300b lovers

Let’s talk a bit about distortion, particularly in a DHT driver. It defeats the purpose of using a very low distortion 45, 2A3, 300B, or 845 if the driver has more distortion than the DHT itself. You might as well use a much cheaper power pentode or beam tetrode and save yourself a lot of trouble. Not only that, but pentode and ultralinear-connected output sections requires feedback across the whole amplifier, which cleans up the mess from the input and driver sections. This is why the driver section is an afterthought in Golden Age PP pentode amplifiers. In the Dynaco circuit, there isn’t even a driver ... the power tubes are driven from the split-load inverter, or "concertina" stage.

But ... if the goal is the most linear possible amplifying stage, with zero feedback, a completely different approach is required. True, push-pull substantially reduces distortion, but in reality it only reduces even-order distortion ... 2nd harmonic, 4th harmonic, 6th harmonic. etc. It has no effect on odd-order distortion ... 3rd harmonic, 5th harmonic, 7th harmonic, etc. Most tubes have dominant 2nd harmonic distortion. But not all. The 6DJ8/6922, an RF tube, has dominant 3rd harmonic. Tubes designed for video amps can also have dominant 3rd harmonics, since distortion in a video tube has almost no effect on the picture (even 5% distortion would be barely visible).

Aside from the subjective sonics of dominant 3rd harmonic, this has consequences for a PP amplifier. The PP circuit will have no effect on 3rd harmonic; it will not reduce it even 1 dB. So tubes with dominant 3rd harmonic should be avoided.

What about more "normal" tubes with dominant 2nd harmonic? Well, one of the requirements for PP cancellation is the magnitudes and phase of the cancelled harmonics line up with each other. The magnitudes of the (even) harmonics from each tube should be within 1 dB of each other, and the phase of the (even) harmonics should be within 45 degrees of each other. Normally this would be of no concern with reasonably matched tube sections, but it does matter if the capacitive loads are not the same. The odd harmonics, of course, do not cancel.

The notion of harmonics having phase might seem a little odd, but keep in mind that square waves and triangle waves have identical magnitudes of harmonics; the only difference between the two signals is the phase of the harmonics. So the phase of the harmonics is not insignificant, and actually reflects a different transfer curve. In graphical terms, you want complementary transfer curves; if kinks appear in the curve of one tube but not the other, that kink will not cancel.

The full scope of the driver requirements for DHTs is probably now coming into view.

* Three times as much swing is required, compared to pentode or beam tetrode.

* Linearity should be better than the DHT itself.

* The load is mostly capacitive, consisting of the 60 to 80 pF Miller capacitance of the DHT grid. Reactive loads increase distortion at high frequencies, where it is most audible, and also reflects the power back to the plates of the driver tube.

* If the driver is PP, 2nd harmonic should be at least 10 to 20 dB greater than 3rd harmonic, and the two loads should be symmetric to keep the phase of the driver harmonics the same.

* There should be at least 3 dB of driver headroom so the amplifier doesn’t all clip at once. More headroom is desirable.

* Momentary sags, or program-correlated noise, in the power tube B+ supply should not interact with the driver tube B+ supply.

Well, I tried that with an amp I once called the Aurora. SE input, and if memory serves, the conversion to PP prior to the driver, which was PP. The problem is the input tube has a high output impedance, which enormously complicates the transformer operation.

You see, the input transformer of the new amplifier, as well as the Mark I Karna’s, is driven with a low source impedance ... the preamp. There are a handful of ancient preamps with a Zout of 8k or so, but they are hard to use because a Zout that high makes them very susceptible to rolloff from cable capacitance. Most tube preamps have cathode follower outputs in the 400 ohm range, and if feedback is used, quite a bit lower. Transistor preamps can be as low as a few ohms.

Transformers like to see a low impedance on either the primary (input) or secondary (output). It doesn’t matter which end. The problems start with an interstage, where the secondary is driving a grid, which has a near-infinite impedance that is somewhat unpredictable, and a primary connected to a plate. The nicest sounding tubes tend to be the old octals, or even the true antiques, the five-pin tubes from the Thirties.

They all have pretty high output impedances, 7.7 k or higher. This is a really high impedance for a transformer. The lack of bandwidth wasn’t a problem back in the day, since AM radio bandwidth was never higher than 8 kHz, optical movie soundtracks the same, and shellac 78’s were mostly noise above 6 kHz. And program sources didn’t go lower than 50 Hz. Modern bandwidths of 30 Hz to 15 kHz didn’t arrive until the mid-Fifties, with magnetic tape, modern LP’s, and FM radio. By then, transformers were used for line level applications in studios, and for output transformers in power amps. This was the all-analog vacuum tube era, of course.

Interstage transformers are a very special use case. We are reviving a 1920’s and 1930’s technology to modern high-bandwidth applications, but there are still limitations, mostly the result of using high impedances. I did use what I call "Interstage 1" in the Karna amplifier, but that was really putting the transformer right to the edge of what can be done. Adding phase splitting to its task list means I will likely see phase spread at the top of the band due stray capacitances not matching between sets of windings. This is a solved problem for studio line-level transformers, but asking interstage transformers to do this results in a not-very-good interstage transformer.

And frankly, for what benefit? It isn’t like a SE input tube is all that awesome. In phono preamps, sure, SE circuits make things easier, what with RIAA compensation combined with noise considerations. But for the input stage of a power amplifier? Where’s the benefit, except for tradition?

So I restrict phase splitting to the easiest location, the input, where it isn’t doing much else. This is proven studio technology that’s been around since the 1930’s, and well-refined by the 1950’s. The interstage, a far more difficult task, is confined to the driver/output interface, and is fully balanced on both ends.

Yes, there are interstage transformers on the market that are SE to PP. I would not use them. It is very difficult to get HF symmetry on the secondaries, but modern transformer designers can do a lot that wasn’t possible even ten years ago. I wouldn’t be surprised to see a 300B SET driver interfaced to PP 300B outputs through such a transformer. I won’t be the one designing it, though.

A power amp with three 300B’s would have a certain visual appeal, and you could weave a fun story around it to match the visuals. At the hifi shows, you could hang pictures of famous trimotor airplanes, like the Ford Trimotor.

What’s odd is that I have trouble describing the sound of my own gear ... this applies to Shadow Vector, my loudspeakers, and my electronics. I aim for natural, open sound that is free of electronic artifacts. If there is a residue of coloration, I’d like it to be pleasing, but as low as possible.

That’s the goal. What I hear is a very spacious sound, no surprise there, but also an unexpected "trippy" kind of feeling that slowly deepens over the first ten to twenty minutes of relaxed, non-critical listening. I have no idea what causes it, to be honest. Many don’t experience this, but some do, and it’s fun to watch them process the experience. My only guess is the disjunction between the auditory experience and the visual experience is so strong that secondary emotions are invoked.

My degree is in Psychology with a focus on Perceptual Psychology. I grew up in Japan and Hong Kong, which gave me a cultural experience different than most Americans, and am a bit familiar with Buddhist, Taoist, and Vedanta Hindu world-views. Partly as a result of that, I take Western audiophile pronouncements about what can, or can’t, be perceived with a lot of skepticism. Different cultures experience the world in profoundly different ways, and there’s a lot of individual variation, too.

The task of the audio system is to get out of the way of what can be a profound emotional experience for the listener. "Accuracy" in this context is absurd, since the goal is to facilitate a trance experience, a dreamlike state of consciousness. Artificiality (colorations that do not occur in nature) is distracting and can prevent the experience from happening.

By minimizing coloration at the level where it originates, the device itself, there’s less need for post-facto processing, which can induce dynamic colorations that are unnatural and a signature of "electronic" sound.

This discussion is kind of "meta", but it is my experience audio designers need to have a goal they are aiming for, otherwise you will never get there. And not a mechanistic goal, but a perceptually subjective goal.

The house I lived in back when I was in Portland had a ceiling like that, with a panoramic picture window overlooking the Oregon Coast Range behind the speakers. The very first PP 300B amp, the Amity, had a vast, CinemaScope soundstage. Once I heard that, no more SETs for me, nor PP pentode. All done.

That’s when I contacted Harvey Rosenberg back East and told him that interstage transformers were The Way. He never did get to hear my amps, but he got his hands on the Japanese Sun amplifiers (with ITs) and never looked back. Harvey was powering his gigantic Tannoy Westminsters, but from what he told me, the Sun amplifiers took charge of the big Tannoys. No surprise there.

IT-coupled PP DHT’s have a power and majesty unlike anything else in audio. Bell Labs and Western Electric knew what they were doing back in 1935. It probably didn’t hurt they had Leopold Stokowski as an in-house musical consultant.

Hi, CuriousJim!

I am kind of dubious about any KEF speaker being really 91 dB/meter/watt. That’s almost 1% conversion efficiency, and believe it or not, that’s quite high for the mainstream market, and especially KEF. True efficiencies between 85 and 88 dB are much more common. Efficiencies in that range need 100 to 200 watt amplifiers, which is very large for tube amps. When people say XYZ speaker needs 200 watts to "come alive", they are not joking.

1% efficiency is 92 dB/meter/watt. 0.5% is 89 dB/meter/watt. 0.25% is 86 dB/meter/watt. This raises the question ... where does all that power go, if not making sound?

It heats the voice coil, which eventually radiates its heat to the magnet structure, which radiates its heat into the cabinet. Copper increases its resistance with temperature, which leads to a type of dynamic compression as the voice coil heats and cools. In addition, voice coil heating eventually damages the cylindrical former it is wound on, leading to driver failure over time. That’s why Nomex and other fire-resistant materials are commonly used for VC formers.

This is the charm of high efficiency speakers: for a given SPL, much less power is wasted in heating the voice coil. More is turned into sound, which is the goal.

P.S. I agree, it is somewhat annoying to realize that 99%, or more, of the expensive watts we buy do nothing more than heat a voice coil, but that’s what’s really happening. It’s kind of shocking: hundreds of watts from the AC wall socket end up as milliwatts of acoustic power. The rest ends up as heat, and not in a great place, either.

A good way to visualize the difference between even and odd-order distortion harmonics is to imagine a sine wave ... a perfect, happy little sine wave. That's the original signal.

* Now, use a single diode to clip one side of it ... say, the positive side. In addition to generating a DC offset, if you run a spectral analysis of it, you'll see a series of harmonics ... 2nd, 4th, 6th, etc. etc. If you look at the "transfer curve" ... a curve mapping the input/ratio at different levels ... you'll see a diagonal line that is perfectly straight (the linear portion) that also has a sharp bend in it at the top, with a flat-topped region beyond the bend. The transfer curve is actually the true distortion mechanism; the spectral analysis of it is an indirect indicator that is (relatively) easy to measure.

* Let's use two diodes to clip the top and bottom sides, both positive and negative. This is known as symmetric clipping. If the flat-topping is at exactly matched levels, there will no DC offset. Similarly, if the clipping is precisely symmetric, the spectral analysis now shows no even-order terms (2nd, 4th, 6th, etc) but only odd-order (3rd, 5th, 7th, etc.). The transfer curve now has TWO kinks in it, at matching plus and minus signal levels. If the levels precisely match, there will be no even-order terms, but odd-order terms are abundant.

The diodes create hard clipping. Vacuum tubes, properly biased, create a softer "knee" region, but rest assured distortion is still there, just not as much, and with less high-order content. The property of symmetric circuits is they cancel transfer curves that are precisely opposite in shape ... a "C" shape that is inverted in the other phase of the circuit. But that depends on symmetry and precise phasing that tracks as levels go up and down.

It is difficult to predict real-world distortion from idealized triode or pentode models. The models assume tubes with perfect physical assembly and ideal emission characteristics. In practice, grid windings are not evenly spaced, grids are tilted a little bit, coatings on the cathode are not perfectly uniform, and there is always just a bit of residual contamination. Tubes are not built by robots, but skilled technicians, and as a result, they are all a little different from each other. By looking at spectral distortion measurements, patterns that are unique to each manufacturer emerge, and none conform exactly to the tube model. (The map is not the territory.)

Successive stages multiply distortion terms as more and more kinks end up in the transfer curve. Of course, this applies to the entire transmission chain from microphone to loudspeaker, with everything in-between.

Models are useful for finding bias points and the expected high-frequency response, but predictions of high-order distortion can be way off from the tubes you can actually buy. Low-order terms like 2nd and 3rd harmonic distortion may conform to the model, but I wouldn’t trust it further than that, not with real tubes. Think of the models as first-order approximations.

I was there. Thom’s amps did not use interstage coupling (at the time), so 8 watts is pretty much all she wrote. Frankly, I was kind of surprised ... I knew Thom’s NiWatts were maxed out for all they were worth, but audibility was surprisingly low. No flabby bass. No clipping as such, just really loud sound from an absurdly low-efficiency speaker. No screech, but not dull or muffled, either. A shocking amount of punch and bass slam. We both knew in advance that the combination of no feedback (thus no hard-clipping or saturation) and the split supply would protect the amp from things getting really out of hand.

Thom kind of did it on a dare, trying to get the fuse to blow. We both wanted to hear what the NiWatts, still in rough prototype form, did when given an impossible situation. The split power supply really, really came through. The 300B was getting hammered while the rest of the amp just sailed right through. It acts like an 8-watt compressor with surprisingly subtle action.

It reminded me of a little practice guitar amp, which if you have heard one, can get insanely loud, and retain its basic character.

P.S. Yes, Karna Mk II, or Statements, have independent regulated B+ supplies for input+driver and output section, with massive overdrive capabilities. That’s something the Mark I’s, the Statements, and the NiWatts all share. It makes them sound many times more powerful than the nominal wattage rating would suggest. Surprisingly, the 300B survives this gross abuse with no apparent damage, not something I expected.

Pindac has described the stage Don and I are at now: tuning the subjective balance ... there are a couple of nodes in the circuit where parts selection is quite audible, and we’re fine-tuning that.

I wrote an email to Don a couple of days ago that this topology is uniquely susceptible to parts coloration at the critical nodes. You get the same parts sensitivity in non-feedback SET amplifiers, but the much lower distortion of this circuit, compared to SET, exposes parts coloration more vividly. Fortunately, the right parts are available and are not super-exotic.

The circuit is inherently transparent, so there is almost nothing we can do to take that away, nor would we want to. But subjective tonal balance can be adjusted at the critical nodes. Surprisingly, the tuning has no effect on measurements, since topology, operating points, gain structure, tube loading, and bandwidth all remain the same.

Cloud Sessions asks a good question. Why does it sound this way? The best I can give is:

1) The circuit avoids both local and overall loop feedback, so there are no issues with hard clipping (transient overshoots in the FB network), stability margin (running out of gain and/or phase margin at high frequencies), or sensitivity to load reactance (which decreases phase margin and increases settling time after a large transient).

2) There are no differential stages to current-limit when one tube saturates or clips, taking the other tube along with it ... instead, the paired PP-mode tubes are functionally in parallel, helping each other out when the opposite-phase tube saturates or clips. Avoiding series-mode operation has a big effect on subjective dynamics. No SRPP’s, no split-load inverters, no long-tail pairs, no cathode followers.

3) Last but not least, each stage is individually optimized as much as possible for intrinsic linearity over the audio band. This is a matter of optimizing loads and minimizing the effect of Miller capacitance on the preceding stage.

Interestingly, Class D amps are free of Class AB transition artifacts, so there’s an entire class of coloration that just isn’t there. The big issue for Class D is nanosecond precision of timing for the pulse-width modulation (Class D is pulse-width-modulation, akin to FM, and not PCM), and insensitivity to reactive loads affecting the PWM modulator.

If a cool room is important, you want Class D (90% efficient or better) and speakers that are more than 1% efficient (a true Theile/Small 92 dB/meter/watt), as well as switching (not linear) supplies for all the other audio components. That’ll make no more heat than a transistor radio.

Class A vacuum tube is the exact opposite. Constant power draw regardless of signal, from silence to clipping, plus heater current of a few watts per tube, and a few watts of excess heat from the linear regulators. Similar to a 1963 all-tube color TV. My Panasonic 58" plasma HDTV consumes 500 watts, a bit more than a pair of Statements.

To put that in perspective, the same as four or five 100-watt conventional light bulbs, or the heat emitted by two people at rest. A light to moderate additional load on the A/C system, less than 1/10 of its capacity. Of course, if your A/C is running more than 50% to 70% of the time (a 50~70% duty cycle), probably not a good time of day to use the oven or turn on vacuum tubes.

Not going anywhere near an 833 or those Eimac transmitter tubes. No thank you. Those things light up a room with a dazzling white light and get serious, room-heater hot. They are designed for transmitter use, with forced-air cooling that ducts to the outside.

And if somebody chooses speakers that are 0.2% efficient, well, what did you expect? Class D or Class AB for you, as well as a dedicated AC power outlet that goes straight to the circuit-breaker box.

Very good article about how the brain distinguishes between reality and imagination. Since audiophiles are endlessly arguing about perception and reality, this is a useful read:

Quanta Magazine

I agree with Atmasphere. Some things cannot be accurately modeled. Put it on a bench, attach a dummy load, light it up, and measure square waves. Tune for nice-looking leading (and trailing) edges. Measure both low-level (below 1 watt) and also just below clipping.

Although a glance at a schematic might lead you to think it is simpler than a classical (Golden Age) PP KT88, the parts are more expensive. Way more expensive, just as 300B’s (as a group) are more expensive than KT88’s (as a group).

Another factor is sensitivity to parts quality. An amp with 20 dB of feedback (which is nearly all Golden Age amps) tends to wash out differences in the sonics of different parts. This is exactly what feedback is meant to do ... 20 dB of feedback is a 10:1 reduction of all sources of coloration. A zero-feedback amp, by contrast, reveals the sonics of every single part, particularly at critical nodes in the circuit. This raises costs compared to the PP KT88 equivalent.

Depending how you feel about the sonics of solid-state and feedback, you can travel a continuum between modern Class D, with sophisticated and complex feedback, to Class AB transistor or MOSFET with 20 to 40 dB of feedback, to Class AB push-pull pentode with 20 dB, to Class A with zero feedback. Each type sounds different and has different distortion spectra.

Comparisons between modern Class D and all-triode Class A are not absurd, despite radically different technologies. Class D and Class A both skip over the many difficulties with Class AB device switching, whether bipolar transistor, MOSFET, or pentode (each device type has different artifacts associated with the AB transition). The difference is Class D switches at 100 kHz or higher, with pulse width translating to signal level, while Class A is non-switching and like a preamp, fully analog from start to finish.

Show Update: Don, myself, and Spatial team will be at the Seattle show. What you will see and hear will be pre-production prototypes, sonically close to what we plan to manufacture. I can tell you neither Don, nor I, will tolerate any backward steps in sonics. I’ve lived with original Karna amps since 2003, and they remain my personal reference standard. The production amps must match or exceed that standard.

Prices, and names, are still up in the air. I call the preamp the Raven, and the amps Karna Mk II’s, or Blackbird, depending on my mood. Don calls them the Statement 300B’s. Spatial will probably come up with own names for the preamp/amp combo. Don and I are encouraging people to buy the set, but they are flexible enough to interface with industry-standard components.

At the expense of a small technical quibble, I don’t see Class D as entirely analog. Without the pulse-width modulator, it is a 100~500 kHz AM transmitter that transmits a silent carrier. The pulse-width modulator is what makes the whole thing possible ... the pulse widths are precisely (and I mean very precisely, down to parts per million) 50% positive and 50% negative, if the input is zero.

Deviations from exactly 50/50 alter the pulse widths (to 51%/49% or any other ratio) but the pulses themselves are rail-to-rail, and the output devices are purely switches. It is a modulation system like AM or FM, which are also analog, but it is a modulation system nonetheless. Without the PWM modulator, there is no signal that can get through the amplifier.

Class D has been around commercially since the early Seventies. The trick is extremely fast switching with no overhang, resistance to load reactance, and a (very) low-distortion modulator. An FM transmission chain that achieves less than 0.1% distortion is at the limit of the art, and PWM modulators inevitably have their own set of distortions. PWM is not inherently distortionless, any more than AM or FM. Yes, it can be transmitted, but it would be very sensitive to multipath and group-delay errors ... both would cause distortion. With both FM and PWM, small time errors translate into amplitude distortion after demodulation.

Interestingly, SACD/DSD, at the native 2.8 MHz switching rate, is a type of quantized PWM. Since the pulses of true PWM are variable width, they cannot be recorded on digital media. DSD uses dither encoding and digital feedback (noise shaping) to quantize the PWM pulse train into fixed widths and provide the closest approximation to true PWM on playback.

Snapsec, the unstated point of the Quanta article is that objective reality cannot be experienced directly ... the entire ear/nervous system/brain/mind system processes everything into mental images and impressions.

This is no little man watching a movie theater inside your head. It’s all signal processing, from the neurons in the ear to everything else, continually pattern-matching against memory, expectation, and emotion. Surprisingly, the brain can actually physically alter the hair cells in the cochlea with feedback mechanisms, altering perceptions right at the sensory level.

This is happening all the time ... there is no such thing as passive listening. For that matter, hearing and signal processing are still going on as you sleep, ready to wake up the rest of the fight-or-flight mammalian brain at a moment’s notice. Over the last 30,000 years, the powerful social experiences of speech, story-telling, and music gradually overlaid our mammalian brains, and made our species into the humans we are now.

This is why the notion of an Absolute Sound is inherently absurd. That’s like saying an Absolute Dream, or Absolute Taste. The entire hifi system is an illusion (and emotion) generator, and the quality of the illusion depends a great deal on the set and setting of the listener ... at that moment. Considering how vastly different our individual realities are, it’s amazing we can agree on anything at all.

Reviewers and consumers are at a great disadvantage in assessing the sound of a product, because they have never directly auditioned single-parameter changes. For example, a change in operating point (quiescent current) of 20% up or down, or a shift in topology in one part of the circuit. I grade these with a simple metric of:

1) No audible change, or at least nothing at threshold level

2) Different, but neither better or worse, just different

3) Worse (and how quickly is it noticed ... 5 seconds, 5 minutes, or an overall quality of discomfort or dislike)

4) Better (how quickly is this noticed, or is it a change in mood)

There are probably twenty or more points of adjustment in an amplifier or loudspeaker where these changes can be made. Some affect measurements, but most don’t ... they’re purely subjective. Also, conflicts can occur ... a better measurement may result in worse sound. At that point, something has gone wrong, and you need to stop and see where you have gone off-track.

We have to address what can be measured at the current state of the art and what can’t. Here’s just one example: for the the purposes of electronic design, nearly all modern capacitors are perfect. There is nothing to choose between them except voltage capacity and long-term reliability. Distortion is vanishingly small, at or below the threshold of measurement.

But ... in a high-resolution system, they all sound different. They are not neutral sonically. Mylar sounds different from polypropylene which sounds different than Teflon which sounds different from waxed or oiled paper. Metallized film sounds different than solid foil. In a vacuum-tube circuit, there are circuit nodes that actually exaggerate the coloration. Worse, DA, DF, self-inductance, or even price have little or nothing to do with sonics.

Perhaps worst of all, the notorious "burn-in" phenomenon where XYZ parts sounds really bad for the first 5, 10, 20, or 50 hours. With no change in DA, DF, or distortion measurements, and no plausible physical mechanism responsible for this. Anything that slow must be electrochemical, but what is it? Just a lot of hand-waving and supposition from the manufacturer, with no data to back it up. But plainly and clearly audible.

Not only that, some parts have essentially no break-in at all (paper and wax) while others can take 50 hours or more (polypropylene or Teflon). No explanation offered, no measurements, no underlying physical mechanism. Well, it’s not ghosts or psychic energy. It’s physics. But what physics? Nobody’s saying anything.

In loudspeakers and vacuum tubes, break-in is real, measurable, and the reasons are well-documented in papers going back to the Fifties. Caps? Nope. Why does copper wire sound different than silver? Again, no explanation. I accept this, but it is not satisfactory. Something is going on, and it is not self-hypnosis or expectancy effect. Often, the most expensive part sounds the worst, and the cheapo part sounds quite good.

I have had situations with break-in on my own amps, which are very revealing of parts quality, where the sound got better and worse on a two-hour cycle. While there was an overall upward trend, the up and down kept going on, with the cycle timing varying between a half-hour and two hours. At the twenty-hour mark, I finally lost patience and threw the questionable parts in the trash. That experience has made me wary of all parts that require subjective break-in.

My rule now is twenty to thirty minutes, tops. If the part can’t make up its mind in that time, in the trash it goes, no matter how expensive, or what the reviews say. I don’t want unstable parts in my system. Now if the fancy audiophile part requires a half-hour on a burn-in gizmo, fine, but I will regard it with some suspicion.

I think audiophiles and reviewers are too tolerant of this hokum. If it sounds bad for days on end, it is bad. Something is wrong. There is a design or fabrication error.

My concern for the customer experience might sound big-hearted, but actually it’s purely selfish. Neither Don nor I want ’em coming back. I am 100% retired, Don is thinking about it, and we both want the preamp and power amp to be reliable and good-sounding right out of the box. People tell their friends, etc. etc. So every part going in has to earn its keep in terms of reliability and sonics.

My concern is reliability and unknown, undocumented processes going on in critical components. This is just bad engineering. Imagine an expensive car where the horsepower and handling changed from day to day, and the manufacturer had no idea why. Cars are 100x more complex than an audio product ... the average car has 35,000 parts, 5,000 of which are moving parts, nearly all of which are critical to performance, reliability, and safety. A single-part failure can make the entire car useless.

What’s the capacitor manufacturer’s excuse? "This is how we’ve always done it" isn’t good enough. "We don’t understand what we are making" is even worse. There’s good engineering, making reliable products that people enjoy, and bad engineering, where mysterious things occur and nobody knows why. Regrettably, this is the situation for much of high-end audio.

For all I know, Don, Thom, and I might end up designing a capacitor conditioner for one of the cap companies. Back when I was working with Gary Pimm, he came up with a gizmo that did that ... pushing through 10 kHz square waves at 1/4 of the cap’s rated breakdown voltage. After twenty minutes of that, it either survived or not. Actually, nearly all survived, and they were "broken-in" for sure after that treatment. And it weeded out the parts that were going to fail anyway.

By contrast, using the cap in a normal circuit, with music stimulus, is barely tickling it. No wonder it takes forever.

There’s nothing quite like twiddling with a speaker crossover or the critical components on an amplifier. That brings the guesswork to an end. You know what increasing the slew rate of an amplifier sounds like ... it’s a pretty distinctive sound, actually. Likewise, if the tweeter crossover is screwed up, you hear tweeter distortion ... lots of it.

Which makes show-going by and large an unpleasant experience (although I love to meet people). I walk by a room, without going in, and I hear problems. Big ones. Multiple problems. And yes, I know how to fix them. But I’m not going to do it for free, and besides, most exhibitors really don’t want to hear unsolicited advice from a notorious busybody like me. So I just keep on walking. There might be one, two, or three rooms where the equipment is working OK, and I’ll spend most of the show there.

This might sound cynical, but seriously, I’ve been designing speakers since 1975. I can’t stand speakers with wonky response, multiple resonances, or gross distortion. And that’s most of the famous-name speakers, so they’re out. And they certainly don’t want to hear my wisecracks at first hand.

I got into tube-amp design around the early Nineties, and joining the staff at Vacuum Tube Valley was a fabulous learning experience, especially from Charlie Kittleson, a great guitar player, and John Atwood, an engineer’s engineer, having worked at Intel, Tandem Computing, and Apple. Sadly, Charlie passed on around 2000, but I’m still good friends with John Atwood. We’re both big fans of the history of technology, and can talks for hours on end about AC distribution systems in Japan and the arcana of NTSC, SECAM, and PAL color television. His current project is restoring a TeleType machine ... because why not? He’s also a ham radio enthusiast who builds vacuum tube rigs from scratch.

At any rate, those of us who do this for a living (although I am technically retired) can spot design errors pretty much immediately. My background is speakers and electronics, which also means I cannot ethically review anything, because all I would do is criticize, which isn’t fair to the manufacturer. Besides, the various designers all disagree with each other ... my designs are completely different from Atma-Sphere, or Audio Research, or Krell, etc. etc. We are all very opinionated.

Charlie reached out to me and asked if I would like to be Technical Editor of Vacuum Tube Valley, and I happily agreed. First class bunch of people, including the irrepressible Eric Barbour, one of the most notorious curmudgeons in the industry ... he made me look like a mild-mannered moderate. But all of them were, and are, great people.

Back when I was on the Editorial Board of Positive Feedback (a few years earlier), Eric sent in an article so outrageous that we couldn’t publish it, but it was the funniest thing about the biz I had ever seen. No, it still can’t be published, here or anywhere else. We’d get sued for sure. Eric is one of a kind.

Well, OK, there’s Harvey Rosenberg, but his sense of humor was much more gentle. Meeting him in person at the Consumer Electronics Show in the late Nineties was pretty memorable. He would drop the clown act and get quite serious, but if he saw someone he recognized, he do something outrageous right on the spot, then switch right back. The clown act fooled a lot of people, as it was meant to. He was a very sharp observer and knew what was going on.

Once you get away from the oh-so-serious gatekeepers, there are some remarkable people you meet in the biz. The fun thing about Eric, Harvey, and myself is we were outsiders, and we didn’t have to take it seriously.

Frankly, that was part of my motivation to design the Amity and Karna ... to show, by example, that things could be done differently. At the height of the SET mania in the Nineties, Harvey was the only one who (very strongly) encouraged me to follow my own path. That encouragement, from an old pro like Harvey, made all the difference.

I was proposing "Blackbird" as a product name for the amps, and Don said it is not just black like a raven, but "Intergalactic" black, which cracked me up (literal LOL) when he told me over the phone. Looking forward to meeting him at the show in person.

I can say the Amity, Karna, and the new amps sound nothing like other tube amps, whether SET, OTL, or conventional, or like transistor amps, either. I find them hard to describe, partly since I’ve been so close to them for the last twenty-five years (the first Amity came to life in 1998 or maybe earlier).

I’m trying to think of parallels ... maybe the first time you heard an electrostatic speaker? Kind of like that.

Yeah, no way of knowing what Seattle will sound like. Might be stunning, might be not-great, no way of telling in advance. Shows are unpredictable.

The Raven is something else, quite unexpected, really. Don did a superb job on it, no question. The Khozmo volume control, with a L/R balance control thrown in, is the extra deluxe touch.

45’s are wonderful tubes. By measurement with a spectrum analyzer, possibly the best tube of all time, with a beautifully clean harmonic structure. Which is why the Mark I Karna’s used them as drivers.

But ... NOS examples are now very rare, so it’s down to a few vendors as a specialty item. I am hoping they become more broadly accepted, though. They really are something special, and are much easier to drive than a 300B or 2A3.

Yes, folks that try the 46 (a direct-heated pentode that is often connected in triode) really seem to like it. By general consent, it sounds very much like a 45.

How different does a triode-connected 6V6 sound from a 45? Good question. The 6V6 was designed as a replacement for the 45 back in the mid-Thirties, and has very similar operating points (by design).

The Raven was the surprise for me. You see, I’ve been listening to the Karna amps for 20 years, now reborn as the Blackbird. Yes, there are differences, but what’s at the show is very very close to what I know so well at home ... maybe a teensy bit better, with a couple more improvements Don and I are hiding up our sleeve. But the show amps are 95% to 98% of what the shipping amps will be (yes, the production Blackbirds will be even better, Don and I promise that).

But I’ve never heard a Raven before. Hoo boy, is it good. Wow. I didn’t expect that. I knew what it does, of course, because I designed it more than twenty years ago, and Don refined it and took it a few steps further. But I never actually heard it, just other people’s impressions of it. I was quite surprised when I hear it for myself, in what is a very good room.

I’m not all that good at writing puff pieces, but the initial impression is speed and power, followed by vividness, and depth of tone, an impression that grows over the course of a few minutes, and after listening to a few different selections. We keep telling visitors, no, there is no 200-watt subwoofer (or two) tucked away in a corner, and yes, the new Spatial speakers are entirely powered by ONE Blackbird per channel, with passive crossovers (sample on display at the entrance to the room), and no electronic EQ, no active crossovers, no multi-amping, and no DSP. What you see is what you get.

As for source, Don and I are joking around about our favorite DACs. He’s all-in on the LampizatOr family, while I was dazzled by the Bruno Putzey-designed Mola Mola in a nearby room. I was expecting to be bowled over by the Holo May, but nah, it’s not for me. But there’s no way I’m going to cough up the $13,500 the Mola Mola costs, especially considering how fast DACs become obsolete ... five years or less.

Tube amps, by contrast ... the tech never becomes obsolete. For that matter, all of the tubes used in the Blackbird, and the basic circuit, were first designed in the 1930’s! Talk about not going obsolete.

What. A. Show!

Just got home ten minutes ago from Seattle. The most intense thing I've done since I went to the European Triode Festival in 2004. At least I didn't have to fly to Zurich this time ... only a 2.5 hour flight to SeaTac, and quick trip in the hotel bus. But man, the SIZE of the concourses in Denver ... combined with THREE gate changes.

Lots of fun meeting people at the show, and there were three or four really outstanding rooms, including [blush] our own. The Spatial team are great people and thoroughly professional, and several pre-orders are coming in. I have a feeling several vendors will be using a flight of Ravens and Blackbirds in their rooms next year.

I’m truly sorry you had that experience. The staff in the (fairly small) room rotated between Don, myself, and the four younger people at Spatial. (Don and I were the obvious old farts.) The Spatial people were manning the room (with a nonstop 4-hour playlist+personal requests) and Don and I were there to explain the electronics, tell various origin stories, and provide comic relief.

Reviewers popped in and out, and the famous Amir from ASR even showed up for a minute or two. Crowds came in surges, particularly on Saturday, when it was jam-packed. I was not aware that anyone had emptied the room for a reviewer, that’s certainly not what I saw when I was there. They had to take their chances like anyone else. There might have been special cases after 6PM, with private showings arranged in advance, but after-6 was mostly Social Hour for the exhibitors ... for example, I dragged all of the Spatial staff and Don to hear the Songer/Whammerdyne room, just down the hall from us. I’ve been old friends with Matt Kamna, the designer of the Whammerdyne 2A3 amplifier, so that was my after-hours excuse to get them to open the door ... "hey, here’s all my friends, can we listen for a while?" ... kind of thing.

I don’t know who the Famous Name Reviewer was. I wasn’t aware of anyone dropping by, much less closing the room for them. Our plan is to build a customer base (that’s you, folks) before we let any product get formally reviewed by the magazine or YouTube reviewer. Speaking only for myself, we’d like the Big Guns to give us a little breathing room, while we get the production machine going.

Don and I already have market identities ... we’ve been doing our respective things for thirty-plus years, with our names attached to what we do. We don’t need to build something that’s already there. So I’m not in a hurry ... at all ... to get in a review queue. People have heard about the Karna since 2003, well, here it is, in a room that’s about half the size of my living room, so please make allowances.

The day before, Don told me about Whitestix, his oldest customer, and said he was trained as an economist. (!) So as we sat down to a delicious dinner at the Thai restaurant across the street, I was curious and couldn’t help asking: "Chicago School or Keynesian? How do you feel about MMT? Freshwater or saltwater (terms of art to economists)?"

Economists famously disagree with each other, even more than audio designers, and the stakes are far higher, since their assessments steer entire economies. Worse, there is a general consensus in the economic community that economists have probably wrecked more economies than they have helped. When economists get it really wrong, people starve, are sent to gulags, or revolutions start. By contrast, all that bad audio design does is make people a little unhappy.

And things kind of went on from there. Not downhill, fortunately!

I’ll third Don Sach’s opinion. I’ll go further and say the Songer speakers are the best full-range drivers I have ever heard, better than the Feastrex (which I have heard), and remarkably, free of the upper-mid roughness that usually plagues drivers with whizzer cones. I did not think that was possible.

They sound more like a really good coax with an electrically perfect crossover. If money’s burning a hole in your pocket, get the dipole version. You didn’t really need that BMW, did you? Just drive the Lexus a couple more years. You’ll enjoy the Songers more.

Don’t get me started on the differences between Mahayana Buddhism and Vedanta Hinduism, and subtle (but fairly obvious) Buddhist influences in the Sermon on the Mount.

Here’s one for you: when I was a little tyke in Japan, no more than ten years old, I was preoccupied with one burning question: "Where Do Thoughts Come From?"

My parents deflected me, of course, because they had no clue, and probably thought it was a ridiculous question. But as an adult looking back to those days in midcentury Japan, it would have been pretty entertaining if this little brown-haired Western kid asked a Japanese monk that same question.

Our family went to plenty of temples, after all, and English was fairly widely spoken at the time. The monks would have went bonkers to see a little Western boy ask, in all seriousness, a Buddhist koan. They’d be running around and shouting, "Proof of reincarnation! Bring the abbot! Take a photo!"

Sometimes I think my Higher Self has quite a sense of humor. I may not get the joke, but it certainly does.

P.S. So, sixty years later, where do thoughts come from? If I met him now, I’m not sure I could honestly answer the little boy.

Coming back to Earth, I heard the Purifi Class D amplifier at the show. Quite good, and free of the usual transistor Class AB metallic coloration. The Raven preamp would make an excellent match for the Purify, since it is very transparent, has plenty of drive current, and is also an ultra-quality ground and RFI isolator. It might seem like an odd pairing, but for somebody that needs cool-running Class D power, the combo would probably work really well.

However ... the way the Raven is configured now, it can only drive one output at a time, either XLR or RCA. Well, technically it can drive both at once ... no harm is done to the circuit or any parts ... but the load is then unbalanced, with more cable capacitance on one side than the other. So the amp with the balanced inputs will receive a signal that doesn’t quite match at high frequencies.

This is probably true of other preamps with XLR and RCA outputs ... unless the literature comes out and says it has independent and isolated output sections for each set of outputs, you should assume the RCA is simply connected to one side of the XLR output. If that is so, then the added capacitance of the RCA cabling, plus the input capacitance of the RCA power amp, will unbalance the XLR output.

I should also mention when a transistor power amp is turned off, the input section of that amp is not isolated from the cabling, and the turned-off input transistors can create a nonlinear load for the preamp. Once it’s on again, feedback and power is restored, and the amp operates as specified. But when it’s off, best not to have another amp connected in parallel.

This is an important difference between tube power amps and transistor power amps. The input grid of a turned-off tube amp just sits there with a few pF of capacitance from the tube socket. Nothing else happens. It would take hundreds of volts to arc-over the tube, or break something in the input section.

By contrast, the base or gate of the bipolar or JFET transistor has a nonlinear input capacitance that remains when the power is off, and there is serious risk of damage if a 20 volt transient comes by ... this is a real hazard with conventional cap-coupled tube preamps and their associated turn-on thump. (The Raven is transformer-coupled and cannot pass DC transients, but there’s a small click when the VR tubes snap on, but small enough that even the most delicate Lowthers would not be harmed.)

The Raven is particularly well-suited to driving transistor amps because:

1) The transformers prohibit the transmission of DC pulses to RCA and XLR outputs, under all conditions, including total failure.

2) Unlike solid-state preamps, there are no DC servos to fail. There are no coupling caps on the output to store turn-on pulses, unlike cathode-follower designs. No muting relays are needed, or used, so the signal path is direct from transformer secondary to output.

3) All internal circuitry is fully balanced. The faint click when the VR tubes light up is only due to small residual imbalance (no more than 2%) in the 6SN7’s. Balanced operation alone reduces noise by 30 to 35 dB, then the VR tubes another 20 dB, and then 130 dB from the regulated power supply.

4) In addition to internal balanced operation, phase splitting at the input and output is entirely passive, not using discrete transistors, op-amp inverters, or split-load tube inverters. The custom transformers are precision balanced through 30 kHz, ensuring accurate phase balance for the power amplifier, whether it is solid-state or vacuum tube.

5) The input connections are floating, breaking any potential ground loops from the DAC or phono preamp. The transformers only respond to differences between signal pairs, with 80 dB or better common-mode noise rejection. The transformers also reject radio-frequency interference (RFI), passing only audio-band signals.

I wish I had heard the Dutch & Dutch 8C’s. The exact opposite approach of our room, but 100% valid: internally tri-amped with DSP for the side-firing drivers to create a near-perfect cardioid radiation pattern. Digital-in is limited to 48/24 PCM (higher-rez content is accepted but internally downsampled), and the triple internal amps are Class D with 400W each. I was reading Amir’s ASR show report (third section) and the Dutch & Dutch exhibitor was measuring 15 Hz during setup.

This might be anathema for some readers, but I like the massive engineering effort that must have gone into this bookshelf speaker. DSP to shape the polar response with side-firing drivers is not trivial, and from the reviews, they actually tamed the hotel rooms that were driving the rest of us nuts. Of course, with DSP tailored for each driver, the speaker would be flat as a pancake, not a bad thing.

For analog fans, the prospect of converting every precious LP on-the-fly to plain old 48/24 PCM is appalling, and for high-end DAC fans, the standard-issue ESS or AKM converters hidden inside the Dutch & Dutch are probably not acceptable. So you pick the approach that you like.

Surprisingly, Amir, of all people, liked both Spatial rooms. Did not expect that. The real invective on ASR was reserved for the absurd cables in the ballrooms ... to be honest, have to agree there. The Anticables we were using worked just fine, and were unobtrusive, as a cable should be.

I’ve been a little quiet since I got whacked by Covid on my return to Colorado. Got an intense sore throat on Tuesday, tested positive on Wednesday, and started Paxlovid that afternoon. Doing better today ... actually, mostly better, and my mind is finally clear again. The metallic taste Paxlovid is famous for hasn’t been too bad, and it is fast-acting. (I tick three of the boxes for "high risk", so Kaiser Permanente of Colorado prescribed it right away.)

Strongly suspect the Denver or Seattle airports as the culprits. Total zoo in both places. If I go next year, I will ask for airline assistance getting to and from the gates. Two things that did work well were TSA Pre-check and Express Bag Check-In.

Don continues to evolve the Blackbird. Full-size chassis, some excellent parts choices, and bringing elements of the much-loved Raven to the Blackbird. The people on the Spatial Audio pre-order list are going to like them ... a lot.

Matt Kamna, the designer of the Whammerdyne, is an old Tektronix guy I’ve known since the late Eighties. We were at the second meeting of the Oregon Triode Society, and were both so impressed by what we heard at the meeting we set aside a project on a transistor amp .... for the simple reason it is much easier to get good sound than with transistors. Tubes are simple to work with and sound great, right out of the box.

Back in 1996, when I lived in Portland, Matt physically built the first Amity amp, although it was 100% my design from the ground up, and my first amplifier project. At the time, Matt went down the path of gigantic and extremely rare Western Electric power tubes in the transmitter class, I moved to Washington State, and then to Colorado.

Matt approached me a few years back and wanted me to write the user manual for the Whammerdyne 2A3 amp. He had definitely moved in a new direction. Matt’s an old friend and I didn’t mind some spare change, so yes, I agreed. The Whammerdyne 2A3 has some interesting user setup settings, and I had fun coming up with various names for the internal features of the Whammerdyne. Yes, I know what’s inside, in order to write the fairly complex user manual. (NDA’s were signed, etc.)

Internally, the Whammerdyne 2A3 couldn’t be more different than the Amity, Karna, Raven, and Blackbird (which are all closely related). Matt’s design is almost the anti-Blackbird, but what can I say, it works really well. I know from personal experience these things don’t design themselves. They look simple, but I can tell you, they are tricky to get right.

It was delightful to hear the latest Whammerdyne at the 2023 PAF. It physically looks more or less like the original, but sounds much more refined and elegant. It’s clear they’ve been working on it, and good for them for doing it. And it works really well with the Songer Audio speaker, another Portland company.

It’s interesting the SET folks are all-in on the single-ended concept, and have been since the early days of Sound Practices magazine. I certainly fell in love with direct-heated tubes, but felt from the beginning that the colorations of push-pull amps (compared to SETs) could be solved with a little creative thinking, and a return to the well of Western Electric research. I credit Vacuum Tube Valley and John Atwood for pointing me in the right direction.

Since I’m on a roll (and thankfully nearly out of the woods on Covid), let’s talk about PP coloration. It exists, it is real, and if you spend time with top-quality (not junk) SET amps, the absence of that coloration is wonderful and refreshing. Compared to Golden Age amps, everything is so clear, so open, so natural with wonderful tone colors. (Again, you never hear this with junk SET amps, which are murky and dark.)

The magazine conventional wisdom would tell you that clarity and beauty is "euphonic coloration". That’s complete horse****. Euphonic colorations can’t add detail, resolution, more depth, and more in-the-room presence ... colorations can twiddle with subjective tonal balance, and usually adds mush, murk, or grain. They don’t remove it. Build your own amps and you find this out right away (I can see Don nodding his head). The magazines have had this wrong for forty years.

Back to PP coloration. The mainstream "alternative" view is that it is inherent to to all PP, so just build SET and forget about it. Entirely aside from power considerations, SET has its own universe of colorations unique to SET, and they can be quite severe. What I call "junk" SETs sound like old antique equipment. So designing a really good SET isn’t quite as simple as it appears at first glance. At the top of the performance spectrum, it gets fiendishly difficult, with costs and complexity reaching into the stratosphere, and all charm and simplicity lost.

I’m in the small minority that feels the colorations of PP are very real and not imaginary, but can be solved. The Golden Age amps of the Fifties and Sixties, and the modern copies from the big name vendors, all have remarkably similar circuits, boiling down to three types (Williamson, Mullard, and Dynaco, with a handful of variants). These were adopted because they were well understood, responded well to feedback, and were cost effective at the time. Even Marantz and McIntosh were part of the watts-per-dollar race, which only ended when the Crown DC300 and Phase Linear 700 came on the market, putting all tube amps in the shade. From then on, if you wanted Big Watts, you got a transistor amp. Still true today.

But Golden Age is not the only way to build a tube amp. There were other, pre-war circuits, before the Williamson wiped out everything in 1948. The pre-war "floating paraphase" phase inverter ... not as perfectly balanced as later circuits, but more powerful. Transformer coupling, which passively inverts phase, but demands ultra-performance transformers, and also rules out global feedback. And other methods.

The prewar era had a lot of interesting byways and interesting circuits, which all disappeared by the 1950’s. And they do sound different, and get away from the 1950’s monoculture which dominates tube audio. The tubes don’t care; they cheerfully work in any circuit, provided you pay attention to operating point, loading, and stability (part of any amp design).

At any rate, when you stop using Golden Age circuits, the "PP sound" changes. It’s no longer a thick blanket that lays on the sound. It might be a brand-new coloration, or might go away. That’s where the fun starts.

Once you hear the difference an IT makes, you realize: "Aha! So that’s what RC coupling sounds like!" And then you start hearing that coloration everywhere in mainstream audio equipment, and can never forget what it sounds like.

This is an experience no reviewer ever has, even if they have $250,000 systems. Because why would they? Everything has been handed to them on a platter. They don’t know what’s inside the pretty box... it could be elves doing magic tricks for all they know.

What drives this exploration is curiosity. What will this sound like? Well, you don’t know until you try, and why take somebody else’s word for it? That’s no fun.

Grounding in the physics of tube function helps, because you get a feel what the device "wants to do". Pilots have a saying, "the airplane wants to fly". If you’ve ever been at the controls of an airplane as it lifts off the runway, you feel it. The vibration stops, the wings lift, the ground falls away, and you keep moving forward, a creature of the air, not the ground. You are now in a different realm.

Tubes are the same. They want to amplify. It’s what they’re made for. Our job is to get out of the way and let them do that, so we get into load-lines, the most linear region, what happens as the current dips downward, staying away from trouble spots, etc.

I made a little post a few days ago on Facebook, quoting a meme from my Ukrainian friend, Misho Myronov. When asked what his amplifiers amplify, he replied: "Happiness! My amplifiers amplify happiness!"

All of us who create tube amps get this. We do it because it makes us, and our friends who listen to them, happy. In the comments to Misho’s post, I added:

"I was just talking to Karna, and said, an amplifier is like a dance partner for the speaker. If the partner is dull and disengaged, the speaker will be bored and not interested. If the dance partner is lively and fun, the speaker will light up and come alive."

All of us who have solved a difficult technical problem, or just got a better amplifier, will hear the speaker really come alive and surprise us. That’s the happiness.

IT coupling dates back to the 1920’s Atwater Kent and radios of that vintage. Like field coil loudspeakers, it is the oldest form of coupling of all. But transformers have always been labor-intensive and expensive (going right to the beginning of electronic amplification), so when coupling caps became practical in the late Twenties, ITs mostly went away, although Western Electric and high-end radio builders used them as late as the end of the Thirties.

What pretty much ended them in all applications was the universal use of global loop feedback, with the landmark Williamson of 1948. You can wrap feedback around one set of coupling caps and an output transformer, but two transformers are out of the question. So IT coupling was moribund until zero-feedback amps saw a comeback in the early Nineties.

In the mid-Twenties, the only signal source was AM radio. Electrical phono pickups were just coming on the market, and movie sound was brand-new and experimental. By the mid-Thirties, movie sound was universal, with a bandwidth topping out at 8 kHz. Electrical phonographs could reach 8 kHz, but users often used "scratch" filters to soften the sound of noisy shellac 78’s. AM radio reception could go higher, but people often used the narrowband setting to get rid of interfering whistles from adjacent stations. The only truly wideband source was the Armstrong "Yankee Network" of FM stations in the 42~50 MHz band, which was limited to a few stations in the Northeast. These were the only FM stations in the world, and could be received by the high-end radios of the day.

It was only in the postwar years that wideband (30 Hz ~ 15 kHz) sources became widespread, with LP records in 1948, pre-recorded tapes in the mid-Fifties, FM broadcasting in the postwar 88~108 MHz band, and 70mm movies with magnetic soundtracks in surround sound. By then, all amplifiers were medium to high feedback designs, and used RC coupling throughout.

Modern wideband transformers were in studio use from the early Fifties, and the triode designs of the early Nineties opened the market for more unusual products, such as interstage transformers. True, it’s a 1920’s technology, but they didn’t have modern bandwidths back then ... the recording technology was unforeseen and decades in the future. What we hear now, with our ultra-wide band, ultra-low distortion sources, is a brand new sound, running through new-tech devices.

Hi, Lewinskih01!

You bring up two different approaches to system building. One is taking full advantage of modern multichannel DAC chips (8-channel is a common default size) and letting DSP do the heavy lifting. Taking it a bit further, tuning each amp for its own driver, rather than using a AV multichannel amp of marginal quality.

It depends on subjective priorities. Does the speaker need DSP to reach its full potential, and is DAC coloration small change in the overall scheme of things? Can’t say I blame you. Speaker colorations are obvious and gross, and DSP is the most direct and powerful way to attack them.

I have friends who own Altec Duplex 604’s and they don’t like it when I tell them the only way to straighten a 604 Duplex out is DSP ... no physically realizable crossover can fully correct it. Otherwise, you learn to live with the coloration, as Lowther owners do.

DAC coloration ... hoo boy, let’s jump into that rabbit hole, shall we? I feel most audiophiles can barely hear DAC coloration for modern delta-sigma designs ... and measurements are essentially perfect, far exceeding the 44.1/16 Red Book PCM specification. If a modern AKM or ESS converter with a circuit board full of op-amps is perfect for you, you can save big money, and jump on the DSP train with confidence. Do not pass GO, collect your $200, and enter the wonderful world of DSP. Amps built to taste are entirely optional.

Only a few people can hear differences between modern converters, and if you can’t, don't feel bad, you are part of the vast majority of audiophiles. Just buy a $700 Topping or S.M.S.L. and explore DSP. It’s what headphone jocks do these days. No shame in it.

Differences between DACs are weird and extremely subtle, and frankly you have to train to yourself to hear them. I can’t honestly recommend audiophiles go down this rabbit hole. It’s extremely expensive to pursue and full of deliberately confusing technobabble from slick marketers. Maybe not as bad as cables, but still pretty bad. Trust nothing when it comes to DACs, no matter how famous the name, or how glowing the review,

I was shocked and disgusted I could hear what sounded like "big" differences between my antique Monarchy DAC, with its Burr-Brown PCM-63K converters, and the latest confection from the Berkeley DAC (which any Topping will take to the cleaners these days). I also have the exaSound DACs which are ESS based.

I find DAC chasing neither fun nor enjoyable. The best are insanely expensive, and they go obsolete really fast. I might love the $13,500 Mola Mola I heard in the Songer Audio room, but the Mola Mola won’t be worth as much three to five years from now. DACs should be thought of as consumables that depreciate the moment you buy them.

Amps and speakers ... ah, now that is good value. Buy or build a good tube amp (and that certainly includes the $5000 Valhalla from Spatial) and it holds its value indefinitely. Similarly for speakers. The good ones cost more because the parts themselves cost more, and it takes serious design work to make them perform.

Well, enough of the rant on DACs. Addressing the question in the post by lewinskyh01, what does a really good tube linestage bring to the table if the DAC can directly drive the power amps?

A sense of ease, dynamic impact, and sometimes more vivid tone colors. How? Partly better cable drive, partly signal conditioning, scraping off RFI and noise induced in the cables. On paper, op-amps can do an amazing job driving a cable, in practice, not so much. If the preamp passes a quality threshold, yes, it can improve the signal compared to a direct connection to a DAC. Found that out the hard way with first Amity amp.

Love ’em or hate ’em, DACs have gone a long way in the last thirty years, and continue to evolve pretty quickly. The internals of the AKM and ESS converters run at 90 MHz, with stupendous processing power. It’s what makes 4K TV and digicams possible.

That kind of speed makes up for many sins, and lets the noise-shaping algorithms operate much, much better than earliest days of SACD and single-bit MASH converters running at 2.8 MHz. In a lot of ways, it makes the endless upsampling discussions on the forums moot, since the internals are upsampling everything to 90 MHz anyway. Might as well let the chip do it, rather than play games in Roon. (Although converting PCM to high-rate DSD forces the chip to use different algorithms, which will definitely sound different.)

It is a consciously retro decision to use antique Eighties-vintage Philips TDA1541A converters, or late-Nineties Burr-Brown PCM-63 or PCM-1704 converters. Those are true once-through flash converters, with no signal processing or noise-shaping involved. But the least significant bits are kind of marginal, since it took R2R to the limit of what can be done with laser trimming and ultra precise fabrication. Nowadays, speed and good algorithms are the answer.

Which leaves the current-to-voltage converter as the last domain of audio tweakery. Op-amps are way, way better than the 1979-vintage 5532/5534 from Philips/Signetics, but you still find these antiques in consumer DACs. That’s probably where tuning happens in modern DACs, since there is little left elsewhere in the design.

And if you want to "sweeten" things, do it in the power amp or speaker. Much easier to tweak. I think making records sound like ultra-quiet, ultra-precise digital, or making CDs smoothed-out and "analog", is taking away from the strengths of each medium. LPs sound like LPs, and PCM sounds like PCM.

PCM to DSD256 is fair game, though, so why not? It’s what my Marantz SA-KI SACD player does to incoming PCM (it has S/PDIF and Toslink inputs), and an interesting "alternate view" of PCM sources.

Due to package heat-dissipation limitations, most op-amps operate in Class AB. Now, they have a stupendous amount of feedback, and it takes a difficult load to excite the AB transition, but it’s still there. Speed is the friend of op-amps, so the high slew rate versions (more than 20V/uSec) often sound cleaner and smoother than the slower versions.

It is difficult to build a discrete solid-state circuit with distortion specs that exceed an integrated circuit, but it can be done, and they are often seen in the pro recording studio world. These do operate in Class A, and that is always mentioned in the sales literature.

I'm a bit surprised that ARC uses op-amps in their "Reference" series. Why not just buy Topping or SMSL and get even better performance, or if you insist on made-in-America, Benchmark, who live up to their name in performance standards?

Why buy an high-end audiophile component, with audiophile pricing, made from off-the-shelf $5 parts. What's the point? I don't see the value proposition. Now, if there are a zillion discrete transistors, and it does 1000V/uSec and delivers 500mA mA into a 300 pF load, that's insane, but still in the realm of engineering possibility.

Tube gear costs a lot because transformers and vacuum tubes are inherently labor-intensive, and the parts are not inserted on circuit boards with pick-and-place machines. I'm one of those madmen who think zero-feedback circuits are interesting, and I like tubes. Nelson Pass is your man if you like zero-feedback JFET/bipolar transistor circuits.

If you are more sensible, read ASR reviews, ignore the comments section, ignore the single-dimension SINAD number, and look at the noise floor of the multitone IM distortion graphs. That is the true wideband IM distortion, and multitone is the most severe test of the entire circuit. The Shenzen group of manufacturers have some really good engineers, and it shows in the IM distortion measurement. From what I can see, Bruno Putzey and the Shenzen guys are at the top of the game, if specs are at top of the list. They also know how to "tune" a power supply to get a subjective result.

I guess the micro-rant above is about audiophile components where the case costs more than the parts in the audio circuit. It doesn’t make sense to have a $500 case housing $50 worth of parts (in the audio circuit), unless the look is the main reason to buy the product.

I didn’t mean to imply any connection between Bruno Putzey and the Shenzen group. Bruno works in Europe, and designs top-class DACs and Class D amplifiers, which are inherently complex and very hard to get right.

The Shenzen engineers have slowly but surely improved their game, and the Chinese have been quietly building complete OEM products for high-end European and American famous-name manufacturers. Many deluxe and high-end raw parts are built right there in China, so they don’t have to go far to design and build their own high-end components, Whether you love or hate ASR, they have uncovered some terrific Chinese products.

On the international scene, a lot of truly remarkable products are coming out of Eastern Europe these days. Some real talent there.

The Karna Mk II/Blackbird is bit by bit evolving towards the original Karna, but without the madness of a four chassis design. Having a separate chokes and power transformers for #1 B+, #2 B+, and the filament supply gets really heavy and awkward. Don’s monoblock approach is much more sensible, and more important, he has real-world experience of what is reliable in the field.

I design things as a thought experiment, just to see how it works out. About one design in three is a flop, and gets abandoned. You have a sound in your mind, and wonder if the real thing will sound like you imagined. Sometimes it does, sometimes it doesn’t. You never know in advance.

The Raven and Blackbird were, and are, thought experiments to explore what minimum intrinsic distortion would sound like. Zero loop feedback, and zero local feedback, with all cathodes fully bypassed. Balanced, but not differential, with passive transformers doing the summations and cancellations.

It is not SET, which require skillful arrangement of various colorations and very, very careful component selection. But it still requires careful selection of components because there is no feedback to minimize and wash out colorations. If XYZ tube has a certain sound, well, that’s what you’ll hear. If XYZ cap is imposing a coloration on the cathode circuit (which is a very sensitive circuit node), yes, it will be audible.

A big difference between the solid-state world and vacuum tubes is capacitance. Capacitance with tubes is essentially linear, aside from Miller capacitance, and even then, the delta in the capacitance is very small (no more than a percent or so). The transit time through the circuit is constant, regardless of signal. Part of making the transit time constant is passive (not active) phase inversion and summation.

Solid-state capacitance is known for varying with current and temperature, so it pays in transistor circuits to get the (nonlinear) capacitance to the lowest value possible ... if it can’t be linearized, get it as close to zero as possible. Modern transistors are much faster than previous decades, so this really helps. Build a very linear video circuit, and many problems are solved.

A bit of background on cost to the consumer: if a company isn’t charging Bill of Materials (BOM) cost times four, they won’t be around very long, one or two years at most. This rule-of-thumb has been true since the Fifties (for hifi manufacturing in North America).

Not true for cars, of course, since that is a hyper-competitive, extremely price-sensitive industry that has enormous capital barriers to entry. In electronics, the Chinese are able to shave it down to two to one, most likely due to a wide range of hidden subsidies that favor exporting.

So a smart DIY’er can indeed get serious high end for medium (not low) cost, partly by pricing their labor at zero. But even a very experienced DIY’er is going to find that building a Blackbird from scratch is the same as the price of a good used car, setting aside labor and debugging time. I know several people who got stuck halfway through building a Karna and wanted many hours of my free help completing it. No, that’s not how it works. You want a Heathkit, go buy one. If you can design and build an amp from scratch, more power to you! Have fun! Be glad you don’t have to use a slide rule any more, like the bad old days.

(Yes, I have used slip-sticks. They are no fun. You’re lucky to get 2% precision, and you have to do the calculation twice because it can be off by a factor of 10 or 100.)

Back to circuits. A differential and balanced circuit are not the same. A differential circuit has a current source or high-value resistor in the common cathode (or emitter) circuit, which is why they are called "long-tailed pair" in the literature. This forces differential operation, but has a limitation because the two tubes (or transistors) are effectively in series. If one device cuts off (impedance goes to infinity), then the other device is hard-limited to 2X the quiescent current. It can never go further, because the long-tail or current source hard-limits total current to both devices.

By contrast, a balanced circuit, without a long-tail or current source, can turn on the "on" device as hard as it likes. That can be as high as 5X the quiescent current or even more. It effectively slides over into Class AB if it needs to, unlike a differential circuit, which will hard-clip if too much current is demanded. The phase splitting is done by transformers, not a long-tailed pair.

By the way, if you are looking for value, you really should audition the Valhalla from Spatial. It took on every other high-bucks big-name tube amp at the show and came out ahead, often by a good margin. It is a seriously good amplifier at an absurdly low price.

Now, if you are looking for 100 to 200 watts of tube power ... hate to break it to you, but paralleling arrays of power pentodes does NOT improve the sound. Rule of thumb for PP tube or transistor: no more than two devices if you care about quality. Once you start paralleling arrays, there’s always just a bit of mismatch to trip you up. And that’s just DC matching, which is completely separate from matching transfer curves (AC matching). That’s a lot harder, and there’s always the issue of tubes drifting apart as they age.

The other issue with arrays of pentodes is the grid capacitance for the power tubes is multiplied, which then requires high-current cathode followers, or separate power tubes as drivers. This gets into No Fun territory as the design complexity multiplies, all just to squeeze out a few more watts.