Wednesday, June 14, 2017

Love, Pain, and the Finicky Audio Gods (final update, Sunday June 18)

I fell in love from the first moment, opening the clean and nicely packed box.  The sound exceeded my best expectations: clean, organic, potent, now even FM (which I digitize for DSP) sounded high definition.  Digital files from my Sonos system now sounded as good as the best high definition discs played on my Denon DVD 9000.  Wow was this a good choice, I felt about the Audio GD Master 7 Singularity for the first 5 days.

As they say, it's not love unless it sometimes hurts.  The hurt began on the 6th day, when I was setting up sound for the living room TV for a party.   Sadly the Tact requires me to cycle through the 5 main digital inputs rather than just press the one I want (note: it is always best to provide a button for each input, rather than having to cycle through).  As I hit #2, which connected to my HDMI Audio stripper (which strips the audio bitstream either directly or converted to SPDIF, they both work the same for me because the Oppo BDP 95 which provides the video is set ONLY to output SPDIF) the most horrid snapping sound emerged.  I quickly pressed the Digital Input button again to select the next input.  Of course in addition to being scary it was embarrassing.

It was a chamber of horrors, each inactive or disconnected unit connected to the Tact delivered a very loud, potentially system destroying noise.  And when I started I had only 2 out of 5 inputs active, and the Oppo source (fed through about 65 feet of RG6 cable from the kitchen) was not fully started yet.  And sometimes the lights on the Tact aren't very clear, so it's not clear which input has actually been selected (sometimes it looks like several inputs have been selected, because the Oppos is low to the floor, and may need some internal repairs).  After a couple cycles through I made it back to the safe Sonos digital input which I had been playing for 5 days, but decided in future it might be best to switch to an unused input on the Master 7 itself.  I tried that and that worked perfectly, selecting any unused input on the Master 7 itself gave me perfect muting immediately.  But to get there, I had to crawl on my knees around the Acoustat where there is very little room.  But I soon mastered that task.

Now during this party, I played the movie sound from the Oppo, it might have been 48k or 44.1k.  There was no problem at all.

Also still at this time, I had the Master 7 hooked up via the coax input number 3, using the same AES to Coax converter as I had always needed to use before with the Audio GD Dac 19 Anniversary, which I had purchased a couple years before.  (Actually, I was first using the Dac 19 with Toslink, because the Behringer DEQ has Toslink and AES outputs only, and no coax, and I thought the optical isolation from all my noisy digital stuff might be nice.  But later I found the noisy digital stuff was not a problem at all for the digital coax, and the Toslink had such poor definition I could not read the jitter on my digital audio tester, whereas the coax had effectively none, I could connect earlier in the chain and get the exact same jitter number.)

I had not once ever had a muting issue with the Dac 19 anniversary.  If I selected an unused input in the Tact when I was using the Dac 19, it simply muted without a sound.  I had heard people warn me that with an Audio GD Dac I might have muting problems, but I had none at all with the Dac 19.
Other Dacs I have running might clink as the kHz light goes out, but no dacs I've had in the past also have done anything but always mute perfectly, no matter how complex the input chain.  Because of that muting, however, in the living room system, if I erroneously selected a bitstream output, I would simply get nothing (and not the terrible noise you get on some systems).  So I was completely shocked and unprepared for the strange noises I got at the party, but there wasn't any more technical delay than usual.

The movie party (3 videos) went OK, but not without blowing one speaker fuse, and hurridly finding one in my "laboratory."  I was grateful I could find one in just a few minutes, I hadn't needed a speaker fuse in a long time, I had fancy $49 (discounted to $29) audiophile speaker fuses too.  I was afraid I was going to have to use another of those, but then I found my package of 5A fuses from Radio Shack.  I looked online to confirm the Acoustat needed Slow Blo.  Problem #2 solved.  BTW, the speaker fuse had blown sometime as the Oppo went into stop as the movie was ending.

After the party I found that by removing rarely used inputs and simply capping them with shorting plugs I could stop the noises from unused inputs.  Problem 1 solved!

But then I also finally got around to actually installing the AES cable directly from the midrange Behringer DEQ to the Master 7, bypassing the need for an AES to coax converter.

Now this was different, I thought I'd try an unshorted input again, and now, no noises!  Somehow changing to the AES input of the Master 7 had eliminated all the lack-of-muting noises I was getting with the SPDIF cable.

So, shorting plugs no longer needed!  I could plug the unpowered units back in again too.  Problem 1 solved even better!

However, the Oppo still appeared to be a problem.  After I had played all the party videos on the Oppo, at the very end, relays clicked as the drive stopped, and the Krell amplifier shut down, for the first time since it had come back from Capacitor Service at Krell in February.  I dared not try the Oppo again.  I could probably live without a direct SPDIF connection between the Kitchen Oppo and the Living Room system, I thought then.

Something didn't seem right, however, with the sound after the party.  Now I didn't seem to be getting the new warm and organic sound of the DAC.  It sounded more edgy.

Then I remembered that just before the party, I changed the polarity menu option in the Tact to Normal.  I believed that had been part of an incomplete test during the first few days.  So maybe THAT was part of the magic.

I changed the polarity option to Inverted, and now I was getting the good sound again, it seemed.  (I wrote about this in an earlier post).  To confirm my suspicions, I downloaded a polarity test called Speaker Pop, and found that, indeed, the polarity of my system was correct only if I selected the Inverted option on the Tact.

I did not believe or want to believe that my new jewel, the Master 7, was inverting the polarity.  I figured this was just a long standing problem with the Acoustat speakers themselves.  I had previously gotten ambiguous information from the Tact's own impulse test program as to the polarity of the Acoustats themselves.

So a day later I removed the Denon DVD 9000 from my rack, and turned it into a DAC by turning the front panel INPUT knob to Coax.  I ran the exact same signal (from Sonos to Tact to Behringer DEQ) but also through a commercial AES to coax converter (no, this is not going to invert the signal polarity, which is a property of the data not the carrier) into the Denon, and its output showed the expected correct polarity, that is whatever I had selected by the Tact polarity menu.  The unbalanced outputs of the Master 7 showed the reverse.  But what about the unbalanced outputs???

I found I had no XLR adapter that would let me connect XLR to my scope.  (I ordered one from Markertek but didn't have it until later.)  But right then I figured out a cheat.  I could simply touch the positive probe to the positive output on the Krell.  The Krell is certainly not inverting, and simply reflects the polarity at its input.  I found that the polarity on the Master 7 XLR outputs matched that on its single ended outputs in being the reverse of the Denon and the reverse of whatever I had selected in the Tact.

I also discovered something else: I had sometime a few months ago put the subwoofers in different RELATIVE polarity.  And the subwoofers own PHASE control didn't seem to help.  I figured out a way to fix that.

By now I was really really liking the sound.  I was playing the Crystal Cables sampler from SACD which I had analog resampled to 16 bit digital.  But not playing very loudly, one track caused the Krell to shut down again.  I was simply playing this from the Sonos Connect digital output routed through 9 feet of RG6 cable to my Tact.

Hoping that I could keep my new DAC and work around this problem, I made made new Teflon twisted pair speaker cable to replace the super low inductance but relatively high capacitance Canare 4S11 speaker cables.  The Krell still shut down.  So I then also replaced the Baldur XLR cables that connected the Master 7 to the Krell with Belden 1800F cables.  THAT fixed the shutdown and I was even able to boost the playback level 12dB and play the track over and over.  I thought all the problems were now solved.

I had been thinking, however, that the Oppo sounded better when my system had inverted polarity.  I didn't know whether it was safe to test the Oppo again.  But since I had switched to AES input, muting hadn't seemed like a problem, so I made a recording of the SpeakerPop sound, and put it in the Oppo.  After the track had already started playing, I changed the source selection on the Tact so the Oppo was now selected.  I started at a very low level and really had to crank it up.

No, sadly, my second big polarity guess turned out to be wrong.  The Oppo had correct polarity.  I tried the disc in all my other players, and they all had correct polarity also.  I had been thinking the Integra Research RDV-1 had incorrect polarity just like the Oppo.  Nope.  That makes one good guess out of three so it really looks more like I can't identify polarity than otherwise.  The Sony DVP-9000ES would not play the CDR I had burned on my Lacie, and I'm not sure if it generally has the ability to play burned CD's.

In testing the Oppo, I also wanted to see if the polarity selection in the audio menu made any difference.  It was somewhere doing this, messing with the audio menu in the Oppo, that the Krell shut down again, for the first time in several days.  I had been hoping I had fixed the problem by running an AES cable to the Oppo, but apparently not.

During one of the polarity tests on the living room disc players, the Krell also shut down.  And later, when I was playing the Pioneer F-26 as digitized by the Lavry (for better fidelity than when I usually digitize it with Sonos) the Krell shut down.

These later shut downs were beginning to surprise me.  They didn't seem to be related to any faulty muting on the DAC, as previous ones had been.  But now I thought of something else.

I now figured that the reason for the shut downs was specifically that the Krell was trying to upbias and there is insufficient AC power.  This is one of two possibilities when the Krell automatically shuts down but shows only the "power" light illuminated.  The other possibility is that the speaker or cable has shorted.  I had pretty much ruled out the second possibility long before, the speakers were still working after all (thankfully!).

Now it didn't occur to me at first there was anything I could do about this.  I have a single 20A dedicated line for my entire system, but not for just the Krell (as Krell recommends).  But then I figured out how to do it.  I could power the subwoofers on a different circuit than the rest of my audio system.  Inspired, I set about doing it, and in just a few minutes I had moved cords around so that both subs were plugged into the regular living room circuit and not the 20A audio circuit.  This would leave only the low power components on the same circuit as the Krell.  I didn't want to move THEM to the other circuit because that circuit is incredibly noisy as it also powers the front yard photocell operated lights, and they put tons of noise on the line.  But the subs have digital power supplies which aren't affected by that.

This made a huge difference right away.  Now I could play jazz and indie rock on FM through the L-1000T through the Lavry instead of Sonos, and run it that way all night long.  And I did that just to be sure I could do it.  Problems solved!  Or so I thought.

It still bothered me a lot that I couldn't play the Oppo.  I was thinking, maybe if I activated the PLL jumper inside the Master 7, it would keep the Master 7 from responding in a very bad way to sampling frequency changes and muting from the Oppo.

Just to be sure this was a good thing to do, I emailed Audio GD, and I also emailed a friend.  Neither response really answered my question, however Kingwa said that the default is NO PLL (which I was somewhat uncertain about) and the digital receiver is ALWAYS ASYNCHRONOUS.  Those two bits of information were very useful.  But Kingwa also said I should not be running SPDIF longer than 10 meters.   I replied that I had been and still do run lines 50 feet (actually, 70 feet) with other DACs with no problems at all, and other people I know run SPDIF lines as long as 120 feet.

My friend had only used I2S and he always runs the DAC through his preamp first.

I'll express my strongly negative feelings about I2S in a future post.  Meanwhile, I simply went ahead and changed the jumpers for PLL and tried the Oppo again.

This was different or so it seemed.  With the polarity correct, and PLL enabled, the sound was superb, better than I remembered it going back to the days when I first had the Oppo and Denon 5900 plugged straight into the Lavry into the living room.  The glare and haze had disappeared, leaving the sound a bit darker, but far more real sounding, and with punchy bass.  I was playing the Santana Supernatural DVD-Audio, sending 24/96 across my 70 feet of cables into the living room, and it was great.

Problem solved!  I cranked it up and started grooving.  A half hour went by.  Relays inside the Oppo were clicking on every track, making other DACs mute, but there was not a sound from or through the Master 7.  It was now muting and changing sample rates and everything over a 70 foot cable at 24/96 without a hint of problem.

But then, 40 minutes later, the Krell shut down again.

OK, so maybe I can't use the 70 foot spdif cable anymore.  I started thinking of other options for my monthly party, such as running the Oppo audio through my team of Sonos zone players in every room.  I wouldn't be able to get high resolution, but it would be OK for movie sound.

Feeling somewhat down, I decided simply to listen to a cassette I had recorded in the bedroom from an FM local broadcast the previous week.  Piped into Sonos in the bedroom and back out in the living room in glorious 44.1/16 digital piped through my system.  Pure sonos connections had NEVER been a problem with the Master 7, in 2.5 weeks to this point, so  I could just relax.

But this time, after about one hour, the Krell shut down.

Now it was beginning to dawn on me.  Something had changed since the Party, and it was not just the fact that I had switched to AES cable after the party.  (Actually, in another set of breathless tests I performed earlier, I had switched back to Coax hoping that, despite the muting issue, perhaps it handled other things better.  It did not seem to, and I got shutdowns using Coax just like using AES, and the sound wasn't any better either.  But correcting the polarity made a big difference to the sound.)  It was because, at the party, the loud noises had caused some permanent change to the speaker.  Most likely the HF transformer inside the Acoustat interface had an intermittent short.  That's how these things start.  You get a tiny short and then a fuse blows.  Then, the fuse blows more and more often as the short gets bigger.  Except, in my case, the Krell FPB is so clever it can figure out when the speaker is shorting, and shut down before the fuse has a chance to blow.

THAT was what had been happening all along, ever since the night of the party, I now concluded.  I have at least one bad transformer in my Acoustats that needs fixing.

So to test that, I first ran the left speaker playing FM through Sonos, just as I had done the night before with a cassette tape of an FM broadcast.  I played louder and louder on the left channel, but after 90 minutes the Krell had not shut down.  Back to the right channel, I played FM through Sonos again through the right channel, and in 20 minutes the Krell had shut down.

So I have proven that there is a problem with the right speaker now.  THAT was the only speaker to blow it's fuse at the party, and the issue probably started right then.

So that could explain everything.  Perhaps now using the AES input I will have no more issues with the Master 7.  But it still seemed that PLL helped me get the Oppo to play much longer and sound better, so I'm going to keep PLL enabled.  Likewise for the other changes I made, replacing the Baldur XLR interconnects with Belden 1800F, and putting the subs on the other circuit.  Those changes were all for the best anyways.

I remember now that I had long ago had the same trouble playing the F-26 and L-1000T tuners as digitized by the Lavry AD10 instead of Sonos.   The few times before I had tried to do that, the Krell had shut down.  I figured maybe the tuners had too much low frequency noise resulting from the FM tuners, I just couldn't do that.  But now I think the problem was that I was putting both the Krell and the subwoofers on the same circuit, and the Krell was trying to get more current on some peak and competing with the subs for that power.

So while it's possible that all the shutdowns I saw resulted from a growing transformer short in one speaker, it also possible, and it was part of the reason this took so long to figure out, there there were actually more than one cause, and probably two, for the shutdowns.  One being the transformer short, and the other being inadequate AC power.  At the end of the day, though removing the Baldur cable seemed to help playing Taguita Militar, that might have really represented the "inadequate power" scenario because (1) it happened consistently over and over at the same point in the track, and (2) in addition to extreme highs, that recording also has deep powerful bass, in alternation, creating the situation where the Krell and the subwoofers were competing for AC power.

I haven't actually proven the left speaker does not have a problem, it just didn't show up in 90 minutes of pretty loud (not extreme) testing.  I will continue testing the left speaker and go back to using the Audio GD DAC using the safer AES input with PLL enabled (belt and suspenders) but meanwhile record the right channel output using a Masterlink at 24/96 so I can see problems up to 40kHz.

The interface which blew was the original interface that came with the speakers.  It's brother was already replaced in 2010 with a spare new old stock unit in museum quality which happened to appear on eBay exactly when I needed it.  The unit which has not yet proven to be defective on the left channel was that new old stock unit which I have now used for 7 years.  It has a much later serial number.  So there is hope that it survived the cruelties of the past few weeks.

Ironically just a couple months ago, the leading Acoustat interface refurbisher posted to The Audio Circuit that people should not use 5 amp slow blow fuses anymore.  That recommendation was even a bit risky in 1980, but in 2017 many power amplifiers have enough voltage and current to blow a transformer before they blow the 5 amp fuse.  Actually, Dr. Strickland originally specified 3 amp slow blow, as that had a decent chance of protecting the speaker in most cases.  But there were so many complaints of unnecessary fuse blowing, he gave in to pressure and changed the specification to 5 amp, which would stop most, but far from all catastrophic events.  That just barely passed muster with the 200 watt amplifiers of 1980, but not the 500W amplifiers of today (my FPB 300 actually has 600 specified watts at 4 ohms, and more like 900W at measured clipping, and over 1500W into 2 ohms, mind you, it's so smart it stops the moment it detects a short, making no foul noise in the amplifier or speaker, but it doesn't necessarily stop full power signals into a not yet short).

I went to what was described (and seemed) like the last Radio Shack store in San Antonio to get a collection of fuses, I'm not actually sure which is best, 3a slow blow and 5 amp fast blow, however they didn't have 5 so I got 3 and 4 amp in fast blow.  But I'm using 3a for now.

I played the Left speaker by itself for hours Friday night, playing FM mostly through analog Sonos, not wanting to extend the experiment too far.  But by Saturday I was testing the Dac 19 through system connections to the kitchen Oppo, including 70 feet of Belden RG6 for coax SPDIF.  It was working fine.  I had hooked up one of my DEQ's as a monitor meter (having level and spectrum peak hold, etc etc) and saw nothing wrong,  Santana Supernatural DVD-Audio through the Oppo.

After several days testing to prove the fault with one speaker, and so far only one speaker, I hooked up the Master 7 again.  Now I didn't power the amp yet, I just used my DEQ monitor to check out all the inputs on my Tact.  Everythng was working flawlessly, and no spurious crap, EXCEPT using the Kanex HDMI de-embedder to strip SPDIF from a standard HDMI signal (in this case, from the Oppo BDP-95 set to only output PCM up to 96kHz).  Using the de-embedder in both modes gave a signal with periodic low frequency bursts up to -10dB apparently going down to DC.  Even though off, the Krell seemed to be making noises like it was detecting DC at intervals.  Somehow the likely high jitter of the HDMI signal (which itself comes from a CAT6a house network) is overloading the jitter removing capacity of the Master 7, and the result is bursts of low frequency which would not be good for directly feeding DAC into a DC coupled amplifier.

Well anyway the HDMI de-embedder has always been a troublesome piece.  Quite often I have found it not to work on either position.  But I have never noticed DC pulses.  Still, these -10dB pulses were nothing like the high pitch max level shreak I heard the night I first selected the de-embedder to play the Oppo output at my party.  So from destructive shriek to quasi normal operation with bursts of DC is some progress anyway.  But I'm simply not using the de-embedder now for sure (except testing purposes as it clearly makes an interesting test).

But testing further I also found something to work which had never worked before.  Playing Reference Recordings 88 kHz data discs on the Oppo!  I had run into a brick wall with these discs, they can be downloaded to my kitchen mac, but the only network available from there now (I've de-certified the part optical connection to living room) is Sonos, which doesn't do high rez.  So I had to laboriously convert each file to 44.1kHz, which I did for one disc, but I have others that have been sitting there, and I've never enjoyed the original high rez.

Bnt now I can, now that these 88kHz discs play perfectly on the kitchen Oppo, over 70 feet of RG6 wire (actually in several segments, one going through the attic), into the Tact where it gets level adjusted, output through the Tact AES output to the DEQ, and from there by AES to the Master 7, now it all worked perfectly and sounded wonderful, much as the Santana Supernatural DVD-Audio sounded better on the Kitchen Oppo digital output than ever before and ever through any other of several means I have to play it (but, the Integra Research RDV is extremely troublesome on this disk, it goes through various broken frames of garbage and takes persistance to play the DVD-Audio, so much so, I've given up.  I'm getting another Denon DVD-9000 which is said to have full DVD capability, hopefully full DVD-Audio capability, and I now think it sounds much better thank than the RDV at the analog output anyway.  Speaking of which, before I use any new analong input now, I think I'll certify it as OK by using the Behringer as a monitor as I just did.  (I used the supertweeter DEQ on top as monitor, turning off everything related to the super tweeters and disconnecting them from the DEQ, then changing it to simply monitor the analog input connected to the XLR output of the new unused channel of the Master 7, I've got glorious mono now that one of my Aoustats needs repair and my spare interface in storage was never repaired either.  But I believe I will have my other Acoustat back online before long, better than ever.)

Don't let 'em tell you it can't be done.  For me, it seems AES input does the trick, even if there's an intermediate length of SPDIF back to the source behind that AES.  And I'm not really sure if PLL is helping, but it seems to be working for nearly everything, and I think it prevents stuff from going badly awry,  and sounded better to me.  Now I'm back to the rock solid digital connections I had before, mostly, with one more important one and one less unimportant one.

So now I just need to get one of my Acoustats fixed.  But that's not a problem, as they say, it's an opportunity.

A lady friend of mine knew the explanation right off.  You get expensive new toy, the old toys get jealous, they want money spent on them.

No comments:

Post a Comment