USB DAC shootout featuring JPLAY and the Rega DAC, Beresford TC-7520 (modified + Burson Buffer), Halide Design DAC HD, JKDAC , MK3 JKSPDIF and JKDAC32.
I'm either a
most unlikely or a most ideal person to run a DAC review. You'll have to decide
for yourself which you believe I am. I've been a vinyl junkie for some years. I
flirted with CD, starting out in the mid-1980's with a Philips CD-104B. I was
hoping for a more "lifelike sound" as I progressed via various transports &
DACs until I reached a Meridian 588 not long after the turn of the Millennium.
More recent mid to upper-mid market CD players haven't really delivered much
improvement to my ears. During my journey I never quite left my record decks
behind and over the last 10 years I moved my prime focus back to vinyl again, my
current decks being a Trans-Fi Salvation rim-drive and Garrard 301 idler.
better of my two record decks is the Salvation, it's by no means warm and
cuddly, it digs tremendous detail from the grooves, produces massive 3D images
and lifelike sounds; it's no shrinking violet that's for sure. The Salvation is
close to as good as it gets with a record deck so it's a useful reference as
well as a huge pleasure to listen to. What I've been searching for is a CD or
computer audio system which sounds lifelike and at a price that won't bankrupt
me. Sure the likes of DCS produce great sounds but at great cost too. Does it
have to be like this? It's just technology after all and this progresses really
quickly. The biggest problem I've had with CD sound over the years is a slight
upper-midrange harshness or over-emphasis coupled with a sound that's closer to
2D than 3D. Where folks don't have a good analogue front-end to compare with,
the deficiencies I mention become accepted. The slight harshness can be smoothed
over but usually with a consequent loss of fidelity. A couple of years ago I
setup Foobar2000 with ASIO4ALL for "bitperfect" running on Windows XP. Once I'd
cut out many unnecessary operating system processes I found the sound was better
than my Meridian 588 so I proceeded to rip my CD collection and sell the
Meridian. Even though the sound was improved over the Meridian it was still
somewhat stereotypically CD-like in my view. Having a very good record deck to
compare with meant it was hard to adjust to my CD sound. I was ok to listen to
my file-based audio one night and vinyl the next. Mixing the formats during the
same session usually led to dissatisfaction.
A few more words about my system; it includes a
DHT preamplifier or TVC preamplifier as my mood takes me. I run 300B SET
amplifiers feeding purely Open Baffle speakers with a serious bass capability
assisted by DSP and solid-state clout.
Ok, so now you know something of where I'm coming
from it's time to dig into USB-Audio Heaven!
First Step to Heaven = JPLAY
If we are to use a computer as a music server the
main options include specialized hardware from the hi-fi industry, MACs, Windows
PCs, and Linux PCs. I'm not an electronics designer so I'm not going to even try
to preach what is the best technical solution. Here's my take on some of the
Hi-Fi servers: A known quantity and quality, a
dedicated and convenient solution for a family to use.Generally expensive.
MAC: In standard setup sound quality is ahead of Windows, optical output seems preferred with some reporting issues with USB.
So very prevalent and it is well understood what needs to be done to easily
improve on the sound that can be corrupted by the operating system. Windows 7
SP1 was a big improvement in terms of USB.
Linux: This low cost or free operating system is a tweakers delight and should be technically the easiest to get a great sound out of. The problem is that there's a learning curve and little standardization or pre-packaging.
what's the big issue here? Whilst DACs re-clock the data as it hits the input
buffer, jitter still impacts sound quality. The chances are that no two general
purpose computer setups will have the same jitter profiles; there are so many
variables involved. All the above options will sound different, having their own
personality if you will.
A general purpose computer is not a Real-Time computer so for our data packets jitter gets introduced when it does not arrive in the required timeframe. Real-Time? Imagine a vital industrial process, this could be the control of robots on a vehicle assembly line. Just imagine if the computer couldn't respond in time to install a windscreen (aka windshield). Sorry the disk was accessing some other data and other processes were running so the windscreen became a sunroof. Maybe next time we'll have more luck....
My installation of Foobar2000 with ASIO4ALL,
albeit tweaked, cannot in truth get close to mimicking a real-time computer;
Windows is doing so many other jobs in parallel. I tried Jriver in various
modes, the user interface is sexy and the sound improves a little over Foobar
too. At the behest of John Kenny, one of the DAC producers involved with the
review, I tried JPLAY. JPLAY cordons off its own memory, reading tracks from
disk ahead of playback. In hibernation mode it stops much of Windows operating;
it does all this to enable the computer to respond with data packets with very
low latency. Want to run it on a Mac? I hear JPLAY works well running under
Windows on a Mac.
Finally we have something approaching a Real-Time
computer. There is the capability to run JPLAY under Jriver, Foobar and iTunes
but the very hairshirt option of JPLAYmini on its own sounds best. The
improvement over the likes of Foobar is massive. The 2D nature of my
Redbook (i.e. standard resolution CD) experiences is now banished. The sounds I
hear with JPLAY are realistic. Digital die-hards will have to put up with my
saying the sound is like a really good record deck! 3D and realistic without
upper-mid harshness. Previously I only heard this effect via a very few
ludicrously expensive CD players. As a basic player JPLAYmini may seem pricey at
€99 but compared to a DCS it is amazing value. You will need to have the
temperament of a vinyl junkie to put up with the 20 to 30 seconds to load an
album into memory and the total lockup of your computer while music is played in
the highest quality mode. If viewing cover art is more important to you than
achieving very high sound quality then I suggest you read www.EnjoyTheCoverArt.com
instead of www.EnjoyTheMusic.com.
You can however run JPLAY with the likes of Jriver for your more casual moments.
Media players aren't just for hi-fi nerds so don't
assume sound quality are their number 1, 2 and 3 objectives. A sexy and
intuitive GUI will be more important as may be their ability to play lossey
formats. JPLAY on the other hand will only play WAV, FLAC, AIFF and ALAC formats
and there is no GUI, unless you hook it up to media player but then you lose
some sound quality – at times this may be a trade-off that works for you. I'm
extreme with this stuff, I want the best for my main system; if the family want
to be able use a system they can use an iPod & dock.
a basic DAC via USB - the impact of JPLAYmini was immediately obvious. Redbook /
CD had now grown up, I had not expected it to ever sound this good at anything
other than truly high-end prices. You may dismiss record decks and warm and
euphonic sounding, this not my view of a good record deck. When I say JPLAYmini
bears comparison with a good record deck this is praise indeed!
I was now hearing the effects of what some smart
people have worked out is critical in reducing jitter. Many of us assumed the
stereotypical CD sound was due to the format. It seems not. I could often detect
a CD source even from outside a room, not so now. JPLAY is responsible for
getting rid of the CD sizzle, which to my ears has been prevalent since CD's
inception. JPLAY is not your only option, some relatively costly pre-built
servers are available or building your own server should get you there too.
JPLAY ultimately makes the music realistic and analogue-like there's a
really solid bass performance too providing a powerful musical underpin.
Using JPLAY in my system is a no-brainer but I wondered whether I would hear the difference between JPLAYmini and Foobar2000 with ASIO4ALL through a more mid-range system setup for AV around our plasma TV. The system comprised a Burson Audio PI-100 solid-state amplifier with Visaton Topaz speakers. The amp and speakers are really rather good, most people would be very pleased to own them as their main system. If this space constricted and AV-biased system setup could benefit from JPLAY I would be surprised but I didn't know what to expect. First up was a tweaked XP laptop running Foobar2000 with the Halide DAC HD, the tracks I played sound restrained, laid-back and pleasant, very suitable for background music. Very "nice" but the pacier tracks were done little justice.
I swapped to a laptop with Windows 7 running JPLAYmini and continued to use the Halide HD DAC. I wasn't shocked but I was surprised. The TV slap-bang in-between the speakers does the sound no favours whatsoever but even so the differences were immediately discernible. The sound was still quite relaxed but there was more get-up-and-go, there was a decent amount of instrument separation, bass was cleaner and more tuneful. The overall sound was musically more coherent and something you would listen to rather than have on as a background boredom killer. I'm sure that if the system were setup as a music system with space all around the speakers and especially between them, then there's no question that JPLAY would be a justifiable part of the system.
JPLAY will cost you €99. I've heard comments
such as "you can get a copy of Windows for less money!". This is true but
1) A Windows license is very restricted, probably
a Home OEM license tied to one PC for life.
2) The JPLAY license is for your household, it's
not restricted to one machine.
3) How many millions of Windows licenses do
Microsoft sell? Once their development and support costs are covered they are
making pure profit.
When you buy a Windows license and a JPLAY
license, who do you think makes more profit? I'll give you a hint; it's not
You owe it to yourself to download the trial
version on JPLAY. Do it, set it up with Throttle ON and Fullscale Hibernation
and then decide for yourself. While you're doing this check out the two "engines"
in JPLAY; Beach and River. I find Beach sounds dry (as a beach should be) and
detailed whereas River is err... wet or at least fruity and more flowing. River
suits my taste and system better.
Second Step to Heaven – Pick Your DAC!
We used JPLAYmini for the panel assessments, for
ease of use reasons we did not use Hibernation mode. The differences between the
DACs were mostly very clear to hear and there was generally consensus. What we
heard fell into 3 categories:
1) DACs with standard (adaptive) USB - where the
computer controls the data timing.
2) DACs plus an asynchronous USB converter module
outputting S/PDIF -- where the DAC controls the data timing.
3) DACs with inbuilt async USB capability and an
I2S internal feed to the DAC
Place: Rega DAC, this uses a pair of
parallel-connected Wolfson WM8742 DACs. Considering the expansive soundstage the
auditioning system was capable of we all felt the soundstage was compressed and
the sound simply didn't stand out as special.
Plus async USB Converter Outputting S/PDIF
John is well known for upgrading the M2Tech HiFace async USB to SPDIF converter. The HiFace board used here has various upgrades and is battery powered. The MK3 JKSPDIF and both JKDACs require a HiFace USB driver to be installed on your PC, there is also a 32-bit/384kHz resolution Young version. The Young driver was found to be richer sounding and was preferred by the panel so this was the driver used for the auditioning. Later in my time with the JK DACs I grew to prefer the HiFace driver. It's a choice you can make around what suits you and your system, it's rather like changing interconnects.
The MK3 JKSPDIF, in common with the two JK DACs
also reviewed, is battery powered. These JK units all use Lithium NanoPhosphate
(LiFePO4) batteries which are said to have a 10 to 15 year lifetime, no memory
effect and low noise. Fully charged batteries will deliver around 8 hours
continuous playback. You can extend the playback time by turning off the unit
for a few minutes before the 8 hours is up, this tops up one of the batteries
from the other one. When you've finished listening to music you simply switch
off the unit via the switch at the front to leave the unit on charge. You can
leave the batteries on permanent charge. You will need to supply an external
power supply; 9V to 12V. I already had three suitable power supplies, if you
need to buy one you'll find them easily and at low-cost. I expect these power
supplies are not provided as you may have one already plus coping with all the
different international power plug considerations add unnecessary complications
to the ordering process.
The casework for the MK3 JKSPDIF and both JK DACs
is common to all. It is a small aluminum chassis with plastic end plates. The
quality of the casework is fine though nothing special. Given that the prices
for the units are between €335 and €500 we're not talking high-end
audiophile jewelry here. No doubt a chassis hewn from a solid billet of aluminum
could be produced but I've a horrible feeling prices would increase
substantially, probably more than doubling them. Personably I'd rather prices
stay as they are.
5th Place: Rega
+ MK3 JKSPDIF, this did not disappoint; the
addition of async USB resulted in a much more open sound and released the
soundstage. The sounds we were hearing were very good – spacious and detailed
with instruments playing within their own location in space. Bass was tuneful
and deep. This is the level where Redbook (CD) sound does not disgrace itself in
comparison with a really good sounding record deck. I'm not saying they sound
the same but it's possible to switch between sources and not feel short-changed.
4th Place: Beresford TC-7520 (modified + Burson Buffer) + MK3 JKSPDIF; async USB also benefited this DAC and to assist it further we added a Burson Buffer stage following the "passive mod" capacitors. The buffer stage helped the DAC to be to drive the TVC and power amps. The result was impressively propulsive bass, much smoother treble than before and a feeling of space. The Rega had the greater detail and both DACs handled guitars equality well.
The MK3 JKSPDIF brought a very considerable performance lift
to the Rega and Beresford DACs, if you have these or similar DACs and you use a
computer to serve your music I urge you to consider adding async USB to your
async USB Capability And I2S Internal DAC
3rd Place: JKDAC.
The two John Kenny (JK) DACs share the same enclosure, LiFePO4 batteries and
modified HiFace 24-bit/192kHz async USB device. You get a single input (USB) and
a single pair of outputs presented as RCAs – simple and all that most people
need. Where the JKDACs simplify matters over the MK3SPDIF is in their connection
to the DAC - they skip the S/PDIF connection. The HiFace module connects to the
DAC via very short internal I2S wiring. Not only are a pairing of
S/PDIF transmitter and receiver omitted, I2S separates clock and data
lines thereby reducing data-to-clock crosstalk. This all sounds great in theory
and the technical gurus can argue over the finer points. What I wanted to
investigate is what sounds best.
The JKDAC uses a 24-bit/192kHz Sabre ES9022 DAC whose output I'm
told is strong enough to drive headphones, you would of course need a converter
cable to use headphones. Certainly this raises an interesting high quality
portable music possibility for aircraft trips and the like.
Note the photos below show both JK DACs.
I found the Sabre-based DAC presented a broad central image
with a very expansive sound, far much more expansive than the previous DACs
auditioned. On some systems the bass seems particularly strong but on
systems which dig really deep and do so cleanly with a good room setup (avoiding
room modes), I find the DAC doesn't excel so much deep down. It doesn't quite
deliver the absolute ultimate in bass texture. For bass-bandwidth limited
systems the bass on offer here could work out better than the two DACs in 1st
and 2nd place. The upper-mid and treble is clean sounding, on a clean
or lean system the sound could be a little too explicit over long periods. It's
as though the leading edge of treble notes rises especially quickly. If you want
a sound where you can relax into the music then this isn't your perfect match,
if you want a listening experience that's just a touch towards "active" or "edge-of-the-seat"
then the JKDAC gives you this. Although this DAC sits here in 3rd
place, with some recording mixes it's a contender for 1st place, as indeed is
2nd Place: Halide Design DAC HD. This DAC has its origins in the devilsound DAC. You get more than you might at first expect. If you ever felt the need to buy an audiophile USB cable then there's no need with this DAC as you get Wireworld Starlight USB cable hardwired in to the DAC, I requested the 2m option which comprises 1.8m of Starlight, the DAC and then short silver interconnects terminated with Eichmann Silver Bullet RCA plugs. You can purchase a total length of up to 7m should you require this. The DAC enclosure is CNC machined from a lump of metal, presumably aluminum, which is anodized black and laser engraved. Overall it's a cable integrated with a very slick looking DAC. It seems a shame to hide the DAC behind an amplifier but that's most likely where it's consigned to live.
As the DAC is featured here in the async USB section it's no
surprise that this is a key feature. It runs the Streamlength code licensed from
Gordon Rankin's Wavelength Audio. Like the JK DACs there and I2S connection
feeding the DAC except this time it's a Wolfson WM8716 running at 24-bit/96kHz.
The async USB protocol used here does not require a special driver to be
installed. I noted that the output stage uses silver mica capacitors so I was
particularly careful to ensure a good burn-in as I've found these very good
capacitors need quite a lot of time to settle down.
The DAC HD derives its power from the USB 5V feed, Halide
describe on their site how a lot of filtering and decoupling of supplies takes
place to provide 3.3V, 3v and -4.5V required by the digital circuitry, clock and
output stage respectively. There are 8 voltage regulators employed.
As you'll now be aware I've taken to using JPLAY for playing
my music. JPLAY sounds its best in Hibernate mode but it's still way better than
most other players when not in Hibernate mode. That's just as well as the DAC HD
and JPLAY in Hibernate mode are incompatible. It seems there's something in the
Streamlength async USB and JPLAY code which means they won't work together in
Hibernate mode. The panel auditioning was done without Hibernate mode for all
DACs to ease switching music so the DAC HD was not disadvantaged, indeed you
could say the other DACs lost out to some extent as they weren't able to operate
in the highest quality environment. Anyone who must have a very slick user
interface such as Jriver and doesn't care that JPLAY won't hibernate with the
DAC HD will love the Halide.
It is interesting that the three DACs in this section share
sonic characteristics, notably an expansive sound with very palpable 3D images.
Async USB seems to be part of this as is I suspect the lack of a S/PDIF
connection; I2S being a better method to transfer data and clock
information. I always want to extract the best possible sound out of anything I
use and whilst pursuing this quest I found that via one server (laptop) the DAC
HD improved its sound when running purely on batteries. One of the tricks when
you find this is to ensure the laptop is running an ungrounded mains connection.
Given that with some power sockets you don't have a ground in the first place,
running a laptop without a ground will be safe but if you aren't confident don't
do it. What seems to be happening is that there's potential for a ground loop
between laptop and amplifier as the USB specification includes a ground on pin 4
of the Type A and B connectors. Sometimes this can result in a hum, as indeed it
did with the system used for the panel session.
The DAC HD mid-range and treble was very hard to tell apart from the JKDAC, they shared a slightly clean or etched character leading to an ever so slightly intense listening experience. I do not characterise this as a stereotypical old-fashioned CD glassiness, it's more a slight emphasis that any source could emulate in the some circumstances. This is more an observation / synergy thing, not a criticism; the overall sound quality on offer here is truly excellent. There's great detail and strong imaging in a way that I've not heard with Redbook data -- except for with some very expensive equipment. Then there's the bass which was exemplary; it did everything you could possibly want; it goes deep, it is dynamic, tight and textured.
Bass is very impressive, the equal of the Halide DAC HD, so it's
got a great bass dynamic, it goes deep, it's tight, textured, indeed it's really
propulsive. Where the JKDAC32 wins extra points is that the upper-mid and treble
are sweeter yet still very detailed. This is no treble cover-up job! The next
thing that struck the auditioning panel was the separation between the
instruments; this was a step ahead of the best of the other DACs being tested
here. Imaging was found to be very three dimensional and we agreed there was an
especially strong focus for the central image whilst the expansive nature of the
soundstage was maintained.
The best DAC here? The panel was unanimously in favor of the
JKDAC32. So much so that the owner of the Rega sold it the next day. He
purchased a JKDAC32. So have I.
A Few Cautionary
This turned out to be interesting. My 2m screened but cheapo
cable made treble a little hard at times. The direct connectors made quite a
difference, more substance and a more refined sound with the occasional hardness
banished. This rigid connection was rather inconvenient so I tried various
non-audiophile cables I had to hand. The closest to the direct connection sound
was a screen 0.5m cable from a Toshiba disk drive, this cable has a ferrite core
on the "B" end. The sound was close to the direct connector, not as good but
reasonably close. The JK DACs don't use the power wires in the USB cables except
for the ground connection so I expect these DACs are less influenced by cable
quality than some other DACs. I really don't want these cables to make a
difference but they do. One theory is that many cables don't conform to the 90
ohm impedance standard at the high frequencies employed by USB.
I've heard the argument that as a printer using any competent USB cable will print character or pixel perfect copies there can therefore be no sonic benefit due to audio USB cables as no data is lost. This argument misses the point; the correct analogy would be whether the printer printed the character at the allotted millisecond. This is not about loss of the data; it is about whether the data arrives within the correct time window.
Having heard clear differences between USB cables and a direct USB connector I decided to get my hands on an audiophile cable. I contacted Russ Andrews who very kindly loaned me a Kimber Ag cable. This was a 1m cable and as its name suggest there's silver used in it. The signal conductors are pure silver; power, drain and shield wires are copper with silver plating constituting 6.1% of the wire. The connectors are gold plated.
now compared what had been the best USB connection to date – direct connectors
– with the Kimber Ag. The direct connection gave a more authoritative bass but
was a little lean in the mid-range, at times just a little too explicit. The
Kimber cable gave a sweeter presentation that was more extended in the upper
reaches too, so it wasn't curtailing frequencies to sound sweet. Overall
the Kimber was easier to listen to but there were occasions when the direct
connectors sounded more impressive. For me on balance the Kimber won. Of course
the direct connectors are just a couple of bucks but they are rather
impractical, then there's the considerable risk that at some point broken a USB
port will be a consequence of forgetting there's a direct connection in place.
1) You don't have to spend a fortune to get a sound that bears comparison with a very good vinyl rig – even with "lowly" Redbook / CD as the source.
2) The player you use makes a massive contribution to the
sound. The JPLAYmini in-memory player was found to be vastly superior to some of
the commonly used players. To get the best sound you need to put up with reduced
usability. There's no gain without pain!
3) Removing conversion from USB to S/PDIF, using Asynchronous
USB and using I2S connection internally USB to DAC – these all delivered
highly with the top three DACs auditioned here.
4) I know it's "just" data but don't overlook your USB cable,
it really does make a significant contribution to the sound you achieve.
5) The DAC chip type and the way it is implemented is still important but it's just one factor in this recipe, don't overlook the other factors.
Halide Design www.halidedesign.com
John Kenny google.com/site/hifacemods/home