Tuesday, July 10, 2018

Another mistake fixed

Though many times I think I examine situations far more carefully than others, and think about them far more deeply, nevertheless I often make mistakes.  Big mistakes.  Not long ago I had my system speakers "out of phase" (with each other) for over a week.  In my defense, the audio specialist at my house immediately declared the phase was correct (though it wasn't), and after he left I played music almost entirely in background until discovering the problem.

Recently I made another discovery that pretty much invalidates all the comparisons I've done since February when my 2nd and 3rd dedicated AC lines were installed.

Last week on staycation I discovered there was (barely) audible hum coming from both speakers.  I immediately realized I had not followed through with the plan to put the DACs for the main amplifiers on the same circuit as the amplifiers themselves.  (I have separate DACs for the two amplifiers so I can level match each amplifier in my now permanent ABX configuration, with zero degradation as compared to what I would be doing otherwise, it's just the same thing but now in parallel, nothing was added in series except upstream there is a AES3 digital splitter which I'll say more about later.)

So, I did that, and at first the hum seemed to almost entirely disappear.  I had remembered it being nil in the past, with ear right up to the speaker I could hear nothing through the Krell and Aragon (when everything was on the one dedicated circuit).  Now it wasn't quite nil, but so low it could only be heard with ear to speaker.

Then a few days later I realized the hum was still louder than it should have been.  So I tried also putting the AES3 digital splitter (the digital output of the Behringer DSP is split for the two dacs which feed the 2 amplifiers) also on the same circuit as the amplifiers.

With this change, the hum vanished entirely.  Zero, nada, zilch, with ear right to speaker.  So this is what I'll have to do (I'll need a small power strip for the 2 DACs and the splitter because the wall has only 4 outlets on this--the original--dedicated circuit and there are already 5 things that need to be plugged into it).

Given that AES receivers are transformer coupled, I'm not at all sure why the AES splitter needs to be on the same circuit.  I think the AES3 isolation in the Emotiva Stealth DC dacs is not as good as my Henry Engineering (HE) AES3 splitter--which seems as solid as anything could be, and this is more evidence as to how excellent it is.

BTW in earlier tests I found that the HE splitter added no measureable jitter to a digital signal as analyzed by my Sencore Digital Analyzer.  In fact, that was true of my entire digital setup starting from the Sonos Connect, through the long coax cable to the Tact RCS 2.0, the Tact itself, the cable to the first HE splitter, then to the midrange Behringer unit, and the cable leading up to my Audio GD DAC 19.  I measured the same 220ps jitter at the Sonos output as at the cable connecting to the DAC 19 going through all those pieces of equipment and cables.  This convinced me at the time that 220 ps was just the "inherent" jitter in SPDIF, given the nature of the signal and the impedances, there is just enough deflection of the baseline to cause that much jitter.  Well, I was wrong, I've now measured the lowest ever (for me) jitter at the output of the Oppo BDP-205, actually playing the tougher J Test signal, where IIRC measured under 200ps, 190 or so IIRC.  Now I think THAT is probably approching the inherent SPDIF limits on J Test.  My meter can reliably measure as low as 150ps, which I did using white noise (as I did in the aforementioned full digital system test reaching 220ps) using the Integra Research RDV-1.  Using the tougher J Test signal, the RDV-1 measured about 210ps, not as good as the Oppo, and we would expect the Sonos to rise above 220ps a bit.  Anyway, at the time, I was pleasantly surprised by the low jitter (none measureable) added by my entire "complicated" digital processing system.  And I should add I don't believe jitter as observed in typical audio systems to be of major importance, experiments should it to become important at levels 10,000 times larger.  When I was using Toslink, I could not measure the jitter reliably, the Sencore wouldn't show any numbers mostly, but finally after much fiddling and conversion back to coax I got it to show 600 ps.  That's when I decided not to use toslink in my main system for serious sources, though I still doubted the jitter added to be audible.

Now, with separate AC circuits having a boundary at the HE splitter, some might suspect I'm adding jitter.  It is the input to the HE splitter which must reject the ground loop at the ground at its input.  If it does so imperfectly, SOME of that ground difference might be impressed onto the digital signal, causing jitter.  This is in spite of the splitter having a transformer coupled input which rejects hum by 80dB or more, and in spite of AES3 being a balanced signal "unaffected" by ground loops, and in spite of AES3 being a signal with 2-7 volts PP.

So, one could argue, I've eliminated the analog hum, but I've replaced it with a trace (and it would be barely a trace, for the reasons I just gave above) of jitter.

It's worth doing another jitter measurement anyway, but I suspect the jitter is negligible, and even if the splitter were adding jitter signal, it would then be almost entirely eliminated by the asychronous receiver in the Emotiva DACs.

Sonically, it is better than ever.  But now the slight physical hum from the Hafler 9300 chassis became noticeable, so I tweaked it using a 3 piece set of Mapleshade Heavyhats (which I had still in factory box waiting to be used, I had intended to use them on DACs).  This seems to help, though I wonder if it's mainly caused by expectation bias.  Strangely, I now can't hear the hum from a couple feet away from the Hafler, but inches away it seems about the same.

I also tweaked the speaker cables using 4 CableElevators I bought recently, lifting them well above the power and digital cables.  I bought an 8 piece set of CableElevators, something I'd lusted for about 20 years, on sale.  These don't fall over like other lifting devices, and present several surfaces where cables can be delicately perched as well as the main central grove.

No comments:

Post a Comment