Living Room System

Sunday, April 6, 2014

Crossover Settings

These are the settings of my Behringer DCX 2496 crossover as of April 6, 2014.  I worked on these settings over the past few months, primarily to reduce bass boom from room modes that were disturbing outside the listening position especially, and made the walls rattle so it was not enjoyable at the listening position either.  I think the bass may be a bit too attenuated at 45 Hz on music (but on pure tones, it doesn't sound attenuated) but I haven't been able to come up with settings I like better.  It seems any attempt to roll back the -11dB attenuation at 45 Hz leads to boom in a strange nonlinear way.  However my experimentation has been cut short by my fear that I would lose these magic settings, so I also wanted to make sure I recorded them before further fiddling.

#1 Subwoofer Left w 10dB balanced attenuator

Channel A -9.0 dB
LR 24 80 Hz
EQ 1.  BP 30 Hz +2.0 Q3.5
EQ 2.  BP 45 Hz -11.0 Q4.0
EQ 3.  BP 39 Hz -3.6 Q 5.6
EQ 4.  BP 32 Hz + 4.0 Q 2.0
EQ 5.  BP 25 Hz +3.0 Q 1.4
Delay 0

#4 Subwoofer Right w 10dB balanced attenuator, level at 1220 (slightly beyond straight up)
Channel B -9.0
LR 24 80 Hz
EQ 1.  BP 27 Hz + 1.8 Q 1.6
EQ 2.  BP 45 Hz - 11.0 Q5
EQ 3.  BP 66 Hz -3.0 Q 3.2
EQ 4.  BP 32 Hz +2.0 Q 1.4
Delay Long 0.1m 0.29mS

#2 Acoustat Left unbalanced via Aragon 8008 BB
Channel A -6.1
LR 24 80Hz (no high pass)
EQ Off
Delay Long 0.85m 2.47mS

#5 Acoustat Left unbalanced via Aragon 8008 BB
Channel B -6.1
LR 24 80 Hz (no high pass)
EQ Off
Delay Short 8mm 0.02mS Long 0.85m 2.47 mS

#3 Elac 4pi via Parasound HCA-1000A no attenuation
Channel A -3.0
Butterworth 48 20kHz high pass only
EQ Off
Delay Short 672mm 1.96mS Long 0

#6 Elac 4pi via Parasound HCA-1000A no attenuation
Channel B -3.0
Butterworth 48 20kHz high pass only
EQ off
Delay Short 678mm 1.97mS




Thursday, April 3, 2014

Do we need audio bandwidth above 20kHz?

Previous post linked to articles related to what has been called "audio objectivism" that all we need is audio equipment with 20Hz - 20kHz bandwidth and other similar easily measurable and long accepted "high fidelity" requirements.  Now, with audio downloads going to 24bit/96kHz and 24bit/192kHz PCM formats and DSD (which also has significantly higher bandwidth than 20kHz, though the upper bandwidth may be more corrupted by noise than PCM formats), a particular question is whether we specifically need audio bandwidth extended beyond 20kHz, and if so, how far?  Here's the article I linked previously which is primarily concerned with this question specifically, and it may be worth reading despite the incredible one-sidedness, condescending tone and over simplistic diagrams.  It is the discussion on digital audio sampling here which is comparatively well presented.

…………..


Now I've found a wealth of other information about the bandwidth topic.  And more test material!  Unfortunately, not enough to reach any sort of conclusion, but enough to know this is a serious issue and that many people taking the "higher bandwidth" side are indeed very serious audio engineers and scientists, not the starry eyed kooks and charlatans you would think they were from only reading audio objectivists--who would like you to believe this is an open and shut case and no qualified audio engineer would think otherwise for a second.

Here is one of the best summaries I've found, written by the famous audio reviewer Martin Colloms.  I must say I was very surprised to find him ultimately taking the lower bandwidth side, but his paper is very useful  because he takes the issue seriously and documents many important studies with contradictory results.  I was surprised because of my previous knowledge of reviews written by Colloms, and if you take a look at other articles in his archive, you find that has often not been on the side of the audio objectivists with regards to things like the need for special capacitors (above and beyond what would be required to achieve 20-20kHz bandwidth--for which almost any available capacitors would do fine) and exotic Class A power amplifiers (audio objectivists claim all reasonably well designed audio power amplifiers today, with THD less than 1%, 20-20kHz bandwidth and flat frequency response sound identical, so one might has well have Class AB or D amplifier that meets these specifications than an extremely expensive and high quiescent power consuming equivalent Class A amplifier whose advantages over a well designed Class AB amplifier might not even be measurable in the ways suggested by audio objectivists).

I discovered the Colloms article reading this long and argumentative blog about 192kHz downloads at Steve Hoffman's forums.  This blog is very good reading, once you get past the useless posts, it is full of arguments and evidence on both sides, and having links to articles on both sides.  Unlike many others long blogs I've seen that are often more heat than light.  Just to specifically link some of the things it links, famous audio engineer and scientist (Cal Tech professor) James Boyk has long taken the higher bandwidth side.  In this paper, he merely shows that musical instruments produce significant acoustical energy above 20kHz, and then he simply refers to experiments by others showing this has effects on humans, some of the same studies discussed by Colloms.

Monday, March 31, 2014

More on audio objectivism

NwAvGuy has posted a great and lengthy thesis on what we can (and can't) hear.

I should read and digest this, especially now that since 1983 I've crept back in bed with the subjectivists.   I officially consider myself a "Grey Hat" who looks for plausible scientific explanations of what subjectivists talk about--beyond mere dismissal--while simultaneously saying that anything not proven in blind testing should not be considered "a big difference" as subjectivists are constantly doing.  My hope is to find things that have been overlooked in standard audio engineering, or maybe even further beyond, in information theory.  However I immediately concede that I have not done so, and that many subjectivist claims regarding how the objectivists were proven wrong are not accurate, and that overall the objectivists have been more right than wrong, and perhaps even entirely so.

Here's a similar thesis maintaining that 24/192 music downloads are very silly.


4580 IC's in Onkyo RDV-1



I was greatly relieved to find the Onkyo RDV-1 does indeed accept 96kHz inputs (in either coax or toslink, it turns out).  I run all my digital processors at 96kHz.  When I first opened the box I checked the manual, it said it only accepted up to 48kHz.  Despite my lack of time on Saturday, I quickly hooked up my new Black Lion Micro Sparrow ADC (mk1) to the Onkyo, and found it handled 96kHz fine and even shows a message saying it is doing 96kHz.  The Onkyo does not accept digital inputs higher than 96kHz, when given such inputs it displays the LCD rate, either 44.1 or 48khz.  I did not listen to see what those did.  But 96kHz in several tests over the weekend appears to be rock solid and sounds fine through the "stereo" outputs (not the same as the "front" outputs on this player…the "front" outputs sound very thin as subwoofer use is assumed).  I measured the Stereo outputs with pink noise (which is random, so always fluctuating and never perfectly flat) and it looked as good as it gets with random noise measurements:







I've been pouring over the service manual schematic to see what the circuitry is like.  The output board has each output channel going through a 4580 opamp, then a 47uF capacitor, a small resistor, a number of shunt muting devices, and then the output.  Not wonderful but not too bad either.  I'd like to replace that electrolytic cap with a Teflon.  But what is a 4580 opamp?

Apparently it's similar to 5532, a higher current bipolar amp.  The 4580 is said to be much better than the 4558's used ubiquitously in "mid fi" audio equipment, maybe even some not so mid fi.

While researching this, I found the quotation from John Curl himself, as of 2009 his CD player uses 4558's (!!!)



Trevor, the 4558 is the IC that I have to listen through to compare SACD to DVD to CD. I can't afford an expensive player, because I don't design them.

I also found this objectophile blog saying essentially that 5532's are good enough.

http://nwavguy.blogspot.com/2011/08/op-amps-myths-facts.html

I just checked photographs inside the Behringer 2496 DCX and guess what chips it uses?  4580's!  Somehow that still gets to about 0.1% distortion at full (just below 10v rms) output.  The problem with the DCX may be mainly that it insists on driving the outputs to 10v RMS at 0dB digital, and the only way you can attenuate that is by reducing the digital numbers, which is losing resolution and linearity.  And also that the DCX doesn't really have good enough power supply to drive the chips to 10vRMS with low distortion either.  To really do the 10V output well, it either needs better power supply, better chips, or both.

Meanwhile, my friend Tim is planning a significant chunk of work to add the best sounding of all IC's according to him and other sources, the OPA211, to our Kenwood L-1000T tuners.  That sounds good to me, so I won't be bugging him with links saying all good chips sound alike.  In fact I plan to address the psychology of those claims in a future column.  Tim analyzed a set of top opamps and decided which ones were actually most linear, the OPA211 on top, just as in the list he started with.  Here is the list he started with (not sure from where):

OPA211
AD797
OPA827
LME49860
AD845
OPA2132
LT1115
LT1122
LT1468
LT1469

Compared with 4580's, Tim says OPA211 have 20dB more gain (which is good for opamp circuits, particularly ones with equalizing feedback) 80Mhz vs 12Mhz bandwidth for 4580, and distortion of .000015% vs .0005%--a factor of 30.  He says these specs (gain, bandwidth, THD) correspond generally to better sound quality (as determined in tests by him and a professor of engineering he collaborated with).

He generally doesn't respond when I talk about sighted tests, statistical significance, and the like, so I'm not going there.  It's not like even "white hat" audiophiles blind test everything to statistical significance.  The develop rules of thumb which extrapolate from blind tests, and then follow those rules of thumb.  But there's no telling whose extrapolation is actually better, without actual evidence.  All we really know from serious audio science is what the minimums are for good fidelity.



Thursday, March 27, 2014

How the SL 1000 was packaged

The SL1000 turntable base and attached motor (less platter) were packed in a makeshift oversize box which was made of two layers of corrugated box material salvaged from two other boxes.  As I opened the box from the top, or at least where the shipping label was, I first encountered 3 layers of bubble wrap covering the base and motor, with foam pads around the sides.  The turntable had been upside down, and looking down on the box I could see the bottom feet.  I carefully removed the turntable base and motor from the box, and after removing the bubble wrap, they appeared undamaged.

In several initial emails, starting from before I purchased the turntable, I told Mr. George Heid how I wanted the turntable pieces boxed.  He had agreed completely, and said he would box them as I had said or even better.  I also described how to do proper double boxing, with two inches of solid padding between the inter and outer boxes, and that the outer box should be a very strong box, preferably triple ply.  Even before I ordered the turntable I ordered a motor bracket to secure it.  I told him to wait for that, and he said he would start working on packing the turntable anyway.  It took almost 2 weeks for me to get the motor bracket, and then I sent it to him by 1-day express mail.  He then told me he would get the turntable packed and shipped that week.  At the beginning of the next week he said he was still working on it, but he would ship it by the end of the week, and so on.  Finally he did ship the turntable on the 45th day after purchase.  I only discovered around day 40 that I could not open a case through eBay buyer protection about non-delivery since more than two weeks had elapsed since the projected delivery date.  But he did finally ship a few days after I called the number he had given me earlier oat about day 25 in order to describe some better condition SP10 parts I could buy from a friend of his, especially a better condition dust cover (without the small crack shown in one corner) and asked me to make an offer (I made some low offers, still having confidence in him but worried about buying things off ebay, and also not quite appreciating the value of an SL1000 dustcover).

It was pretty clear from the moment I picked up the boxes at UPS that he had not done as I had said, and it was even clearer after I started opening the boxes.  So I was extremely relieved to see that the turntable base and motor were undamaged anyway.  After doing this, I left home for the workday.

I believed at the time that it was only the base and motor that were in that box, because there were three boxes and I figured the dust cover was in another box.  I only discovered later that day that underneath where the turntable base was, there was also the dustcover, wrapped in bubble wrap, and it had been broken in three pieces presumably by the weight of the SL1000 base and motor.  Unfortunately, at that time, I was so horrified I didn't take pictures of it until much later.








Sunday, March 23, 2014

I have seen data sheets for Burr Brown PCM63 and PCM1702 that call them Sigma Delta.  Never the PCM 1704.  Maybe marketing changed, but all three of these dacs are are now known as PCM dacs, not sigma delta dacs.  They all feature dual collinear dacs summed together.  These collinear dacs uses the same ladder in opposite directions, hence canceling out error.

Here is the official data sheet on PCM 63.

I actually have a DAC that uses these, an Aragon D2A2, currently not working (some electrolytic in power supply failed and leaked).  I loved the fact that it has HDCD, but I hated the fact that the input selector goes back to position 1, which I couldn't use, after every power outage.  I had it wired in for the disk player (then, a Denon 2900 which lacked HDCD) and the Sonus input (to de-code HDCD discs), and the Dish receiver (to get Dish music).  The Dish connection (via Toslink) often sounded underwatery for some reason, so I quit using it and had to run Dish audio through an isolation transformer to fix a ground loop I had then.  (Little did I know, my whole house was ungrounded then.)  Getting HDCD was always problematic too (of course Sonos had to be on max level).  I rarely did the Sonos thing with it, too much hassle.  And before I knew it (a few years after I got it in 2007 or thereabouts) it died (20012?) with leaky capacitor in the power supply.   My Aragon 8008 BB from that same era is still doing fine.

Anyway, I do need to get D2A2 fixed.  It really is nice.  And now I know ladder dacs are interesting.

I saw StephenSank on DIYAudio.com bragging that his Pioneer DV-AX10 uses PCM1704, but that's not what the Pioneer website says, so I posted that to this blog about 1704's.  It turns out earlier ones, or something like that, used 1704's (it's shown in a 2000 service manual) but later Pioneer switched to Analog Devices (as one person saw when opening one).

Here's a great list of which DACs us which DAC chips.

Most of those are nearly unobtanium, or ultra expensive even now.  I saw the Levinson 360 going for $3100, and the Wadia's sell for still more.  Now, 14 years later.

An assemblage 3.1 sold for $1350 late last year.  Those are nearly unobtanium, it looks like.

On Sunday I bought an Onkyo RDV-1, which features two 1704 dacs, and apparently clock developed by Apogee.  It has SPDIF coax and toslink inputs, and can be used as a DAC.  One reviewer in 2012 found it better sounding as a DAC than his Schiit BiForst.  I am hopeful that it will at least allow 96kHz input (most likely it will not allow 192kHz input given the date of manufacture, but 96kHz spdif was common by 2000, and especially since Onkyo went so far as to get collaboration with Apogee, there is a good chance the input will allow 96kHz.).  Details have been hard to find on the web.  This was a DVD-Audio player made to the highest standards to compete with the top multichannel players of Sony, Denon, Toshiba, and others.  It was replaced far too soon with the RDV-1.1, which added SACD (what I now call modulated noise) capability, and the collectible 1704's were replaced with early sigma delta's by Wolfson.  The 1.1 lacked the external inputs too.  Onkyo apparently bowed to market pressure and the Sony SACD juggernaut.  Here's a blurb from the blog linked above:


Integra Research's RDV-1 DVD, DVD Audio, and CD Player
05/14/01 - The Integra Research RDV-1 is a THX Ultra certified DVD player that combines professional audio and video circuitry features to extract the ultimate performance from DVD, DVD Audio, and CDs -- including CD-R recordable disks. It can also function as an outboard D/A converter for other source units. The RDV-1 was designed from the ground up to set new standards for DVD performance and quality; the D/A converters and power supply alone make this product stand out from the competition. The RDV-1 uses 192kHz/24-Bit DAC to provide the most accurate DVD-Audio playback possible. The DAC uses a Vector Linear Conversion (VLC) system with a low jitter Master Clock developed by the professional audio firm, Apogee Electronics, of Santa Monica, CA. The Apogee clock all-but eliminates jitter and provides for the highest quality digital conversion available. The Vector Linear Conversion system completely eliminates the ""sonic unevenness"" inherent with conventional conversion methods.This low-jitter digital clock circuit was first developed by Apogee for the professional music recordingindustry, and is at the very heart of the best equipment used to make the master recordings for music. Now, Integra Research and Apogee have used the same technology to play these recordings back at home. Jitter is the measure of the lack of rhythm between digital sound samples. Unfortunately, the human ear and brain are very sensitive to these tiny timing irregularities. Jitter of just a few nanoseconds can compromise digital audio performance by interfering with the brain’s ability to perceive a stereo soundstage. By using the Apogee clock, the RDV-1 minimizes jitter and insures each digital sample arrives in perfect step with all the other samples. With all the digital signals zipping around inside a DVD player, there is a lot of potential for these signals to go where they do not belong. To circumvent these problems, Integra Researchhas developed high-quality dual power supplies to provide inherent DC stability and ensure that no traces of digital artifacts enter the audio paths or analog ground. The Integra Research RDV-1 is also state-of-the-art when it comes to video. It has progressive scan video output for a smooth, flicker-free image, and compatibility with digital-ready TVs that can upconvert video signals. The video playback system uses a 27MHz/10-Bit video D/A conversion with four times the accuracy of conventional 13.5MHz/8-Bit systems. In addition to a full complement of optical and coaxial digital outputs, the RDV-1 has a multichannel analog output (DB-25) for simple single-cable hookup of multichannel applications.

96kHz is all I need right now, since everything else I have runs at 96kHz, except I can use my Lavry AD10 to resample higher rates from my Denon 5900 or Oppo back down to 96kHz, so it can run through my Tact and 2496 DCX.

I figured out finally how I can get digital output for the midrange to run through a real ladder DAC (and not the reasonably good, but still sigma delta AKM, inside the DCX.  The answer is, bypass the DCX for the midrange!  I wonder why I had never thought of this before.  I can use a DEQ to create whatever crossover I want by combining parametric EQ's !  And the 2496 DEQ does have digital output.  And it can crucially add the needed time delay so I can time align my speakers, as before.  Actually, one could create a super DCX this way with three DEQ's.  But I think it's only worth bothering with for the midrange (which is 80Hz to 20k on my Acoustats, so it's really almost everything except deepest bass and super extreme highs), because it is a bit more hassle than using the DCX, as well as being more expensive (drop in the bucket compared with other solutions, like the $1000 digital output upgrade for the Behringer, which is no longer listed on a mod website, and I wonder if it handles the muting that the DCX needs properly).

So with my spare DEQ, and the new Onkyo, I can upgrade to full ladder DAC in the midrange.

Friday, March 21, 2014

PCM converters

Dan Lavry sets the story straight regarding PCM AD converters.  First of all, non-sigma-delta converters are PCM (or that's the term he uses).  Sigma delta converters can be 1-bit or multibit.  All converters were at first PCM, then 1-bit sigma delta appeared.  1-bit sigma delta had problems, so it has been almost entirely replaced by multibit sigma delta converters.

The Pacific Microsonics Model One and Model Two were PCM converters.  NOT sigma delta.  The actual AD unit in these (?both?) were Analogic PCM units tweaked by them.  The DA they used was from Analog Solutions and designed by Dan Lavry.

Lavry's DA924 converter is PCM, as is the Lavry Gold.  They are not a straight forward architecture though.  They were an improvement on earlier designs he did for Pacific Microsonics, Levinson, Wadia, and others.

Lavry now (2007) thinks multibit sigma delta are OK.  PCM still has advantages, but it is very expensive to do correctly, with hand calibration and the like.

The output of a 1-bit modulator can be copied straight to DSD, or it can be decimated to PCM.  But 1-bit modulators were replaced by multi-bit modulators.  Multibit sigma delta modulators require either an downsampler (AD) or upsampler (DA) to complete the system.

The Analogic AD used by PM was similar to Lavry's ZAD-16 made by Analog Solutions.

Whenever one tries to criticize converters based on specific things, such as architecture, he gets very testy.  Reminds me of the late James Bongiorno, who got testy whenever I tried to ask him any questions.  But we cherish our great designers regardless of personality, and Lavry is the best.  Meanwhile it seems to me completely fair to say 1-bit anything is total crap, multibit sigma delta is better, and PCM done right is the best.

Sadly, the unique (PCM with auto calibration) DA924 converter is discontinued.  So perhaps that is probably why Dan Lavry got testy and refused to say unambiguously what I just said.  He may have already decided to move on to multibit sigma delta, believing it to be better.

But designers are not always the best judge of their own work.  WRT James Bongiorno, while he may have been a brilliant designer, emulated by others, I think many of his designs weren't that good in totality.  The Sumo Nine was one of his favorites, but it had a fatal flaw IMO, it used fans (as did the original Ampzilla).  Likewise nearly everyone who isn't a Bongiorno fanboy thinks The Charlie wasn't the best tuner at the time (I don't have one of his personally adjusted ones with rack handles, but I've heard lousy reports regarding those also).  With Dick Sequerra, his best tuner was likely not the one with his name, but the Marantz 20b, and FM has continued on far longer than he ever anticipated.  And so on.  How can I be the idiot that I am, and know these things that the genius designers didn't know?  I don't know.  All I do is listen and think.

I think designers of analog converters have been far too preoccupied with voltage accuracy.  The most important thing, most likely, is in preserving timing.  And as James Bongiorno told me point blank, the truth is not in the measurements.  It is in the math.  I now believe he was correct.  I was asking him what measurement I could do to find the weakness in a pulse count detector.

I think if I had the money, I would try to get the DA924.