAC chips have differing tolerances to jitter.

 

So..........having said way more than most of you really wanted to know, (but seem grateful for the edification), what you really want to know is:

 

"Can we believe his specs?"

 

Without having ever seen it, but knowing what is possible, realistic, and should be expected, I can say with 100% certainty "Yes". His phase noise numbers are in line, with what one would expect, using a decent crystal oscillator. Using those values, to come up with a single number for jitter yields a number that is valid.

 

Now.......a slight diversion, on that subject.........

 

"Oh, no.........don't you ever shut up!?"

 

Yes. Soon.

 

As in any kind of spec, you have to know all the details, to know if it is good, or just marketing BS. With our unit, it is easy: we have one clock, and that is all we have to measure. But what if you have multiple rates? That means multiple clocks.

 

As you increase the frequency of the clock, the same phase noise plot will yield a lower jitter value. So, one may conclude he is cheating, by using the highest clock rate.

 

No, probably not. Because as you go higher in frequency, it becomes harder to get good phase noise numbers. Without going to exotic crystal cuts, and using more than one crystal to clean up the noise. (No, please don't ask me to 'splain how you would use an expensive ST-cut, and then an AT-cut, as a filter. No one we know is going to spend that kind of money.) So, while he does not say which clock rate has what measurement, I can still say his numbers sound very believable.

 

There. I'm done.

 

Unless you ask about how the magazines measure jitter. Please don't. I do not have much faith in those techniques. Maybe for SPIF, they tell something. Useless for CD players.

 

[pgruebele]

By all means Kamil. The only way to tell if it will audibly improve your setup is to try it!

 

Cheers

 

Philip

 

[pgruebele]

Hi Pat

 

1. It's still not clear to me from your response: Do you measure your jitter at the actual BNC output?

 

2. We specify exacly what equipment and test methodology was used to obtain our phase noise plots and integrated phase jitter values here:

 

http://audiophilleo.com/definitions.aspx?jitter

 

Cheers

 

Philip

 

[labjr]

"...SPDIF RX chips have horrible phase noise plots."

 

Seems like time to get rid of the old DAC and get one with a USB or Firewire input and skip the converter. Are the old DACs so good that it's worth spending $1,000 to continue to use them?

 

[pgruebele]

But only the newer DACs have USB input and unfortunately for the most part I think the general consensus is that their USB input sounds worse than their SPDIF inputs...

 

Cheers

 

Philip