That long away? If I understood Lynn Olson correctly-and he's clearly an authority on the good and the bad of probably every kind of speaker technology to date-horns (or maybe, for the HF range, AMTs as well) will indeed retain "....all the detail and fullness of the soundscape" regardless of volume level.
As I completely trust Lynn's experience and judgment on speaker design and compatibility issues-and because you personally use among the best performing Pass amps, I obviously now have little reason to be overflowing with optimism for the day when I finish build Gary Dahl’s speakers and drive the horns with my J2 amp. And I do recall Lou from DaedaIus saying that he likes how the Pass Labs sound with his speakers amps, but prefers the sound of tubes. If I get lucky my ears may compel me to trade that amp in for a 2A3 amp (though not some $$$$$ PX25); maybe the F4 amp and my sealed Altec 416 midwoofers will like each other. Perhaps the J2 would have sounded beautiful with non-Lowther OBs, but I don’t have at least the requisite 4 ft to spare in back of them. These confirmations are very upsetting as it for it took me a long time to decide on these particular Pass amps, rather than several tube amps that I was strongly considering. At least I can use these ss amps to set up my First Watt B4 crossover when I bi-amp the system-although I’ll probably have to re-adjust the B4 again when I install the tube amp(s).
QUOTE=boldname;4381549]You cannot convert LP into digital without losing something. It is like taking Lance Armstrong and reducing him to a good club rider .[/QUOTE] I'd think it doubtful that much, if anything, would be lost using this Platinum ADC <i>plus</i>
No, seriously, My2Tech, Lavry, Ayre Acoustics and certainly Benchmark Media and others make really great performing ADCs. Just remember that you'd want a DAC of at least equal quality for playbacks.
Uh, hardly. I'm a system integrator more than anything. Other folks know far more about horns, cabinet-edge diffraction, advanced measurement technique, simulation, DSP, active crossovers, dipoles, subtle details of driver construction, the list goes on.
I glean information from others so I can create devices that induce a psychological state in the listener. This approach works for me and a few people I know.
The majority of audiophiles I've met don't "get" the psychological state I'm after; they're into the attractions of the mainstream ... precise imaging, "slam" (whatever that is), a certain type of exciting sound that reviewers like, and other stuff I don't relate to.
I took a vacation from high-end audio in the Reagan/Thatcher years of the Eighties, and when I returned, it had been transformed out of all recognition by the Big Two magazines. It was new and alien territory for me, and a relief to jump into the smaller pond of the vacuum-tube revival.
A couple of years ago, I met a fellow at one of the usual hifi get-togethers, and he was busy was checking off a full-page, twenty-point list. He recognized my name, said "hi there", and asked me how I assess hifi gear. I told him I either I like a system or I don't, and wouldn't know where to start if given a 20-point checklist of subjective qualities. I don't listen that way; it's a gestalt experience, all one thing at once.
I expect a hifi system to induce psychological states in me, and if it doesn't, there's a reason why. This is where it gets technical. Faults jump out at me and break the "suspension of disbelief", the same way as badly-written dialog, a continuity error or a gross historical anachronism will take me out of a movie. It's pretty much the same when I hear grossly nonflat response, driver resonance, poor crossover integration, or a cluttered and "dirty" time signature.
Most of the time I'm just chasing out faults that break the suspension of disbelief. Other people focus on different things, and don't hear the faults that I notice, or notice faults I don't hear.
I'm not very sensitive to absolute phase, for example. It only sounds like a moderate timbre shift on some program material, and isn't noticeable on other kinds of music. But that's just me. I've met people who can't listen to music that's been inverted in phase, so they have a ready-to-hand phase switch for every track. (The same LP or CD can have tracks recorded at different studios that have different microphone phasing.)
The direct-heated subset of the vacuum-tube crowd are into the expressive qualities of a hifi system, and we share a language that is not the same as the language of the magazine reviewers. This linguistic mismatch makes it difficult to communicate with mainstream folks, since we're not talking about the same things, and more importantly, don't value the same things.
Last edited:
Look at the work of Alexander Shulgin, author of Pihkal and Tihkal. Shulgin is a chemist who created hundreds of psychoactive compounds that have effects that range from subtle to absolutely overwhelming.
I'm not a stranger to this; in the late Sixties, before these things were banned and demonized, I sampled synthetic psilocybin and LSD-25, as well as hashish. It goes without saying these are very powerful and anyone will notice a dramatic change in psychological state, and it can be terrifying as well as transformative.
I had already been meditating on and off for several years before this, and noticed that there are a very wide variety of possible states-of-consciousness, all of them seeming very real while they are happening. Dreams seem real when they're happening; it's only when we wake up it all evaporates in the light of day. Serious practitioners of Buddhism and Hinduism assert that the "normal" state of consciousness ... the one you are in right now ... is also a dream, and that "samadhi" is a sudden breaking out of the dreamlike "normal" state into the light of full consciousness.
I view eating, making love, daydreaming at a sunset, piloting an airplane, dreaming, creative reverie, ecstatic states, or listening to music, as discrete and separate states of consciousness. They're all real; they're all part of being a human being. The constant fluctuation of awareness is a basic feature of consciousness itself; meditation is nothing more than noticing this fluctuation as it happens, and not judging it good or bad.
In this context, listening to music, and being profoundly affected by it, is a state of consciousness. It is as real as a dream, a deep emotional experience, a kiss, or watching a puppy playing. It's all real.
From my perspective, a hifi system is a machine to induce a state of consciousness in a receptive listener; I nicknamed it an "Illusion Engine" some years back. Since the goal is a state of consciousness, the only thing that matters is whether or not it works.
The entire electromechanical system of sound reproduction is as artificial as psilocybin ... actually, more artificial, since psilocybin is found in mushrooms all around the world, and there are receptors in the brain that lock-in to these compounds. These chemicals, and the receptors, have been there for millions of years, shaped by the lathe of evolution.
The electromechanical system of sound reproduction is only about a hundred years old, and we don't have all the answers yet. It is thoroughly artificial, made by the hand of man, and is intended for the propagation of art and the creation of musical states of enchantment. Enchantment is meant in the most literal sense, of a chant, phrase, or musical expression that leads to an altered state of consciousness. Temple chants have been around for thousands of years, and shaman's chants and group singing have been around for tens of thousands of years. We are programmed by evolution to respond to music; we can't help it. We are descendants of tribes that out-competed tribes that were silent.
So a hifi system is a machine in the same way that Shulgin's chemicals are machines that alter consciousness. It either works or it doesn't. If the desired effect is only partial, unsatisfactory, or incomplete, there are good technical reasons why.
I'm not a stranger to this; in the late Sixties, before these things were banned and demonized, I sampled synthetic psilocybin and LSD-25, as well as hashish. It goes without saying these are very powerful and anyone will notice a dramatic change in psychological state, and it can be terrifying as well as transformative.
I had already been meditating on and off for several years before this, and noticed that there are a very wide variety of possible states-of-consciousness, all of them seeming very real while they are happening. Dreams seem real when they're happening; it's only when we wake up it all evaporates in the light of day. Serious practitioners of Buddhism and Hinduism assert that the "normal" state of consciousness ... the one you are in right now ... is also a dream, and that "samadhi" is a sudden breaking out of the dreamlike "normal" state into the light of full consciousness.
I view eating, making love, daydreaming at a sunset, piloting an airplane, dreaming, creative reverie, ecstatic states, or listening to music, as discrete and separate states of consciousness. They're all real; they're all part of being a human being. The constant fluctuation of awareness is a basic feature of consciousness itself; meditation is nothing more than noticing this fluctuation as it happens, and not judging it good or bad.
In this context, listening to music, and being profoundly affected by it, is a state of consciousness. It is as real as a dream, a deep emotional experience, a kiss, or watching a puppy playing. It's all real.
From my perspective, a hifi system is a machine to induce a state of consciousness in a receptive listener; I nicknamed it an "Illusion Engine" some years back. Since the goal is a state of consciousness, the only thing that matters is whether or not it works.
The entire electromechanical system of sound reproduction is as artificial as psilocybin ... actually, more artificial, since psilocybin is found in mushrooms all around the world, and there are receptors in the brain that lock-in to these compounds. These chemicals, and the receptors, have been there for millions of years, shaped by the lathe of evolution.
The electromechanical system of sound reproduction is only about a hundred years old, and we don't have all the answers yet. It is thoroughly artificial, made by the hand of man, and is intended for the propagation of art and the creation of musical states of enchantment. Enchantment is meant in the most literal sense, of a chant, phrase, or musical expression that leads to an altered state of consciousness. Temple chants have been around for thousands of years, and shaman's chants and group singing have been around for tens of thousands of years. We are programmed by evolution to respond to music; we can't help it. We are descendants of tribes that out-competed tribes that were silent.
So a hifi system is a machine in the same way that Shulgin's chemicals are machines that alter consciousness. It either works or it doesn't. If the desired effect is only partial, unsatisfactory, or incomplete, there are good technical reasons why.
Last edited:
Imo, if you were deaf, 100 years with measuring equipment would not help to some extent. If you don't hear it, you just don't know the meaning of 2dB high Q peak, do you? Those who can "listen", to make decision they will normally depend on others who also can "listen" but with better knowledge and experience.
I made it separate between musicality and enjoyment, where distortion (mostly non-linear) determines whether a musical sound is sufficiently enjoyable or not. On quick listen, I will find out if the distortion is too high or not.
What some measurement guys do not understand is that we need language to communicate what we HEAR. Of course, without measuring, we cannot say "Oh there is 3dB peak at 10kHz", or "The 5th order distortion seems higher than the third!"...
The language/terms can be meaningless for some, but NOT for some others. When we gather with audio friends and listen for the same system together, when one of our friends says "Wow, the treble is very airy..." then because we can hear the same thing, we at least understand what term is used by others to describe that sound characteristics.
I don't know, Jay. A 3dB peak at 10k is not 'airy' to me, it's out of place and unwanted. It's a distraction from what I want to hear. When what I hear resembles something natural I relax and accept it.
I hope I didn't mislead what my measurement was based on. The FFT was the output of the first SN7 stage, taken differentially. It is not the 4 ohm output presented at the speaker. The measurement was approximately 20V worth of signal, which would be an effective 26W to a speaker when it passes through the remaining stages of the amp. The warm accolades may now cool off, perhaps.
My internal process was to find the lowest distortion input stage I could muster, with the former design driving an IT like in the Karna presented on nutshellhifi. As much as I was interested in avoiding the SN7 (I like using less popular and lower cost tubes), the numbers were strongly confirming that as long as I could dictate the load of the input stage the SN7 was the way to go. So I presented a capacitive load to the stage closely resembling the capacitance of the triode-connected 46. CCS was a plain jane cascode DN2540/10M45S, nothing fancy there. I figured, since I am driving a capacitive load of a few tens of pF, is there really a benefit to using the Pimm design, with its ultra-low capacitance?
Regardless of the tube (SN7, 7119, 12B4, 6N6P, 6N8S) and IT (Monolith, Lundahl, Onetics) combination that I tested, results were consistent in that the use of the IT introduced 2nd/3rd that were of similar magnitude, along with an easily-measurable 5th. The Monolith was preferred, as the 3rd and 5th were comparably less, but still similar pattern-wise. The Monolith, having a bifilar winding technique, provides very tight coupling of primary and secondary windings, with low leakage but high capacitance. You also have to respect the maximum rating of inter-winding voltage capability, which is a limiting factor in bifilar windings. Fortunately, the Monolith carries a 375V rating, and my power supply is a gas-regulated 300V. The bandwidth of this transformer was about triple that of the others; not just in magnitude, but also phase shift - the bifilar winding did not suffer nearly as much from wonky phase shifts as seen on a lissajous pattern. Yves was kind enough to gap the core suitably for push-pull operation; just ask.
Therefore I determined to dispense with the first stage IT, using it only for the 300B driver stage. That's where the real benefit of the IT far outweighed the compromises.
Kevin at K&K Audio has a very slick design of a differential headphone amp that provided me with some inspiration. Where I differed from his topology was the DC coupling of first and second stages. Getting rid of the capacitor was tempting, but I was concerned about the bias conditions straying over time. With the RC coupled topology, I no longer care about bias conditions of the SN7 stage. I fix plate current and grid voltage, allowing plate voltage to fall where the triode chooses. All I am concerned about is the differential output. No adjustments required, no periodic checks (as long as things are functional).
So it becomes practical to hand select envelopes with desirable matching (not of bias conditions, but distortion). I built a down-and-dirty test fixture at identical conditions as the final product, and went through my batch of 100+ SN7's. To keep cost reasonable, I went with the less popular 12SN7. The sound card was able to identify those that produced the most desirable FFT differentially; no need to test individual characteristics. As long as I was doing the test, I did record bias conditions, along with 60/120 Hz hum and individual harmonics to the 7th in a spreadsheet. Any harmonics higher than that were "in the grass", and likely not to be trusted anyway.
Results were most informative. Some tubes provided similar 2-3-5 harmonics, which were nominally low and 'good'. Some provided high 2nd, but odds that were 100-110dB below the fundamental ! This was curious to me, in that theory says a differential stage cancels evens but adds the odds. Some were of course so-so and useful for startup and testing, but not production. Finally, some were so darn beautiful that you had to sit back and smile. Almost questioning the test equipment (is this thing on?)
So once you find a low distortion triode pair in a single envelope, you are golden. Even if the DC parameters are non-ideal. If I was constrained to a DC coupled topology, the lowest distortion tube may not be the ideal tube to use, as it would throw off the operating point of the next stage.
Okay, that confirmed the desired input stage I wanted to use. Operating points dictate that regardless of input level, for all practical purposes the 46 stage will never enter positive grid current, so no worries there. Similarly, no way the SN7 stage would ever see grid current, so my preamp will be happy (it is xfmr coupled with a 100 ohm or so output impedance, so no worries there anyways).
For the 46 driver stage, I was satisfied with the 30V of headroom provided to the tail CCS. Datasheet for the depletion mode FET indicates 20V would be a minimum. I did experiment with adding some additional 20V of negative supply to the CCS, and results were unchanged. That's nice, I can keep it simple with no auxiliary power supply and maintain performance. Bias conditions are again somewhat irrelevant; fix plate voltage and total current; triodes will have to find their happy point based on common grid voltage and current sharing. I repeated the FFT tests, documenting individual triode characteristics in a spreadsheet. When the amp is built to a reasonable degree, I will play around with tube swapping to determine if I want to select push-pull pairs based on bias or FFT.
Tests of the 46 in SE with CCS load were informative. The tube is mainly 2nd (-65 to -70 dB) and 3rd (-75 to -80 dB), with higher harmonics 110 dB or more down from the fundamental. Given the even order cancellation, I am excited to see the final results, with the understanding the IT will add some distortion of its own. Initial trials with AC heating were still good - 5th was -105 dB below fundamental - this is what would be presented to the grids of the 300B.
I have no doubt the use of the 45 triode will be stupendous. As Lynn noted, he couldn't even measure the 45's distortion; everything was generated by the input stage. Just not willing to pony up the coin. I paid $20 each for a bundle of 46's a few years back. Have maybe 40 of them. I have found it really pays to take the time and hand-select. They tend to be a little hummy/noisy (they are directly heated, after all) so the Coleman regulators will be a welcome mod. With AC heating you can definitely see some 60/120 Hz spurs along side each harmonic.
I had an internal debate about going fully differential on the driver stage. I understand Lynn's rationale for going balanced with capacitor from B+ to cathode and cathode bypass. What I found in my first build was that my 300B stage was distorting at the plates well before there was distortion introduced at the grids (even before the onset of grid current). So I'm not sure I need to plan for extreme amounts of grid current. However, I did play around with adding load to one side of the IT secondary to simulate the onset of 300B grid current, and the differential stage seemed to be just fine with it. All I had in my toolbox was the FFT, but the driver stage seemed to be happy. Perhaps if I really drove it super-hard things would change, but I don't listen to music that way. At present I can drive 25W without severe driver stage distortion, so I guess I have made peace with the differential stage. I want the first 10W to be the best they can be, and I am approaching that without going extreme.
Didn't see any real improvements to the output stage, so that remains as is (other than DC heating). Based on the great performance of the Monolith IT, I placed another order for their output transformers. Perhaps that will be another improvement, but I haven't gotten that far yet. Post 12596 has the schematic; you can get the operating points there.
Sorry to be long-winded. Like others have noted, a separate thread might be appropriate, but then again this is one of the more common threads Lynn participates in. Up to him in my view; it's his thread.
zigzagflux/jpak and others in interest:
About CCS...
I noticed a harmonic superiority in the Gary Pimm CCS "core" testings, see: http://www.pimmlabs.com/web/solid_state.htm. The core for SSP used in Solid State Tabor is same core for Gary Pimm CCS, so CCS's will follow same rule. See superior FFT measurement from SSP vs single MOSFET, especially under high Z drive (...like 6SN7 anode, 7k-10k rp is not that low...). Futhermore, despite the capacitances from conventional current sources be low enough for use in theory, using cascoded CCS not only reduces the raw value of capacitance but also reduces its non-linearity at voltage excursions, which may sound less coloured, depending on the case. In some of my projects I used the Gary Pimm CCS and works wonderfully. And call me crazy, but I could never stand the sound from simple CCS's, unless they are used in very low rp valves or power supplies (the commom CCS sound grainy to my ears).
I've never tested the Rod Coleman filament regulator, but based on the results of Gary Pimm CCS's (over-engineering at first glance), is probably the way to go.
About CCS...
I noticed a harmonic superiority in the Gary Pimm CCS "core" testings, see: http://www.pimmlabs.com/web/solid_state.htm. The core for SSP used in Solid State Tabor is same core for Gary Pimm CCS, so CCS's will follow same rule. See superior FFT measurement from SSP vs single MOSFET, especially under high Z drive (...like 6SN7 anode, 7k-10k rp is not that low...). Futhermore, despite the capacitances from conventional current sources be low enough for use in theory, using cascoded CCS not only reduces the raw value of capacitance but also reduces its non-linearity at voltage excursions, which may sound less coloured, depending on the case. In some of my projects I used the Gary Pimm CCS and works wonderfully. And call me crazy, but I could never stand the sound from simple CCS's, unless they are used in very low rp valves or power supplies (the commom CCS sound grainy to my ears).
I've never tested the Rod Coleman filament regulator, but based on the results of Gary Pimm CCS's (over-engineering at first glance), is probably the way to go.
I know what you mean, but I was not saying that 3dB peak at 10k is "airy" 🙂 But may be, the peak existence will make sure the frequency is being heard? 😀
Seriously, we cannot avoid using terminologies like that, even if it is subject to misinterpretation. That is why, it is a good idea to document those terminologies to make sure people will use it more properly.
But you had a point that many of us can get used to a certain sound characteristics or quality. So basically we can choose to get used to a proper sound or the opposite. Surprisingly and fortunately (or not), we cannot train ourselves to get used to dissonance and persuade ourselves that the sound is enjoyable.
I have made my decision myself. Realizing that sound reproduction is no more than an illusion engineering, and that we cannot avoid compromises, I have put "enjoyment" as the main objective. Of course, I have for years tried to find out what is required technically to achieve that. But my point is, I don't care to differ (with the mainstream) as long as the objective is met.
There is perhaps two variables i.e absolute accuracy(technical), and there is perceived best sound. The latter is the problem. Never take another persons view of the best sound without audition. That goes for all of us.
I will save it for the book, i'll stop. I like my DHT amp but valves are not inherently better. Not at all. That is all I am saying. but the constant damning of solid state is now passe, I think the word is.
I know that i can hear past 45khz because i have had my hearing tested that far.
Many people can hear past 20khz but do not realize it.
I also can hear below .01hz but there is no music down there worth hearing
Be honest, most often the first variable (absolute technical accuracy) is needed only for political reason.
Tho I have learned that to a great extent the absolute accuracy is needed for absolute performance.
I think Lynn has mentioned that along with technical accuracy he put subjective performance as one of the criteria. I think this is needed because we don't know the exact relationship between technical variables with perceived quality.
That is why if you have good ears, you can trust someone like Lynn, imho. I don't buy the "trust your own ears" idea. Sometimes we have to trust other people ears more than our own.
It's all subjective isn't it? Point is, you like your DHT amp, and that is important.
It is, in the graph posted, which was an effective 26W operating point of the first stage. If I measure what would result with 1W operating point, the 2nd is 93 dB down, with no other measureable harmonics.
I only provided one FFT spectrum, but described two in words.
Ah! Yes, you are right. I had a friend who liked to ask me to go out watching movies after work hour. But I have learned that I don't like movies he likes and he doesn't like movies I like 😉
So of course, we should know what it means when we have to trust someone's ears (not trusting preference).
Yeek! Please don't do that. I've only met a handful of people over my entire career in audio that heard things in more-or-less the same way I did. The majority of audiophiles are listening to different things than I do, and most certainly valuing different things than I do.
The valuing part is important. Values are part of who you are, it what you care about, and the events and feeling that have meaning for you. I'm not sure values are transmissible except through a process of direct exposure to events that activate them.
My writing is vivid and at times eccentric because I'm trying to convey things that are not easy to express in English. Sanskrit, maybe, since that language has precise terms for perceptual impressions and states of consciousness.
Perception and consciousness are intertwined and pretty much inseparable, so there is a state dependency on how we hear, and more importantly, what we feel when we hear music. We bring all of our personal and shared memories and feelings to a musical experience, whether it's live or electromechanically mediated.
Returning gently back to Earth, I want to thank Zigzagflux for his rigorous research in the performance of the input stage of the Karna amplifier (and others of similar topology). He's confirmed what I've long suspected about the 5687/7044/7119 family of tubes ... plenty of drive-current capability, but a little too close to the 6DJ8 in terms of high-order harmonics.
But ... Gary Pimm and I have confirmed audibility of difference types of current-source. They definitely have a subjective signature ... I regret to say "solid-state", but that's what we both heard on direct A/B/A comparison. The big reason he went with a cascode architecture was to minimize MOSFET coloration (by shielding nonlinear capacitance) as well as doubling the voltage standoff capability. Pimm and I go back and forth whether his current sources have a sound or not; I contend there's still a teeny-tiny trace of solid-state left, but way less than anything else with transistors in it.
The alternatives are audio-grade (not power-supply!) inductors, but they have problems with stray capacitance as well. The 6SN7 (or 12SN7) has to drive the 60 pF of the Miller capacitance of the 45 or 46 as well as any additional capacitance from the current source or inductor. Only plain old wirewound resistors are free of the (possibly minor) concern about capacitance ... and they load down the tube more.
Last edited:
By the way, I see all this detailed chit-chat about vacuum-tube architecture, perception, and various circuit approaches as directly related to the new speaker. Like the Ariels, they have been designed from the outset to be used with high-quality vacuum-tube amplifiers.
Some readers believe that there is no such thing as a high-quality vacuum-tube amplifier, and a well-engineered transistor amp is the way to go. I would suggest that they borrow the elements of the new speaker that appeal to them, and ignore all the vacuum-tube arcana.
Although the Karna and the new speaker are designed to be used together, they comply (approximately) with industry standards for input levels and the expected output impedance of the amplifier, so there's room for mix-n-matching.
Some readers believe that there is no such thing as a high-quality vacuum-tube amplifier, and a well-engineered transistor amp is the way to go. I would suggest that they borrow the elements of the new speaker that appeal to them, and ignore all the vacuum-tube arcana.
Although the Karna and the new speaker are designed to be used together, they comply (approximately) with industry standards for input levels and the expected output impedance of the amplifier, so there's room for mix-n-matching.
Last edited: