Hmmmmm. Coherent 3D imaging at my listening chair or a wide soundstage? If that's the choice I need to make then would you know what's the best compromise to get equal measures of both?
Would it simply come down to a choice between the Azurahorn AH425 or the AH550? Azurahorn -Le Cleac'h Acoustic Horns - Products
But if it's to be the AH550, I'm too much of an ignorant newbie to then know how to design the passive filter that Gary and Pierre added to extend HF response of his Radian 745NeoBe driver beyond 10kHz with his AH425 horn. I'd probably be very foolish not to stick with what's proven to work!!
Thanks Marco, that's super important to know! Thank goodness, "I've seen the light". So its the horns for me.
Still, I really did wish that the Beyma TPL150/h + Altec 416s and my subs could have worked out. They're no less efficient than Gary's horns. But too much of a dispersion hole between them, said PK, who has these Beymas.
Thanks, Gary I was wondering about this here.
So was the filter that you and Pierre designed mostly done to attenuate a resonant bump or a notch in the Radian745NeoBe's response above 10kHz?
Great to know this too. Hopefully, using the First Watt B4 crossover to bi-amp them will make tuning them by ear (LOL?) fairly easy.
My two nieces who'll inherit my estate may have to worry about that more than me.
That's nice for the builder and seller of them. As for customers, aren't the "best sounding" 300B tubes (which die a lot faster than transistors) quite pricey?
http://www.sophiaelectric.com/pages/se/300B_plus.htm
Well, yeah. I mean I would think that even I'm qualified to judge Nelson Pass as being no slouch at what he does.
That much different? Be that as it may, like I'll probably never get to hear Lowthers (no one I know locally has any), its too bad that you probably never heard his J2 or the XA30.5 driving any speakers that you actually like, rather than very often paired the Lowthers, which you clearly do not. Perhaps someone here might attempt for themselves. I certainly know that will when I build Gary's horns.
For the benefit of anyone dropping in on this conversation, Lynn doesn't take issue with the use of transistor amplification in the bass. In his current system, the Karna amplifiers power the Ariels, and a REL subwoofer does its thing below. The upper half of the frequency spectrum is extraordinarily spacious and pure. It plays quite loud before running out of headroom, but the limit is there and Lynn is certainly aware of it. To my ears, the Ariels lose their luster from about 250 Hz downward. The frequency balance is uneven in this range, and sound is less distinct.
The whole purpose of this thread has been to move "beyond the Ariel," with the goal of moving past the Ariel's limitations without giving up any of its strengths. Improving bass performance and dynamic range hasn't been all that difficult, but doing so without sacrifices elsewhere is quite another matter. All of the choices involve tradeoffs, and we all have somewhat different preferences.
Lynn has experience with SE tube amps but doesn't use them in his own system. His Karna amplifiers have plenty of power for the Ariels, but the goal of increased headroom through higher loudspeaker efficiency is a high priority for him.
Gary Dahl
Now boys, no politics at diyaudio.com. No, I think he means me, the pesky wannabe design engineer....just kidding, Gary.
Very interesting post. I agree, especially when you include adjustment of the volume control.
I have always found it fascinating to note that live instruments always sound real, regardless of the room size, acoustics, or listening distance. The consequent variations in frequency balance are probably much larger than the ones we agonize over while optimizing our systems.
But I do have some thoughts about appropriate playback levels for classical music, regardless of a system's limitations.
I have noticed that going beyond a certain level of loudness (with orchestral instruments) detracts from realism. For instance, turn up the playback level of a solo violin. At first, it begins to sound like you are closer to it. But as you continue to increase the level, eventually it just sounds wrong. Put a violin under your chin and play it. The strength of the sound might surprise you. But it won't sound like that violin recording played back at a high level, for many reasons. For example, consider proximity effect, which lifts the lower frequencies as you get closer to the microphone. Had the violin been recorded close-up, the proximity effect would have been in the recording (unless filtered out).
This is an intentionally absurd example. If miking was done at a distance but the playback level is chosen to replicate being close to the instrument, our ears will receive conflicting cues. I think that realism is easier to achieve when conflicting cues are minimized.
When setting the volume for a well-recorded piece of orchestral music, I am really trying (intuitively) to balance several factors: the apparent miking distance in the recording, the recording level and dynamic range of the piece, the relationship between the soundstage in the recording and the soundstage produced by my system in my listening room, and the shortcomings of the acoustics in my listening room. If I turn it up too far, something will sound "out of whack" well before my system begins to struggle with the output level.
Regardless of which high frequency drivers or horns I use, there is no way I will ever recreate the enormous sound field that I hear from the podium. I think what I have been working toward in my own system is something I might characterize as "density." I'm not as concerned with holography as I am with tonal completeness and a sense of solidity. In my experience, the lower half of the orchestral spectrum has been the bigger challenge to recreate in my listening room. The success with the GPA416's is a big reason for my present enthusiasm.
The loudest sounds I encounter as a performer are in the pit during performances of the Nutcracker each year, Act II, during the Andante Maestoso in the Pas de Deux. My timpani and I are mostly surrounded by cement walls, and I am right next to the percussionist. At the climax, I am playing rolls at full force on my timpani, while the percussionist plays FFF cymbal crashes just a few feet from my head...beneath the concrete overhang of the stage, of course. I can't plug my ears because my hands are busy. Ouch. Last December I brought my iPhone into the pit with me (shh!) and turned on a loudness monitor app. I don't remember what the exact measured peak was, but I was surprised that it wasn't higher.
At home I listen for pleasure, so really, my system plays plenty loud enough for me.
Gary Dahl
Broken record - The density and fundamental tones and power of the instruments is in the 80 to 400 cycle region. To get it right in a home system is quite a challenge. Outside of a bass horn the only time I have been able to get it at home where it is double take convincing (as in WOW) is with a couple of lines of eight high sensitivity 10" woofers but compared to a good bass horn it's a bit messy and smeared where bass horns are dense with detail and articulation. The deeper bass needs separated and needs lots of surface area (you really can't over do it as far as quantity and quality of bass drivers) to lower the IM distortion and free up the 80 to 400 region to get the density without smear. Broken record 🙂
Last edited:
Oltos,
We really don't know what Gary has done in his passive xo, but I can tell you he can't be boosting the top end, you can only do that with an active circuit so I imagine he is doing as I say and has actually sloped the response by cutting some of the bottom range to make it appear that you have more high end energy. You only have the output that the driver can put out, unless you are using an active xo you have to cut to bring up another area in a relative balance. I am just conjecturing here, I haven't seen his circuit.
As to the wide sound stage, what I am saying is that even if you can get the upper frequency range to be more extended with an eq type network the dispersion angle is set by the horn and the exit angle of the driver. A larger driver usually has a longer throat length and a larger exit diameter but the angle is rather small compared to a smaller pancake style 1" exit driver, that will normally set the dispersion angle of the top end. If you aren't going to have a couch that is 10' wide where you want everyone to hear the same thing it won't be a real problem and in some rooms is an advantage if you can't do anything about the acoustics of the room.
As to your bass speaker not covering the entire room I don't think that will be a problem, you can't really control where all that bass is going to go, it will be basically omnidirectional in nature. Many of the gurus are now recommending distributed subs around a room to even out room modes but I don't think that most people would notice a problem without learning to identify the difference.
Gary said something earlier that I am not sure you clued into. He said he can not hear the high frequencies anymore. I often hear this and question those statements. What I mean by that is what is more typical is that your hearing acuity is just diminished and slopes down from a certain frequency and not that you truly can't hear these frequencies at all, that would be some more serious hearing loss. But since he is around orchestral music and get the full spl level where he stands he may be correct. I learned that the 20 to 20Khz numbers often used are just an average. Until I used a TAD ET-703 driver I believed the same thing, then I used those devices and realized that most devices just don't work that high, I could hear things in the music I never could hear before, the best high end driver I ever heard. They are just to expensive now to consider, they raised the price more than substantially for those devices just as all the other TAD drivers. I could at one time buy a pair of new TAD 2001 driver for $450.00 per pair from the sales rep. Those days are long gone.
POOH,
That amp surely doesn't look like a SE Cary amp, you understand what I mean by a serious tube amp. It takes more than a single pair of output tubes and some serious power supplies and output transformers. I still remember many of the old tube amps, whether they were Mac, Fischer, and even some of the really ugly green Altec power amps. Not to may designers are doing that today unless you have mega bucks for some obscure audiophile amps. I don't know if Manley has anything in that range but I am getting old!
We really don't know what Gary has done in his passive xo, but I can tell you he can't be boosting the top end, you can only do that with an active circuit so I imagine he is doing as I say and has actually sloped the response by cutting some of the bottom range to make it appear that you have more high end energy. You only have the output that the driver can put out, unless you are using an active xo you have to cut to bring up another area in a relative balance. I am just conjecturing here, I haven't seen his circuit.
As to the wide sound stage, what I am saying is that even if you can get the upper frequency range to be more extended with an eq type network the dispersion angle is set by the horn and the exit angle of the driver. A larger driver usually has a longer throat length and a larger exit diameter but the angle is rather small compared to a smaller pancake style 1" exit driver, that will normally set the dispersion angle of the top end. If you aren't going to have a couch that is 10' wide where you want everyone to hear the same thing it won't be a real problem and in some rooms is an advantage if you can't do anything about the acoustics of the room.
As to your bass speaker not covering the entire room I don't think that will be a problem, you can't really control where all that bass is going to go, it will be basically omnidirectional in nature. Many of the gurus are now recommending distributed subs around a room to even out room modes but I don't think that most people would notice a problem without learning to identify the difference.
Gary said something earlier that I am not sure you clued into. He said he can not hear the high frequencies anymore. I often hear this and question those statements. What I mean by that is what is more typical is that your hearing acuity is just diminished and slopes down from a certain frequency and not that you truly can't hear these frequencies at all, that would be some more serious hearing loss. But since he is around orchestral music and get the full spl level where he stands he may be correct. I learned that the 20 to 20Khz numbers often used are just an average. Until I used a TAD ET-703 driver I believed the same thing, then I used those devices and realized that most devices just don't work that high, I could hear things in the music I never could hear before, the best high end driver I ever heard. They are just to expensive now to consider, they raised the price more than substantially for those devices just as all the other TAD drivers. I could at one time buy a pair of new TAD 2001 driver for $450.00 per pair from the sales rep. Those days are long gone.
POOH,
That amp surely doesn't look like a SE Cary amp, you understand what I mean by a serious tube amp. It takes more than a single pair of output tubes and some serious power supplies and output transformers. I still remember many of the old tube amps, whether they were Mac, Fischer, and even some of the really ugly green Altec power amps. Not to may designers are doing that today unless you have mega bucks for some obscure audiophile amps. I don't know if Manley has anything in that range but I am getting old!
POOH,
I agree there is nothing like a bass horn to bring something to the party. I have done that but now I just don't have the room or have the permission to have them! I have the beginnings of some double 18" bass horns and I am more than sure you would like them, but you have the room for that and the permission to use them. I got rid of one set of neighbors with mine, I like my neighbors now so probably wouldn't subject them to the booming they would hear in their house. Stucco walls surely aren't going to stop 20hz waves, they would want to kill me.
I agree there is nothing like a bass horn to bring something to the party. I have done that but now I just don't have the room or have the permission to have them! I have the beginnings of some double 18" bass horns and I am more than sure you would like them, but you have the room for that and the permission to use them. I got rid of one set of neighbors with mine, I like my neighbors now so probably wouldn't subject them to the booming they would hear in their house. Stucco walls surely aren't going to stop 20hz waves, they would want to kill me.
Fair question. The most annoying thing right now is mechanical buzz from the B+ transformers; these have to be changed out for transformers (and chokes) that are silent.
Also not completely happy with the performance of the 5687/7044/7119 input tube. It has the low plate impedance the first interstage transformer requires, but nearly all the upper-harmonic content is coming from this tube alone. The 45 driver is clean as a whistle ... pretty much nothing above 3rd harmonic (measured down to -150 dB by Gary Pimm) and the 300B's are all good too (and the Chinese meshplates are perfectly good for the purpose).
I also found from Bud Purvine that the first interstage may have as much as 100 pF of capacitance, which is going to load down the first tube too much. It doesn't matter as much for the second interstage, which is driven by the 45, which has current and linearity to spare.
It's the input tube that's the delicate sister. So I'm contemplating a simplification; boring old RC-coupling between a good-quality vintage 6SN7 (they made a zillion of these, they're not that rare) and the 45. I'm not usually a fan of RC-coupling, but in this case, the 45's are never going into the grid-current region, and the total capacitive load (as seen by the input tube) is important. A variant would be a pair of Gary Pimm current sources to load the plates and capacitor-coupling to the 45 grids. The GP current sources have nearly infinite power-supply isolation, which is a plus for the input stage.
The key concept in the Karna amplifier is complete isolation between the power supplies of the power section and the rest of the amplifier. The distorted current draw on the output section does not affect the more sonically critical earlier stages, and the whole amplifier does not clip at once. Since the output stage is interstage-coupled to the driver, and the driver has enough capability to drive the 300B's into positive grid current (by about 30V), recovery from overload is immediate (in microseconds).
By contrast, the usual RC-coupling seen in traditional PP-pentode amplifiers leaves the output tubes in a grossly misbiased condition for about 0.1 to 0.5 second after a transient overload. That's where that "jukebox" or "guitar-amp" flabby bass sound comes from.
Interstage transformer coupling (between the driver and output) solves that problem. Recovery is immediate, and if the driver can linearly deliver 20 mA or more, the output tube will gracefully enter the positive-grid region (Class A2 operation) without a sharp increase in distortion. Subjectively, the Karna has about the same headroom as a Crown Macro Reference, mostly thanks to the combination of soft-clipping and immediate recovery from overload. I've never actually heard it overload ... the Ariels audibly compress before that happens, and I back off the volume.
The power supplies are also absurdly overbuilt. The 300B B+ voltage is 500 volts, the damper diodes can pass 1-amp peaks, and the even the Chinese meshplate 300B's are good for 300 mA transients. That's a lot of peak power.
Pretty much confirms what I have found in my V1 of the Karna. First stage just can't provide low distortion, no matter what IT I used (Lundahl, Onetics, Monolith). Tried tubes from SN7, 7119, 6N6P, 12B4, nothing was up to the task. Finally faced the truth that I needed a different topology, and RC coupling was the simplest. I have attached my V2 that is very promising; extremely low distortion from the SN7 stage, to the point where the 46 is becoming a significant distortion factor (and it already is a low distortion tube). No doubt you will have even better results with the 45; I'm just not willing to put the money into buying a population of them. I can share some FFT's if interested, but really this is the way to go IMO. You get all the benefits of IT coupling to the output stage, and the first stage never sees an unfriendly load; there really isn't a need for an IT after the first stage.
Also significant is the removal of the input transformer, as my preamp is fully balanced. Merely provides a nice high impedance balanced load, while also providing a low grid leak for the SN7. It is very easy and rewarding to set up a test jig and select SN7's that are optimized for a low differential output distortion. With a low cost but quality FT-3, I can live with the philosophical challenge of having a capacitor in the signal path.
I don't mean to answer for Lynn, but maybe my post will help. I am a huge Nelson Pass fan and currently use Pass Aleph 2 with my Quad ESLs. They are great amps, but they still do not have the tonal realism of DHTs. Lynn has a very well written page on his website chronicling his DHT push pull amp journey.
I have heard Pass XA60.5 with Daedalus Ulysses speakers which are very good speakers as far as commercial offerings go. The setup sounded a touch dry and again not as vivid as you'd get with a high power DHT SET amp or I imagine DHT push pull (but I've never heard one). The owner now uses a SET amp and is much happier with the sound. I haven't heard it with his tube amps, but trust his judgement.
To use another analogy I have a pretty good vinyl and digital setup. The digital does everything very well, but can't capture that feeling of a 50s Blue Note recording. When you can really hear the overtones of the alto, tenor or piano. And that is what I hear in real life at a small club with no microphones amplifying the instruments. I am going to look into transferring these LPs to digital, as I'd really like to learn more about the why or the how this is happening. I do not think it is simply "more distortion".
Thank you Lynn for the detailed reply regarding the Karna updates.
Last edited:
You cannot convert LP into digital without losing something. It is like taking Lance Armstrong and reducing him to a good club rider
.
Lynn,
Just a suggestion - if the chassis is steel (particularly on the light side), the problem might not be the transformers. I recently had an issue in an amp I made where it turned out to be the chassis box buzzing in response to the transformer's magnetic field. Some spacers and rubber pads did the trick.
A few years back I got rid of an amp because I could hear acoustic hum and buzz. Now I wonder if I could have just spaced the transformer a little further from the steel chassis on that one, too. 😕
Yup, your schematic is pretty much what I'm thinking. The input tube just doesn't like all that capacitance, so RC-coupling (or using a very low-capacitance current source like the GPimm version) is most likely the way to go.
Since I'm retaining the input transformer (partly because it breaks the ground and filters off VHF crud from the DAC), I'll audition a long-tail resistor on the common cathodes of the 6SN7 versus a cap bypass (of the same resistor). It seems like small change ... just one part ... but changes the operation of the PP pair from differential to balanced. This has a measurable effect on the harmonic structure of distortion and the peak-current capability.
Technical digression: true differential circuits are series circuits. If one tube clips or runs out of current the whole diff-pair shuts off, like Christmas-tree lights when one burns out. Diff circuits are characterized by hard current clipping, and they behave a somewhat differently short of full clipping as well. In comparison, balanced PP circuits operate in parallel, and clipping or current-limit in one device leaves the other device to turn on as hard as it likes.
This is significant because audio tubes are normally biased at a small fraction, sometimes as small as 1/10th, of the peak-current capability of the tube. In a SE or true differential amplifier, the entire circuit is limited by one tube going to zero current; that shuts off the high-current side as well. In a balanced PP circuit operating in Class A, the other tube is free to turn on as hard as it likes ... and this peak current is many times higher than the quiescent current.
That's the major function of the common-cathode bypass caps in the Karna. The balanced tubes have a better-sounding spectrum than a differential circuit, and have far more peak-current capability. Since the primary challenge for vacuum tubes is driving the stray and Miller capacitances, anything that linearizes current delivery improves HF and slewing performance.
One of the not-so-nice tests I did with the Karna was trying to see just where the slew limit might be. I cranked it up to full power ... around 20 watts ... and then just kept increasing the input frequency. When the level started dropping thanks to the LPF of the input transformer, I just raised the input level so the output remained level at 20 watts. I finally lost my nerve at full power at 500 kHz, with no visible sign on the scope of the wave turning into a slew-limited triangle. That's kind of a horrible test, making the amplifier into an AM transmitter. But the tubes seemed happy enough ... no red glows on the plates or anything. No change in performance ... worked just fine. Yes, I did this on an amplifier with new factory-matched 300B's and old-stock 45's.
When I talk about headroom, this is the kind of thing I mean. The amplifier never really reaches the hard edge of the performance envelope, and is safe from line transients or any conceivable input signal, including RFI incursion.
A related topic is Class A biasing. Traditionally, a SE 300B amp runs the output tube at 60 mA. A SE amplifier is always in Class A, of course, since there is no "other side" to the circuit. Similarly, if you simply double the circuit and build a balanced amplifier that operates in Class A, each tube gets 60 mA, just like the textbook recommends. (By comparison, a Class AB amplifier runs each tube between 30~35 mA). As one tube shuts off and goes to zero current, the other tube turns on harder (as mentioned above, a diff circuit hard-clips when one tube goes to zero current).
One of good things about living in Tek-land and having friends with a lot of good test gear is testing out hypotheses, and see if the textbooks are really right. Well, they're not when it comes to Class A biasing. It turns out that distortion went down two times when I increased the quiescent current for each tube to 80~85 mA. That's a honking big difference, and yes, you're going to hear that. This is what I mean when I talk about "deep Class A" ... essentially, taking it deeper into Class A territory, and in this amplifier, with an additional Class A2 drive capability.
So the amplifier can run anywhere from 20 watts Class A1 (normal program material) to 30~40 watts Class A2 (transient peaks) to Class AB2 (transient peaks combined with an elliptical load-line from a reactive loudspeaker load). That's what the 300 mA peak current (at 500 volts!) capability is for.
One nice thing about tubes is that they accept transient (for a few seconds) currents and voltages far outside published specs. What wears them out are unstable fixed-bias circuits where one tube of a quartet starts to current-hog and destroy itself (ahem, Audio Res ...), running the screen at a voltage 150V in excess of the published rating (very common in "ultralinear" circuits), or not respecting the steady-state plate dissipation (75% of rating is considered conservative operation).
Last edited: