[Tachikoma]

Quote:

Originally Posted by DoYouRight /img/forum/go_quote.gif Bitperfect via asio or even clean power from non usb or clock. The Cryoparts Diverter has none of those and is $1000. Dont get confused in power hype, engineering is possible. Converting USB>SPDIF is no where near as power hungry as an actual DAC runnin on usb power alone.

The point is that its a lot easier to get clean power by not taking a noisy 5V line from the PC (the power isn't a huge deal). Yes engineering is possible but why go out of your way to solve a problem you could have avoided entirely in the first place?

 

[tosehee]

Quote:

Originally Posted by Tachikoma /img/forum/go_quote.gif The point is that its a lot easier to get clean power by not taking a noisy 5V line from the PC (the power isn't a huge deal). Yes engineering is possible but why go out of your way to solve a problem you could have avoided entirely in the first place?

If this is such a big issue, I doubt Diverter with $1000 would go then route.

 

[JayDee]

Quote:

Originally Posted by Tachikoma /img/forum/go_quote.gif The point is that its a lot easier to get clean power by not taking a noisy 5V line from the PC...

True, the 5V supply from a USB connection has a certain degree of noise, which is a function of the aggregate internal noise inside the computer, the coupling to other noise sources, as well as the characteristic noise of the computer's power supply. But a cheap external switching power supply is often no better, and in fact may be worse.

The scope plot below below was taken directly from the V+ and ground terminals of a generic USB cable, connected to an equally unremarkable PC. The scope has a 1.1GHz bandwidth, and the probe used is an 2.5GHz active unit with very low loading and an extremely short ground connection.

The next plot was taken under identical conditions, directly from the output connector of an external power supply provided as part of one of my competitor's USB to SPDIF products.

The USB's supply has a slightly higher RMS noise value, mostly because of some subtle, low-frequency undulations - easy for a linear regulator to track. IMO the more egregious offender here is the external supply, because of the high-amplitude, high-frequency noise spikes riding on its DC. These spikes are much more difficult to filter out, and linear regulators won't touch them at all - their control loops simply lack the bandwidth to do so, and even if they did, the parasitic elements in the regulator would likely prevent nulling spikes like this. Keep in mind that a power supply isn't inherently better merely because it is external, yet this concept seems to almost be a fundamental tenet of high-end audio. Granted, this external supply example isn't necessarily representative of all external supplies, but I wish people wouldn't automatically categorize a supply's performance simply as a function of being internal or external, because the assumptions are often erroneous.

 

[glt]

Very nice!. Thanks for sharing. First time I see real measurements of the usb power. Perhaps a laptop would be even cleaner?

This sort of helps me understand why I couldn't hear differences betwee usb power and alkaline batteries with the musiland (digital out). The local regulators will further filter the noise and the predominant noise becomes that of the local regulators.

 

[JayDee]

Quote:

Originally Posted by glt /img/forum/go_quote.gif Perhaps a laptop would be even cleaner?

Certainly a legitimate question. Here's your answer (at least one of them; you'd really need to take quite a few such measurements before drawing any valid general conclusions).

These plots were taken from an old Dell Latitude D600. All other conditions were identical to the ones from the measurements I posted last night.

The first plot shows the USB Vbus running from battery power:

This is indeed considerably better in all respects. The second plot shows the same laptop, but connected to the external AC supply brick:

Here, you can clearly see the ripple from the AC supply riding on the +5V. What you can't see is that the entire DC level shifts as the Li-Ion charge controller goes about charging the batteries in burst mode - there's a charging phase and then a relaxation phase (which is done so that the battery voltage can be accurately read) about every half second or so. In practice this shouldn't be a big deal, as any linear regulator of even marginal quality would be able to track it. Even with these new anomalies, the laptop appears to be better than the PC I tested last night, and is still considerably better than the outboard supply that came with the other USB to SPDIF device.

 

[DoYouRight]

And for those of you who don't know, JayDee is the creator of the Diverter. Wish I could afford one, but JayDee would you think this is the best of the bunch under $500 in terms of USB>SPDIF?

 

[JayDee]

I honestly don't know much about this product, but even if I did I think it would be unprofessional of me, as a competitor, to comment on it.

 

[gevorg]

Quote:

Originally Posted by Tachikoma /img/forum/go_quote.gif The point is that its a lot easier to get clean power by not taking a noisy 5V line from the PC (the power isn't a huge deal). Yes engineering is possible but why go out of your way to solve a problem you could have avoided entirely in the first place?

Exactly! There is nothing magical about USB power, its just a small power line from some switching PC power supply with added EMI/RFI/etc from the components of a given computer.

The graphs above for external power supply are misleading by using a random switching power supply, instead of a decent linear power supply which are available on the cheap too (~20 bucks).

USB power is acceptable for $100 or so devices, but if you're going with a high-end device, like Empirical Audio Off-Ramp, you expect there to be a dedicated/battery power. Unless of course, its a snake-oil product that claims some secret "magic sauce" inside that fixes all problems.

 

[Edwood]

Would really be nice if the M2Tech Hiface were available in the USA.

 

[laevi]

JayDee,

What would the USB Vbus plot look like from a laptop running off of its external AC supply brick but without its battery inserted?

Would it be cleaner than the external brick + battery because the Li-Ion charge controller may be deactivated?

 

[JayDee]

Quote:

Originally Posted by gevorg /img/forum/go_quote.gif Exactly! There is nothing magical about USB power, its just a small power line...

Correct, there's nothing special about the USB supply, but my whole point was that there's not necessarily anything special about an outboard supply, either. I wasn't comparing switchers to linear supplies, I was only addressing USB-derived power versus and external supply, and the mistaken assumption on the part of many the that latter is somehow inherently superior. It may be in some cases, but certainly not always, and this was just an illustration of that. I posted it because I'm frankly rather tired of hearing people simply parroting the idea about on forums that the 5V supply from USB is some dark and evil thing, and that no serious product could possibly use it - so here's some real data to show otherwise.

Incidentally, the USB supply is capable of 500mA - more than enough for any USB to SPDIF conversion I could conceive of, and then some. There's no benefit to additional amperage capacity, unless it it has the side effect of yielding a lower output impedance, which is beneficial.

Quote:

Originally Posted by gevorg /img/forum/go_quote.gif The graphs above for external power supply are misleading by using a random switching power supply, instead of a decent linear power supply which are available on the cheap too (~20 bucks).

I can't see how the graphs are in any way misleading. If I had taken a measurement of a switching supply and labeled it as a linear supply (or the reverse), that would be misleading. What you see is what you get, and this was representative of the supply that was actually shipped with one of the USB to SPDIF converters on the market currently (and not an inexpensive one, either) - hardly "random." No deception here.

EMI/RFI is an unavoidable consequence of the modern world, regardless of whether you choose a linear or switcher supply. A linear supply in the $20 range would likely offer poor filtering indeed for high-frequency interference.

Said another way, my point here was simply that the individual implementation is what really matters, not the underlying design principle. Its is entirely possible to achieve excellent performance with linear, switching, internal, or external supplies. Each has its advantages and disadvantages, but in the hands of a good designer any combination of these can be made to work well.

 

[JayDee]

Quote:

Originally Posted by laevi /img/forum/go_quote.gif JayDee, What would the USB Vbus plot look like from a laptop running off of its external AC supply brick but without its battery inserted? Would it be cleaner than the external brick + battery because the Li-Ion charge controller may be deactivated?

Very likely yes, it would be cleaner. I've already indulged in too many measurements in this vein today, so I can't go take a measurement just now. If the battery were fully charged, the supply would also be cleaner in that case because the charging cycle would be completed.

I wouldn't advocate that people go and take the batteries out of their laptops in order to listen, because this charger noise is relatively benign and easily tracked by any decent linear regulator. Also, it is low enough in frequency that it's outside of the audio passband, so even if you used the straight, unprocessed supply, this particular effect would likely be inaudible unless the charging were severe enough to current-starve the USB supply (unlikely).

If we want to get into some tech details, I should note that there are linear Li-Ion charge controllers, and there are also switching ones. Just as in the case of supplies in general, the former type would tend to be quieter than the latter type. In the case of the measurements I made, I suspect the charge controller was a linear type because I didn't notice an increase in high-frequency noise when the charge controller was active.

The "bad" components of the power supply we'd want to avoid in these plots would be the high-frequency hash, not the very low-frequency anomalies.

 

[Tachikoma]

The thing is you could measure the V+ on 10 computers and get 10 different looking plots (In the end, all USB power lines are derived from one massive switching PSU, and you could easily have plenty of variance in the quality of those). Is there any reason you couldn't have used something fancy like say, a super-reg to power your device?

 

[JayDee]

Quote:

Originally Posted by Tachikoma /img/forum/go_quote.gif Is there any reason you couldn't have used something fancy like say, a super-reg to power your device?

It sounds like you got the impression that I'm powering the Diverter's circuitry from the USB V+ line directly, with no further conditioning. Nothing could be further from the truth; in fact there are three very high performance regulators inside, and quite a bit of pre-regulator filtration as well. My point in these posts is simply that for something like USB to SPDIF conversion, the USB's V+ line is a perfectly workable starting point, and if it's processed properly, it can be the basis of a truly high-performance power supply.

I kind of feel like I'm hijacking this thread, which is supposed to be about the M2tech Hiface. That wasn't my intent, so I'm going to fade out here. I've written quite a lot about the Diverter (including the power supply) over here: http://www.head-fi.org/forums/f7/son...review-445553/ So if you have more questions about it, I think it would be most appropriate to continue the discussion over there.

 

[DoYouRight]

tosehee, any news on the x64 drivers and stability/sq? I am all for one when Windows Seven x64 is working flawlessly like XP.