Revealing the Denon DVD-9000 HDCD bug, in pictures

[For background, see the preceeding post Curiouser and Curiouser.]

To prove the DVD-9000 HDCD bug, I first played a track on the Mephisto and Co. HDCD, which has been ripped onto my harddrive library.  It plays on the DVD-9000 operating as a coaxial SPDIF DAC (selected by front panel switch), fed coaxial SPDIF from a Sonos ZP80, being fed through the Roon streaming software.

With the gain on my Emotiva XSP-1 preamp set to -5dB, the level peaked considerably above a normal CD, as shown by the LED's on my Lavry AD10 converter.  -1dB to be exact.  That corresponds to nearly 4V output on the Denon (because the Lavry is set for 0db at 3.88V, and the Emotiva doubles the voltage of single ended inputs at its balanced output).

Next, I played a track from  non-HDCD, the album Out of Sight by Pancho Sanchez.  This was also showing -1dB output, same as the HDCD, which should not be happening.

Next, I played a special CD I had burned years ago, with a 0dB recording of 1000 Hz.  This was created with Cool Edit Pro, and reaches the 0dB level precisely at peaks, but with no clipping.  I do not store this CD on my harddrive to avoid playing it by accident.

This showed a level of -7dB, indicating a normal 2V single ended output from the Denon.

Then, I switched the Denon back to Coaxial input, and played the same track from Out of Sight again.  Now, it's only reaching -6dB, like an ordinary CD (but 1dB higher than a 0dB fixed tone recording because of inter-sample-overs, which I may discuss in a future post).

So, to recap, playing the 0dB disc appears to have changed the DVD-9000 so it is back to normal, playing normal 44.1/16 PCM files at 2V output, instead of the 4V output it was reaching immediately after having played an HDCD with peak expansion.

Further tests indicated I could also "reset" the stuck Peak Expansion mode simply by flipping the Source switch on the front panel of the Denon from Coaxial to Disc and back.  And also that the bug does not occur when playing a CD disc after playing the HDCD disc, it only occurs while the DVD-9000 is being used purely as a SPDIF DAC.

Curiouser and Curiouser

Yesterday seemed kind of a dud.  Mostly just writing up various things discernable and claimed about HDCD.  Was it a breakthrough, or a scam?  Well, I still don't know, probably a bit of both I'll concede, but I've discovered some amazing things.

For one thing, and here I may have the smallest scoop of all time, but it is actually most likely a scoop because I doubt anyone has reported this before:

The Denon DVD-9000, when used as a SPDIF DAC, has an HDCD bug.

If I play any track from Mephisto and Co., one of the most justifiably praised HDCD's ever, and something I've long used as demos and tests, while using the DVD-9000 as a DAC to decode it, which it seems to do better than any other DAC I've experienced, it apparently sets the Peak Expand state of HDCD, which is used throughout this disc according to ffmpeg.

THEN, if I play anything else, even if not using HDCD, it gets peak expanded also.  The peak expand mode stays set (or, perhaps it was because I didn't play Mephisto all the way through...not yet tested, just a few seconds is apparently sufficient to enable the peak expand bug I am describing).

So it's a bug in general, if you happen to be using the DVD-9000 as a DAC, which is probably a significant fraction of the small number of DVD-9000's still being used, and playing some fraction of HDCD's.  In that case you, as I once did, might consider the DVD-9000 one of the best DAC's ever, because it's cheating, applying peak expand to everything.

BUT, it's also a feature!  I can now apply peak expansion to basically anything.  It's like having Carver's famous (or infamous) Peak Unlimiter on tap.

And, I can tell you, it sometimes really ups the ante.  It really adds a lot to other albums, especially it seems, HDCD albums which don't actually use the Peak Expand feature.  Such as Mannheim Steamroller Fresh Aire II.  Or even non-HDCD albums, like Pancho Sanchez' Out of Sight.

This is a Damn Good peak expander function.  We can already hear how fine it sounds on Mephisto.  If you listen to that HDCD disc on a non-HDCD enabled player, it sounds perfectly fine and wonderful.  You would just never know you are listening to a disk with at least 4dB of compression.  Then, when you listen to it on an HDCD player, it knocks socks off.

But that's exactly how a GOOD compression scheme should work.  You don't want your socks knocked off while your driving down the freeway or on your bicycle.  You want your socks knocked off when you're seriously listening to your big rig, with the HDCD equipped player.

As I started to say in the last post, this is exactly how it should be, or should have been, to end the loudness wars.  FM stations and car drivers can play the compressed version, but turned up to a higher average level.

But if you want to turn the Peak Expand state off, you can simply turn the Source switch on the DVD-9000 back to Disc for a few seconds, and then back to Coaxial.  And by the same token, the bug does not appear to be triggered by playing the Mephisto disc, and then any other disc.  It seems to only occur if you are using the DVD-9000 as a SPDIF DAC.

Anyway, the story of how I discovered this and many things, was like falling down the Rabbit Hole.

It started with the delight of getting my Denon DVD-9000 back online as a DAC, so I can play HDCD albums on my hard drive through Roon with actual HDCD decoding, from the very best HDCD decoder I have ever experienced, the Denon DVD-9000.

The DVD-9000 is not just a 40 pound monster Statement Piece by Denon as the first foray into serious competitive DVD-Audio (this player DOES NOT support SACD, though some had believed a PCM conversion would be permitted, Sony was never willing to license it for this player, which has a the last and best PCM DAC chip, the Burr Brown 1704, used in it's best dual differential mode, much like the top Mark Levinson 360S DAC did for 10 years afterwards, and many other famous DACs of note).

It is also apparently the first generation HDCD product to not include a PMI digital filter chip.  It uses the second generation HDCD specification, which was never made into a chip, within a 32 bit digital filter system, which Denon called AL24.  And, the development of all this probably occurred before, or contemporanously with, Microsoft's purchase of HDCD from PMI.

If any DAC has the "transient filter" feature enabled, this might well be it.  Maybe it doesn't, maybe no HDMI players did, but I can state that this HDCD DAC sounds better to me than any other HDCD DAC I've ever heard, by a large measure, and it could be because of a Transient Filter feature, which I had always believed existed.

******

Just after setting up the Denon DVD-9000 with SPDIF input through a Roon enabled Sonos ZP80 Zoneplayer, I wanted to be sure this was actually working to correctly decode HDCD albums in my library (if not others).

It's damning that the DVD-9000 does not (usually) light the HDCD light when playing HDCD albums through the SPDIF input.  That bug had been observed by others.  It appears to only do so if you are playing an HDCD disc.  But it also appears, I have confirmed, to do the HDCD decoding for 16/44.1 SPDIF signals as well, if they are HDCD titles.  It just doesn't light the HDCD light.  (And, yet, sometimes it does light the HDCD light, but I have not been able to duplicate this occurance, a peculiar series of albums might cause it, or perhaps I incorrectly observed this when an HDCD disc was loaded.)

So, how does one confirm that the HDCD decoding is actually being done, in the absense of a reliable luminous HDCD indicator?

Well, before I knew of anything better, I could measure the output of the DVD-9000.  I do that automatically when I listen to the analog outputs, which are buffered and level adjusted transparently by my Emotiva XSP-1 preamp, and thence fed in balanced form to my Lavry AD10 analog to digital converter.  The resulting digital signal then feeds my digital crossovers and equalizers.

The Lavry has a wonderful array of LED's, with Peak Hold that stays held until you flip the reset switch.  Which means I can measure and/or compare any level within 1dB.

I've known for some time that HDCD's put out higher levels than regular CD's on the DVD-9000.  This was what the original HDCD license required, and the DVD-9000 implements it.  Later HDCD players tended to drop this feature, which reviewers and fans of other equipment (especially Sony equipment) tended to depricate.  Justifiably, non-fans of HDCD thought this was an unfair trick, and indeed, it was.

Many falsely claim that CD players were supposed to put out only 1V maximum for CD's, and 2V peak extended maximum for HDCD's.

What the Denon does is put out 2V for regular CD's, and up to 4V for HDCD's.  It's easy to see this difference on the Lavry LED's.

I generally try to adjust the gain on the Emotiva preamp so that the peak signal is -1dB on the Lavry indicator.  That way I'm sure there was no clipping.

However, as I was playing various HDCD and other files from my library on the DVD-9000 through SPDIF, it seemed I kept having to turn the level down on the Emotiva to stay away from the 0dB light.  Ultimately, it seemed, that if I set the Emotiva gain control to -5dB, I could do this effectively.  Playing regular CD's on the Oppo UDP 205 and other players, I can do this with the gain control set to -1dB or so.  So we can see there's at least 4dB difference in peak levels.  When I play LP's, I have to set the gain to +4dB to +7dB.

(I have set the Lavry input maximum input level to 3.88V, but the Emotiva automatically doubles the output voltage from the single ended Denon to the balanced outputs feeding the Emotiva.  So a single ended output reaching 2V would require a gain setting of about -1dB, as also with the Oppo balanced outputs reaching 4V.)

Anyway, I thought I could verify that the DVD-9000 was properly decoding the HDCD sent over SPDIF by playing an HDCD, and a CD, and comparing their levels shown on the Lavry.

I started with Mephisto and Co.  I couldn't actually remember the gain setting required for HDCD's with this setup, I remembered it seeming to differ from disc to disc, so I started with a gain setting of -3dB.  That didn't work, so I tried -4dB, and ultimately -5dB gain setting.  That eliminated hitting the 0dB indicator LED on the Lavry with this disc.

Wanting to be sure I had found the maximum, I also tried Fresh Aire II, and it was also hitting the -1dB LED more consistently.

I was sure a regular CD would not reach higher than -4dB or so, because of the level settings I've just described.  I was sure that I had already proven the Denon was playing HDCD's at an elevated output level.  So I didn't bother to compare the levels of regular CD's, so much as CD's I had always wondered might really be HDCD's, whose potential might previously have been untapped.

I somehow had this idea that Poncho Sanchez' Out of Nowhere might have been mastered in HDCD.  (Actually, I was confusing it with Arturo Sandoval's Hot House, I remembered later).  So I tested that, and it was hitting the -1dB Lavry LED even more frequently than the previous two discs.  So this had to also be an HDCD, despite not being marked as such.

This was a shock.  I knew that not all HDCD's were labeled as such, but exactly how many of my CD's are really HDCD's that haven't had their full potential revealed, I wondered.

I decided it was high time I use one of these programs, like Foobar2000, to test my files and see if they have HDCD features enabled.

That was a long trip into computer frustration.  The initial (and perhaps ultimately fatal) problem is that I use a Mac mostly, and Foobar2000 is solely a Windows program.  Well, I'll try it under the Wine emulator.

I was pleased that I managed to get Foobar2000 running.  And then I downloaded the HDCD plugin, and installed it, which required learning a lot about Foobar2000 and even a bit more about Wine.

Loading the HDCD plugin probably didn't work completely.  Either that, or it is broken on the current release.  It seemed to be installed, it shows up in black in the list of available plugins.  And I was able to add the HDCD reporting to the default display, as described on websites (the actual HDCD plugin has no documentation itself).

But every HDCD I played showed no HDCD features enabled.  I edited the default display so the contents of the HDCD variables are ALWAYS being shown, and they always show as ?.

It might have been that this was caused by an incomplete installation of the plugin.  When I pressed the INSTALL button, Foobar2000 was supposed to reboot itself with the new plugin.  It did not reboot, I had to restart it manually.  It's possible that Foobar2000 missed picking up the last bit of the HDCD plugin installation process.

Well, I figured, I'll just have to wait until tomorrow (it was already past my getting ready for bedtime time) and install it on my actual PC.  But this didn't keep me from another futile hour of futile fiddling with Foobar2000 on my Mac, all of which failed to reveal anything about any HDCD.

By the next morning, I decided to take a different path.  I happened to stumble upon the other recommended way of dealing with HDCD, ffmpeg.

Now, ffmpeg is definitely more of my kind of program! For one thing, it runs natively on Mac (and many other systems!).  It does all sorts of conversion tricks.  And there's no need to fiddle with plugins or recompiling (which I was afraid of at first) because HDCD is included in the standard Mac binary distribution!  (And if that isn't good enough, one can download the source code.)

And it runs from the command line also, another thing that I, a retired Unix programmer, really and truly like and appreciate.  That means I can easily roll shell wrappers and things (which I have already done--I created a script which simply examines files for their HDCD flags, deleting the pesky output files I don't want yet).

And, the results of ffmpeg were quite shocking.  I was shocked to see that neither Fresh Aire II, which is clearly marked HDCD and which I always believed was so much better on HDCD players as to not be worth playing on any other kind, uses no HDCD peak expansion or gain adjustment features!  It is fully embedded with the HDCD Transient Filter flags, which some claim to be ineffective, and that is all!

Out of Sight, a more recent addition to my library, showed no HDCD flags at all!  So how it the world was it reaching the nearly 4V output level when being played on my Denon DVD-9000 ???

Eventually, the theory I described at the top became clear to me.  The Denon must have a bug, I concluded.  And only then I set about systematically to prove it, as I shall show in the next post, with pictures!

HDCD

HDCD augments CD quality in just the required ways to make it slightly better.  It might not be as good as true high resolution PCM at 88/24, but close (I guess HDCD gets about 50% of the audible benefit of 88/24).  Most people think about the dynamic range expanding.  That's part of it, but another part is the variable digital filter*, which optimizes either low distortion (mostly) or transients (when they are important).

One way or another, CD's using HDCD encoding should best be decoded with HDCD.  This, from the beginning, was the biggest complaint.  It's not "truly" compatible.  But the same limitation may apply to nearly everything "compatible," and then there are lots of things like DVD-Audio which are not compatible.

Nowadays I play back HDCD's on my Denon DVD-9000, which is one of the best HDCD players ever.  It has the second generation HDCD decoder...but it's not an actual HDCD chip.  It's embedded in the firmware, which also includes AL24, Denon's upconverting oversampling system.   Along with that, the DVD-9000 has dual differential Burr Brown 1704 Dacs, the best true PCM DAC chip ever made.  After the 1704, everyone went to the cheaper Sigma Delta approach.  1704's continued to be used in Levinson and other high end DACs for awhile.  The power supplies and build quality of the 40 pound DVD-9000 are beyond reproach, however the analog IC's could be better, though they are a notch above the preceding DVD-5000.

But, to make this work with digital crossovers and DSP, I must convert the analog output of the DVD-9000 to digital (using my Lavry AD10) in order to get the decoding.  I convert the output to 24/96 high resolution digital so the HDCD decoding is fully captured.

Starting last year, I added a new way of "streaming" HDCD's from my library or music services.  I put a dedicated Sonos Zoneplayer (a ZP80) sending PCM (with the HDCD bits if they happen to be there) to the digital input of my DVD-9000.

But sometime last year I borrowed that Sonos unit to serve as a second Sonos analog input in the kitchen.  That never really worked for various reasons.   The main reason for doing it was so that I could have a Sonos input dedicated to my Kenwood L-1000T tuner.  But about the same time, I decided to switch to the slightly quieter and always pleasant sounding Sansui TU-D99X, the last TOTL tuner from Sansui featuring their custom made Walsh MPX decoder.  It only has one output.

In over a year of "borrowing" the second ZP80 for the kitchen, I don't recall ever actually using it beyond the first test.  Now I also have a 3-way switch feeding the main ZP80 in the kitchen, so I could send audio from the Mac, the Pioneer DVD recorder, or the tuner.  It should not be necessary to send audio from the Oppo BDP-95 because it rides with the HDMI signal, which is sent to other rooms through a 4x4 matrix switch.  I capture the audio from HDMI in the living room (as of last month) with a de-embedder, and I capture the audio from HDMI in the bedroom by digital optical from the TV to a DAC.  The digital optical means that TV (and all it connects to) are electrically isolated from the stereo to prevent hum.

Re-installing the living room de-embedder enabled me to remove the second ZP80 from the kitchen and put it back into HDCD-streaming service in the Living Room, which I was happy to finally be able to do last night.

But one curious thing is that when using the DVD-9000 as an HDCD decoder and DAC, it does not necessarily display the HDCD logo when a digital signal with HDCD is being decoded.  In fact the last time around I never saw the HDCD logo light up, but I did tests to be sure (by testing dynamic range) that the HDCD decoding was actually being done.  This is a quirk of the player that other people have commented about online. When playing HDCD discs, the light always comes on.

This time, I noticed something different.  I noticed that while playing some titles, the HDCD light DOES come on.  It comes on particularly for the Mannheim Steamroller Fresh Aire II HDCD.  But on several HDCD's from Reference Recordings, which you'd think would be the most HDCD of all, it does not come on.

When using the player as a DAC, it seems the light is triggered only by the presence of certain HDCD features in the digital signal.  It may indeed require the "dynamic range upward expansion" feature, in order for the light to come on.  If the digital signal has only the variable digital filter feature, it's possible it does not come one.

And that may be why Reference Recordings tend not to make the light turn on.  They might not use the upward expansion feature as much--precisely in order to be "more compatible."

For titles in which the HDCD light comes on, it comes on identically for using both the Sonos controlling software, and Roon.  Both Sonos and Roon are leaving the bits alone.

I believe I've seen as much as 3dB dynamic range expansion with HDCD's.  Last night I measured 2dB expansion from Fresh Aire II.  Unlike some later players, the Denon DVD-9000 actually puts out more signal voltage when the dynamic range expansion is occurring, up to 3.5V RMS or so.  A friend of mine complained about this in 1990, and panned HDCD players as a result.  Later players simply play HDCD's at a softer level to allow the peak expansion to occur.  You can imagine that helped "kill" the format as discs seemed to play softer on average, and louder always sells better.

Hydrogen Audio has a good, though slanted, overview of the HDCD format.  Their view is mostly derogatory, but their technical information is mostly excellent and otherwise hard to get.  It's hard to read this without seeing how they're trying to get in a dig whenever they can, such as by first mentioning Microsoft as the Owner instead of the original inventor and owner, and then, when they finally get around to mentioning the famous inventor, they misname them "Pacific Microsonic" (PM) instead of "Pacific Microsonics Inc. (PMI).  If someone doesn't even bother to pronounce or spell your name correctly, it's pretty much a giveaway that they are disparaging you.  And then, what exactly does it mean to call a format "dead."  A factual but non-derogatory description would simply be that new titles and hardware are not beling released.  (For the folks at Hydrogen Audio, even claiming that anything is audibly better than CD format is basically forbidden.)

The HDCD wikipage at Hydrogen Audio alleges to echo the late Charlie Hanson (a competitor having his own slow and leaky "linear phase filter") in claiming that HDCD could not use variable playback filters because of a Meitner patent.  I'm not sure if this story is 100% true.  If you take the time to read Hanson's original comments at Hydrogen Audio, he nevertheless recommends using a hardware decoder (as I am doing) as "the gold standard" because of the fully correct implementation of dynamic range expansion.  I spoke to Reference Recording people around 2010, and that is exactly what they recommended, and they also agreed with my approach of analog sampling the output of an HDCD decoder in high resolution digital for downstream DSP processing.  Using Foobar or HDCD.exe to do the dynamic range decoding is known not to be as accurate (or at least claimed, by Charlie Hanson, others dispute this).

I have not seen a fully convincing debunking of the claimed transient filter selection anywhere.  Along with Hanson, there are small number of audio objectivist cranks (who don't believe high resolution is needed either) say it was never used because of the Meitner patent.  But they don't actually show any proof that HDCD players don't use the transient filter selection feature or link to verifiable information.*  Meanwhile, there are numerous text and graphs show the effects of the dynamic range expansion, which is relatively easy to implement in conversion software (but apparently not always perfectly).

In this long thread about HDCD at Steve Hoffman's site, there is only one commenter, Jedi Joker, denying the utility of the transient filter function.

The "TF" flag is meaningless on playback. Why it is encoded into the HDCD signal remains somewhat of a mystery. The transient filters do not exist on the PMD100 and PMD200 decoder chips that were used in playback gear, nor do they exist in current HDCD-capable playback solutions utilizing alternate decoding methods, but only on the Model One and Model Two ADDA converters. The filters are engaged on the A/D side during recording, meaning whatever effect they have is digitized as audio into the resulting files. For whatever reason, a flag is simultaneously encoded into the HDCD bit, but has no effect on playback in any player

So, are we going to the words of the Jedi Joker as the final words?  I'm still trying to verify this.  The post on HDCD at Hydrogen Audio extends to 40 pages!  Here's another discussion of HDCD at Roon, and a poster who calls himself Andrew J Shepherd reiterates the claim that the Transient Filter is not implemented in any hardware device.  At least we can see Andrew J Shepherd as a real person with apparently solid engineering background, whose main interest seems to be telecommunications.  But where did he get his information on this?

HDCD might have been very very useful...if it had been used like this: a single CD has two versions, the undecoded version with little dynamic range (like all standard CD's these days) and the HDCD decoded version with much more dynamic range, for those who demand it.  That kind of feature is offered nowhere other than HDCD as far as I know.

The transient filter sounds like a very good idea also.  That might capture 50% of the benefit of 88.2kHz sampling rate above 44.1, if in fact it were implemented.  That Meitner--an early supporter of Sony's SACD project--would block full implementation of HDCD is dastardly.  My only experience with a Meitner product was at a hifi society store demo where the Meitner amplifer society members wanted to hear burned out driving Apogee speakers.

And there are some very good sounding HDCD's.  That's why it's worth having some kind of HDCD decoding.  One of these is the Reference Recordings Mephisto and Company.  Without decoding, it sounds very good.  With decoding, it's a one of a kind sonic knockout.  Most other HDCD's show similar differences with and without decoding.  So for me it's worth having HDCD decode capability, even if it was somewhat of a scam in the beginning.  (And strangely, Mephisto and Company does NOT light the HDCD light on my Denon while streaming.  And yet it appears to have the maximum amount of HDCD expansion, approaching 6dB peak expansion, while the Mannheim Steamroller discs do light the light, with only about 3dB expansion.  I suppose I should get one of the HDCD exploration tools mentioned in the threads above to see what's actually in the files.)

(I see now in one of the threads I've linked, the Mephisto HDCD is mentioned specifically as using lots of dynamic range expansion.)

I suppose one justification for calling HDCD "dead" is that Charlie Hanson himself did, in the above linked thread.  He commented that since HDCD is not on the feature list of the "latest Oppo player", we can say officially it is dead.  (People love to call things they disparage dead.  They shouldn't.  Dead doesn't mean "not growing.")

Weirdly and sadly Charlie Hanson, a highly touted High End audio designer who created good sounding but high priced and needlessly high distortion CD players and amplifiers, himself died not long after writing those words.  I did like one thing he did--a very good technical denunciation of SACD, wait a minute, maybe I could question some of his assertions on that one a little more now too.  I remember very much liking his linear filter when he first described it, but when I saw the measurements in Stereophile I was not impressed...another leaky "linear phase" filter with non-negligible distortion.  I was actually reading this HDCD column at Hydrogen Audio just and before his death.  I was about the same time considering getting the Pono which he designed--and thinking that to be a plus.  But it could not have worked out as well for me as my Oppo BDP 205, which I didn't realize was going to be a great streamer in combination with Roon.  Pono does not support Roon as far as I can tell, it doesn't function as a digital output streamer, and it doesn't play any kinds of discs.  The high res service didn't go very far then, but now with Qobuz it does, on any Roon device.)

HDCD lives on at my home in many titles, with excellent decoding.

But maybe it could be better still...either if preconverted to 24 bit (to avoid a needless DA/AD conversion which I believe essentially harmless because done at 24 bits) or, an even better HDCD decoder actually using the TF flag if, as some claim, existing units don't?

(*Test:  Resample the output of an HDCD DAC with the TF in the material, or with the TF expunged from the material, and show they have the same HF spectrum and transients.  Or written words by the original owners and engineers of PMI, such as Keith Johnson.)  It's the TF that should restore some of the 88kHz sampling rate quality, with sharper transients and so on.

I had known that Microsoft had purchased PMI for the IP, which many are claiming as licensing fees on the many HDCD discs, but I had also previously heard Microsoft wanted the HDMI patents regarding sticking extra code bits in an audio stream, for other purposes.

Anyway, I would think all these patents, the Meitner and HDCD patents, have all expired.  What's left are trade secrets we may never know for sure.  But what is patented is public knowledge.

While true high resolution negates the need for quasi high resolution systems like HDCD, a compatible CD quality format (or streaming!) format with standard (dynamically compressed) and audiophile (uncompresed) still might be an excellent idea if it could end the Loudness Wars, at least for people who care.

On Hanson, notice how he asserts here that:

1) Computer-based digital volume controls work with 32-bit floating point operations. These only have 24 bits of precision, as the rest is in the exponent. When you turn your volume down by (say) -48 dB, you are literally throwing away 8 bits of resolution. If you are starting with 16-bit data (99.9% of all music files), you are only left listening to 8 bits of resolution. Not good.

First, the meaning of floating point is that when you lower the magnitude, the exponent changes.  Within the 32 bit domain, you are not losing any resolution!  So this is error #1.  The fact that floating point is used does not make the difference he is describing, in fact, exactly the reverse

When the ultimate value is mapped back to 24 fixed bits, then you may lose precision, because only 24 bits are available.  At that point, if you have reduced the magnitude by 8 bits below maximum, you still have 16 bits of resolution left, and if that was what you have started with, in principle you have lost nothing.

The actual penalty in this illustration is only paid in the ultimate conversion, which lacks true 24 bit resolution.  But it may have close-to-24-bits "resolution" whatever than means, it's just not distortion or noise free.  And, besides, the lack of distortion and noise isn't very important at levels that low.

And then he goes on to claim DSD starts with only 1 bit and therefore 6dB of dynamic range.  That is true at the DSD sampling rate only.  At musical frequencies of interest, there is already far more dynamic range.  Though it is native dynamic range falls below the 16bit equivalent at some midrange frequency.  But once again that's far more complicated than his simplistic brush here.

Then he makes an outrageous claim that dither is inferior to rounding, a slap in the face of thousands of engineers and decades of digital audio engineering.  And then that nothing doesn't make a difference, including the purity of the copper.

If Charlie Hanson is the ultimate source of the claim that the HDCD Transient Filter feature is nowhere supported, even on PMI (which he calls PM) chips, and players designed prior to the Microsoft purchase of HDCD (like my Denon DVD-9000), I wouldn't find it convincing at all.  However, I would not be surprised if it were true of players designed AFTER the Microsoft purchase of HDCD, as they had no particular interest in HDCD itself, as has been obvious by their actions.

One thing is clear, my DVD-9000 does respond with about 6dB higher potential output on HDCD discs than CD discs.  Later players do not do this, as apparently Microsoft backed down on making that an HDCD requirement.

And, yet, the DVD-9000 does not actually have a PMI chip.  I believe it has the equivalent of the PMI 200, which was (as far as I have been able to determine) not an actual chip, but a design specifiiation, that included the 6dB higher potential level, and possibly an implementation of the Transient Filter changes, because this was a unique "reference" implementation.

Update: Despite all the claims by Charlie Hanson and a few other individuals who had no direct working knowledge (Charlie claims he talked to one of the designers at "PM", which he always misnames (it was PMI), and not naming anyone) that there was not switchable Transient Filter on playback, it appears that none other than Christopher Key, the author of HDCD.exe, and very knowledgable about all things HDCD, apparently did think there was, at least in 2007, in this blog (which ironically was linked by the HDCD wiki denying that Transient Filters, by bdo, who I now worship for implementing ffmpeg's HDCD, even if he wrote a disparaging and incorrect wiki about it).

Christopher Key went so far as to prepare two files, one with an impulse train, and one with a high frequency sweep 10kHz-22kHz, and asked anyone to try it with their hardware decoder.  This was after discussing several papers about the filters that (might) be used.  (This was not a nothing...there were at least papers written about it.)

Sadly, look for yourself, I don't see any evidence in the above discussion that anyone actually did the test the Christopher Key was asking for.  This is exactly what I want right now to test, that perfect, instead of trying to ferret out the effect in actual music.  Unfortunately the files do not appear anymore to be downloadable.  I have messaged Christopher Key.

Archimago on Ultrasonics

I think this is excellent reporting, as always from Archimago (even when I don't agree).

I do ultrasonics because I can.  As Boyk showed, it's there.  What does it do?  Nobody knows.  But I want my system to "do it all," not just to what has been proven necessary.  The ideal bandwidth would be DC to Light, but I'm willing to compromise on that.  I would agree that 40kHz is all one ever needs (along with 96kHz sampling rate).  That's not so unreasonable, I think.  There have been many many legendary systems with ultransonic response as good as that or better (including "massless speakers" like Iverson's Corona, and Diamond Tweeters).  Are they really better because of that?  Who knows, but I want to be in that league.

 (Meanwhile, I think it is unreasonable to badger people into using planned obsolescent Personal Computer based delivery systems like USB in order to escape the potential -145dB sidebands caused by the SPDIF, as Archimago does.  I think it's fine if he personally chooses to do things that way, but not so fine to stir up audiophilia nervosa regarding jitter to justify it, all while claiming that jitter doesn't matter at the same time--thereby having it both ways.)

In my fully sighted testing, I always prefer having the super tweeters on.  I know this does not prove anything.

One whole avenue unexplored by Archimago is this:  What if the ultrasonics influence the non-ultrasonics?

I tend to find that enabling the super tweeters removes hardness and glare.  I think what may be involved is that the ultrasonics have a beneficial masking effect, which is of course there in the live performance as well.  What's masked is the painful frequencies like 6kHz and 12kHz that sound unpleasantly "metallic."

In live performances, brass instruments never sound overly metallic, just nicely metallic.  In recordings lacking extended highs, brass instruments can sound positively harsh.*

In this process of so masking, somehow ultrasonics help also help delineate the bass..

So it's not so much about what you hear, and no I don't hear even loud tones much above 16kHz, but how ultrasonics affect what you do hear.

And this isn't so much dependent on preserving linear phase and all that.  Though I'd like to agree with my friends that "it's not the frequencies, it's the high slewing," actually I know I haven't come close to anything like phase linearity with my super tweeter deployment, and it still works, I think, whatever it's doing.

Just like in other frequency ranges, preserving the actual signal envelope is key, the phase relationships normally hardly matter, if at all.

I don't care to prove I need high frequencies.  Failure to prove I can hear it wouldn't mean I don't.  I'd prefer to keep it, so long as it's not too hard, and keep making it better.  Maybe eventually I'll be able to prove I need it.  I think 2x clearly audible bandwidth is reasonable (for me, that's actually 32 kHz).

Regarding the 28kHz spurious tone in one recording.  I don't think that's necessarily a good thing.  But if that was the mix the artist and producer signed off on, and if what you want to hear is the mix the artist and producer signed off on, then that is it, defects and all.  Without the "defects" it's not the same thing, and potentially could be less satisfactory due to the masking I have described.

(*It could be this is entirely caused by more mundane effects, including the need for a Linkwitz Dip.)

Archimago on SPDIF

Archimago seems to want it both ways.

1) Jitter, caused by decent quality modern devices, is not audible.  (This is a standard audio objectivist position.  A JAES paper written long ago showed that jitter must be 100-1000 times higher than produced by modern digital devices to be audible.)

2) Nevertheless, you should always use USB, if possible, because it has lower jitter.

I will not hide my feelings here.  I hate USB audio interfaces.  And I am angry that Archimago is hypocritically using his platform to promote FUD (Fear, Uncertainty, and Doubt) of SPDIF type interfaces, while at the same time claiming it's an unimportant issue.  This is weasel hypocrisy.  (Note: in other ways, I truly appreciate Archimago's audio journalism.  He is the best I know.)

I hate USB interfaces because they are a product of the tyrannical Computer world, and will be endlessly subject to planned obsolescence, mandatory "upgrades," computer failures, software failures, complicated hardware, firmware issues, IPP, UPP, Wepp and so on.  It might be possible to make those systems needlessly technically better than the alternative, but at the cost of complexity and brittleness.  When things like this work, fine, but the problem is there may be times when instead of working needlessly but slightly measureably better, they simply don't work at all.  That's what I mean by brittle--rather than bending they break.  That could mean dropouts, or it could mean hours of frustrated fiddling followed by another high dollar trip to the computer store to get another mandated upgrade, including possibly a whole new computer.

SPDIF is a simple and open and unchanging interface which is Good Enough.  Period.  (And this applies to AES as well, but not Toslink--always avoid Toslink if you can--though sometimes it is useful for breaking ground loops.)

SPDIF interfaces also inherently permit an endless chain of devices, with essentially zero loss.  That is mainly why I love them.  I used to dream of having power like this.  You can add any number of DSP processors in line, including devices like Behringer DEQ 2496 which are cheap and powerful, and miniDSP.  Devices like these tend to be open, like the SPDIF interface itself, and let you play with them, rather than being a magic box which does stuff you are just supposed to sit down and like.

Generally, with computer interfaces like USB, everything is supposed to be done by your computer, which generally also means proprietary software, such as Archimago's beloved Accourate, but also things like USB Drivers and USB Firmware, which often require updates on the day of purchase, and months if not years into the future.

There is little or no flexibility in the hardware configuration with USB.  You can't chain devices endlessly (or, plug in devices to record the output of other devices--in fact that is the whole reason why things are sometimes supported on USB and not in free and open interfaces--it's all about the IPP).

With USB, your hardware setup is going to be like this: computer connects to DAC over USB, end of story.

Archimago complains that the Oppo doesn't permit SPDIF or HDMI output when USB sources are used.  Yes, that's kind of the nature of the USB beast, generally speaking, and getting around it requires fancy footwork rarely seen (and I wonder if IPP would get in the way also).  Archimago hints at what Oppo should have done for USB, to permit SPDIF and HDMI downstream and it doesn't sound trivial.

Possibly Archimago's endless pushing of USB interfaces is they are the ones best suited, at this time, for multichannel audio.  There are already existing standards for high resolution multichannel audio over USB.  (Which will change every few years, of course, which is back to one of the big reasons I hate USB.)  And multichannel audio is one of his big things.  I think objectivist audiophiles tend to oversell multichannel audio.  I think it's barely worth the effort for music.  I'm working on it myself as a fun extra thing, but only that.  Not worth the bother, at this time, for most people, IMO.)

Just take a look at Archimago's peerless and wonderful review of the Oppo BDP 205, where he shows the performance of the UDP 205 as a DAC.  Here you can compare the performance of his beloved USB and the SPDIF/Coax interface.  Is there a difference?  Yes but smaller than with practically anything else.  We're talking a handful of tiny ticks at the reaching from the -155dB noise floor to about -145dB.

24/96 is also Good Enough.  What is the theoretical S/N ratio of 24 bits?  144dB.

Is there an asynchronous type interface which I like for audio?  Yes.  Ethernet and wifi, which are also basically open interfaces.  Ethernet never locks you down to a specific computer like USB does.  It isn't specifically part of the Personal Computer world.

I don't like using computers directly as audio sources either.  I like dedicated players, streamers that connect via Ethernet, and things like that, not things that are going to tell you that another update is going to be needed today.

I suppose, if you are going to be using a computer as your audio source, not mediated by a player connected through ethernet, etc, then USB is fine, and yes it may be the way to go.  So use USB if it's applicable like that.  I can't say you should never use USB.  I use USB audio solely for doing audio measurements because computer based measurement systems are almost unavoidable.  And every time, it's a pain, because the computer needs an upgrade or whatever.  Unlike traditional audio gear, which is near timeless and universal, computers are cranky and immediately obsolete after purchase.

But you should not be badgered into buying a new DAC simply to attach via USB because "it's better."  Or abandoning non-Personal Computer playback systems to reach that goal.  The alleged benefit in lower jitter isn't worth it.  In fact, it isn't worth anything at all.

While Archimago dabbles in the like of DSD and 24/386, which may require USB (or, gasp, HDMI), he also believes that nobody needs a sampling rate higher than 50kHz.  Well then SPDIF is just fine, again.

It should be understood, anyway, that I truly appreciate the very fine work Archimago continues to do.  In this incredible blind test, involving a large number of listeners, he shows, once again, that the differences between DACs, even much larger and more obvious differences than the ones caused by SPDIF jitter, are so small that they appear to be inaudible in fair tests.  But this, once again, shows the tiny differences caused by the "jittery" SPDIF interface are inconsequential.

Next Priorities

Before putting away the measurement rig, which I've decided I must do at least once a week to maintain sanity, I made one last measurement of the Left channel, with 0 degree calibration file loaded.  (Actually, part of the purpose was to be sure I was loading the calibration file correctly, which requires remembering to tick the box "Use with frequency response" and the "OK.")

It's only slightly different from the uncalibrated response, with some differences above 10kHz and especially below 20 Hz.  To see below 20 Hz or above 20kHz with ARTA, one has to remember to use the scroll or zoom controls on the right side.  It extends to respectable limits beyond.  Until I discovered that, I was cursing the limited 20-20kHz range.

The frequency response graph shows a few problems that should be tackled before moving on to phase linear crossovers, at least with regards to likely audible consequences.

Probably the most important if the rear wall reflection notch at 200 Hz.  I thought I had fixed that by moving speakers farther from the wall.  Indeed, the notch used to be at 250 Hz and bigger.  Also it gets covered up when the region around 100 Hz isn't equalized properly.  But when 100 Hz is equalized to roughly flat, then it is obvious there is a large notch at 200 Hz.  Currently this problem is limited to the left channel for some reason.

It would be nice if I could ameliorate this without going to greater complexity, such as using the midrange driver in the supertweeter box for fill-in just in this notch range.  (That would make the linear crossovers thing even more complicated.)

Another bit that bugs me, but probably not very important, is the depression just before the supertweeter takes over.  That relates strongly to the high frequency crossover.  I possibly need to equalize the Acoustat output around 14kHz where there may be some kind of resonance, so I can make the acoustic crossover "tighter" without causing a huge peak.  That's been the limitation so far, I can't fix the dip without creating a huge peak at 14kHz instead.  That needs to be fixed before trying to make the high frequency crossover linear phase.

Finally, the Linkwitz dip from 2-6kHz is not well done on the left channel.  Instead of a dip, it has a bulge around  4 kHz.  Most of the EQ tuning for that was done on the right channel, and I haven't used separate L and R settings.  I will need to have separate Left and Right equalizations going forwards to deal with things like this.

But meanwhile, I also need to apply time alignment to my two other subwoofered systems, the kitchen and the bedroom.  The time alignment for those systems has never been properly measured or adjusted.  (That was true of the living room system also, until last week.)

I also want to check out the digital jitter resulting from the use of the miniDSP's and other new features, such as my Kanex Pro de-embedder which is now used to extract PCM from HDMI while I am playing SACDs.

Time Alignment using ARTA (description and pictures)

Time Alignment showing microphone at listening position

ARTA running on laptop and Focusrite Scarlett

For many years up until now I had not been satisfied with my methods of measuring time alignment.   Measurement is the hard part.  Actually adjusting the time alignment is trivial, I simply dial in time delay values into the separate Behringer DEQ 2496 units that digitally EQ the signals for my SVS PB13 Ultra subwoofer (the bass below 100 Hz), the Acoustat 2+2's (the midrange from 100Hz to 17kHz), and the super tweeters (two small dome tweeters with response to 40kHz mounted on a wood box).  Now that I am also using miniDSP OpenDRC-DI units to perform the actual crossover functions for each of these 3 ways, I could dial the delays required into the MiniDSP's instead, but I find it easer to make adjustments with the Behringer DEQ units because they don't require a computer to be attached.

Stack of MiniDSPs on top of stack of Behringer DEQ's

Previously I had used the measurement program that is part of my Tact Room Correction System 2.0 Preamp (I don't use the Room Correction itself, but I have often used the RCS measurements).  That is strange and hard to use.  Last year I started using Room EQ Wizard (REW), which is OK for doing loudspeaker+room measurements, but didn't seem helpful specifically for the time alignment.  Just last week I downloaded and registered the ARTA program used by many loudspeaker designers.  For only $100 it is a far better program IMO than REW and after a week of fiddling around with it, I came up with a very intuitive, repeatable, and I believe accurate methodology for time alignment.  The method has two parts:

1) Align the leading edge of the acoustic output of each driver.  I measure each driver separately with the Signal Time Record feature of ARTA.  For midrange and super tweeter, I use a pulse with width "1".  I expand the vertical and horizontal scales of the time record display, and put the cursor at the exact position where the signal begins, as differentiated from room noise.  This is easy to see with the scales expanded, though less easy for the subwoofer output.  I adjust the delay times in my Behringer DEQ's so that the signal beginning times for each driver are the same, within about 0.02 ms for the panels and 0.06ms for the subwoofer (which is hard to see as clearly).  When measuring the subwoofer, I change the pulse width to 1000, otherwise it doesn't show up at all.

Leading edge of Acoustat signal

The ARTA Signal Time Record is an unprocessed display, similar in principle to a storage oscilloscope.  It can be used with a simple up and down pulse, which results in the measurement shown above.

The Impulse and Step displays are derived from longer term signals using FFT, and therefore show the room response as well as the speaker response, and not just from the very first instant.  For the first phase of time alignment, we need to examine ONLY the very first instant of the signal.

2) For the subwoofer and electrostatic panel alignment, I further fine tune the delay adjustments to either maximize the output at the 100 Hz crossover frequency, or minimize the output with the polarity of the panels reversed.  I think the minimization method is the better one, and IIRC was specifically recommended by Linkwitz himself.

When the polarity of the panels is reversed, the subwoofer and the panels are "cancelling" each other, because normally the LR4 crossover has them "in phase" at every frequency, especially the crossover.  I was pleased to achieve deep null using the cancellation method.

Deep Null at 100 Hz Crossofer Frequency (Green Line)

To measure the output at 100 Hz, I run an Impulse measurement with ARTA, and then run the Frequency Response and Distortion analysis.  This gives a remarkably stable frequency response graph, so stable that averaging his hardly needed, but I usually averaged 3 runs (a setting in the Impulse Response dialog) to be sure the measurements were not contaminated by ambient noise.  Two runs in a row typically show the same value at 100.3 Hz with 0.1dB precision.

*****

I had never been able to do the leading edge alignment before with any feeling of confidence.  No other tool has given me as clear a display of the signal vs time as ARTA.  You may be able to find pictures of my previous attempts farther back in this blog.  In every previous case, a lot of guesswork and interpretation was involved.  Not so with the Signal Time Record feature.

Although the ultimate alignment is done with the frequency output at the crossover frequency, the leading edge alignment is still important, because it is a good place to start.  When starting the alignment on a system like mine, one doesn't really know the latency of processors and DACs in line.  It all has to be included in the "starting point" which can then be further optimized at the crossover frequency.

Sometimes I run the signal time record feature several times if the transient looks affected by nearby noise.  I didn't at first realize how to zoom and scroll the ARTA window, so my earliest pictures are not as revealing.

Here is the final determination of the leading edge of the Acoustat panel output.  I put the marker where the signal begins abruptly downwards first.  (I do not know why it goes downwards first, by all other measures all my speakers are unambiguously "in polarity."  I've spent much time verifying that by various technical means, including and hand-made asymmetric signal and a smart phone app.)

Leading edge of Acoustat signal

Here is the final determination of the leading edge of the supertweeter output (with the miniDSP OpenDRC-DI running at 24/96 using minisharc 4x10 plugin).  I line up the marker with the leading large edge, ignoring smaller pre-ringing.

Leading edge of Supertweeter signal

Here is determination of the leading edge of the subwoofer signal, after I finally figured out to expand the horizontal and vertical scales first (previously, it was looking for the first pixel).  Prior to exampding the scale, it was virtually impossible to see where it started closer than 1ms or so, because the initial start is so low in level.  There is still some ambiguity because of noise, and sometimes I will try another run to see more clearly.

Leading edge of Subwoofer Response

The subwoofer start time does not actually optimize its phase at the crossover frequency.  That is best measured at the crossover frequency by putting one way (the midrange) out of polarity.  The supertweeters are deactivated.  At the crossover frequency, if the levels are set properly, there should be a null in the response.  I was pleasantly surprised that there was a deep null.  And then that null is made deeper by moving around in 0.02ms increments until the optimal delay value is determined.  First I tested positive and negative 0.02ms increments away from the leading edge alignment, to determine which direction to go in.  Then larger increments in delay time are tried, backing up when the notch starts getting less deep.  Ultimately, the optimal delay adjustment causes a very deep null, differing significantly from 0.02ms forward and backward delay adjustments around it.

In this picture the green line is the system response, and the lines below are % distortion (which is very high where the fundamental cancellation is occuring, an artifact of the measurement being done).  I set the marker as close as possible to the crossover frequency (it is set to 100.3 Hz).  Then I can read the amplitude very precisely (from one run to the next, it is often exactly the same to 0.1dB, except at the deep null where it goes lowest, it actually varies a lot, but always lower than the surrounding points).

You may note the notch in the response actually appears to be centered below 100Hz.  This is because of room acoustics and rear wall reflection.   I measure the level as close as possible to 100 Hz, ignoring any deeper notch below 100 Hz.

The high end response in the graph above rolls off because supertweeter is disconnected.


Here are the delay adjustments dialed in on June 19, before converting the supertweeter miniDSP to 96kHz operation.

Here are the ultimate delay adjustments dialed in, after the miniDSP for the supertweeter was switched to 96/24 operation on the afternoon of June 20.  Note that the super tweeters require 1.3 ms more delay than before because the 96/24 path has lost 1.3 ms latency for some reason, compared to when it was running at 48/24.  The change in latency might be from the miniDSP itself, or in the following Behringer EQ, or in the DAC, or all of these combined.

The final delay adjustments on June 20

I was unhappy to see the following left channel system response when all was done.  It was not very flat in the bass, with 100 Hz itself still in a notch (after correcting the Acoustat polarity), with a huge rise over 15dB from there throughout the deep bass below 100 Hz.  This was not my intention (I intend to have "flat bass", or what I call "electrostatic bass" even though realized with a dynamic subwoofer).

After adjusting the magnitudes of my pre-existing parametric EQ's, but NOT their frequencies (which were adjusted by hand tuning an oscillator, and are therefore "real" resonances and not digital artifacts) I was able to make the bass much flatter.  I still couldn't totally flatten the deepest bass without creating undesired holes there.  So I left it like this:

This graph is also showing the effect of the super tweeter in the rise abov 15kHz.  The super tweeter is optimized through other means, mainly using the microphones of my smartphone (which use algorithms to simulate a perfect spherical omnidirectional response).  The rise in on-axis response shown here above 15kHz is not representative of the average room response, or even at ear positions a few inches to the sides, because of the high degree of beaming from the supertweeter at those frequencies.  The on-axis level must be exaggerated at the exact center (where the microphone is) to create the equivalent auditory sensation as a live performance.  This is the opposite of the situation which calls for the Gundry (aka Linkwitz) Dip, which is another useful alteration of flat on-axis response,  because small rooms reflect unnatural amounts of 2-6kHz directly into the ear from side reflections.  Also, I have not as yet incorporated my microphone calibration into ARTA, and my microphone has a slight HF resonance contributing to the rise shown here.

Whenever I've lowered the supertweeter level to make it appear flatter with some kinds of measurements (and not so much my smartphone held at the listing position, which shows almost flat response at 20kHz) it sounds wrong to me, dull and more irritating actually.  The supertweeter takes away any sense of "strain," making things more rather than less listenable, stridency is reduced when the super tweeter is active.  But when the super tweeter is too low, the strain comes back, the magic is gone, and I'm just listening to 30 year old electrostats not a live band.

Here are the new PEQ adjustments for the left channel bass:

Notably I reduced the notch filter at 106Hz down to 0.5dB, to help neutralize the dip around 100 Hz.  I didn't zero out the notch at 106 Hz so it could be increased again if it later seems this decision was not altogether for the best.  I also increased the depth of the notches at 45.3 and 71 Hz and the broad cut around 28.3 Hz to help flatten the deepest bass.


The right channel looked better, especially in the bass, so I didn't make any EQ changes there:

MiniDSP now in 24/96

I am now running the MiniDSP product OpenDRC-DI (digital I/O only) for my super tweeters at 96kHz.  A helpful poster at the miniDSP forums gave me the correct information.  I needed to obtain the miniShark 4x10 plugin.  That was not the plugin linked at the OpenDRC-DI product page.

Once I got the correct plug-in, it should have been simple, but I messed things up and I spent 4 hours sorting it all out.  Now it is running a LR4 crossover at 17220 Hz just as with the previous plugin, but now at 24/96 instead of 24/48.  This is more suited to a super tweeter setup which (supposedly) responds up to 40kHz.

Once I finally got things sorted out, i re-measured the time alignment and readjusted the time alignment of the leading large edge of the supertweeter response (assuming the tiny edges to be digital pre-ringing).   This seems to require 1.30 ms more delay than with the 24/48 plugin, possibly partly because of reduced latency downstream.

There is little use in doing a "cancellation" type of delay adjustment with the super tweeter, and the crossover isn't really that well worked out.  The acoustic crossover of the Acoustats is a combination of their own falling response (down at 18kHz) and the LR4 lowpass I have added at 20kHz (to avoid wasting amplifier power into the capacitive load of the Acoustats).  This approximates, but not perfectly, a lowpass around 17 kHz, but it needs to be fine tuned better.

Anyway, the leading edge of the output of both Acoustats and super tweeters are fairly easily measured, and I believe it is quite close to exact time alignment required for these drivers (the subwoofer needed 1ms less delay than the leading edge alignment, but subs are different).

  It's possible the best type of delay optimization for the super tweeters would come from looking at the impulse response.  But that should be done in combination with analyzing better crossover curves.

New Time Alignment. This Time for Sure

In my own thinking, the time of first air movement was more important than alignment at the crossover frequency, and I "figured" if there were any correction, it would be additional woofer delay.  I tried testing the other direction only briefly with a big difference.

Anyway, Thursday night's alignment ended up in total chaos, things were not looking correct at all (though it sounded ok).

So I decided to start all over on the subwoofer/panel alignment.  Starting from the time of first air movement.  THEN, from that point, I "fairly" looked fore and aft, starting with the smallest increments in delay adjustments so as not to miss something close (as I had apparently done on Thursday night).  The smallest increment is 0.02ms.

On the right speaker, I optimized the total output at the crossover frequency, 100 Hz.  At first I worried how I would do this, knowing acoustical measurements to be hard to make sufficiently precise.  But I simply decided to use my new program in the ordinary frequency response calculation, and see what it calculates for 100 Hz.  It is precise to 0.1% (in linear voltage) and apparently quite repeatable.

Technically I run an impulse, using the sweep signal, and then convert it to frequency response (with distortion harmonic estimations also, an added bonus).  To help improve the repeatability and accuracy, I set it to run 3 sweeps and average.

It was quite clear from the beginning that I had been optimizing the bass in the wrong direction.  Instead of adding MORE delay to the bass, to move the alignment further out past the "phase lead" or whatever it is, it turned out I needed to subtract even more delay from the bass.

After seeing which direction to go, I scaled up the changes and then scaled them back down as nearing what appeared to be the optimum, peak response at 100 Hz.

Satisfied with right channel, I moved on to left channel.  But this time, after a few measurements, I decided to try something different.  I inverted the panels on the panels (sadly I have no easy way to do this for the bass) by swapping speaker connections, and then maximized the null.  I was surprised and very pleased that in fact there is a very deep null at the crossover frequency when I do this.  And I made it as deep as it can be, within the closest 0.02ms of delay.

The nulling approach is really the better one, I think, and if I remember correctly it's also what Linkwitz recommends.

The result of doing one channel by peaking and the other channel by nulling nevertheless makes sense.  The difference between the two seems about what I would expect based on the fact that one subwoofer is about 5.5 inches further back (a situation not easy to change, it's very cramped with equipment in the front of the room).  Correspondingly, the delay for the closer subwoofer is 0.3ms more.

(On the previous day's adjustment, the delay difference was opposite from expectations, leading me to believe something was wrong.)

But then, seeing generally how poor the curve below 200 Hz was in the left channel, I decided to optimize the parametric EQ settings (PEQs).  I didn't add or subtract any PEQ's, just change their magnitudes, in some cases by fairly large amounts.  There is little doubt that ARTA is showing a more accurate frequency response curve than my phone's RTA app, which is generally what I have used before  (though, combined with very slow hand sweeping, which I still believe is the best way to find the critical points where adjustments should be centered).

I wasn't sure this was going to work out well, because it was already late and I wasn't going to have much time, but it did work out pretty well.  There had been about a 20dB rise below 100 Hz, 100Hz being at a low point (even after time delay adjustment and putting the polarity back to normal).  I lowered the excess deep bass, and raised the area around and just above 100 Hz, so it's much closer to flat overall.

I've also been working on getting the miniDSP used for the supertweeter changed from 48kHz sampling rate (response to about 24kHz at best) to 96kHz sampling rate.  These units convert all inputs to their internal sampling rate--whatever it is.  Using the standard "plugin" supplied by miniDSP, you get 48khz.  Actually the website says it differently, it says a different plug-in in standard and will get you 96kHz, not even mentioning the old standard 48kHz one I got last year.

I ordered a new miniDSP a few weeks ago for experimentation, and I thought therefore I would get the new 96kHz plug in.  But the website info was grong, I still got the old 48kHz plugin.  I tried buying the one they indicated for 96kHz.  It only costs $10 so I didn't want to wait to try to send get them to me for free.

Well, sadly, that didn't work at all, I just got error messages.  I also tried another plug in that seemed close to what had been recommended in forums.  That didn't work either.

So finally yesterday I posted my problem as a question to the miniDSP user forum.  Within a few hours, a veteran user suggested what I needed.  So I've obtained that plug in, and hope to get the 96kHz working today.

That will require me to readjust the delay because likely the new plugin will have different latency.  But now I have good ways of setting the time alignment.

Problem getting miniDSP to do 96kHz sampling rate

I have just now posted this at the miniDSP help forum.  [Problem was resolved later by a helpful poster who told me I needed the minisharc 4x8 plugin and he provided a link.]

I have four OpenDRC-DI units, including one new one purchased this month. I would like to use one of them at 96khz, which seems to be an advertised feature now, but the (formerly?) standard OpenDRC-2x2 plug in does not support it. The product page for the OpenDRC-DI now says that when you buy an OpenDRC-DI, you should get the miniDSP-4x10 plugin. However, when I purchased a new OpenDRC-DI this month, the only thing that actually appeared in my personal download folders (and I looked at each one, couldn't this be simpler???) was simply the same old OpenDRC-2x2, which I already had from previous purchases. So I clicked on the link to the miniDSP-4x10 plugin on the OpenDRC-DI product page and purchased the miniDSP-4x10 plugin for $10. I then go to the my personal download folders and it appears that I can download the 4x10 plug in. BUT when I click on the image which says "miniDSP 4x10", I don't get anything that says 4x10, what is actually downloaded to my computer (through Chrome browser) the miniDSP 2x8 plugin. I can't figure out how I'm supposed to obtain the miniDSP4x10 plugin ! When I run the miniDSP 2x8 that I downloaded as above (trying to get the 4x10), it fails to communicate with any of my OpenDRC-DI units, including the brand new one. I get this error message: Incorrect Device connected to this plugin software! Please use the correct plugin software HWID supported by the software: 1 HWID of the device:4 I also tried purchasing and downloading the nanoshark 2x8 plug in. That simply fails to find the device and doesn't give error message. I have no trouble using the OpenDRC 2x2 plugin. But it gives me 48kHz sampling rate, and I want 96kHz sampling rate for my super tweeters. I am using MacOS computers though I do also have a PC if needed.