Monday, October 3, 2016

Handling the DSD noise

I was noticing that the Krell FPB 300 was clinking a lot, yet I was only playing at a fairly soft level, playing the RCA Living Stereo hybrid SACD of Wallenstein Symphony of the Air.  I think that these old analog recordings often contained HF noise that was ignored, and in DSD encoding there is no low pass filtering.  And I think this range of early manufactured SACD's and/or this specific one is notorious for HF noises as described in the pages of The Absolute Sound.  HF noises that can potentially be destructive to amplifiers, etc.

So I thought, what if I low pass the Krell at 20kHz?  This might not be a good idea in general, but since I am adding in the supertweeters at 20kHz, it might be the right thing to do anyway.

So I tried it, and even ignoring the great reduction in amplifier clinking I thought it sounded better too.  More relaxed sound.  BTW I noted in a wonderful performance of the San Antonio Symphony on Friday that live music has a particular relaxed sound that is often more like LP's than CD's.

I fine tuned the adjustment and the pink noise is flatter than ever, and notably dipped at 20kHz when I turn the supertweeters off (it was a bit like that before also).  So that the high pass and low pass match I now have them both set to 12dB/octave, which would increase the supertweeter because it used to be set to 24dB/octave (LR) with two LC filters.  But I didn't delete the second LC filter, I moved it down to 9kHz because the first filter seemed to run out of attenuation (maybe or maybe not, I'm not sure the meaning of the -15dB setting when using LC, does that limit each filter to -15dB attenuation?).  The supertweeter also has a notch 4-6k to remove any metallic sound.  AND it has it's own built-in filter which I've never really accounted for, I think it's around 10kHz.  But you see my filters have already attenuated it greatly by then (and perhaps my second filter is redundant).

On the panel DEQ, the HC (12dB/octave presumed for HC, this is a cut!) seemed to cut less highs (by the graph) than H6, which would be 6dB/octave.  That's true, the shallow filter would have to start
earlier and cut more prior to 20kHz than the 12dB octave.  One more reason to prefer HC to H6, though one might argue a 6dB octave cut would be less sonically harmful.  In a quick audition I couldn't hear the effect of either filter, though it was my impression from extended listening that applying HC/20kHz to the panels sounded more relaxed.  Actually it's hard to know the true acoustic cut because the speakers themselves roll off as steeply around 18-20kHz.  Meanwhile the amplifier is pushing harder and harder due to the capacitive load.  Certainly doesn't make sense when a supertweeter is covering the range with wider dispersion and response to 35kHz.

I had been planning to just play this disc in CD mode to keep my amplifier from blowing up, but it sounds so much better in SACD, and now I can play it at +3dB on the Krell (since the 9000ES output is a bit low) at it sounds fabulous.

I'm wondering how much time the amplifier will run before the next repair.  And how it will go when it goes next time.  It is clinking more.  That could just be loose panels, but when I sturdied the amplifier current draw, the clinking is generally after-the-fact but often does follow high current use.

The way the Krell works is that more bias current is demanded based on a comparator, so linearity is always maintained.  However, when the amplifier ages, more current is demanded more often until fixed limits apply.  So you get sharper bursts of high energy, which leads to more clinking.

I'm keeping the 18 year old Krell plugged straight into the wall with the factory power cord which is quite robust.  That way I'm operating by the required standard for safety listing.  No more monkeying around with switches and current testers.  However it goes now, it should be safe.  It could go for a another decade or a minute, nobody knows.

No comments:

Post a Comment