300b lovers

From my perspective as an amp designer (not as a consumer or reviewer), the Sweet Spot in tube amps is from 3 watts (Class A SET) to 60 watts (Class AB PP pentode). These are all simple circuits with an emphasis on sound quality and reliability.

If you MUST have 200 watts, consider combining a modern Class D amp with a preamp like the Raven. The new Class D amps don't have the irritating and fatiguing Class AB sound, while a good tube preamp lends the sound some charm and likability.

I’m always interested in boundary conditions ... what happens when the amp leaves its happy place and a surge of current or voltage is required. Does a circuit saturate and hit the wall? Does a transistor fail? Does it store charge and "stick" for a few milliseconds? How smooth is the transition in and out of the Bad Place?

I mention this because speakers are badly behaved much of the time. They store energy for tens to hundreds of milliseconds, then throw it back to the amplifier. The feedback network might, or might not, keep correcting this, but the error overshoots can be very large and can saturate an input section.

Many power amps do not accept boundary conditions gracefully. Not just the output section, but the driver as well. Driver transistors fail when SOA is exceeded by transient reactive loads (failure to accurately read a SOA graph almost bankrupted Audionics). In tube amps, drivers can’t push enough linear current into the Miller capacitance of the output tubes. The voltage-amp section of a transistor amp can’t charge the dominant-pole capacitor fast enough, resulting in slewing.

These are all boundary conditions, and they are audible not just when they reach 100% failure, but well before that, when nonlinearity just begins. The previous point about Class A operation in a differential stage still holds: what happens when more than 100% of the current programmed in a current source is exceeded?

This is a boundary condition problem. When current is exceeded, what next? What’s after that? Does anything fail? What does current clipping look like? Are there any energy storage mechanisms that result in "sticking", a well-known problem in solid-state power stages. If sticking happens, how long does it take before it gets unstuck? In milliseconds?

The approach in the Blackbird/Karna does not use current sources, nor differential stages. Each side is parallel, but in antiphase, and all phase splitting, and re-summing, is done by passive devices, which do not have slew limitations. The 1:1 interstage transformer makes sure that recovery from A2 grid-current events happen in microseconds, not milliseconds.

Overload happens in the tubes, mostly in the output section, and the overload condition is not affected by local or global feedback, so the overall boundary characteristic is that of a (very) fast-recovery limiter/compressor. There is no hard boundary between Class A, where it remains most of the time, and A2, AB, or AB2, depending on current or voltage demand.

During the development of the Karna amp, I was in a kind of perverse mood, so I was curious just how much abuse the circuit, and the tubes, could take. I set the oscillator level so the scope display was just below clipping, around 20 watts, and the 8-ohm test load was nice and warm. I increased the drive frequency beyond 20 kHz, and as transformer gain started to fall off beyond 50 kHz, I just increased the input level to keep the output at a steady, undistorted 20 watts. Because why not?

I finally lost my nerve at 500 kHz. The scope display was still an undistorted 20 watts, with no sign of triangle waves or flat-topping, but playing around with an AM-band transmitter (with 500 volts inside) was asking for trouble. I wasn’t trying to kill anything, but sooner or later some part was going to fail. (If any of you customers try this stunt, yes, we will void the warranty, so don’t do this. Ever.)

Not many transistor amps would survive full power at 500 kHz. Some would, some wouldn’t. It’s an absurd test, with no relation to audio use. But it’s interesting to know the development prototype survived it. No, I would never do this to the current production model, and don’t you guys try it, either.

 

Quick recap: actually, vacuum tubes are far from saturation when set to normal bias points. Look at a 300B, or any other power tube. Normal quiescent bias is set between 60 and 85 mA, if Class A operation is desired. If Class AB is desired, 35 to 40 mA is more typical. With 400 volts from cathode (or filament), that's a steady-state plate dissipation between 14 and 34 watts, well within the 40-watt rating.

But that's nowhere close to the peak current emission of the cathode. I've measured 250 mA from a generic 300B, and the exotic European 300B's can slam out nearly 500 mA (transient). The only time I've ever seen a 300B current-limit around 80 mA were some particularly weak Chinese tubes from the mid-Eighties ... they sounded and measured pretty bad, and were near-defective. Other vacuum tubes are similar; the recommended quiescent currents are set by plate dissipation limits, not cathode emission maximums (which are left unspecified). Transistors will melt the internal copper links, but damaging the cathode in a vacuum tube is really hard to do unless the tube is operated with no B+ present.

It's transistors that have Safe Operating Area (SOA) curves that are log-log in both current and voltage (with an additional time dimension), not tubes. The current saturation mechanisms are totally different and have nothing in common.

Unlike transistors, vacuum tubes have very large areas of peak current emission that are left untapped by most circuits. Of course, plate heating goes up when these areas are explored, but unlike transistors, tubes do not fail in milliseconds (this is shown in the SOA curves of transistors, and must be respected). It takes sustained abuse, over many seconds, before mechanical deformation dooms the plate.

I think Ralph will agree that Class AB operation is not "false". In Class AB, one device cuts off (goes to infinite impedance and conducts no current) while the opposing device goes to a large multiple of the quiescent current. In conventional Class AB transistor amps, the idling current is a tiny fraction of the peak current, and in Class AB tube amps, it's still a small fraction.

Let's look at what happens in pure differential circuit, either tube or transistor, with a current source setting the quiescent current. This circuit must always operate in Class A. Unless something fails, the current source will always deliver the programmed current ... that is a hard limit that cannot be exceeded under any condition.

The late Allen Wright actually built a PP 300B power amplifier that had a current source under the pair of VV52B's (massive Czech power tubes). He stayed at my house during one of the VSAC shows, and we compared his amp to my early version of the Karna (which has bypassed cathodes and can operate in Class A, Class A2, or Class AB, or even Class AB2, depending on current demand). Allen's output stage was true differential, and true Class A, with a powerful solid-state current source running around 160 mA (if memory serves ... this was in 2003 or so).

The two amps sounded completely different. That's when Allen, and I, realized that differential, and balanced, are not in fact the same. This is a common illusion, a hangover from the Fifties. The question is what happens when one device cuts off.

When this happens in a current-sourced differential circuit, the "ON" device can never pass more than the total current programmed in the current source (by definition). That's a hard limit. It is a brick wall. The circuit, as a whole, will always pass whatever the current source is programmed to do ... no more, no less, always the same. This is why this circuit is seen in the Mullard topology as a low-power, medium-voltage phase splitter. Allen, as a big fan of differential circuits in Tek scopes, took it all the way and used it in a power stage.

This is quite different than a Class AB, or conventional Class A, power stage. Whether cathode or fixed-bias, current flow through the output pair is dynamic. IF (a very big if here) the output tubes were distortionless, perfectly matched, AND never voltage-clipped or driven into Class AB, yes, it would behave the same as a current-sourced pure differential stage. Only then are they the same.

But we don't live in a world of Platonic ideals. Tubes are not actually the same as the tube models, they are not perfectly assembled in perfect factories by robots, loudspeakers have odd ideas when they want lots of current, and bass drivers in particular are notorious for nonlinearity and very long energy storage .., all of which affects output stages.

So a power amplifier must deal with speakers as they are, not as we want them to be. So peak current excursions can be accommodated when necessary, without the amplifier grossly departing from basic design assumptions. The loudspeaker conforms to Theile/Small equations most of the time, but both Neville Theile and Richard Small warn us that these are only small-signal approximations. They are not valid once the voice coils start to move significantly. Speakers are only linear on average, not all the time.

My goal with Class A output is to synthesize a fixed output impedance that remains constant with real-world loudspeakers, which I have been designing since 1975. I know how awful speakers are. Most power amps use 20 to 50 dB of feedback to synthesize a perfect voltage source, and they do a pretty decent job of it. With zero feedback, the best I can hope for is a fixed, moderate-value equivalent resistor, about 2 ohms or so, which a low-Q vented or closed box speaker can deal with. And an output stage that does not have a hard current limit, but soft-clips in both voltage and current, without requiring protection circuits.

Glad you’re enjoying it! As I might have mentioned earlier, in person or in this forum, I’ve been listening to the Karna/Blackbird amps for twenty years. It’s just what an amp sounds like to me.

But the PAF is the first time I ever heard what my slightly older design, the Raven preamp, actually sounds like, and it was quite a revelation. Ultra-fast, vivid, and of course, dead silent. Many thanks to Don and the Spatial team for taking it to the next level, rolling in a lot of good ideas of your own. This has been a very enjoyable collaboration, and I look forward to more.

I should also thank Don for suggesting the best way for the preamp and amps to work together, as a unit. This is the special XLR direct-in mode for the Blackbird, bypassing and disconnecting the input transformer. This lets the Raven output transformer do the phase-splitting chores with its matched split secondary windings, one of the charms of custom iron, made for the purpose.

Let me tell the story about how misreading a graph almost bankrupted a company. This is a true story.

Our chief engineer, a Tektronix veteran, designed a power amplifier called the Point Zero Three, or PZ3 (hey, I didn’t name it, OK?).

A 100-watt/channel transistor amp. It measured great, sounded so-so, but had a little fault ... it blew up without warning, and worse, took out all the power and driver transistors, scorching the circuit board as it went. Take out the circuit board, replace the power transistors (all of them), and put in a new one. Repeat as necessary.

There were days when more came back than were shipped out. Obviously, this couldn’t continue. Word was getting out, and a failure rate approaching 50% is unsustainable.

Our new engineer, Bob Sickler, looked more closely at the SOA curves for the driver transistors. Bob told me these curves are intentionally hard to read, and it usually escapes notice that both voltage and current axis are log scale. Not only that, a straight load-line is assumed by most engineers, but with real loads, that line opens into an ellipse. Once that ellipse touches the no-go line, and stays there for more than 10 milliseconds, boom!

The chief engineer, despite being an old Tek hand, had missed this tiny little detail. It turned out the driver transistors were undersized by a factor of three, so we had to parallel them in a 3-stack with individual emitter resistors to have a stable amplifier. The fix worked; but we had to recall every amplifier in the field and update the circuit board with the new driver stack. It wasn’t cheap, but it stopped them coming back, and they kept working.

This incident resulted in Bob Sickler, the new guy who saved the company, getting permission to design a new amp of his own, which became the Audionics CC-2, their most successful, reliable, and best-sounding product. They sold more than 1,000 with a failure rate of less than 0.3%, the lowest in the industry. The CC-2 was their most profitable product.

All this happened because one part, a driver transistor, had poorly understood behavior in a boundary condition. Ask for too much current for too long, go past the SOA boundary, and blooey! A scorched circuit board with mostly shorted transistors, all thanks to DC coupling propagating the single point of failure through the entire output section in less than a second. Engineers love DC coupling, but it can propagate failure very, very fast, in less time than it takes to jump across the room and turn it off.

That experience is why I am wary of dismissing boundary conditions. Poorly understood boundary conditions can destroy a product, destroy consumer trust, and take down a company. All from not reading a data sheet carefully enough.

Most audiophiles do not realize how stupendously inefficient speakers are. By way of reference, 92 dB/watt/meter is about 1% efficient, or put another way, 100 watts of electricity is converted to one acoustic watt (which is plenty loud).

So where does the other 99% of these pricey watts go? Voice coil heating, which isn't great considering how tiny voice coils are, and how poor thermal coupling to the outside world is. First the voice coil has radiate its heat to the magnet, which is the closest thermal sink, then the warmed magnet has to transfer its heat to the inside of the enclosure.

Since the goal is to create X amount of acoustic watts, not a clumsy form of room heating, even small gains in efficiency are worthwhile, since less voice coil heating is occurring for given acoustic output.

Aside from outright failure, another problem with VC heating is copper's change in resistance with temperature. The resistance goes up with temperature, which might be acceptable, excerpt the time constant is fairly slow, on the order of several seconds, This creates a dynamic slurring which is pretty audible.

I should add I am completing a large-format 2-way speaker this summer, a collaboration with Thom Mackris of Galibier Designs, and an entirely separate project from Don Sachs and the Spatial Audio team. It’s a culmination of the extremely long "Beyond the Ariel" thread over on DIYaudio.com, and the first version was built by Gary Dahl, of Silverdale, Washington.

The woofer is an Alnico-magnet 416 (15" midbass) from Great Plains Audio, the successor to Altec Lansing, using Altec staff and tooling. It’s in a low-diffraction (4" radius curved edge) 4.2 cubic foot closed box. My version will have Bubinga (African rosewood) veneer on all sides.

The high frequencies are from an Athos Audio Yuichi A290 wood horn, with a to-be-determined 1.4" exit monitor-class compression driver. Crossover will be around 700 Hz, most likely Altec-style 2nd-order. The RCF 850 and 18Sound drivers are candidates. I also have a pair of Altec/GPA 288’s in house as fallbacks.

Efficiency will be a true T/S value of 97 dB/meter/watt. With a 27 watt/channel amplifier, headroom should be, in the timeless words of Rolls-Royce, "adequate". Alternatively, sufficient for a studio monitor application.

A 20-watt amplifier and 97 dB/meter loudspeaker was pretty typical for a serious high-end system in the mid-Fifties, so it’s not as weird as it sounds. It’s only weird in the modern context of 200 to 500-watt Class D amplifiers and 85 to 87 dB/meter audiophile speakers.

Ouch.

None.

As you might imagine, Allen was pretty shocked at the direct comparison, since his amp had much more powerful tubes than mine, which had generic Sovtek 300B’s, good and tough, but nowhere in the same league as Vaic’s finest. I mean, a quartet of top-of-the-line 300B’s ain’t cheap, so I never went down that road.

And Allen had just given a presentation at the VSAC, only hours before, on the power of this secret circuit, which he did not fully reveal. It was a very large current source with heat sinks and all. Yes, he could have cranked up the current even more, but the heat sinks and power transistors set an upper limit on the current. It was already close to max output.

He expected that I, an old Tek hand, would be thrilled with Tek-scope type circuit. But I disappointed him. Driving deflection plates (at very high speed) on a CRT is one thing, driving a loudspeaker is quite another. And I’d been designing speakers for Audionics several years before joining Tektronix in 1979.

Scopes are about speed, and the load is a very well-defined capacitance. Cascode differential circuits are the right answer for that problem ... they’re very fast, ideally suited for square waves, and linear enough for the purpose.

Speakers are orders of magnitude slower and are inherently vile loads. The best speakers are the worst loads ... the ones that have near-resistive loads are planar-magnetics with very low BL product (which is magnetic coupling). As you raise BL product, efficiency goes up, they get snappier sounding as the coupling gets better, and ... they also get more reactive, for the simple reason the amplifier is in more intimate contact with the big, sloppy, electromechanical system. Few amplifier designers are aware of this ugly reality. They keep hoping for speakers that can never exist.

The worst thing speakers do is insert speaker colorations (through back-EMFs) into the feedback loop, where they do not belong. Feedback is great at correcting amplifier nonlinearities ... it’s fast and responds in microseconds, just what you want. Speakers have inherent high-Q resonances that are an inescapable part of an electromechanical device. The better the magnetic coupling, the worse it is for the amplifier, which has dirty spurious currents injected into the output node by the speaker.

My approach is to brickwall-isolate these back-EMFs to the final output stage, and not expose the rest of the amplifier to them. I think of the speaker load like attaching a vacuum cleaner motor to the output section ... a source of noise and garbage, nothing good about it. The amp has to ignore this racket and continue to do its job. Feedback amps can get into trouble when the error voltages get very large; this can saturate the input section, and induce additional distortion.

In a more conventional application, like a long-tail Mullard phase splitter, differential circuits have a subtle imbalance that is not obvious at first glance. On the top, or front, side of the circuit, there is the expected Miller capacitance, as per expectation. This is the inverting side ... grid goes down, plate goes up, just like you expect.

The non-inverting side can be drawn (and is better understood) as a cathode follower driving a grounded-grid stage. Rotate the other tube by ninety degrees and it becomes more obvious. This side of the circuit has very little Miller capacitance, making it ten to twenty times faster than the other side. The beautiful symmetry falls apart at (very) high frequencies. As mentioned earlier, it can never enter Class AB drive when one side cuts off, although this is not a problem if the diff-pair is not used as a driver. In a scope, you see clever bootstrap circuits and cascodes to give that extra push at high frequencies.

This is Nelson Pass’ speciality; high speed cascode differential circuits. If that’s your thing, he has an amp or preamp just for you. If you’re using transistors, this is an attractive path.

I should mention that Allen Wright liked a very different sound than I do; he liked fast, snappy, and what sounded to me like thin bass. I like a big, lush, spectacular, CinemaScope sound, the sound I heard in 70mm theaters when I was growing up. (Which had Altec Voice of the Theater speakers behind the screen, along with Altec amplifiers.)

The same applies to my brief encounters with Nelson Pass. He likes it a lot thinner than I do, but with a different tuning than Allen Wright. Kind of hard to describe, actually, since this was all a long time ago. Allen liked the sound he was getting, and he liked his own amp, even at that meeting all those years ago. What I thought was a disaster seemed OK to him. In all honesty, it was a split decision.

I mean, I didn’t like it, nor did Gary Pimm, but we were on a different wavelength than Allen Wright. His designs, like mine, are tuned to his own tastes, and we found out they were surprisingly different. Similarly, I was surprised at Nelson Pass’ tunings, very different than my own.

As it is, Don and I have a bit different preferences, but at least we are still on the same planet, so we get along. From what I heard of Allen’s designs, no way, they are too different, no good way to reconcile the two approaches. But he was a really fun houseguest, and Gary Pimm and I had great discussions with him about everything under the sun.

I miss him very much. He was really funny and one sassy dude with total disrespect for the high and mighty poo-bahs in the industry, which I very much shared.

There was a funny incident a few years back at the Dallas Audio Show. Back then, it was a little bitty thing, just a few exhibitors, but very much a home-town thing where everyone knew other. New to me, of course, as a former West Coast guy fairly new to Colorado. Never been to Texas before.

I wander aimlessly down the halls, no real goal in mind, looking for interesting tube gear. I walked into one room, and whoa, that’s Nelson Pass! Now people joke about me being Mr. Natural, but Nelson really looks the part. You can’t miss him. Me, my only trademark at a show are the Hawaiian shirts I like to wear.

Nelson had actually built an open-baffle speaker around a Lowther and a 12" guitar speaker called the Tone Tubby that I had written about some time ago. Well, that’s different, but why not? As I turned towards the door, Nelson blocked the exit.  how do I get in these situations? Me and my big mouth.

It turned out the two drivers were bi-amped with a simple low-level crossover. Oh, now I get it. Four knobs, two for level, two for the crossover frequency. Nelson wanted me to tune the thing ... by ear.

Now I really want to escape, but Nelson is still in the way. Fine, anything to get out. Twiddle, twiddle. Too little bass. Mo’ bass, man. Turn that knob up. A bit less Lowther, but not too dull. Mess with the crossover overlap some, so that mellow hemp cone transitions into the characteristic hard paper Lowther cone. A few minutes later, sounds OK, as good as I can get it right now. (Did not sound OK when I walked in.)

Escape permitted. Afterward, Nelson allowed as to how he saw that article I wrote about the charms of the Tone Tubby and wanted to build a simple open baffle around it, with a Lowther on top. So he figured if anyone could tune it on the fly, it would be me. Well, he had me there, but I allowed that he might have different preferences than I did, so feel free to mess with the knobs, although he might want to mark the current positions before changing anything.

These weird things happen to me at shows. That’s how I met Nelson Pass.

Ralph, I sincerely invite you to build your own 300B amplifier. (The name of the thread is "300B lovers", after all.) You’re smart enough to bring Allen Wright’s topology back to life, and give it unique improvements of your own. Seriously, if anyone can do that, it would be you, not me. The only way to explore the 300B sound is design your own amp around it, and I have no doubt it would turn out well. I think you would be quite pleased with the result.

Don, myself, and the Spatial team are developing the Raven/Karna/Blackbird architecture discussed in this thread, with really good suggestions from the whole team. Much appreciated, I can’t say that enough. You guys have taken it way beyond the 2003 thought experiment that Gary Pimm and I built.

What folks heard at the 2023 PAF show was a good preview of the full-on production model, which Don is charging ahead with as we speak. (Well, actually, I was just on the phone with Don, and he was at the beach, but tomorrow, OK?)

Alex Berger, what you have sounds fine to me. The only refinement would be separate power supplies for the input+driver and the output section. And maybe a dedicated filament transformer for the 300B, whether AC or DC powered.

These pictures give me a bit of a chill. It isn’t a Raven, not quite, but it’s pretty dog-gone close. Both date from the Twenties, and they come out of Bell Labs. In the first, note the archaic nomenclature for the direct-heated tubes and the weird little capacitor ... and what the heck is it doing, dragging the lower grid off-center?

Western Electric 7A Repeater Amplifier

This might look familiar - the WE 42A amplifier

Build either with modern parts, and it sounds like it’s from outer space. It’s like discovering a working spaceship under a long-forgotten tomb.

On a more technical level, here’s the relevant discussion, a deep dive into audio archeology:

Western Electric, the Rosetta Stone of Audio

I second what Don just said. Isolation is the key. You can get away with a power transformer that has dual isolated secondaries, but this is a specialty item so don’t bother tracking it down. At the DIY level, just get another B+ transformer that runs at a voltage suitable for the IT-connected driver tube, and take it from there. In terms of rectifiers, damper diodes are the quietest of all, but they consume a lot of heater current at 6.3 volts.

The dual power supply approach is surprisingly rare in consumer equipment, even at extremely high price levels. It is the single biggest improvement you can make to any tube gear, from push-pull 6L6, EL34, and KT88 to SET amplifiers of any type.

Extremely large banks of electrolytics are popular over in transistor land, but they are frankly mediocre sounding caps, compared to good film caps of more moderate values from 50 to 200 uF. In the Karna, I used banks of industrial-type motor-run caps from ASC and GE.

These are precision parts designed for extremely severe duty outdoors. I prefer them to audiophile parts in that application. A minor audiophile tweak is to bypass the industrial array with a single 0.1uF cap of very high quality, such as copper foil. (Also use copper foil for the RC or LC coupling of the 6SN7, but be aware that wax caps are not suitable for under-chassis use.)

Location is important. Keep the wiring, especially in the cathode circuit, as short as possible. This is more important than the type of wiring, although if you want to go nuts, use industrial Litz wire for the critical audio path. Litz does require a solder pot to get rid of the enamel coating, a minor annoyance when working with it. A close second choice is tinned stranded, which is super easy to work with.

This approach, if done right, will take your SET performance to the mid to upper tier of Audio Note, in the $20,000 to $50,000 price range.

The very first one, the Amity, designed by yours truly, and built by Matt Kamna on an open breadboard-style chassis. Matt has since gone on to co-found Whammerdyne, a company that makes 2A3-based SET amplifiers, and exhibited in the Songer Audio room at the PAF show this year.

Amity at night, 1996, Aloha, Oregon

We’ve gone a long way since then!

Here’s a picture of Gary Dahl (seated) and Gary Pimm, shortly after Gary delivered the newly built Karna amps to our living room in Silverdale, Washington. 2003.

Silverdale, 2003

Gary Pimm and Gary Dahl look under the chassis

New Karna amps, 2003

Ariel speakers and Karna amps in Silverdale, 2003

Full schematic, 2006

The dual B+ power supply chassis is external, connected by the aviation-grade Amphenol connectors at the rear of the chassis. The glowing VR tubes are at the front of the chassis. The EL34 on the left side is part of the high voltage current source that feeds the VR tubes. The audio tubes are 5687 input, 45 drivers, and 300B power tubes. Interstage transformers are under the chassis.

This picture gives an idea of what Don accomplished over the past year, reducing this behemoth four-chassis prototype to something that could be practically built, and then exhibited at the Pacific Audio Festival.

How would you compare the sonics of your 300B amp to the the Class D amp you make now? You’re the creator of both, so you’re in the best position to evaluate and compare. I only spent a half-hour of casual listening to the Purifi at the show (and Audio Group of Denmark), so I’m hardly an expert on the subject.

The thread is called "300B lovers", so tell us what you think about a 300B (of your own creation) vs your latest Class D amp. At the risk of thread derailment (I ask the forgiveness of the Audiogon moderator), and then I’ll return to the walk down Memory Lane.

P.S. Karna and I actually lived in a house on Memory Lane when we were in SW Portland, up in the West Hills, just off Sunset Highway. It's a real place.

Alex, I wish you every success with your revised 300B amplifier. An isolated power supply for the input+driver, and replacing all the electrolytics with arrays of 440VAC (630VDC) industrial motor-run caps, will make a difference that will astonish you. Not joking here. It will definitely take it into the top class of SETs.

Be prepared for a pretty large and heavy chassis for an 8-watt amplifier. 18" x 18" and 50 lbs or more would not be out of line.

As for cap-value tuning, set the RC (or LC) frequencies between 3 and 4 Hz. The beat of most music is between 1.3 and 2 Hz, and you do NOT want any of the RC networks interfering with that. Anything slower than the beat of the music will give a kind of seasick, unsteady feeling, so don’t go there.

Leave the banks of electrolytics to the DC-coupled transistor guys. A lot of their tuning (and amplifier sonics) comes down to the brand of the electrolytic. (Oops, did I let the cat out of the bag? Sorry, guys.)

Don accomplished a miracle of miniaturization for the show, but we’re dialing it back a bit for the production models (to simplify assembly). We’re expecting 18" wide, the exact same width as the preamp, and maybe 16" to 18" deep. Weight ... yeah, maybe 50 lbs or so. Depends on what the transformers weigh. Specs and overall performance will be the same as the show models, so if you like what you heard, that’s what you’ll be getting.

 

 

OK, I didn’t know how much Don was keeping under wraps, so I was a little vague about our continued progress.

The new full-size chassis of the Blackbird (compared to the models at the show) gives us the freedom to "open up" the Blackbird ... ultra performance caps in critical locations, a bit of Raven tech in the front end, and more rigorous isolation between the high-voltage power supply and the audio circuitry. All of these need more room, which is why the production chassis will be 18" wide. Sonically ... well, I haven’t heard it yet, but they’re all good things that move in the same direction as the past year of collaboration.

One the things about the Raven/Karna/Blackbird that is frustrating, but also very gratifying, is the circuit is extremely transparent and revealing. The frustrating part is that parts quality is revealed in a relentless glare, at least in the critical nodes of the circuit. This is the downside of any zero-feedback design; there is no clean-up crew of servo feedback to tidy up afterward. You hear things as they are.

But the transparency is also a gift, because "minor" substitutions are immediately apparent in the first minute of listening. I think Don, Cloud, and the rest of the Spatial team will agree on that point. I feel we are fairly close to the upper bound of what the circuit can do, but I keep being surprised.

The Raven, which I had never heard before, took me aback ... that was not what I was expecting. It is super fast and resolved, with sounds flying out of dead-black space. You can practically see the shapes of the notes as they fly by. No exaggeration, no tipped-up HiFi sound, no artificial edge sharpening, but boy, it’s all there. If it was on the recording, you will hear it.

I find it kind of shocking that a late-Twenties Bell Labs/Western Electric telephone repeater, built with modern ultra-wideband parts and quiet MOSFET cascode power supplies, sounds like that. There ain’t nuthin’ retro about the sound at all.

Oh, that wasn’t technical enough for you? Well, chew on this, the way I got my first job at Audionics, by inventing this little gizmo:

Shadow Vector Quadraphonic Decoder

Sadly, only one prototype was ever built, but it did get demonstrated at EMI and the BBC in 1975. After that, loudspeakers, Tektronix, and various magazines.

Believe it or not, there’s a programmer in the UK who actually built this in software a couple of years ago. Now, that’s impressive. The 1975 hardware prototype took nine circuit boards plugged into a backplane ... the UK programming genius took it to the next level, and made it into an eight-band decoder, all working in parallel, thanks to the wonder of modern DSP.

Since this is (mostly) a 300B thread, I’m still curious about the general circuit of Ralph’s 300B amp. He’s been doing this a long time, so his design choices are of considerable interest.

I have my own way of doing things, and that was strongly influenced by my research when I was writing for Glass Audio and Vacuum Tube Valley. John Atwood, in particular, showed me the Bell Labs archives and other primary sources. Charlie Kittleson, the magazine’s founder, had a treasure trove of working 1930’s electronics, which sounded very different than anything I’d heard before ... not like the Fifties sound at all. They clearly had different priorities back then.

John Atwood and I have a lively interest in early technology, particularly early monochrome and color TV in the USA, the UK, France, Germany, and Russia. Developments in color TV filter technology went on to influence Neville Theile in Australia, modern crossover design (Laurie Fincham in the UK), and the time-switching technology used in the GE/Zenith FM Stereo multiplex system.

The 300B filament circuit is a very delicate circuit node. The high voltage windings of the B+ transformers see switching pulses of hundreds of volts with very steep rise times. It only takes a few pF of winding-to-winding capacitance to transfer that 120 Hz switching noise straight into that expensive 300B. Unless you know that the low-voltage winding is electrostatically screened (with copper foil), don’t do it. Use a separate transformer just for the 300B alone.

You wonder where low-level buzz comes from? Winding to winding stray capacitance. Whenever you hear buzz instead of low-frequency hum, that’s a capacitive coupling, not magnetic. The spectrum gives it away.

I would NEVER use vintage caps in a power supply. Never never never. Use modern parts. They’re not that expensive, and used in air conditioners all over the world. Vintage is OK in a crossover, where failure is no big deal. In an amp, just say no.

Not sure I see the merit of mixing films and electrolytics. If you have the space, use the industrial parts, and bypass caps to personal taste. In terms of location, the cap bank can be several inches away from the tube socket, but the little caps (0.1 uF or less) need to be close by, an inch or less.

You can do a lot of sleuthing just by listening to the spectra of noise. Magnetic induction is going to be pure 50/60 Hz and fairly hard to hear. Capacitive coupling is high frequency only, and will sound like buzz, usually harmonics of 100/120 Hz switch noise from the rectifiers and transformer secondaries.

Ground loop noise can be isolated by shorting the input plugs of your preamp or power amp. If the input is shorted, and the noise persists, it is inside the component itself, and is usually a design or layout error.

If the noise is the result of two components connected together, that is a ground loop. This can be confirmed by disconnecting the interconnects between them, turning both on (with volume down), and using a DVM to measure the AC voltage potential between the two chassis. Scrape through the paint or anodize if you need to, then measure.

The AC potential between the two should be less than 1 or 2 volts. If it is more, then you have a ground loop. This is caused by capacitive leakage from the power transformer to the chassis. It can cured by reversing the AC polarity going into the power transformer on ONE of the components, but this is not a DIY job.

What causes this is that consumer AC power is not balanced; instead, there is neutral, which is only 1 or 2 volts away from safety ground, and hot, which is 120 volts in North America and 220 to 240 volts elsewhere. Power transformers are not symmetrically wound; one side has lower capacitance to ground than the other, but unfortunately, the leads are not marked, so they can randomly assembled in production. Ideally, the low-capacitance side of the primary should go to HOT, and the high-capacitance side of the primary to NEUTRAL.

If all your components were assembled this way, you would never have ground loops. Unfortunately, the phasing of the power transformers is random. The capacitive leakage from primary to transformer case will let the chassis float to a high value relative to safety ground, which is the true ground. The only real solution are medical-grade power transformers, which have extremely small leakage to chassis.

Short of that, you can hire a skilled technician to wire all of the power transformers in your system for minimum AC HOT to chassis leakage ... which is a good idea from a safety perspective anyway. No more little shocks when you touch a component (which should never happen in equipment built to code). Safety code requires that the fuse, then the power switch (in that order), always be on the HOT side of the line.

Yes, plenty good enough for the job. Don’s comments above are right on the mark.

What you want is isolation. Stage-to-stage, and isolation of the critical filament supply.

And if you really want to get hardcore, make sure all the cathode circuitry, of each section, comes down to a single star ground on the main ground bus-bar. So, star ground for 6SN7 cathode components, a few inches away, the star ground for all 6F6/6V6 cathode components, and another few inches away on the main bus-bar, all the cathode components for the 300B. The idea is keep all audio currents local to that stage, and to that stage only, and only have DC return currents on the main ground bus-bar. You would be surprised how few high-end components do this.

The presentation below is for advanced students. You know who you are:

European Triode Festival Presentation

ETF Part Two

Just a quick update that Don and I are continuing to refine the Blackbird, with a bit of Raven and Karna Mark I thrown in. The chassis will be 18" wide to give a more spacious layout, a simplified build procedure, and visually match the Raven.

The circuit continues to be balanced throughout, with a tube lineup of a 6SN7, a pair of matched and balanced triode-connected 6V6, and a pair of matched and balanced 300B’s. The power supplies (both of them) have a slow-start circuit that protects against hot-start transients if the AC power flickers off for a second or two, as well as controlling tube warm-up.

Sonically, Don and I are prioritizing depth and realism of tone color, like 300B SET amplifiers, combined with a clarity and directness usually associated with high-performance Class A and Class D transistor amplifiers.

Since nobody is reading this thread right now, I’m going to throw out my wish list, my note in a bottle, to the wilds of the Internet:

* I’d like to see LinLai or JJ or any of the other tube vendors, try something a little out of the ordinary. A true triode, using an octal KT88 socket, that biases up exactly like a triode-connected KT88. An indirectly heated triode, in a KT88 package, with only three elements ... cathode, control grid, and plate. With no screen or suppressor grid, and the control grid correctly spaced so the whole tube mimics a triode-connected KT88, so it can plugged directly into a KT88 socket in an existing amp and work right away.

What is the benefit over a standard KT88? Well, with no useless screen or suppressor grid, the one remaining grid can be optimized for lowest distortion ... in particular, the lowest proportion of high-order harmonics, like a direct-heated triode.

It’s not the direct heating of the M-shaped filament that’s responsible for the very low distortion of DHT’s (compared to triode-connected pentodes and beam tetrodes). It’s the clean, uniform grid structure, and the carefully chosen spacing from the cathode (or filament). So there’s no reason a purpose-designed true triode can’t be designed to fit a standard KT88 (or EL34) socket that has the same DC bias characteristics as it’s more complex brother, but also much lower distortion.

Literally, a simple plug-in improvement for all the hundreds of thousands of conventional PP-pentode amps out there. No change in bias, no change in cathode circuit, no change in fixed-bias operating point, just lower distortion, ideally approaching DHT performance if the grid is correctly designed.

You could call the new tube a TR88 to distinguish it from a KT88, while signifying it is plug-in compatible (thanks to the same DC biasing). Or TR34 if it replaces an EL34.

The sonics would be interesting. The target device would of course be the 300B, but with an octal socket with a 6.3V indirect heater, and KT88 biasing.

The 300B is more physically fragile than a KT88 because of filament sag, which can happen if the amp is tipped on its side while the filament is hot. If it sags enough, the filament will touch the grid, and ZAP! the tube turns into a full-power diode, which destroys it and the cathode circuit.

Indirect heated tubes have the heater coiled up inside the cathode box, or cylinder, so it won't go anywhere if the amplifier is tipped on its side. This is why guitar amps are so rugged ... it takes enormous abuse to damage a KT88.

The customer base for a TR88 would be much larger than a 300B. If the sonics and harmonic structure were like a DHT (not a KT88), a lot of people would be interested, assuming the price would be in the KT88 range. If it had a 300B price, that would greatly diminish the market, since it would be far outside the KT88 class.

People have gotten the weird idea that somehow the filament of the 45, the 2A3, the 300B, and the 845 are responsible for the ultra-low distortion, and the super-vivid tone color, of the DHT family. Wrong. It isn’t.

It’s the grid. DHT’s have a physically large grid, well spaced away from the whirling cloud of electrons called a "space charge". Surprisingly, electrons are not directly emitted, pass through the grid, then strike the plate. Instead, they whirl around in the space charge, find a passage through the venetian-bland repelling field of the grid, and are accelerated to the plate.

It’s the grid geometry that sets not only the DC characteristics of the tube, but also its linearity (especially high-order terms). This is the most critical part of the entire circuit. If you need pentode or beam tetrode characteristics, fine, you’ll have very low Miller capacitance, very high output impedance, and easy drive characteristics. This makes an excellent RF modulator, where distortion doesn’t matter.

But if low distortion comes first, and you’re not asking for 20 to 50 dB of feedback to linearize the whole amplifier, a true purpose-made triode should have the lowest distortion. Since there are already lots of octal sockets in PP power amps, why not make a special triode tube just for them? There is some design work to optimize the grid structure so DC biasing is the same as a triode-connected pentode, but the absence of all those other grid wires should help. If the design is good enough, it could rival the 300B without the hassle of direct heating and the complex filament circuit.

I am impressed you took the project that far, but the minimum requirement for 10,000 units on an unknown tube is a steep hill to climb. Even at OEM prices, that’s a half-million dollars on a gamble. The "X" tube would have to be very very good, and very very popular, for that gamble to pay off.

In other news, the all-IT, no coupling-cap version of the Blackbird is the best version yet. The LC coupling on the 6SN7 is going away and getting replaced with a custom IT with 18 Hz to 35~40 kHz bandwidth. This gets rid of six parts - a pair of 100 Hy inductors, a pair of copper-foil coupling caps, and two 220K grid resistors. No RC coupling, no LC coupling, and no current sources, either as plate loads or in the cathode circuit. The signal path is copper wire, high-nickel magnetic cores, and vacuum tubes.

As Don mentioned on a recent phone call, there is no direct electrical coupling between any of the stages, which filters off any RFI incursion before it gets amplified. Homes are much noisier in the RF spectrum than they used to be, with Bluetooth and WiFi everywhere.

Tube lineup remains 6SN7, matched pair of 6V6, and matched pair of 300B. Tube rolling is welcome so long as pairs are matched.

I have to agree with Eddie about the ST-style 45. Aside from driver use, the charm of the 45 is that it is really easy to make a superb, low-power amplifier with it. It has the cleanest distortion spectrum of any tube I’ve measured (closely followed by the 300B), but unlike the 300B or 2A3, it is super easy to drive. This makes two-stage amps simple ... in fact, I’ve yet to hear a bad 45 amp, while I’ve heard plenty of not-great 2A3 and 300B amps. Good 300B amps in particular are quite difficult.

45 amps also sound louder than you’d expect ... they easily keep up with 2A3 amps, despite what the numbers say. If 45’s were more abundant and at different price points, they would quickly find a market.

I have a dark suspicion that modern 2A3’s are simply below-spec 300B’s, and not related to original 2A3’s at all. Of course, that’s not really true, since 300B’s have 5V filaments while 2A3’s have 2.5V filaments. Modern 2A3’s are quite different than the bi-plate RCA originals ... I’m not sure where they fit, actually.

Returning to the 45 triode, I really think people would be surprised just how good 45’s sound, and how crisp and vivid they are. They are very different sonically than 2A3’s or 300B’s. They actually sound more like 845’s than anything else. Very clean and fast, no murk at all, and nothing like any pentode (no hash or grain). If you can get your hands on a good pair (or quartet) of 45’s, they are quite impressive and worth exploring,

These are the kind of conversations I used to have with Harvey Rosenberg, who definitely "got it". Unless you knew him, you didn’t realize his clowning around, and occasional jaunts into speculative metaphysics, was an act designed to chase away mainstream audiophiles (which it did). Of course, I can speculate about metaphysics for hours on end, as folks who met me at the PAF discovered. But the clown act was pure Harvey ... I’m too sobersided to joke around like that. But we did discuss the subtler aspects of musical perception, and how they interacted with culture and worldview.

I am not sure where capacitor coloration comes from, or what’s causing the hours-long "burn-in" process. It can’t be measured, and the plausible mechanisms are pure guesswork. But it’s there. In a positive sense, direct-heated triodes, in the right circuit, have very special qualities.

Not everything is revealed by the spectrum analyzer, and the usual audiophile lingo falls short of describing what triodes really sound like. Sanskrit or Japanese might be better, but I don’t speak either, although I am conversant with a few basic Zen, Taoist, or Vedanta concepts. English is not particularly good at discussing subtle aspects of perception and consciousness.

Now that Don has built a version with zero coupling caps, he’s having the same experience I had when I first heard the Amity in 1996. The Karna was a development of that, and the Blackbird is a development of the Karna.

Agree 100%. Don is describing what I hear.

The metaphysics are optional, but they have a good toolkit for describing perception.

Umm, hard to describe. More "there" there. More sense of a physical presence, and a better feel of the performer’s musical intentions. More hall sound. Most of all, a feeling of the performer being in the room instead of somewhere "out there".

Not the usual audiophile verbiage, but it’s immediately evident when you hear it.

By now, the Blackbird is very close to the original Karna, but with greatly improved power supplies and a much more practical monoblock construction, which also improves performance. Full credit to Don for pulling off what I thought was impossible.

As you approach the highest levels of audio, the sound kind of takes a right-angle turn from the usual path of improvement. Instead of just getting more and more clear and vivid, suddenly, there’s a physical sense of musicians being right there in the room, instead of sounding like a very good recording. The instruments have physical size and feel like you can reach out and touch them.

A piano sounds like it’s five feet away, and it sounds BIG. Even "bad" recordings sound like this. Is something being added? No, I don’t think so. Instead, a mechanical quality is no longer there.

This phenomenon is well known amongst triode practitioners. It’s not a secret. Unfortunately, it is almost never heard at hifi shows, so don’t expect it there. I’ve heard it several times at the advanced DIY level. Never heard before from any mainstream or transistor system, which is why I started working with direct-heated triodes in the early Nineties, after hearing and reviewing the Audio Note Ongaku and the Herb Reichert Silver 300B.

See page 21 for a construction article on a push-pull, zero-feedback 300A/300B amplifier. This is about the same time as Columbia introduced the LP microgroove record, so it’s very early days for high fidelity. What surprises me about this article is that 300A’s and 300B’s were even for sale to the public; they must have been pulling them out of prewar theater equipment

At the time of printing, the only records you could actually buy were 78’s, and FM radio was very new. Note the primitive state of tonearms and phono cartridges ... also, 78’s had no standard equalization, so preamps had to cover several curves, including "acoustical" for pre-electronic records.

The reign of the monophonic LP record was surprisingly short; ten years later, almost to the month, stereo LP’s were announced by all the major labels, and stereo cartridges and stereo preamps were also on the market.

Audio magazine, July 1948

Audio magazine, August 1958, the Stereo Issue

Audio magazine, November 1982, first CD player

This is entirely up to the transformer designer. They need to know the Zout, or Rp, of the tube driving the primary, and the load on the secondary, which will either be a pure capacitance in the 60~80 pF range, or paralleled with a load resistor, typically 100K or so.

The method of extending HF bandwidth is interleaved windings, and this falls into the realm of modern computer modeling. Back in the old days, this was cut-and-try, now, it can be modeled. Interleaved windings extend bandwidth, but the interleave pattern has to be carefully chosen so there is no HF ringing into the intended circuit. (This is why they need to know the Zout of the preceding stage and the load of the following stage.)

They will want to know your expected bandwidth, power handling within that bandwidth (particularly below 40 Hz), and how much square wave overshoot you will accept. And the DC parameters ... if SE, how much quiescent current does the tube run at, if the circuit is balanced, how much DC imbalance do you expect from the pair of tubes. This affects core gapping, which in turn dictates core size and transformer size. A small air gap linearizes the transformer, but can also double the required core size, which in turn affects HF bandwidth. The DC parameters are critical for the entire transformer design.

As you can see, this isn’t a matter of selecting an off-the-shelf part, but consulting with the transformer designer and telling them what you need (and what they can do). They *might* have an off-the-shelf part, or they might not. If not, what is the minimum order, and how long will that take?

I haven’t mentioned sonics yet. Aside from meeting minimum technical specs (which you and the transformer designer both agree on), there’s the matter of subjective sonics, and how it fits with the sound you are aiming for. This might sound trivial, but if the amplifier designer has no subjective sonic goal, you will not get there. "Perfection" is not a goal, it’s a marketing term, like "Perfect Sound Forever" for CD’s back in the Eighties.

Are you familiar with the subjective difference between RC coupling, LC coupling, active current-source loads with capacitor coupling, and interstage transformer coupling? (With this amplifier design, Don built and auditioned each one.) This is very useful to know as the amplifier is tuned subjectively.

Similarly, the driver stage design has a major effect on amplifier sonics, aside from inter-stage coupling. The driver section affects slew rate, HF distortion, and subjective colorations in the mid and upper frequency range. It’s useful to know the sound of a 6DJ8, 12AU7, 6SN7, and a power-tube (45, 6V6, KT88) driver ... they sound quite different from each other, and can dominate the sound of the entire amplifier.

I apologize for making this sound confusing, but I wanted to give the readers of this forum a taste of what Don has been through. There are many possible ways of getting a zero-feedback amplifier wrong, and from the standpoint of mainstream audio engineering, all zero-feedback amps are wrong ... not just in practice, but in principle.

An all-transformer coupled amplifier is especially wrong. The accepted path is DC coupling throughout, using transistors, with lots of excess gain for plenty of feedback. Modern feedforward techniques (Bruno Putzey, THX, et al) can get distortion into the parts-per-million range, so why look elsewhere?

Unless your goals are subjective, and you have a weird hypothesis about linearizing each stage, as much as possible, without using feedback. That’s why Don took a gamble on the Karna topology, a circuit out of the late Twenties and mid-Thirties.

Don is being modest. The last year, going through the present, really made Don pursue every obscure byway of amp design, building and listening as he went, every step of the way.

Don started with an obscure version that I called the Symmetric Reichert, which was literally a Reichert 300B done twice, with a phase-splitter transformer at the input. All RC-coupled. He built that and called me out of the blue, about a year ago.

Don then tried separate B+ supplies for the input+driver and output section, and an interstage transformer between the driver and 300B’s. A few months later, Don used a triode-connected 6V6 instead of the hard-to-find 45 driver. Thom Mackris and Don independently tried this at just about the same time, pretty much on the same day. Don (but not Thom) then used active current-source loads for the input 6SN7, instead of resistor loads. That was the Stereo version Don built and shared with the Spatial team and the first customers.

Next, replacing the active current source loads with custom Cinemag inductors designed for the purpose, and using the shoebox-format monoblocks that became the show amps. My Colorado neighbor, Thom Mackris of Galibier Design, has been following along in a parallel project, with a SET architecture, but with passive CLC B+ supplies and damper diodes for rectification.

That’s where all of us were a month ago ... Don Sacks, the team at Spatial, and Thom Mackris. The latest from Don is an IT between the 6SN7 and the 6V6, replacing six other parts with a much simpler approach ... provided the IT was up to the task, which it is. The IT has turned out to be superbly designed, exceeding expectation, and also making our lives simpler. Don and I have gone full circle, and re-invented the Karna (after trying every alternative), with far more advanced power supplies that were not available in 2003.

Don really has tried every topology, one after another, and carefully measured and auditioned each one. RC coupling, active loads, LC coupling, and now, IT coupling. By lucky coincidence, Thom has been walking a parallel path with his SE topology. All four groups ... Don, Thom, Spatial, and myself, have been exploring this zero-feedback approach for several years now.

If other folks want to build transistor Class A, Class AB, or Class D, more power to them. Those designs have an entirely different set of challenges that have nothing to do with triode amplifiers. In triode amplifiers, the devices themselves are exceptionally linear, and the appropriate circuits take advantage of that.

I’m no expert on the MC30, but it is very unusual. The cathode feedback (from a special tertiary winding in the output transformer) results in very low gain for the 6L6 power tubes, so the driver has to swing 100 volts, putting extreme demands on linearity. And I think it operates in nearly pure Class B, with a very small Class A region. This requires substantial feedback (which it has) to linearize the output section. The Class B operation requires very close coupling between the tertiary and primary windings, otherwise the circuit will have tube cutoff glitches with every zero crossing. So the whole thing is very much a package ... multiple feedback loops, a unique output transformer, a wide voltage swing from the driver, and cathode feedback for the power tubes.

Almost the polar opposite of the Brook 2A3 amplifier, which relied on the linearity of the 2A3 power tubes instead of massive feedback. The high-power (30 watts!) Brook amplifier used sliding bias to keep the output section in quasi-Class A.

It should be mentioned there was no awareness of slewing distortion at this time, because signal sources had very limited HF bandwidth (12 kHz) and limited peak energy. Phono cartridges were very primitive and could barely track LP’s at 5 grams.

When the first writings about slewing distortion appeared in the late Seventies (25 years later), things were very different: moving-coil cartridges with exotic styli were flat out to 50 kHz, and cutterheads could put down tremendous levels on the disc. Tweeters were much better as well.

Interesting circuit ... thanks for posting it, Ralph.

It’s kind of weird the actual gain/driver tube is a powerful 12BH7, while the cathode follower is limited to 1mA or so from the 12AX7, which is about the worst possible choice for a driver. Very strange. Maybe an intentional current-limit for the power tubes?

Also was not aware the global feedback network has its own winding, making four secondaries on the output transformer. Most everybody else samples the 16-ohm tap, which captures the entire secondary that powers the speaker. I kind of wonder if McIntosh did this to make the circuit hard to copy.

From this distance, I wonder about circuit stability. I see at least two feedback networks, one nested inside the other, and the outer loop not actually sensing the voltage on the speaker terminals, but a separate winding (which will never be exactly the same, especially at high frequencies). The circuit has a massive amount of forward gain (12AX7 -> 12AU7 -> 12BH7), so the feedback networks are definitely active.

They all had a different "house sound". Marantz was crisper and more resolved, a more hi-fi sound, and Fisher sounded like a really good console. Not sure if the H.H. Scotts had a consistent sound or not ... they were mostly known for their FM tuners.

Dynaco was always the "value for money" brand, like a VW Beetle.

Playlist

Here’s what I’ve been listening to for the past year. Probably zero percent overlap with Don Sachs, and only 10% overlap with the folks at Spatial. It’s a mix of dense heavily-produced rock, London Sound, and German techno.

Joywave: Traveling At The Speed of Light (7:34 version)

Joywave: Smokestacks

Big Data: Unglued

Big Data: Put Me to Work

Disclosure: Help Me Lose My Mind (featuring London Grammar)

Soul II Soul: Keep On Movin’ (demo quality)

Elderbrook: Inner Light

Elderbrook: I’ll Find My Way to You

Ulrich Schnauss: Far Away Trains Passing By (whole album)

Phosphorescent: Song for Zula

Tears for Fears: Elemental

Tears for Fears: Woman In Chains (demo quality)

Preferred Listening Format:

* Shadow Vector, Surround Master, or DTS Surround system

* Open-back planar headphones

* DHT triode system

Probably half these songs would get me thrown out of most hifi show demo rooms ... certainly the first four, for sure. They’re all really emotionally intense, which is why I listen to music. Both "Keep On Movin’ " and "Woman In Chains" have an unworldly beauty unlike anything else, and Ulrich Schnauss’ "Between Us and Them" is quite a sonic excursion.

I’ve been lucky enough to have several experiences that changed my life ... when I was ten years old and heard stereo for the first time, and heard a live performance of The Messiah the same year ... fifteen years later, hearing Shadow Vector quadraphonic for the first time ... twenty years later, hearing the Ongaku and Reichert DHT amplifiers on my new Ariel speakers ... and ten years after that, my own Karna amplifier. Profound, deeply moving, out-of-body experiences.

That’s why I find audiophile discussions of "accuracy" to miss the point. I don’t listen to music to sip a single-malt whiskey, smoke a Cuban cigar, scratch my chin, mark up a twenty-point checklist, and write long reflective essays about this or that subjective aspect of the sound. That’s silly. What’s the point? Who cares what XYZ critic thinks?

It either gets me flying or it doesn’t ... and if it does, how high?

 

Without making a big deal of it, Don and I have definitely taken the Blackbird several steps (with new parts, and some new circuits) beyond the amps at the show. You guys are in for a treat.

What makes the 45 or 300B "hard to drive" is that it is good practice to have the majority of amplifier distortion in the final power device, not the driver. In other words, the driver should be cleaner than the 45 or 300B ... which are the lowest distortion tubes ever made.

In a feedback amplifier, the source of the distortion doesn’t matter much ... the feedback sweeps it all away. Which is why the substitution of the higher-distortion 12AU7 for the lower-distortion 6SN7 in the mid-Fifties didn’t matter much, since feedback was in universal use by then, and it didn’t show on the low-resolution distortion analyzers of the day.

To be honest, zero-feedback amp design is kind of a cult audiophile thing. For that matter, any kind of tube amp is a cult audiophile thing. If distortion numbers come first, THX or Class D are the answer, end of story. Don’t mess with tube amps, just buy the solution off the shelf.

It’s an esthetic decision to build zero-feedback amps, whether bipolar transistor, MOSFET, or vacuum tube. I think it is good practice to design low-distortion zero-feedback driver sections, but I have seen (but not heard) all-transistor amps, with lots of feedback, used as 300B driver sections. Which begs the question, why use a 300B at all, if it’s just an expensive distortion generator?

Returnng a little more seriously to the original question, there are a lot of SET amplifiers with marginal driver sections. I’d go out on a limb and say the majority of 300B amps on the market sound mostly like overstressed driver sections, not like a 300B.

That’s why some of this discussion might sound like we are at cross-purposes. If state-of-the-art SINAD numbers are your goal, please look elsewhere. Forget all tube amps, whether pentode, triode, or hybrid amps.

If you want a taste of "tube flavor", get a preamp with a 12AU7 in it. I own a charming little Xduoo TA-10R, which is a AKM 4493 DAC, a 12AU7 gain stage, and a simple two-transistor Class A output stage. 2 watts per channel, sounds great, and all for $320 from Apos Audio. Lots of power to drive planar headphones, and a fun alternative to the usual Topping or SMSL.

Designers of zero-feedback 45 or 300B amps have different goals, which are also different from designing re-creations of Golden Age PP pentode amps. A lot of it comes down to esthetics and design philosophy.

Good read on the 6L6, designed by the RCA team that also created the 6V6 a year later.

Vacuum Tube Valley on the 6L6

And of course the 45

Vacuum Tube Valley on the 45

 

The tube Don, myself, and many other manufacturers would like to see go back in production would be a 45. NOS examples are astronomically expensive now, and surely production costs would be similar to, or less than, a 2A3. It’s basically a very simple tube, unlike a 6SN7 or a 6EM7, although "simple" is still plenty expensive relative to capacitor or transformer manufacturing.

In the meantime, we’ll be using triode-connected 6V6’s, which operate in the same power range as a 45 (275 volts at 32 mA = 8.8 watts), have the same (triode-connected) Rp = 1800 ohms, and have excellent performance. Many choices with NOS and current production.

Vacuum Tube Valley on the 6V6

45 Datasheet

 

Good, although brief, discussion of the 6EM7 on Thomas Meyers’s Vinyl Savor page. It looks like a TV tube, not designed for audio, so NOS examples might be somewhat variable (the application in the deflection circuits of a TV would only have moderate requirements for linearity).

No current production for obvious reasons ... vacuum tube TVs disappeared fifty-five years ago, along with the tubes that went into them. But there’s probably plenty of old stock.

Analog vacuum-tube TV’s, particularly the inexpensive B&W models, had pretty bad picture geometry. Once integrated circuits took over in the early Seventies, picture geometry got a lot better, and all of the many different service adjustments went away. NTSC and PAL color TV is a lot easier when all the complex signal processing is inside a single chip, instead of several tubes with many adjustments.

Collectors prize old 21" round-tube color TVs (1955 to 1967 vintage), but they are not easy to keep running (with many adjustments and 26 to 28 hot-running tubes) and CRT refurbishing services went out of business about ten years ago.

Would I use a TV-only part in a new design? No, I would not. Zero chance of LinLai or JJ putting it back in production.

I second what Don just said. Building a kit is far more rewarding than messing around with cables or component-swapping. You get to hear for yourself what XYZ capacitor sounds like, instead of reading random comments on the Internet. I can’t emphasize enough you just can’t trust what people say on the Internet ... but you can trust your own perceptions, first, last, and always.

It’s super educational, and focuses your attention on tuning the power amp, which is where it should be. Likewise, if you build your own speakers, you get to tune your own crossover. This is where the big payoff is, and it trains your ear in what to listen for.

Amp and speaker tuning set the sound of your system. Find out how to do that, and you are most of the way home.

Lot of solid-state in there, probably not my first choice.

It’s a safe guess it’s about the same ... in the 60 to 80 pF range. Most true 3-element triodes, triode-connected beam tetrodes, and triode-connected pentodes fall about there. So-called ultralinear mode is probably about half that. Pure pentode is a little higher than the socket capacitance, so about 6 to 8 pF.

The 45 and triode-connected 6V6 are similar, with preferred operating points: 250 to 275 V on the plate, 28 to 32 mA cathode current, and dynamic plate impedance (Rp) around 1800 ohms. The big difference is gain and bias voltage ... the 45 has a gain of 3.8 and negative 45 to 50 volt bias, while the triode-connected 6V6 has a gain of 8 and negative 14 to 15 volt bias. But otherwise similar, including Miller capacitance.

P.S. If anyone does bring back the 45 at a sane price, I’d love to see it in Arcturus blue glass. That would be very stylish, and a nod to its famous predecessor.

Arcturus 27

Vinyl Savor (Thomas Meyer) discusses the 45

 

Negative feedback is the right choice for the vast majority of amps, particularly direct-coupled solid-state, where you can pile on the gain and use that "excess gain" to minimize distortion via feedback. To oversimplify, if you have 20 dB (a 10:1 voltage ratio) of excess gain, you can have 20 dB of feedback, which will reduce the distortion in direct proportion to the feedback ratio ... in this case, ten times. Pretty slick trick.

In practice, as the excess gain goes up, and the feedback ratio increases, problems with stability creep in. Marginal problems with stability result in overshoots on square waves, and as it gets worse, brief periods of near-oscillation, and then full-power oscillation, which usually destroys the speaker. So you have to take account of the total phase shift on both ends of the spectrum, which includes the output transformer if it is included in the feedback loop. The phase shift of an output transformer typically limits tube amp feedback to no more than 20 dB, but this can be evaded by having multiple nested loops, as in the Citation amplifier deigned by Stu Hegeman in the early Sixties.

But now we get into the (much) deeper waters of both slew-rate limiting and settling time, which are interrelated. That’s beyond the scope of this discussion, but they are limiting factors in any feedback amplifier. Multiple feedback designs can achieve impressively low distortion figures, but settling times can be much longer, since each nested feedback network has to leave saturation, return to controlled operation, and return to zero with its own time constant.

These are not trivial design concerns, and made more complex by load dependence ... a reactive loudspeaker load will decrease the phase margin of the amplifier, and that in turn leads to longer settling times. As the phase margin erodes, settling times get longer and longer, until the amplifier breaks into self-oscillation.

The other consequence of loss of phase margin is an increase in distortion, mostly at high frequencies, with the limit case of oscillation, which can be considered 100% distortion, with the output effectively decoupled from the input.

For obvious reasons, great care is taken in the design phase to avoid oscillation, but there are amplifiers where stability is conditional on the load, with transient overshoots visible under some conditions of load and input stimulus. This was a serious problem with first and second-generation transistor amplifiers. (Which were designed with nothing more than slide rules and nomograms, so you can’t really blame the designers back then.) Nowadays, software modeling programs allow designers to avoid the stability problems of the early transistor amplifiers.

If you want to jump down into the rabbit hole, read about "Nyquist Stability Criterion", followed by "Slew Rate Mechanisms" and "Settling Times in Feedback Circuits". For advanced practitioners, read about "Mixed Feedback Designs" and "Combining Feedback and Feedforward".