Here is a index to this section:
[6.3] Standardizing to normal Consumer Output Levels
[6.4] Level 2 Mods
[6.5] Blak_I/O
[7.3.5] 2nd Generation dAck 2.0
[7.4] Non-oversample/1x-Rate Sampling Converters
[7.5] dAck News
Links to other PeAK Audio Pages
Audio has always been my first love. It started when I was 12 years old when my brother began to shop around for stereo equipment. By the time, he was onto his second stereo, many a store had been visited and Toronto had been blessed with many good ones. One in particular called "Ring Audio" stood out. These guys claimed they could hear differences in turntables (i.e.record players for non Ring types), and amplifiers in spite of the low distortion numbers and wow/flutter numbers. I followed their notion and eventually bought my first stereo in 1984 and still use it today. The NAD 3020 was a landmark design with good sonics at a reasonable price but it was, I think, much overrated in its standard form. A more recent review was recently made by TNT-audio comparing it to the current status quo. I began modifying the phono section of amplifier (RIAA time constants) for a school project. While I was at it I replaced ceramic caps with polystyrene film types in the signal path. The worst caps are electrolytic caps in terms of their memory effects (dielectric absorption). Invariably, these were to be found in sections of the amplifier topology that were popular selling features: loudness controls, tone controls to name a few. Ultimately, removing the electrolytics and adding a wire was the best solution. This sometimes required that the circuit topology be changed so that the DC component be eliminated. The NAD 3020 today has no tone controls, no balance control, no loudness button, and no LED volume indicators... it eventually sounded "better". Audio is one of those things where if you improve one part of the chain...the overall result might sound worse to the untrained ear. It is one of those two wrongs making a sort of right. Things got simpler towards the 90's as better caps and better "wire" removed more of the fundamental flaws. I found that making A/B comparisons to be of little use. The ear seems to "fuse" two like sounds together so that difference...unless really obvious...are hard to detect. The best way to gauge a "mod" is to pick a well recorded and simply miked piece of music and to play it through before the modification and to then play it through after the modification. If your gut says it is better...then it is better. This works amazingly well because I think you tend to be more relaxed when comparing things in this manner. You tend to focus on important things such as pace, rhythm, and space. Try it ! Its been a decade of audio mods for me...I'll sum up the ones that make the most difference:
5 A Brief History of the State of the "Art"
DI/O DAC
This is just my interpretation of the DI/O history by reading between the lines of many reviews and posts. I realized that information is still missing below and welcome suggestion from others on key links and information that they would feel would round up events surrounding the inception of the Art DI/O. Cheers.
[5.1] Background
The Art DI/O was
birthed in
the town of Rochester, NY, not from Niagara Falls by an engineer
named Mitch Milton for the professional sound market. An early
review
(Wayback )
appeared in February 2001 in the the professional music scene. In
March 2001, a LA based recording engineered gave a factual dump on
the DI/O claiming that it gave better sound from the DAC than that
in his professional DAT machine. For the next 3 months of April,
May and June, the DI/O lived a relatively osbcure life of a yet
another audio product.
[5.2] The DI/O Upgrade Buzz
This DAC was first mentioned in the Audio Asylum community in
early April without much
fanfare
and the beginnings of a craze began by a posting from
LA6.It was soon adopted by consumers shortly after a
review (Wayback )in
July 6 2001 by Frank Alles of Stereotimes. The buzz had started
and three key individuals (Todd Krieger, Sean H, and Gary Bonner)
of Audio asylum took on the DI/O mamtra, barely a week later, in a
name
key
threaded post of their findings . I think they were the
original proponents for the diode upgrades and the op-amp upgrade
to an LT1362CN op-amp.
The fermentation of different hobbyists continued for about two months into the month of September when the wealth of information was blossomed. Two aspects of the DI/O design led to modifications as being necessary: First, the outputs of the DI/O DAC are not compatible with normal RCA ended cables but take the form of 1/4 inch phono plugs found used on guitar amps. So custom cables or modifying the DI/O output jacks is necessary. The second apsect of the DI/O is the "hotter than normal" output level of the DI/O which is anywhere from 3x to 4x the level of a typical 2V RMS output level of a typical CD/DVD player. The maximum theorectical output level of a standard DI/O at 0 dB recorded level is 11.34 V peak (8.12 V RMS)...large enough to overlaod most preamps with input buffers. Preamps designed with low level input feeding into the top of the volume pot fair better but the result is the "pot" operating in the region of the pot that has the most channel tracking error and lower signal to noise ratio. The DI/O was ripe to mod for this one aspect alone. This led to a cottage industry of specialized RCA cables with built in attenuators that are still commercially available...as it turned out, the output level was just the crack in the dam before the flood of DI/O mods poured forth.
The DIOFAQ ( archive ) was born at or near the beginning of November courtesy of Craig Fraser. Those who read the FAQ sheet owe it to themselves to understand the prioritization of the mods was for the most part driven by and engineer/audiophile from France named Claude G. The previous link outlined his comparison of the Art DI/O in unmodified form going head to head against the PT-P3A DAC, A2C Hermes DAC, Bel Canto DAC, Wadia 27i, AudioNote DAC ONE, and the Audiomat Tempo 2 DAC. The following are other key postings by Claude G:
Much of this knowledge about upgrades and preferred digital cables has been distilled to a point where the upgrades can be purchased from Boulder Cables (Wayback ). At the Consumer Electronics Show in Las Vegas of 2002, Brian Cheney put together a system to drive his VMPS RM40 speakers using a modified Art DI/O form Boulder and ending up winning "Best Sound of CES 2002"...not too shabby!
[5.3] Acknowledgements of Key Contributors
As mentioned in the DI/O faq, another key contribution to the
popularization of the DI/O was Bob Fitzgerald founding of a Yahoo
"groups" based bulletin board and information center around the
same time frame of November 2001 called
Diomods.
Together, the triumvirate of the DI/O Faq, Yahoo Diomods group,
and Audio Asylum form the backbone of the DI/O community. After
having gone through some of Audio Asylum archived forums for
myslelf (the FAQ is too easy !), I can say that without a doubt,
other voices such as
Todd
Krieger,
Gary
Bonner , Jon
Risch, Jack DuMoulin, and
Bob
McNeice
have made key individual contributions adopted by the group as a
whole.
[5.4] Membership and Priviledged Design
Information
The present DI/O FAQ is fairly complete and has been going off in
a direction of passive output filtering via transformers. The
other significant development is that becoming a
member
to Yahoo Groups entitles you access to two tools that are
indispensible to the Art DIY upgrader in the
Files
section of the Diomods site:
[5.5] Miscellaneous DI/O Links
These links are included here because some of them jumpstarted the DI/O. The Asylum post are ones I recommend you read and are made by people with trusted hearing (IMHO). There are some reviews of other equipment that are listed because the DI/O served as a basis of comparison.
6 "Volks_DI/O" Standard
[6.3] Standardizing to normal Consumer
Output Levels
[6.4] Level 2 Mods
[6.5] Blak_I/O
I will use the designation "The volks_DI/O Standard" to describe the upgrades which I feel significantly upgrade the DI/O sound with the criterion that there be ease of access to the average neophyte or as I put it..."to average volks". The following additional criterion also apply:
While it possible to spend upwards of $275 on the cost of a commercially modified stock DI/O (2x the cost of the stock DI/O), I'm limiting the changes here to things that could have be done by the manufacturer with no more than a 20% premium to the cost of the stock DI/O to achieve the above aims. In the future a "Level 2" classification may be defined for those mods that additional improvement based on medium cost and/or major amounts of soldering or soldering skill.
The following changes (IMHO) upgrade the Art DI/O to a level significantly enough to a have some of the best attributes of hi-end analog playback (lack of harhness/steeliness, detail and intertransient silence) while combining the strengths of digital playback.
This "volks_DI/O" represents the culmination of a suggested sequence of mods based upon my personal upgrade experience and my personal satisfaction with the overall net result with it tested on several different systems. That is it. A compedium of information on almost all possible modifications can be found in the DIOFAQ pages of Craig Fraser that should be consulted whilst performing these mods. With the criteria specified earlier, the information presented here is the material is written up for a different "percieved audience" that would like to see more direction and opinion/weighting on the mods. The existing DIOFAQ's mods get skinny on detail/references/tradeoffs needed by most average "volks" who just want something better than stock. As Candice used to say..."the most for the least" in terms of my effectual priorities. Before going on further, I'll delve slightly into the "grey" mods not presently included in the "volks_DI/O" level one mods...yes a back door!
The greatest omission by me is not having tried "other" cost effective AC transformers that would fit well into the category "volks_DI/O" based upon the comments of others who "hear" well. My reluctance is due mostly to not having tried it. Another aspect is that a large variety of these transformers are available, while the current recommendation is particular to one or two brands. Both the current rating and the voltage have been said to affect the DI/O...more science needed beyond a particular brand recommendation. Perhaps a list of good sounding 9V AC transformers could be the start to the science of their sound. Better yet would be a mod to isolate the DI/O from this dependence short of a new Utility company. So for now, the "volks" represents how far the DI/O can go without a transformer transplant.
The other omission by me is not having tried the "opamp" upgrade. The cost is in the spare pocke money cost of a CD level, but I'm trying to keep myself in the manufacturers position of offering a DI/O with upgraded sound/compatability assuming I have a warehouse tied up with transformer, op-amp, and capacitor inventory. The "volks" admits that the sound of the DI/O is pretty heady in default form. My experience with the TL074 is that you can get a pretty good level of sound out of this one horsepower engine. It might be what I call a level 2 upgrade and is pending "again" or me auditioning a unit. Wholesale changes to "all" the electrolytics are a lot of work and can hurt the future resale value ....I think the ones in the DI/O are not bad...based on what I hear today. Maybe they are reponsible for some of the transformer senstivity...again a level 2 upgrade. Onto the "volks_DI/O" Standard.
[6.1] Start with 1695a Digital Interconnect
Cables
You can purchase this cable pre-assembled at very reasonable
cost for under $20 from
Cousin
Dupree with Canare 75 ohm connectors without the external
braid. Delivery to Canada was about 3 days. His work is very
good. He will also add both the external braid and techflex for
an additional $15.00. Based upon my listening tests of two
cables (with and without the external braid)...I say go for the
extra external braid.
Note: As of Dec 2002...beware that a
number of people
have
not recieved cables that they paid for on the web. Until
this is resolved, he will get a conditional recommendation.
For DIY types who insists on building these cables
themselves,
www.takefiveaudio.com is a place that will sell you short
lengths of 1695a cable.
[6.2] Near "Free" Upgrades
These mods are each under $10 and can essentially be consider
pocket change or "free". Some are no cost at all and simply
involve the insertion of a shorting wire. Cost is mostly your
labour....sweet! I have found the following alterations to be
useful prior to warming up your soldering iron:
This simple mod on the vertical "output" board involves removing a large value DC blocking capacitors (C18 and C19) following the mantra that the best capacitor is no capacitor at all. We remove the cap "electrically" by adding a shorting wire across it. Here we rely on the existence of a DC coupling cap at the amplifier input to take care of any low level DC offset coming directly from the opamps of the DI/O. Certain amplifiers that pass on and are flat to DC will pass any offset on to your speakers...you have been warned.
Removal of input coupling caps C14 and C15 will remove connection of the input ADC jacks from the ADC and stabilize the current draw of the ADC section so that the other high voltage tube diodes can be left in.
The following section is a "weird" one at best. They/It are things that I recommend that you not alter. It has been based on carefully evaluating the DI/O in different setups. You can always try them (as I have) and over the course of making the mods, they may seem to make worthwhile improvements, but are steps back when used in conjunction with fundamental modifications such as the 1695a cable. Just my two cents on this fleeting standard called the "volks_DI/O".
[6.2.1] Upgrade supply diodes to
Schottky
This mod involves replacing six physcially adjacent diodes (CR1 to CR6) with Schottky diodes. I used 40V 1A 1N5819 diodes . The DIOFAQ recommends 100V units made by Internation Rectifier. CR7, CR8 and CR9 can be left in as these supply the tube circuitry which will now be drawing no current/power. Some people have removed unused components as if they were cancerous cells using the mantra of "even simpler" has got to be better. The C14 and C15 input coupling cap removal,above, will minimize any noise induced by the ADC processing the output DAC signal. Click the picture for a larger image.
[6.3] Standardizing to normal Consumer Output
Levels
If you want the to ensure the DI/O can work into your preamp input
stage, attenuating the ouput to more modest levels will allow you
to two potential advantages and one slight possible tradeoff. In
the table below, the "stock" output level is on the order of 8V
RMS...about 3x higher than nominally specified. This can cause the
volume potiomenter to operate early on in its range around the 9
o'clock position where channel to channel balance is often its
worst. The other aspect is that the topology of the preamp has the
very first stage set as a fixed active gain stage buffering a
volume potentiometer. This gain stage sees the whole "8V RMS/11V
peak" input signal. This stage must be powered off at least 12V
rails in order to successfully handle the "0 db" peak output level
from a CD. Even if this condition is met, the output distortion
begins to rise as the inputs approach the rail voltages. This
means that supplies are often set at twice the maximum level input
level (i.e. +/-24V) to handle this signal directly. The one slight
tradeoff reported (Risch) in attenuating the DI/O output level is
with the "larger" signal levels swamping any fixed residual
interconnect cable "noise" induced by a combinations of cable
funnies due to mechanical feedback, reflection and dielectric
absorption.
|
|
|
Cfb |
C24,C25,C33,C34 |
|
Cgnd |
C29,C30,C31,C32 |
|
Rin |
R14,R15,R20,R22 |
|
Rfb |
R13,R18,R19,R24 |
|
Rser |
R16, R17, R21, R23 |
|
In the table,below, we have a listing of four Low Pass second order filter configuration types relating the resistor and capcitor component values to the key filter parameters of gain, cutoff frequency (F3db), damping factor (eta). The default stock DI/O can be described as an "overdamped butterworth" Low Pass Filter alignment with "eta/damping factor" (DF) number set to 0.729 .
(Note: Updated and corrected erroneous information here on Oct 7, 2002 at 2:00pm...basically the correspondence between R_ONE, R_FOUR, and R_THREE (Rin, Rfb, and Rser, respectively) and the PCB board silk screen names was wrong. If you have been planning/comtemplating mods based on information herein prior to the above date...you better take another look at the updated text. )
The picture below is a labelled diagram outlining 2 groupings of same valued resistors that need to be changed. Components labelled in "red text" such as the input resistor "Rin"(R_ONE group consists of R14,R15, R20, and R22) do not need to be changed (stay at 4.7K). The feeback resistor "Rfb" (R_FOUR group consists of R13, R18, R19 and R24) are labelled in YELLOW text and should be changed to 4.7K. The gain is set by the ratio Rfb/Rin. The Rser lines up with the resistor group called "R_THREE" and appears in series between the gain setting resistors and this group should be changed to 20 Kohm (non E12 series component...actually E24 series). These are labelled in BLUE and consist of R16, R17, R21 and R23. This modification preserves the default DI/O roll off characteristics and only alters the gain to a more typical 1.91V RMS level.
I found that my source of metal film resistors may require significant burn-in time to sweeten up its sound...boutique caps by HOLCO or RIKEN-OHM may shorten this and are said to offer other sonic improvements both in the short term swap and possibly in the long term. Your mileage may vary depending on the resolution of your system. Click on the picture for a larger image of the three resistor groupings.
[6.3.1] More on Output Low Pass Filters
A music waveform can be thought of a multitude of sinewaves with
both precise amplitude and precise phase realtionships to
the loweest frequecy component. An ideal low pass filters pass
frequencies below some cutoff frequency (F3db) while those above
are removed. The maximally flat filter response of the
"Butterworth" is damped slightly less with eta=0.707. So all
frequencies on the DI/O below 50.3 kHz are ideally passed
unscathed. In real life the band of frequencies around F3db are
caught in the middle and are neither passed thru (when below F3db)
or removed when above F3dB. Compared to the stock DI/O, the
Butterworth and lower damping factor (lower eta numbers <0.729)
do a better job passing the "amplitude" below F3db. Note that at
F3dB is defined as the point at which the output sinewave will
reduce to 1/2 its power (.707 of its amplitude) from its very low
frequency/unattenuated ouput level. Above F3dB, these lower eta
(less damped) filters also attenutate more (better) of the signal
and "seem" to behave more like an "ideal low pass filter" as
defined above. So you might conclude that an underdamped/low eta
filter would be best...not so fast.
Equally important is that a filter preserve the "phase" relations or "time realtionship" of various sinusoidal components that make up a input waveform. Distort this, and shape of a resultant output waveform will get changed...possibly to an extent that it is more important than preserving the amplitudes of these components. The filter type that does the best job here is the Bessel filter with eta=0.866 with constant group/time delay characteristics within the F3dB bandwidth. You can read more about this by typing in the keywords [Bessel Butterworth filter] in Google. These second order filters are a well documented and understood discipline.
Consult the "File" section in the Yahoo DIOMODS group and find the Filter_dio_v12 spreadsheet.
This spreadsheet can generate different realizations (resistor component values) of filters with different gains, F3dB cutoff frequencies and filter alignments. The table below outlines the electrical and physical characteristics for the stock, Volks_DI/O, and two pending experimental prototypes.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
In the above table, we have a listing of four Low Pass second order filter configuration types relating the resistor and capcitor component values to the key filter parameters of gain, cutoff frequency (F3db), damping factor (eta). The default stock DI/O can be described as an "overdamped butterworth" Low Pass Filter alignment with "eta/damping factor" (DF) number set to 0.729 .
Expt-3 are an updated view on a Bessel Alignment commonly used by early adopters of Gary Bronner's spreadsheet picking a bessel alignment with Gain set to 1.7 and F_3dB cutoff frequency chosen until Rfb=20K (Result was 97 kHz). I was curious to see if Rfb could be modified to more commonly available E12 values (like 22K vs 20K)offered by Riken-ohm and Audionote. The answer is "nearly", as the results are a slight increase to the gain(1.833), "eta=0.84" vs 0.866, a lowering of the cutoff frequency (88 kHz), and an increase to the Rser resistor to 1K.
Expt-1 and Expt-2 are future experiments are pending to explore tradeoffs to the filter alignment and gain. The gain has been deliberately set a bit higher. The resistors Rin and Rfb have been chosen at 4.7K and 6.8K. The Rser, Cgnd and Cfb will also be set using commonly available values and this might mean that I'll need to allow cutoff frequency some room to vary. No final values yet, but a preliminary set has been calculated that will be undergoing evaluation over the fall when boutique resistors are ordered so as to maximize the chance of the getting it right the first time. This is to partly to minimize damage to the tender DI/O board. Come back here to follow developments on this front.
Note that the resistors that I will be using will be the more commonly available E12 series in future filter experimental filter alignments. These resistors differ from their nearest neighbour by about 20%, so that for any desired value of resistor, you will be away "at most" by 10% from some desired design value. However, do not confuse this with accuracy as these values and steps can be very precise and repeatable...sort of like pints, quarts and gallons. If you want a pint and a 1/2...you are out of luck but I can give you a pint accurate to the last drop. So if you get a lower quality 10% accurate pint, you may get a couple of ounces more or less compared to a 1% accurate pint.
[6.3.2] RCA Interconnect compatible Jacks
Once the gain mod is done it is best that the jacks be changed to
RCA so that we can utilize the vast number of RCA cables available
for analog output. This section will eventually include a sequence
of photos/illustrations showing how the existing phono jacks are
removed via solder wick (not easy). Similarly the installation of
the RCA jacks will be eventually done in similar manner. For
connections from the RCA to the I/O board, use a single strand of
Kimber PBJ and be sure to find RCA jacks that have hex nuts on the
outside of the chassis to facilitate future mods requiring removal
of the main board.
[6.4]Upgrade Omission Info and Optional/Level 2
Mods
As mentioned earlier, some mods commonly made have not been made
purely on philosophical grounds related to the "volks" designation
and what it means. The other aspect for me is the "Why" of a mod
"taking". It just comes down to science. This is especially the
case where one experienced listener has a different conclusion
about from another. For the mods below, some can be done easily
via swaps as in the case of the transformer. For the op-amp, this
is also true once the socket is added. For the PS caps, the
availability of a second unit and a ping-pong upgrade strategy
whereby one is changed in one area only and compared to its former
twin is how this will be addressed.
[6.4.1]Transformer Upgrade...some new info
The Stancor upgrade is so entrenched in the DI/O community that I
am afraid to go against the grain and not recommend it. As I said
earlier, I am of the opinion that "both the current rating and the
voltage have been said to affect the DI/O...more science needed
beyond a particular brand recommendation".
Some people have recommended going to an Atari 9V adapter with roughly the same current sourcing capability as the Stancor based upon "more must be better". On a stock DI/O, a DIOmod's post was noted that the Atari was not the equal of the Stancor. My friend and I have obtained data points that indicate that a large part of the sound of the DI/O is correlated to the raw AC supply voltage and how close it is set to 10.2V AC . In theory the raw voltage coming into the DI/O should not matter due to the existence of regulators on all 5V, 12 and -12V supplies, but in practice this seems to not be the case. Short of modifying/winding your existing/custom 9V AC transformer and taking out windings in the secondary (to lower the voltage) or the primary (to raise the voltage), a Variac is highly suggested by my audiobud. He uses a 12V 10A transformer with the primary adjusted to around 86V AC to get an output of 10.2V AC. Go figure.
[6.4.2]Op-amp Update
The "volks" admits that the sound of the DI/O is pretty heady in
default form. My experience with the TL074 is that you can get a
pretty good level of sound out of this one horsepower engine. As
unbelievable a statement as what I'm going to make, the jump in
level for the DI/O using the existing op-amp when powered via
10.2V AC has my audiophile friend making a statement that "he
hears nothing wrong with his system" using the Volk_DI/O powered
at 10.2V AC. Pretty strong stuff coming from him because he has
never been content with his system. So both him and I are going to
sit with the TL074 for a while at 10.2V. The level two datapoint
of using a LT1362 is still to be run. I'm wondering if the reports
related to the LT1365 sounding harder and less sweet have to do
with the larger current draw and attendent voltage drop on AC
voltage coming in.
[6.4.3]Power Supply Electrolytics
Wholesale changes to "all" the electrolytics are what I did a
decade ago wth Panasonic HF electrolytics. These were recommended
by Walt Jung in a seminal article about raising the CD playback
quality via class A biasing of op-amps and changes to
electrolytics in the output filter...sound familiar? I did just
such a thing to the preamp power supply of my integrated amp and
have never looked back. Today, the common recommendation for the
Panasonic FC series and sometimes even enlarge the value. The
axiom being used is that "more" is better based upon less voltage
ripple. Again, the thing holding me back is creating this
datapoint myself and the fact that the DI/O sounds pretty damm
good as is!
The aspect of this mod that I want to make sure of is that it will take on a stock DI/O. I think it is so significant a change that it ought maybe to be your first mod to make on a DI/O.
[6.5.1] Sonic Impression in a nutshell
I've basically gone through my collection of evaluation CDs and
have found that each has the following attributes:
At present, the above applies to the DI/O with Volks DI/O mods using a 1695a based co-axial digital iterconnect and JR twisted co-axial cores. At the present time I will limit the service to people in the area where I live by allowing them to bring in their units to demo on my system in the Toronto area (Ontario, Canada). I realized that the phone jacks and high output level will deter many from considering the DI/O as a long term source. I like to gauge the interest in the above avenue by having you send me private email on your thoughts and interests.
[7.3.5] 2nd Generation dAck 2.0
[7.4] Non-oversample/1x-Rate Sampling
Converters
The dAck is a battery powered/rechargeable DAC introduced around May 2003 by Ack! Industries to the delight of audiophiles; some comtemplating their third or fourth DAC.
In the era of the PC, a certain insecurity/mentality occurs in the "buying consumer" with yearly changes to component specs. These are spearheaded by well-meaing "Standard's Bodies". For manufactureres this means continuous updates to product line ups to keep products "feature competitive". Whether or not we want Dolby 7.1 is immaterial...consumers will choose it (the higher number) over Dolby 6.1. This sometimes mean change for changes sake...just because we can. The question about whether new technology or features are necessary becomes mute. This is very similar to the way unknowledgeable consumers are now percieve to purchase upgraded computers...when in doubt...buy the bigger number. In CD playback-arena, a certain evolutionary path has evolved whereby the original Redbook standard has come to use oversampling, gentler filter slops, easier to manufacture 1-bit technologies, and boutique parts as an answer to the question of getting "improved tone" out of CD material. The reality is that the medium is capable of very good results in its native form but a combination of things need to be done "together" to get there. Upsampling has its own issues with increased RFI, SACD/DVD-A are enslaved to mostly one box solutions (no external DACs), and the world still needs "yet another DAC"....yaDAC anyone?
In an about face, the dAck challenges some of the fundamental paradigms relating to CD playback. On the outside, this DAC exudes quality ...and as we will see on the inside. Unusual is the decision to forego upsampling and oversampling by using a 44.1 kHz sampling rate orginally defined for CD playback. The DAC uses a zero-oversampling "filterless" topology currently favoured by Audio Note, Zanden, 47 Labs and Scott Nixon.
The designer,Chris Own, is a trained musician with a penchant for instrument tone sometimes duplicated in highly evolved moving coil based turntable-centric audio systems. The unit is hand built using top quality components that electrically/sonically settle down within minutes of turn-on...key point for a battery powered unit. So much for amps that require 300 hours plus of breaking in...eternal power-on for sonic splendour not required.
[7.1] Background
In the 80's, there were good examples of audio engineering. Vinyl
was king, some low powered amplifier stages for low output moving
coil cartridges ran off battery power. Farad Azima of Mission
Electronics extended the idea further downstream and introduced
the world's first battery powered pre-amp in the form of the
Mission 776. Using Google I just barely found the picture of the
unit below at a
German
hi-fi site. An old idea that has resurfaced in the dAcK! and a
key element to its sonics.
The industry has evolved over considerably during the twenty years since Farad introduced the 776. There have been major developments in wire technology, discoveries about polarity, better coupling caps and improved resistors. These items were not yet understood to make a difference....the only thing people agreed upon,then, was that metal film resistors were better than carbon film....and we know in hindsight that "we" even got that wrong.
Batteries have made sporadic appearences in components and in the DIY camps over the years. Below are a couple of links that should convince your to pull out that dead car battery of yours and try it in your system:
In that era, when you listen through the pops and clicks of vinyl playback...there was/is gold in the grooves that CD playback looses. The dAck! can mine some of that gold. The dAck has garnered a mushrooming attention from various audio forums for its low cost and quality. Oversampling simplifies the post filter phase problems cause by the octave filter requirement (pass 20 kHz but filter out 44 clock) by shifting the clock to 352 kHz (8x). The result is that the filter can be gentler with less phase problems. The one issue is that noise in the system at multiples of 352 kHz can now be amplified down to the baseband range by frequency convolution. This is the issue of RFI sensitivity. Other approaches, like DCS, has been strong proponents of upsampling. These have the potential of higher RFI noise floors so that the overall result is a tradeoff with the theoretical gains of upsampling resolution.
[7.2]dAck Links
These links are included here to help you decide if the dAck is right for you. The Asylum post are ones I recommend you read and are made by people with trusted hearing (IMHO).
[7.3]Ack! Products
The classic first product of Ack! Industries was the dAck 1.0. It
was refined slightly to address some issues of extended bass
response with later sub-revisions. I have owned this device for
over a year and have recommended it to some of my hard core
friends who have also been seduced by it voicing and tone. Coming
in from the DI/O waters, I should know that products are "voiced"
in conjunction with amplifiers and cables. My system's has long
used JR digital cables and analog interconnect designs as a
reference. These worked well with the dAck but it was not until I
received Ack!'s digital cable and analog interconnect that my
system (and me) started to sing...some of this has to due with
burn in. So 3 new sections have been added (end of January 2005 as
I write) to cover these items.
[7.3.0] dAck l.3 Listening
Impression
This is my second DAC. The DI/O is in its many evolving forms is a fine piece of equipment built to a cost point. Many of the issues that have plagued CD playback is that at the heart of the matter is that it is inherently very analog in some critical aspects. I heard about the dAck! DAC via Chris and the Asylum grapevine. So the plan was hatched to order his in kit form as my Xmas present to myself.
Day Four:
My first impressions after having put the dAck! kit together
was that the unit sounds wonderful on voice. It has a very
analogue sound to it that I think I need to get adjusted to as
compared to the DI/O which by comparison slightly "hypes" the
leading edges of notes to the extent of eventual fatigue. Some
nights things work out well in lack of fatigue. Tweaking is a
two edge sword. I think I have some cable maintenance to do
before I do some more listening.
Day Six:
Old habits like the sound of the DI/O are hard to break.
Thinking that different synergy might be the answer, I tried a
cable
swap
and by the end of the above post...things are "seeemingly"
better.
Day 14: I dropped by my friend's place after work on a Friday to compare his OSCON bypassed DI/O to your unit . He is running a 50 Watt tube amp driving ATC SCM20 and a professional CD/RW from Marantz (same basic mechanism used in Mark Levinson transport). The green locking LED proved to be working. He was basically very satisfied with the overall tuning of his DI/O and was skeptical of whether any sort of change in his existing sonic balance could be tolerated by him. He noticed the less in your face presentation immediately and the smooth midrange, but things were close. I had just come in from the "cold" and needed dinner...I left the dAck on and after some noodles came back to listen to a warmed up dAck. Somewhere in the first cut, things just opened suddenly. It was magic for the rest of the evening. I put on the opening track of Keith Jarret's Koln Concert...the illusion of a grand piano in a large acoustic space was captivating. I think Chris has another customer.
Day 400: Got hold of Ack! digital cable and analog interconnect (silver wires) in conjuction with a burn-in unit alonging for extended continuous burn-in. See Section 7.3.3
[7.3.1] Battery Care for dAck
The sealed lead acid battery technology is quite different from Nickel Cadmium in terms of how one goes about recharging them. Sealed lead acid batteries provide long life when kept topped up so that they do not go into "deep discharge" cycles. Most people who are aware of "memory effect" are told that NiCads should be used until they run out (i.e. deep discharge) and then recharged fully before using them again for maximizing life.
The dAck comes with a intelligent charger designed to optimally charge and condition the SLA (sealed lead acid) battery. It does the following:
The "float charge" will optimally chage the battery and condition it. SLA batteries have a holding voltge which changes with temperature and the charger is designed to accomodate this. There are some additonal notes regarding run time that were posted on maximum runtime. One manufacturer is known to run his unit for 12 hours at shows...he could run it another six hours and be fine.
NOTES:
[7.3.2] AC Powered Burn-in unit for dAck
(Note: the following can permanently damage your dAck. Those
willing to attempt these steps do so completely at their own risk)
There was a time when equipment was reviewed without the benefit of burn-in by many reputable Audio magazines. Nowadays, it seems that with the more expensive the component, owners/reviewers wait longer and more patient for the day of rekoning when "all will be revealed". I have heard of some people stating 600 hours (24 days) for a high-end cap to burn-in and reach Neverland. Small wonder that hi-end audio is in such a mess...I'll take 7 channels of 100 watts, instead, please.
So how does one reconcile burn-in with a battery product. Either the product requires little or no burn-in or you need to somehow be very patient or go back to good old rectified AC for DC.
Preliminary schematics are being finalized. My own experience seems to suggests that prolonged burn-in beyond a week will generate minimal gains. The good news is that reverting back to battery power maintains the benefits attained with burn-in and peak operation is achieved within about 10 minutes. Stay tuned.
BONUS: As the unit burned in using the burn-in supply, it was listen to from time to time...not the same level as the wth battery, but more than just useable. A testament to Chris's layout and design. You could actually tune the voltage up or down using a Variac to dial in the sound to some degree. Best near the maximum +/-13V limit of the op-amps.
[7.3.3] Ack! Analog Interconnect
It took me a long time before I tried changing the Belden based
Jon Risch twisted co-axial interconnects (easier for me to try one
of three DACs) to something more synergistic with my year old
dAck. I'm now using the Ack! Industries own interconnect, called
"The Analog
Connect" with the Revision 1 dAck (Rev 1.2d).
I know I could go on in detail and describe the improvements (which I will in a more formal cable review) but it has got me enjoying and singing along to some of my favourite music CDs (a first). It has even given a new found appreciation of one CD that I had high expectations for that did not pan out. Case in point is the follow- up album of Norah Jones called "Feel's like Home". With the JR cable the sound was sort of distant and un- involving with a lack of sparkle.. The bass sounded as if it was equalised to sound louder but it lacked detail...strange thing as the first debut album sounds fine.
With the "The Analog Connect", I ended up playing the whole CD listening to a rhythmic bass and a piano tone and hammer attack that was just plain captivating. I tried the same cable in the DI/O and found it worked but just not the same synergy as with my old dAck. In SuperBowl-speak, "The Analog Connect" and dAck stay in and the JR/(DI/O) are on the sideline cheering on. I think it has to due with the bad fit of the of characteristic pre-echo signature of digital post filters with the revealing nature of silver cables. In contrast, the dAck and no post filter, a good silver cable comes into it's own. I've also tried Silver Bullet 4.0 with similar results. I think a separate post on silver cable and pre-echo might be the order of the day...perhaps when I put up the more formal cable review.
In summary, if you have an older dAck and sense something "missing", I suggest you try a good sliver interconnect first. Some I know are bright and zingy but the two mentioned above are good ones to try.
[7.3.4] Ack! Digital Cable
The 1.4m cable, christened
"The Digital
Connect" only comes with BNC ends but I think this might be
the optimum way to construct a 75 ohm cable. I think the cable is
a good quality Belden from the way it bends (perhaps 1506 or
1695a). The connectors are very high quality with gold plating on
the parts that count. For my setup, both my dAck and Toshiba CDP
do not have BNC connections (yet!). So in addition to the
BNC-to-RCA male adapter included with mine, I had to go down to my
local electronics store and pick up another one. I found that
things were best when I used Ack's adapter on the transport and my
cheapy (75 cents) unit on the DAC end.
With the cable oriented one way their was a bloom that obscurred the attack and decay of a piano. Voice were more reverberent but less defined. Flipped around the other direction (keeping the adapters locations set at the source and DAC) brought back the attack and bounce to the music. A breakup on loud vocal passages now dissapeared and was replaced by a soaring powerful vocal that just grooved. Overall, I find the cable to complement the dACk very well providing that you play with it a bit. (Note: cable auditioned mostly on a dAck 1.2d continuosly powered for about 500 hours using Ack! Analog Connect)
[7.3.5] 2nd Generation dAck 2.0
I got a chance to listen to
2
different Beta Prototypes of the dAck 2.0. There is a
refinement in the new found detail in the treble and bass. When
partnered with Chris's own Ack! brand of analog silver
interconnectrs and digital cable...things got even better. I found
them preferable in one directional in my system and some
experimentation is in order. This is a new section so I'll just
add info as it comes in.
dAck 2.0 Reviews
Ack 1.3, which remains a stunningly good dac, Ack 2.0 is significantly better yet. In regards to points 2 & 3, the new Ack now comes even with my analog rig, which is a PRAT champ. Ack 2.0 is a winner"
dAck 1.x/Classic
dAck 2.0 Postings and Skinny
dAck RAQ(recently) and 2.0 Vibe
[7.4] Non-oversample/1x-Rate Sampling
Converters
Instead of brickwall output filters, the dAck (along with Zanden, Audio Note and 47 Labs) used very little or "vestigal" filtering. These units tend to "listen well", but measure poorly. Recently an interesting set of measurements were made using a scope to view a digitized 2 kHz and 20 kHz sine wave and discussed on Audio Asylum. This section will try to shed some light on the matter.
In Ryohei Kusunoki's classic article about non-oversampling DACs, there is a figure showing a digitized 20 kHz waveform that looks like a 4 kHz amplitude modulated sinewave. The Figure below shows this over 10 cycles lasting 1 millisecond. Only careful setup of a scope will give a stable trace...more on this below.
With this type of filtering, attempting to look at distortion products, visually by feeding in digitize sinewaves, will result in a family of responses due to the quick response and asyncrhonous nature of the data with respect to the sampling clock. This is best understood by the example, below. The figure below shows a 20 kHz sinewave being sample at :
The interval between samples is precisely 50 units (units are in % of the 20 kHz period) between samples as indicated in the plot. I will refer to this as the "perfect interval".
In the figure above, we have data samples at the extreme amplitude only when we sample at precisely 2x 20 kHz or 40 kHz and carefully setup the data edge to sample on either the valley or peak of the signal. It is possible in this case to also generate a string of "0" values by shifting the first sample to occur at time t=25 in Figure 1. Note that the data samples marked "1, 2, 3, 4 and 5" will shift to "1', 2', 3', 4' and 5'" . Note that point "5" and "5'" are separated by precisely 1/4 period of the 20 kHz waveform with point "5'" occuring 2 and a 1/4 periods (225%) from the start and "5'" occuring 2.5 periods (250%) from the start. So the 44 kHz red line will repeat and be periodic in 10 periods.The actual culprit is the scope and non-synchronous sampling.
[7.4.1] Filterless Interpolation and the Step
Response
So far, we have only established two different sets of sampled
signal points whose amplitude depends upon the starting point and
relative rates of data changes to sampling clock changes. For the
second case, the table shows the values in the table will be
cyclic after 10 periods for the 20 kHz signal (i.e. 2 kHz pattern
repetition). A near filterless 1x DAC will generate step response
transitions (red line) between each of the data samples.
[7.4.2] Scope Measurement Artifacts
Anyone trying to measure a frequency swept sinewave knows that the
response can look like a mess. The spacing between the rising
edges of the "above Nyquist" sampled waveform will also be varying
as seen in Figure 1.
By setting the trigger level on a scope to the rising portion of the "-1V" level, we can capture a scope trace similar to Figure 1 that shows a 10 unique rising and 10 unique falling edges lasting 1000 time units with a pattern period of 2 kHz. The average frequency is approximately 2 kHz. The waveform resembles a amplitude modulated 20 kHz carrier . Only in this manner can we can capture a unique trace using a conventional oscilloscope. For any other trigger level, a family of "non-overlapping curves" will be generated.
Suppose the trigger level was set to "0 Volts" and that the scope was setup to trigger on every positive rising edge. Assume the trace-time( the time taken by the scope to trigger ,sweep, and flyback) is less than the period of a 100 units. The scope's finite persistence would reveal a family of curves where the second red rising edge at t=100 would be translated back to the first rising edge.
Figure 2 is a closely related case just described with the scope trigger set to just below zero volts. Point "a" (t=95) at which the second rising edge of the red waveform crosses -0.1V is transported to overlay the 1st crossing of -0.1V (t=5). This gives the "green" line. Similarly for the third crossing (point "b" at t=180) and the "blue" dotted line. Not shown are seven other "shark fins" that make up the family of responses. To re-iterate, to see the true nature of the "family of curves", it would be best to set the scope to capture a period of 10 cycles and using the trigger level set to -1V to uniquely capture the one (of ten waveforms) that does not confuse the scope retrace.
To the unitiated who is expecting sine waves, there seems to be no advantage in having this highly "distored" response (relative to a single sinewave).
[7.4.3] Digital Filters and
Oversampling
Besides the "shark fin" step response, many different approximations can be used to try to recover the analog waveform from these digitized sample points. One can draw straight lines between sample points. This works out best if sampling rate is much higher than the signal rate. At Nyquist, we are only 2x higher and assign one data point to describe each valley and peak. A bit like going on a trip to Everest and taking of the blindfold at the summit and at sea level. Interpolation is like the guide beside you that describes the journey/scenery in between with words.
For the Nyquist case in Figure 1 (black boxes), we will get a triangular wave at the right frequency with the proper peak to peak level. Not shown (for the black boxes) is the step response approximation typicaly of what you get from op-amp gaining stages with settling times much less than the period of sampling. In this case, we would get a steps very similar to the 44 kHz Sampling red line but with all the red "fins" at the same height.
If reproducing sinewaves from the minimal Nyquist data set is important to you. Then upsampling/interpolation might be for you. This happens automatically at low frequencies where we have no problems. At the highest frequencies, the near Nyquist information on the disc still only offers two points. To generate a nice rounded sine wave from two points would require someone or something to be psychic. This is roughly what upsample filters achieve. They take past samples and use them to estimate how a data between two sampled points . In effect we are influencing the future based upon the past.....Hmmm, jazz is not like that...it just changes from moment to moment. However, if you are playing sinewves, then this might be the ticket.
With the emphasis on reproducing sinewaves (frequency response) with low distortion, mathematical techniques such as "spline" interpolation can be used. The necessary knowledge to behave like a sine wave must however be gathered in the samples preceding the current samples. It is this weighting of prior samples which also gives rise to a memory effect of sorts. For a step function, we will see this in the form of a pre-step wiggle. For an impulse, we also see a pre-echo due to this attempt at being psychic for a input with basically no correlation to the past zero values. The zero "oversampling" design's simplicity prevents this pre-echo from happening.
Jan 20, 2005:
Jan 15, 2005:
2004:
The dAck! DAC has been to 2004 CES and the
2004 Rocky Mountain Audio
Show in the GR Research rooms in the same company as
electronics such as the Halcro, Dodd Audio.
2003: In 2003, we saw the pairing of Dodd Audio, GR Research and the Ack! Industries dAck! DAC.
8 dAck - DI/O Worthy Transports
Here is the link that discusses ways to
address the
performance limitations of transports via modifications in the
DI/O spirit.