Maximalism

 Many audiophiles, perhaps even most, brag about something like minimalism.  They de-contextualize Keep it Simple Stupid (KISS) to mean whatever approach they are using that they think is simple.

But if you then suggested, perhaps, something even "simpler," such as, say, an all-in-one phono player, or a cactus needle, you'd surely get scolded.

The full KISS principle is qualified.  

Einstein is often alleged to having said "Everything should be made as simple as possible, but no simpler."  What Einstein actually said was:

“It can scarcely be denied that the supreme goal of all theory is to make the irreducible basic elements as simple and as few as possible without having to surrender the adequate representation of a single datum of experience.”

Not surrender a single datum???  In audio terms, that would mean we should not do without anything that might make the audio experience better to even the slightest degree!  This isn't minimalism, it's maximalism.  We do more until the last datum is resolved!

Ok, so you may say that the "more" I am referring to could be eliminating extra "unnecessary" things.  But who is to say what's necessary and what's not?  How about doing without the preamp and just plugging your moving coil cartridge into your power amp?  That's simpler, isn't it?

I have never fallen in line with the simple-audio-system meme.  I've cynically believed its "simply" to get you to buy more costly stuff, with the claim that you won't need as many other things.  But then it turns out, in the long run, you may want those other things back.  I've seen that.

Instead, I seek to make the whole system as good as I possibly can.  That could, in principle, mean eliminating things, but that's almost never possible to do while retaining the same high performance.  I basically never added unnecessary things in the first place, at least if I could help it.

Perhaps I could design and build new things that combine just the exact stages, features, and so on that I need, with nothing else.  Trouble is, I have neither the time nor patience nor even knowledge yet to take on such formidable tasks.  Instead, I generally work with things I can buy, ready made and reasonably prices, and connect them together to realize my system ideas.  There may be more circuits, boxes, etc, than I need, if I could create some dream product doing all the things I need, and nothing more.  But no such product(s) exist, and even if they did, they would be incredibly expensive, as we can tell from products that do exist and are almost there.

And so it is that I have 6 DSP boxes, one set of 3 to handle my phase linear steep crossovers, and another set of 3 to handle other EQ, limiting, and display functions--because the other type does those jobs better.  And 3 of each kind because I need one for each "way" of the system.

Now I believe there is a system sort of like this, though not as flexibly as I would like, and it costs around $20,000.  My 6 DSP boxes cost under $400 each, for a total of about $2400.

The system has 3 ways because no one driver can cover the full audible spectrum well.  At least none that I can afford...

Anyway, now I may be adding a new gizmo, a self described "Buffer", the Musical Fidelity X-10D Version 3, at least to the path originating from my living room tuner, the Pioneer F-26.

I'm adding it because it seems to sound better.  I have reasonable ideas why it might sound better also.  The Pioneer actually has a fairly high impedance output.  Even though it has separate fixed and variable outputs, they all originate from the same opamp, the "fixed" outputs (which most but not all audiophiles assume is better on most source devices) go through a voltage divider with 43k on each side, resulting in a source impedance of 22k (!!! this has always appeared to me as a deliberate way of creating high frequency rolloff), the the variable outputs go through a pot in series with a 330 ohm resistor on the ground side, resulting in a minimum of 330 ohms minimum perhaps, with fixed all the way up.

I had first assumed then primitively tested the fixed was better, and I might have tested it too.  So since my usual connection was to Sonos, I ran the fixed outputs to Sonos.  So when I connected first to the PMD 580 (whose conversion to digital sounded horrible to my ears) and then to the Lavry AD10 (which sounds wonderful) I was using the Lavry.  I have the "fixed" output all the way up, so it should be an output impedance of 330 ohms and a little more.  This is fairly high as solid state equipment goes, and could cause high frequency rolloff with long runs of some kinds of cable, and destination equipment.  But it seems it could be far worse with the fixed outputs (which raises some more interesting questions, perhaps a less is more freak would find some adjustment of the output control yielded "just the right" sound by affecting the high frequency cutoff.  but achieving a similar effect with an external filtering and buffering active device could actually have many advantages, including higher ultimate slopes of high frequency cutoff, combined with steep low frequency cutoff, and loading the originating op amp circuit including the output capacitor the least possible).  Interestingly, the cutoff from the fixed outputs at 22000 ohm output impedance into a typical 100 pF of cable capacitance plus other stuff would be 72kHz.  This is not far from the high frequency cutoff of the X-10D.  So using the fixed outputs without the X-10D would have about the same high frequency cutoff as using the variable outputs with the X-10D...

I'm thinking that the X-10D serves a function similar to the "Noise Filter / Buffer" MA Cotter NFB-2, which was designed to improve on the output stages of nearly everything.  In addition to providing a very high input impedance and a very low output impedance, the NFB-2 was claimed to remove out-of-band noise without adversely affecting the audible spectrum.  This was said to reduce the demands upon later equipment, thereby enabling it to perform better.

"Sure, every equipment maker could have an output stage as good as the NFB-2 on their equipment.  Fat Chance." said Peter Aczel of the Audio Critic before he became a full audio objectivist possibly eschewing such differences.

What was in the MA Cotter NFB-2 ?  It was never revealed to the public, and I've never been able to find out online.  They were marketed to and sold by the audio salon and modification studio I worked at known as Audio Directions / Audio Dimensions.  I believe I saw Mitch Cotter himself demonstrating the products to the staff, including co-owner Ike Eisenson, who then took Mitch to the back room for a long discussion.   A senior technician later remarked, "It probably has Nuvistors in it."  Ike was a big fan of Nuvistors and wrote glowingly of them in the Audio Dimensions magazine and Tu-be modification manual.  FWIW, Tony Michaelson was in our orbit as well.

Interestingly enough, the milspec pencil tubes in the X-10D Version 3 are very similar in characteristics to Nuvistors, ,the main difference that the pencil tubes are made of glass rather than the Nuvistor metal.  The pencil tubes are similarly compact, which makes them similarly (if not quite as) robust, immune to vibration and long lived.  In both cases projected life can be around 100,000 hours, which is at least one solid decade of running continuously, and possibly more in intermittant use.  Playing up this similarity, Musical Fidelity called them "Mu Vistors," after years of making equipment with actual (but very limited supply) Nuvistors.

"tubes" as we call them intrinsically have very high input impedances, and generally have sufficient inter-electrode capacitances (combined with internal high impedances) to have inherent stability and limited response.  It's also easy to limit the low frequency response with small high quality film capacitors before, after, and mid-stage, because of the high impedances.  Cooked up just right, or at least like the X-10D Version 3 has them, tube resistance coupled amplifiers naturally work as buffers and as out-of-band filters!  You don't even need to add anything else.  The very nature of its operation filters out-of-band signals.

Now people might ask, why not use tubes for everything?  The answer may be, that tubes are not necessarily the best for everything.  It is hard to make good power amplifiers with tubes unless you are going to use an output transformer...and that's a huge limitation.  And so on.  They just happen to be the best for this one role of "noise filter / buffer" because it's a natural one for them.

So it makes perfect sense to have tubes in one and only one spot in a high performance system.  And that is as the high performance output stages of especially noisy equipment, such as tuners.  As a noise filter / buffer for the preceding stages.

Interestingly enough, Mitch Cotter also worked on the famous Marantz 10B tuner, which has wonderful sound by many reports.   People generally look to the impressive 6 tube "Butterworth" (not really) IF stage, which not even Marantz has even attempted to duplicate.  (Marantz did include similar in concept IF stages in the 20/20b/120/150 tuners but using transistors instead of tubes.)

But one thing also, the Marantz 10B also has a tube output stage, and that could explain some of the "good sound."

Now what I'm suggesting here isn't at all a new idea.  Musical Fidelity cooked up the first X-10D in the 1990's, and for much longer than that people have been sticking tube outputs on nearly everything you can think of that may use no other tubes.  Sometimes they add tubes just for a light show.  There was one guy who did modifications where he stuck tube outputs on nearly everything.  He went by the name Lampizator.  And now there are endless Chifi tube buffer devices you can buy on eBay.

It's only radical for someone like me, who never used tubes for everything, and finally abandoned them altogether around 2002, and whose closest friends haven't used tubes since the 70's or earlier.

Tubes are widely perceived by non-tubeophiles like some of my friends as "coloration devices."  But a well designed tube noise filter / buffer can be "perfect" by audio objectivist standards, adding no perceivable audible coloration according to audio objectivist interpretations, and still improve the sound by reducing out-of-band noise, or reducing impedance mismatches.  Generally speaking, tube equipment can be designed to meet every specification as transistorized equipment, even power amplifiers (but at much larger cost).  Ancient tube equipment did not often do this (but see, for example, Western Electric amplifiers, which were very good).

It might, in some cases, add an audibly euphonic amount of noise.  That might actually help in some cases (similar to dithering noise used in digital).  But it would have to be very small to be considered "not a coloration."

I think a tube noise filter / buffer for a high quality system needs to be designed to a very high level of performance.  It should have at least 20-20kHz bandwidth, and 20-20khz distortion below 0.01%.  The X-10D V3 is claimed to do better than that, 20-60kHz and 0.004% distortion.  That looks good.

The bandwidth, signal to noise level, and dynamic range requirement may depend on what equipment it is being applied to, the type of music being played, or other factors, but weighted noise should probably be below -100dB of peak level.

Most people who are interested in such things eschew all specifications in the first place, and maybe even think devices with around 1% distortion or so to be "ideal."

In my limited experience with such things, amplifiers with 1% distortion or higher sound awful.  I don't think even high order harmonics are audible below 0.01%.  Few power amplifiers have distortion that low, but their harmonic profiles may be relatively innocuous, mostly weighted toward the lowest orders.

Now a general purpose device to do this might have settings for bandwidth and noise level.  The high frequency response could be varied with an air capacitor adding to appropriate elements, varying the high frequency cutoff from 200kHz to 20kHz.  I don't like the sound of low pass filters that cut below 20kHz.

Because of their inherent stability, feedback need not be eschewed with tubes.  Or with well designed transistor equipment either.  But many tweak designers still live in fantasy land, and zero feedback, which typically yields too much distortion to be inaudible distortion, is deployed.

For many years I personally believed that resistance coupled tube amplifiers (like the PAS-3X) should have low impedance cathode follower outputs.  Higher end preamps like Marantz Model 7 tended to do that (though IIRC they included feedback around the cathode follower too).  But many audiophiles long believed cathode followers had a bad sound quality.  While the cathode follower is a simple and stable circuit, it cannot be made to have as flat frequency response as a dual triode stage with loop feedback.  Or as uniform output impedance with frequency.  The more "complex" dual triode with feedback has long been loved.  The feedback works to overcome the tube imposed bandwidth and impedance limitations with the restraint (the resistors and capacitors of the feedback network) being almost perfect as compared to the cathode resistor which is one step removed from the actual output.  The PAS-3X line amplifier was legendarily good in itself, and the subject of endless additional modifications.  Very few added a cathode follower--even though it is a rather high impedance circuit measured in kohms as compared to the 33 ohms of the X-10D Version 3, which achieves that due to the type of tubes and high degree of feedback.  The high frequency cutoff is limited to 60kHz instead of the 100 kHz of a 12AX7 or 12AU7 circuit, but as a noise filter buffer for an FM tuner that may be a good thing...maybe even lower.  I'm not sure exactly where the cutoff should be to be transparent, but 60kHz seems like a good start and work down from there.

I'm now planning to implement an even more ambitious maximalist concept.  Something I've had in mind for a long time, but I've never had enough time to do it correctly.

Automatic background music.  Whenever the system is silent for more than 50 seconds or so, the FM tuner will drop in as replacement background music.

One of the key issues here is that regardless of what level was chosen for the previous input, which may even have included above 0dB digital gain (for low gain sources, such as multichannel movie discs where "summing" is used to generate 2 channels for the living room system), the tuner should switch in at a background music level, or at least no louder than I would like to listen to the tuner, which is somewhere between -10dB and -23dB because of high volume compression on FM, even on the classical station.  You don't want to listen to FM as loud as other sources because the FM is continuously pushed up to the broadcast limit.

But now I have a perfect way to do this.  I have already wired this up: the FM tuner is wired to a dedicated Analog to Digital converter.  It would be nice if I had another Lavry AD10, but I bought something cheaper and nearly as good a few years ago, a Black Lion Sparrow, the original which has no level controls.  Fortunately, the sensitivity (claimed 1.6V) is perfect for my Pioneer F-26 at the more transparent sounding variable output turned all the way up.  I was worried I'd have to turn the F-26 down, but in fact it's already only hitting about -8dB, could even be cranked up a bit, but the Black Lion is a very quiet (-120dB claimed "A" weighted) converter.  And that low level helps me achieve the desired background level of -23dB in combination with the maximum -15dB attenuation of a Behringer Ultra Curve 2496 DEQ  (I wish Behringer had allowed far more attenuation, and in 0.1dB increments for each channel).

So the new setup is like this:  The variable output of the F-26, cranked all the way up, feeds a Musical Fidelity X-10D V3, my latest minty one (looked new in box), which feeds the Black Lion Sparrow (Using RCA to mono 1/4 plugs as an unbalanced to balanced adapter), from which the AES output goes to the Behringer DEQ, which provides some EQ tweaking capabilities and up to -15dB reduction, from which the AES output goes to a Hosa AES to unbalanced SPDIF adapter, and thence to a 4-way video switch (on which the video jacks have sufficient bandwidth for SPDIF) which feeds digital input on my Tact digital preamp.

The plan was to get this new digital path sounding about as good as the one where the X-10D V3 feeds my Emotiva preamp and thence my Lavry AD10 converter.

I think the setup above does this very closely, with only a bit more bass overhang than the Lavry path, and that could possible be adjusted (I'm already doing so to a limited degree) with EQ adjustements in the DEQ.

Also important, the attenuation reaches the required level that I can feed the digital signal directly to my crossovers without attenuation.

So now, I can get a 3-way electronically controlled AES switch.  One "way" connects to the output of the Tact for everything except FM.  The second "way" connects to the new FM-tuner-to-digital-output path described above, except directly from the AES output of the DEQ (no conversion to coax SPDIF is required).  The third way is not connected to anything, making it a "mute" selection.

To that, I add a "silence detector" at the analog outputs of the Tact--which I never routinely use--which causes a contact closure or open after 50 seconds of silence which either controls the AES switch directly, or feeds an Insteon low voltage interface to send an Insteon home control signal, to which another Insteon low voltage output interface communicates with the AES switch.

Running the whole thing through Insteon lets me add buttons elsewhere which enable the "mute" selection on the AES switch, in case even the FM is bugging me too much, or switch back to FM or the main program.

But I find the new digital conversion on FM, and also listening at lower levels like -20dB from peak, cause me to feel like muting it less.

There are many times when you want information loss.