Friday, June 9, 2017

Polarity !

For a number of years now, I've disappointed a friend of mine by disbelieving in the crucial importance of Absolute Polarity (both stereo speakers having the same correct, vs incorrect, polarity).  Nobody disputes the importance of Relative Polarity (speakers having the same or reversed polarity relative to each other) because it causes easily measurable bass reduction, but Absolute Polarity has no frequency spectrum consequences, and the best research published in the Journal of the Audio Engineering Society suggests that Absolute Polarity (which I'll mostly just call "polarity" in this post) can pretty much only be heard on test tones and solo brass instruments on headphones, and once you get beyond that testing music with larger ensembles or loudspeakers in a room the more complex texture of the music and/or room reflections confound the accurate determination of polarity by listeners.

I've still had enough respect for polarity (which is easy to measure in most equipment) to always try to get it right, but not so much that I've bothered to assess polarity differences among recordings subjectively, as my friend does, as he claims that 92% of all recordings have incorrect polarity, and sound best when a system is set to invert absolute polarity.  He has made lists of which recordings have correct polarity and which don't.  He has made similar lists for equipment, also claiming that equipment which inverts polarity is very common (my tests do not show common polarity errors in equipment, including many of those my friends claims are wrong, and equipment is easy to measure in this regards).  The lists are entirely based on listening tests, not objective measurements.  I've been thinking it ought to be possible to assess the polarity of at least some recordings by visually inspecting the waveforms of brass instruments, for example, but I've never thought it important enough to try.  I have also believe that nearly all audio devices preserve absolute polarity,  which has proven true in the past for the small number of devices I had chosen to measure (with the exception of an early iPod).

I may be taking polarity more seriously now, after it appeared that incorrect polarity was making my system sound worse than correct polarity.  And now my measurements have revealed that both of my Audio GD DACs, my DAC 19 as well as my new Master 7 Singularity, invert polarity.*  I was especially shocked to find that the Master 7 Singularity inverts polarity at both the balanced and unbalanced outputs.  This appears to be no mistake.  It suggests that Kingwa shares the belief that my friend has that most recordings sound better in inverted polarity, so Kingwa has simply made inverted polarity his standard polarity.  It would not seem possible that this resulted from oversight, when so many other details have been attended to perfectly.

(*The DAC 19 has a "setting" button which corrects polarity when pressed in.  I was considering the "out" position to be the default, and always kept it that way, but apparently to be "in" correct polarity you need to press the button in.  The Master 7 doesn't have a polarity button, which my friend has always considered a fatal flaw.)

Well that's OK, I suppose, for those who just hook it up and listen, but what's not OK for someone like me who is keeping score is not to know about it.  But now I do, after a week of audio experimentation and discoveries (ultimately leading to the best sound I have ever had).

Back on the first day I hooked up the new Audio GD Master 7 Singularity, I wondered whether the balanced XLR outputs were hooked up the "normal" way (with pin 2 hot) or the "old" way (with pin 3 hot).  So almost after hooking up the dac I tried using the polarity controls on my Tact 2.0 (now hidden in the Polarity menu).  After switching to inverted, I decided to leave it that way for awhile.  Then by the next day, I had forgotten about this setting, and left it that way, during all my rave reporting on the Master 7, until the next Sunday when I had to set up the A/V for my party.  Then I noticed the inverted polarity setting, and changed it back to normal.

But after the party, something was clearly Not Right.  The wholesome sound of the previous week was gone.  Now the sound was just edges around where the wholesomeness had been.  As if the whole had been just sucked out.

Here I should explain that most of the time, and more than anything else, I listen to the local classical FM radio station.  And I do so...using my Audio GD DAC...because all analog sources in my system must be converted to digital for DSP processing, which does room correction and crossover for all the speaker drivers.  The analog output of my Pioneer F-26 FM tuner (one of the best, of course) is converted to digital either by Sonos (using the INPUT jacks of the CONNECT or ZONEPLAYER) or by my Lavry AD10.  For many years I had only listened to FM through Sonos, taking the digital output either on the same or other zoneplayers.  Somehow, when I tried the higher resolution connection through the Lavry things never worked out.  But as described in a recent post, about a month ago I tried again, first by setting up a 10 analog input selector so I didn't have to remove any other inputs to get the FM plugged in.  And since it was all done in a fairly open rack, I could make sure everything was being connected correctly, and the variable level set correctly, ultimately resulting in the best FM radio sound of all, first digitized by the Lavry AD10 to 24/96, then passed through my Tact preamp which only selects and sets level (I have room correction bypassed) and then through a trio of Behringer 2496 DEQ's, one for each speaker, and with all digital I/O.  Ultimately, the digital signal originally created by the Lavry and processed by the other devices is converted back to analog by the Audio GD Master 7 Singularity DAC, which drives the Krell FPB 300 through 18" of Nordost Baldur XLR cables.

Since I listen to FM radio as background music more than anything else, it was no surprise that it was this source that I first identified as being "wrong" somehow.  It had sounded so good the week before, so what now?

It wasn't long before I remembered changing the polarity during my party.  And setting polarity back to inverted brought the wholesomeness back.  This was the shocking discovery of the importance of polarity.  But was I setting the polarity correctly by doing this, or setting it inverted?

I wouldn't have time to do an objective test of that until the next weekend.  And it would turn out to deliver many interesting results leading to further improvement.

Although I already have several useful test tones, I decided to download the Speaker Pop app for my iPhone.  The test tones for the app can be generated by the phone itself, but for my complicated system I decided to download the files.  I used the wav file called SpeakerPop-long44.wav.  I first incorrectly burned a disc (as a data disc, which is now the default if you burn from Finder).  I will eventually burn a CD and also a DVD for future testing, but I decided to skip that for now and simply copy the file to my Ztunes folder (stuff which should not be played automatically), updated my Sonos library, and played the file through through the SPDIF output of a Sonos Connect which is itself wired through Ethernet.

The Sonos Connect puts out a SPDIF stream which becomes Input 1 on my Tact preamp which selects digital inputs and sets gain and/or makes other adjustments (I have room correction and EQ bypassed).  The Tact puts out 24 bit digital in AES/EBU format, which goes to a 4 way powered AES splitter from Henry Engineering, which then goes to 3 Behringer DEQ 2496 units, which do the crossover and room correction functions for each driver using only digital I/O using AES input and AES output.  The AES output for the mid range (100-19,000 Hz) used to be converted to SPDIF by an powered AES-to-SPDIF adapter from HOSA, and that used to feed the Audio GD DAC 19.  With the Master 7 I added the AES input option so it can take the AES straight from the DEQ, but I did not set that up right away, for one thing I didn't have any 5 foot AES cables, so I ordered two different kinds which arrived during the week.  Meanwhile I continued to use the HOSA adapter and the SPDIF input of the Master 7, in the same way as I had been using the SPDIF input of the DAC 19.

I was finding it hard to believe that the Master 7 was set up for inverted polarity.  Instead, I was believing that for some reason the Acoustats simply sounded out-of-polarity possibly because of phase shift in the Acoustat patented dual transformer system.  (My thinking now is that this should be minimal.)  For years I had wondered about what appears to be the impulse response of the system as measured by the Tact RCS software.  There is a tiny negative blip before the main positive part of the impulse appears.  Is that tiny negative blip showing that the whole system is out of polarity???  I'd been wondering about that for years and hoping to use some other tools to check the impulse response.  The negative blip could also be an artifact of the Tact measurement, or digital processing in general.

Finally on Saturday night I had time to run the test.  I simply played the SpeakerPop signal on my Acoustats (after shutting down the subs and super tweeters) and tested it with the app running on my phone.  It showed that, after having inverted the polarity in the Tact polariy menu as I had done a week before, the Acoustats were now putting out the correct polarity.

So I had changed the polarity to sound better, and "better" turned out to be the correct polarity.

But I still did not believe the new Master 7 DAC was responsible for the inverted polarity.  I mostly thought it was the speakers themselves that were inverting the "effective" polarity by phase shift.

Or, it could be the input chain: I take the digital output from a Sonos Connect, run it to my Tact RCS 2.0 Preamp through coax, then the Tact is connected via an AES digital cable to a Behringer 2496 digital EQ, which is then connected via an AES digital cable to the DAC.  I don't think any of these invert signal polarity, but I could be wrong.  In particular, the Tact itself HAS a polarity control, so you would think that would reflect the actual polarity, and in my experience it does.

I tried making a disk and playing it through my 1986 Sony CDP 507 ESD cd player, and using the digital output from that player, my reference for polarity stuff, but the disc I made wouldn't play (it turned out to be I had burned the speaker pop file as a CDROM data disc, not an audio CD).

So then I tried sending the digital signal that I had been playing on the Audio GD to my Denon DVD 9000, which can accept external SPDIF digital signals from coax, through an AES to coax converter.  The Denon originally cost $4500 and was a "statement" product by Denon, a brand known for top quality engineering.  You would not think Denon would get the polarity wrong in their statement DVD-Audio player.  I tested the Denon and the Audio GD with my Tektronix 100 Mhz scope.

And the Denon did indeed output the Speaker Pop signal in the correct polarity, unlike the Audio GD.
The Denon is cap coupled, so it shows the speaker pop signal as little positive bursts only a few hundred milliseconds wide.  The Audio GD is DC coupled and shows a step response--going negative then slowly rising up for seconds until the next pop.  I could measure the Audio GD output only at the single ended outputs, so to test the balanced outputs, I hooked the scope to the positive output terminal of the Krell.  Since the Krell is also DC coupled, it showed the same waveform as I saw at the RCA jacks of the Audio GD, so it appears both the RCA and the XLR outputs of the Audio GD invert polarity.

So I have now proven to my satisfaction that the Audio GD Master 7 inverts polarity (or, all my other stuff is wrong), and it is seeming like it is very important to set the polarity correct somehow, as I can do with the polarity switch on my preamp.

I suppose it could be coincidence that I thought the sound was bad, and the polarity just happened to be incorrect at the same time.  Then, when I fixed the polarity, I was inclined to think it should now be better, so there was a strong expectation bias that this would sound better.  But for now I've decided that in my system as it exists today, polarity is important to get correct.  I do now consider this crucial.


Thinking about phono stages, I found myself reading John Curl's description of how Harry Pearson wrongly tested his Vendetta SCP-2a phono stage (which had been retooled to give Pearson the 47k load he demanded).  According to Curl, Pearson's assistant didn't get the instructions and didn't invert the speaker connections to correct for the inversion of polarity in the phono stage as required.  So when Pearson described the SCP-2a as sounding "cartoonish" he was describing the sound of polarity inversion.  And that description indeed sounds like a colorful description of what I heard.

But there's another thing I haven't mentioned so far.  When I was testing the polarity of each speaker in my system using the Speaker Pop app,

I discovered that while the right subwoofer had the correct polarity, the left subwoofer did not.  Now this is no longer the case of absolute polarity it is the more widely accepted problem of relative polarity.  When stereo speakers have different polarities (relative polarity), the bass tends to cancel out, and the imaging is destroyed, especially the center image.

This brought back a few memories and mistakes over the past couple years.  When I replaced the plate amplifier in the SVS PB13 with the new Sledge version (because the original BASH version died) I wrongly connected the speaker lines.  This wasn't a mere oversight, it was complicated by the fact I also needed some new wires SVS had forgotten to send me at first because the connectors didn't match.  I still don't remember enough to remember how exactly it happened, but it did, as I found out after I had the plate amplifier fully screwed in and the subwoofer and everything else put back into place.

So at first I tried correcting the problem using the "phase" control on the subwoofer.  With the new Sledge amp this control is in the digital menus, you scroll through numbers from 0-180.  The problem is, that it doesn't appear to work for me.  I think this is because I don't use the crossover in the subwoofer.  I use my own crossovers upstream.

So back then, I inverted the polarity in the DCX crossover I was using for the subwoofers then.  (I had already migrated to using a DEQ for the Acoustats).

Then, when I redesigned my system to use 3 DEQ's instead of a DCX and one DEQ, I assumed that the earlier connection was because I had been experimenting (as I had at one time) with an out of polarity operation in one subwoofer to help remove bass boom caused by room modes.  That experiment was not deemed successful for very long, but I had forgotten to undo the change made for it, or so I thought.

But you see, that wasn't the real story, the reason why I had inverted the polarity of one subwoofer in the DCX was because the plate amplifer is internally connected out of polarity, which can't be changed without taking the subwoofer out and apart again.

And so it goes, mistakes are made, and sometimes stick around because of additional mistakes.

Anyway, needless to say, when I was working on the absolute polarity issue, I also fixed the relative polarity.  That quickly improved the bass image focus.  Now the sound may be slightly too bassy because I tuned the EQ's for the previous relative polarity difference.  But definitely the bass is more powerful and punchy and has more potential to be great.

It would be good to fix the problem at the source, but the issue of moving the subwoofer out of the corner is now greatly complicated by all the new equipment set up since then.  I figure I'd have to devote a day to the project of fixing the polarity inside the subwoofer.

So I'm back to finding a way to fix it, which could lead to a future mistake if I forget why I did these things.  And it's especially difficult because the Behringer DEQ 2496 does not have any polarity controls (a serious oversight, IMO, along with not being able to adjust gain in 0.1dB increments).

And the Emotiva DAC doesn't seem to have any way of correcting the polarity either.

SO, the trick I do is I do a relative polarity inversion on the whole system using the Tact, and invert the relative polarity again for the Acoustats and the super tweeters, because those speakers are externally wired to their powering amplifiers and can be easily changed.

Now you see it will be easy to make the mistake of forgetting I did all this for a good reason.  Having the cables connected in different polarities at the back of the Krell looks wrong, but it actually correct because it matches the relative polarity change in the Tact preamp.

So here's how it's all done now:

Device, Channel, Polarity (Plus means In Polarity)

Tact, Right,Minus

Krell Speaker Cable, Right, Plus
Krell Speaker Cable, Left, Minus

Supertweeter Speaker Cable, Right, Plus
Supertweeter Speaker Cable, Left, Minus

Nothing else can easily be changed.

But since I corrected the relative subwoofer polarity on the same day as I corrected for the absolute polarity inversion of the Master 7 Dac, you could argue that my great sense of how the system has improved since then might be coming more from the fix to relative polarity as from the absolute polarity change.  This would not explain how I liked the system during the previous week with the changed Absolute polarity, because changing the system absolute polarity would not change the relative polarity.  But perhaps I just imagined that, it was a one-off thing, and everything else could be explained by the undeniably large audible change in relative polarity in the subwoofers.

So am I going to do better controlled tests to find out for sure?  The problem is, you never know anything for sure, especially these kinds of things, and I know that hearing difficult-to-hear things on demand like that is bound to be difficult, even though I now imagine the difference is huge.  (I've tried to hear absolute polarity in blind tests going way way back previously without success.  I didn't even know what to listen for except from a lot of self-contradictory descriptions I had heard.  Now at least have one "case" where it seemed extremely important, and while no one should take that as proof of anything, at least I have an auditory idea as to what it might sound like, enough to be somewhat interested in.

As I did before, I'm going to assume that absolute polarity is important enough to be concerned about, but not be obsessed about.  If I think I hear it again, I'll check it out (bad form experimentation, for sure, but more fun).  I'm not going to routinely test recordings before I routinely listen.  I still don't really believe it's so huge as to alway be readily apparent, but more as a thing you might bump into once and awhile.

No comments:

Post a Comment