HDMI De-Embedder

Here is a great discussion on HDMI De-Embedders to extract high resolution audio from DVD-Audio's and SACD's.

As for now, I get full resolution from the SPDIF output of my BDP-95, and I think SPDIF is likely better than HDMI for transmission anyway.*  I am lucky to have BDP-95, one of the few players to do this.  AND my current matrix switch seems to be blocking high resolution audio through the EDID--I could probably change that somehow, and when I next come across the manual for my matrix switch I will try.  Meanwhile I have the recommended Kanex Pro de-embedder installed, and it works for standard resolution audio.  If I could get HDMI to work for high resolution audio, I could use the SPDIF line exclusively for the Mac connection, and then use a superior F connector cable (see previous post).  But my current thinking is that full resolution through SPDIF is likely better than that though HDMI, so I would be disinclined to use the de-embedder for any other reason than convenience.

(*The HDMI is a 100 foot connection using HDMI to CAT6a conversion.  So it's hardly a "simple" connection either.)

RCA? BNC?

For a while I debated about whether to change the connector of my SPDIF line from kitchen to living room that now carries high rez audio from my Mac/Amarra and BDP-95 to the main system.  This line uses all RCA connectors on the cables now.

This line is particularly complicated for other reasons, when connecting to the Mac.   It works, but I now go through two levels of optical/coax conversion.  First, a very fine Inday Toslink splitter splits the Toslink output from the Mac into 4 Toslink outputs.  This Inday active splitter has always worked perfectly, whereas passive splitters never work.  From the split outputs, one goes to kitchen receiver, another goes to a DAC to provide analog to my hard drive recorder.  The third gets converted from Toslink to Coax via another very fine M-Audio CO2 converter.  As I explained previously, this combination of splitting and subsequent conversion to coax requires two levels of optical conversion (or 4 levels if you count both the optical emitters and the receivers).  It also goes through at least one transformer for the coax (in the CO2).

With all that conversion going on, it's a wonder that it works at all, but in fact it has always worked perfectly, as far as I can tell, though I worry about jitter.  What did not work very long was a different converter I got to split one Toslink to two Toslinks and one Coax.  I had lots of problems with that converter, then after I did get it working, it stopped working after one day.  So back to the Inday Toslink splitter, and the two levels of optical conversion, which bother me but always work perfectly.

Well I figured I could optimize this a bit by trying to make the coax connection better.  Currently that is very complicated.  From the CO2 I run a 12 foot budget video cable from spare box (vinyl cable with stranded wire--about the same as Radio Shack's lowest grade AV cable from 2006) to the kitchen connection panel.  There it runs through an RCA to F adapter, into the panel, where it then runs through 50ft of Belden Precision Video cable.  In the panel in the living room, there is a second RCA to F adapter, followed by two short lengths of Monster Video 3 cable joined with an old RCA barrel adapter.  This is what stuff in the real world looks like.

One way to make the connection better would be to eliminate as many RCA connections as possible, particularly the ones that must go through those questionable impedance RCA-to-F adapters.

The original plan for this SPDIF line was that it would only provide data from the Mac.  Then I would simply get a cable with RCA connector on one side (to connect to the CO2 converter coax output) and F connector on the other side (to connect to the F connector on the kitchen patch panel).  In one fell swoop I would be replacing the low grade 12ft video cable, remove one RCA connector, and remove one RCA-to-F adapter which might be even worse than just one RCA connector.

But once I started feeding the SPDIF line from both Mac and Oppo, I needed a connection which would be easy to change.  Screwing and unscrewing F connectors is a big pain.  For awhile, I looked for SPDIF switches, but there's hardly anything like that available and often it is way overcomplicated with re-clockers and the like.  So back to just using the patch panel as a patch panel, and changing the cables to switch the playing device.  So then back to finding a good connection for patching.

On a lark, I looked for a push-on BNC connector.  Sure enough, Neutrik makes a special push-on BNC connector that was intended for patch panels.  It appears that I could special order a cable with this connector on one end through Markertek, as they are a Neutrik dealer.  The I could get (and already did) a BNC to F adapter for my existing patch panel.

That might be an optimal solution, but I wondered if even a push-on BNC might get a bit bothersome after awhile also.  One thing about RCA connectors is that they ARE easy to plug and unplug.

I also realized I was not seeing the big picture.  I could get just as much benefit by replacing the cable in the living room with one having F connector on one end and RCA on the other.  That would also eliminate the need for the F-to-RCA adapter in the living room, as well as the barrel adapter.  And all this business about impedance matching in the coax line is probably insignificant compared with the two levels of Toslink conversion.

So, for now, I've decided to stick with RCA connectors in the kitchen, where I must change the cables often.  I ordered and have now received from Blue Jeans Cable a 10 foot Belden 1505 with RCA on both ends for the Mac-to-Panel connection.  I used that cable to verify that a 10 foot cable would also be correct for the living room, and I will order that second cable with F connector on one end.

I've got a new plan for the Toslink splitting.  I'll use two (now out of production) M-Audio CO2 converters in series connected together with coax.  It's hard to know if that would actually be better than the current setup though.

Multipoint Room Correction

One of the features I want in my next low frequency room correction system is multipoint correction.  That means that a correction will flatten a primary listing position while at the same time not letting secondary listening positions get too far out of whack.  I need that because the quasi-central listening position gets cancellation but much of the room around the boundary gets huge boom from room modes.

DSPeaker Dual Core has this feature.  One thing I don't like about the DSPeaker, however, is that it only has Toslink digital IO.  The Toslink output of my Tact preamp doesn't seem to work (didn't work with Behringer DEQ 2496 when I tried that a few months ago).  So I will need to convert the Coax SPDIF output to Toslink with an adapter.  Then I will need to convert the Toslink output of the DSPeaker into AES for input to the Behringer DCX 2496 which I may still need for crossover and time delay functions.  I might be able to use the analog output of the DSPeaker if I can program in correct inter channel delay and crossover, and it looks like I might be able to do.

Here is the manual for the DSPeaker Dual Core.

On the other hand, AcourateDRC does not seem to have multipoint room correction.  I was looking at that alternative, as it can run on the MiniDSP OpenDRC-DI, which has AES, SPDIF, and Toslink inputs and outputs (for only $299 w/o software).  MiniDSP claims you can get a license to extract filters from Acourate for only $99 but I have not confirmed that.  It looks like the full version of Acourate is more expensive than that.

Another possibility for the MiniDSP OpenDRC-DI is Dirac.  I haven't yet determined whether it is possible to set an upper frequency limit for the correction by Dirac.

A third possibility for OpenDRC-DI is REW (Room EQ Wizard).

Switching from Tact to something Else

What's Best Forum addresses the switch from Tact room correction to something else.

A consensus seems to be that room correction is best kept below 250 Hz.  Unfortunately, my earlier Tact 2.0 RCS can only do full range correction.  Later versions of Tact, in a late late update, got a top correction frequency setting.  For the longest time, the designer of Tact maintained that full range correction was essential, so a top frequency setting was not allowed.  That probably doomed the company, which appears to have vaporized (it had a good run for about 10 years).   You could (and I've never bothered to do this) set a correction curve that matched the upper frequency curves of both speakers well enough that there would be little or no correction.  I've never really mastered drawing target curves that well.

That's why I've been using manual room mode correction, for my subwoofer only (except I just last week added a minor notch to the panels).  I'm planning to try the DSPeaker Dual Core correction soon. It apparently has a low top frequency…though it's not clear you can set the correction top frequency.

But even without top frequency for correction, I can fake it by applying correction only to the subwoofer channels (after measurement).  Then whatever correction it might have applied to upper frequencies won't matter much anyway.

Since I upped the bass slightly (reducing the steep notch at 45 Hz) in the last time alignment I have been noticing excess bass boom around the room (but not at listening position).  My plan is to correct both at listening position and at a wall position, so the wall position (with maximum nodes) won't get exaggerated.

HDCD described

Here's a good (partial) description of HDCD.

It explains the amplitude processing features:

Peak Extension (up to 6dB of compression, matched by identical expansion during HDCD decoding, as with all other amplitude processing features.)

DSP Gain (+12 to -31.9dB)  If Peak Extension is also used, DSP gain is limited to a maximum of +6dB.

Low Level Extension, in normal and special modes.  Special mode does much more low end compression (raising the lowest amplitudes up to 7.5dB) compared to normal (which raises the lowest levels 4dB).

The post doesn't explain the variable filters, however.

Basically these features let you squeeze about 20bits of resolution/amplitude into 16 bits.

I like HDCD, and HDCD's are some of my best sounding recordings.  It is a bit problematic, however, that without HDCD decoding, what you actually get might have been better without the HDCD (contrary to claims HDCD proponents are always making) because, obviously, you are losing dynamic range if the amplitude processing features are used.

However, the producer might well have chosen to limit dynamic range the same way.  And therein, HDCD offers a quasi dual disc solution.  Non-audiophiles can listen to HDCD undecoded, and get the flatter dynamic range that may be more suitable for casual listening.  Audiophiles can enjoy the full dynamic range with HDCD decoding.

Now that I can play back high resolution files from my hard drive, I'd like to convert HDCD to 20 bits. There are playback software programs (including the standard Windows playback) that will decode, but I'm not aware of any way of saving the decoded data to a new file.

Hi Rez on Oppo BDP-95

I'm very happy I can get 96kHz digital from the SPDIF output of my Oppo BDP-95 playing commercial DVD-Audio discs, like Hotel California and Rumours.  But is it the full 24 bits?  I was thinking I might do some kind of test to figure it out.  Not easy to do for various reasons, especially in the case the output might have 24 bit dither even if not 24 bit information.

Anyway, famous audio reviewer John Gatski has already done the tests.  Not only does the Oppo put out 96kHz, but it's also 24 bit, from the SPDIF output.  He used ATI ASDAC which has bit depth indicator to confirm.  The Oppo BDP-105 does not do this, it dumbs down the SPDIF output to 16/48 like most DVD-Audio players.  So what's with the BDP-95?  I think possibly the DVD-Audio consortium wasn't answering their phone when the BDP-95 was being designed.  But afterwards the RIAA got to them and told them never to do that again.  However even on BDP-105 you can get the full resolution stream from HDMI as well (which doesn't seem to be working in my system, possibly because I have programmed my matrix switch to only allow 2ch audio).

Now even if he used a bit rate meter, it wouldn't prove that the BDP-95 wasn't just dithering to 24 bits. However, if that were true, why didn't Oppo do that in the BDP-105?

So the best information that I have is that I'm getting the highest resolution my system can handle, 24/96, from the SPDIF output on my BDP-95 when I'm playing DVD-Audio.  The best of everything!

The Best of Everything

It's been a harrowing six weeks since I returned from vacation, particularly haggling with the guy I bought a semi-functional Denon DVD-9000 from.  It didn't play most DVD-Audios.  Out of a stack of 8 it played only one.  The seller agreed to pay for repair.  I got estimate, but before I got his approval (he took more than a week) I went ahead and paid to have it fixed.  But replacing the laser didn't help.  In fact, afterwards, it wouldn't play the one DVD-Audio it played before.  Still plays CD's.  So now I was stuck.  I offered either to have the old laser put back in (no guarantee that would restore original mostly broken operation either), take $125 off the refund, or settle for $250 partial refund (out of $699 purchase price) for unit that only plays CD's (and not DVD's).

I might not have offered $250 except I figured that with the $125 off (because the repair made it slightly worse) the amount I was effectively getting off was at least $375 compared with sending the unit back. And if you count the cost of shipping and materials, make that $450.  Then, if you count all my effort and patiently waiting…you see I actually got the player for free compared with sending it back.

Actually, of course, I paid $699+shipping+$89 (for ineffective repair) - $250 partial refund.

Anyway, it rarely happens that I've gotten any such satisfaction in previous cases, so eBay's buyer protection system works (even though I didn't elevate this case to the full eBay buyer protection, the seller was under effective threat that I might).  It's unclear what they would have done in this case.  I might have gotten full refund (but still out shipping and ineffective repair costs).  Or they might have maintained that the ineffective repair made the unit unreturnable.  I think it would have gone my way.

I still look at getting the unit at low cost after some mistakes (like the repair and the minor damage it caused--which I feel correct in taking responsibility for).  And for one thing it does, it might have been worth having at any cost.  For this may well be the Best HDCD Player Ever!

It has the dual differential PCM 1704 for the real PCM conversion.  And the separate power supplies, etc.  And it's built like a tank (which, sorry to say, often has little to do with actual reliability).

Many of the famous HDCD players used predecessors to the PCM 1704, such as the PCM 1702 and PCM 63.

You really want a real PCM converter, I believe, especially for something sampled with a real PCM converter like the Pacific Microsystems Model One and Two.  So you can rule out pretty much all the other universal players that have been made since the DVD-9000, including the estimable Oppo's BDP-95 and BDP-105.

I actually did and A/B/A comparison (I hate doing those) with my DVD-5900.  The difference was as I expected.  Through the DVD-5900, which uses high grade sigma delta conversion chips, the HDCD sounded like you were hearing the music play from inside a jello mold.  The dynamics were inverted, so that transients were convex rather than concave.  The DVD-9000 seems to have measurably higher transient output on HDCD's, maybe by 1-2dB.  I had the change the gain setting on the Lavry to compensate.  I have also worried the extra output might be a sign of some sort of deterioration.  But the DVD-9000 is by far the better sounding.  It's been a revelation like hearing everything the first time.

This is exactly the sort of difference I was expecting from my "information" analysis.  The real PCM player preserves the full information, the sigma delta player shaves the original information off by giving a successive approximation, which only looks good in slow measurements.

So I now have by far the best HDCD playback I've ever heard, and I've been enjoying that on Reference Recordings and an HDCD encoded set of Mannheim Steamroller Fresh Aire I-8.  Which do finally and really sound fresh and not canned as they do on sigma delta HDCD players, or worse without HDCD decoding.

But what about DVD-Audio.  I cherish my DVD-Audio's and in fact just bought a new one (!) this month.  This format isn't dead at all.  Am I still forced to play those through sigma delta derived analog output?

No, and in fact, I can do that even better than I expected, I have found!  My Oppo BDP-95 is the rarest of models that can put 96/24 on the spdif output.  And I can now route that to the living room on coax (as I also do for high res audio on my Mac).  I first tried that last weekend, and was blown away.  It was even better than my DVD-Audio played on Onkyo RDV-1, resampled by Lavry AD10, recorded from digital on a Masterlink, then transferred to my Mac.  This was not a resampling of the DVD-Audio, it was the recording itself!  The bass had even more impact and tunefulness.

So between the DVD-Audio and HDCD, my house was playing wonderfully this weekend, and still.

Actually I don't know what might happen, though, if the DVD-Audio has a 192kHz stereo track.  Does the Oppo digitally convert that to 96kHz?  Then I've got digital conversion…though probably fairly benign.

In addition to the eBay fight I was having (which created a continuous feeling of stress…even going back to before I placed the order as the eBay buy-in-now page malfunctioned and I had to pay from checking, which screwed me up for several days until payday I had to keep adding money to checking), I finally got around to doing the time alignment on my system.  Last time I did a full time alignment it was 3 years ago.  Everything was way out of whack, including one subwoofer being out of polarity (at one time, I was testing that as a way to reduce boom).  The time delays were way off too.  Following the Tact measurements I also tweaked the EQ a bit, reducing the huge notch at 45 Hz a tad, and even adding a slight notch in the Acoustats low bass but well above the crossover point (but nowhere else).  The result measured nicely and sounded wonderful.  The new DAC (Onkyo RDV-1) for the midrange helps greatly, and I'm running the digital typically with little attenuation (maybe 3dB in midrange) due to the low output of the Onkyo, which works beautifully in my system.  This also means, btw, that I can't turn up the volume very much more, though the Tact can play up to +7dB, I'd rather not go above 0.  But having extra volume control range below 0dB which you don't use--is actually a bad thing.

Almost all kinds of attenuation (except amplifier gain switching) are information losing, so having less attenuation also means that much less is being lost.  My typical master gain was more than 10dB lower because the Behringer DCX2496 has 10dB more useless output.  Resistor ladders also reduce information in an analog system--6dB less output from a passive network means 1/2 as much information down to either the noise level (the conventional, but wrong IMO, way of thinking about this) or down to a quantum level (probably around 10-30 below 1V).  Either way of measuring information, you get this loss through attenuation, either resistive or in digital.  (Only HDCD avoids the information loss when going to lower levels, though it's level control features.)

I'm also copying the Reference Recordings HRx files to my Mac and converting them, using Triumph with the Izotope resampler, to 88kHz for playback through Amarra through my digital link to the living room.  They didn't seem to get resampled to 88kHz by the Oppo.  None of my digital stuff goes above 96kHz.  Lavry argued that 192kHz wasn't better anyway.  Anyway, my resampling is "the best" also.