Home |
Search |
Today's Posts |
#2
|
|||
|
|||
Yo, Frank, if you intimate enough with my mind to build one of my amps,
you should be on first-name terms as well. Dear Mr Jute, I have carefully read all the fascinating chapters of the KISS Amp story you published and referred to the relevant sections of Langford-Smith's book. That's a lot more than the kibbitzers have done. Forgive me if I sound like a thick old automobile engineer but I have some questions. You are too modest, sir. The end result of your KISS Amp 300B project is not one amp but two amps, right? Right. The T44bis Populaire is a standard good, indeed excellent, 300B design with a two-stage 6SN7 front end. The T39 Mk VI Ultrafi is what it says, a 300B with a 417A front end designed to the tastes of the minority SE niche I christened the Ultrafidelista about a dozen years ago. How should the person who builds his first 300B amp choose between the two designs, one with a "cascade" of two 6SN7 and the other with a single 417A? The builder of a *first* 300B amp should not choose between the two. He should first build the Populaire, otherwise he will have no reference. That is why I kept calling the Populaire the *standard good*. This is an amp that can compete on level ground with silicon numbers but that sounds better. Once the 300B virgin has built this elevated quality reference, he can build the Ultrafi and understand what I am talking about when I say it lifts your music onto a different level. If you don't do it stepwise, you will not understand that the Ultrafi is as much an engineering marvel as a marvel of taste. I am not at all keen to encourage the mysticism which has conquered a significant part of audiophilia, nor am I keen on people doing what I say because I say so; I'd rather they took the journey with me and decided at each stage that I am right and therefore they will take the next step with me. Comparing the two schematics you published and the related frequency response graphs for the two amps, they are tuned differently, right? Right. That is the only reason for building two amps. (Boredom is not a good reason for working with high voltage!) I assume from the graphs on your netsite that the flat amp with 6SN7 you call "Popular" is intended to work with any relatively sensitive speakers and the "Ultrafi" with 417A is tuned to your preferred Lowther horns. But can they be tuned the same? Both will drive my Lowther horns well, though the Populaire (not Popular as you have it) will require rolling off higher in the LF to protect the speakers. Whether they *can* be tuned the same is one of those interesting but ultimately useless questions that will run and run on RAT. The short answer is yes, they *can* be tuned the same. As always in tubes, the short answer hides more than it illuminates. The two amps are *deliberately* differently tuned, the Populaire to be flat (near enough DC to daylight) and have the lowest possible THD within the constraints of the available transformers and of course absent NFB; it is a precision instrument. The 417A was chosen for the Ultrafi for the quality of its sound regardless of the fact that it is less linear than the 6SN7 (the 417A is still a very linear tube). To make the 6SN7 work like a 417A is to pervert its character, though it would be easier than the other way round. To make a 417A work like a 6SN7 is simply an engineering waste, considering how much cheaper a good 6SN7 is; it would take an enormous amount of development work. In 1998 I built your "Triple Threat" push-pull design for EL34 tubes. I have owned three well-regarded commercial tube amplifiers since and always go back to your amp. Did you ever publish a similar description of the thought process behind this fabulous amplifier? I'm glad my design continues to satisfy, though I am not at all surprised. The T113 Triple Threat is the best value amp I ever designed. Operated in triode mode with the NFB turned off, it is also one of the best sounding. It is extremely speaker-tolerant in UL mode, in fact a most adaptable and amiable all-round amp. You have to spend a *lot* more to get a better-sounding amp, another fortune for suitable high-sensitivity speakers because the better amp will almost inevitably be less powerful, plus possibly more on your sources as well to be able to hear the marginal improvement at all. I published a short article with the circuit about 1996 but can't now find the article. The thought process behind the T113 can anyway be summed up in a single challenging sentence: *Design a good amp with all components except transformers straight out of the RS catalogue.* Thank you very much for sharing your thoughts and hard-won experience so generously. I doubt I'll do it again. When I came into electronics I was very generously helped by my seniors, men imbued with the old ARRL ethic of service and mutual assistance, and I have tried to put something back. As a libertarian I also welcomed the internet as enabling people's opinions but I was soon disappointed by the nastiness. Now I'm becoming bored with being a target for every sneering, jeering little scumball who thinks that abusing his betters will bring him fame beyond his desserts. Yours sincerely, Frank B. Good luck. Andre Jute wrote: Reach The KISS Amp through http://members.lycos.co.uk/fiultra/ or directly at http://members.lycos.co.uk/fiultra/T...mp%20INDEX.htm on which KISS 190 is an index of schematics and illustrations. All text and illustration is copyright property and may not be reproduced except in the thread KISS xxx on rec.audio.tubes KISS 124 Shaping the Ultrafi driver stage by Andre Jute By the time we come to design the driver stage of the ultrafi 300B amp there are so many hard points settled by the process elsewhere in the amp that the stage almost designs itself. Let us look at some of those hard points. For historical reasons I decided on: -a triode -a Western Electric tube -a single tube to do duty both as input and driver -a 300B output tube which has a particular signal voltage requirement All of this together fixes the input tube to a choice between the 417A and the 437A, which, given their relative availability and price is no choice whatsoever: the 417A is as WE tubes go inexpensive. Choosing the 417A is no hardship: it is a gloriously musical tube, a little wonder. For reasons of taste I have chosen to use: -a tube rectified input, which drops much more voltage over the diodes than silicon rectifiers -a choke input power filter, which requires two chokes rather than one, so doubling the voltage drop in the filter, which further requires -a big bleed because swinging chokes are not available these days; I would anyway want the big bleed for stability and safety For reasons of common sense, aesthetics and cost, I have chosen; -to buy my power transformer the same place as my output transformers -all of that fixes the voltage available before the power isolation filter for the driver to 396V From consideration of the necessary slew rate current and the Miller effect, I want to put a good deal of current on the plate of the driver for the 300B. Experience leads me to the same conclusion: high current is a club to linearize even recalcitrant tubes and the 417A, an RF tube, is not among the best behaved, though it is linear enough. But I selected it for its sonic qualities (if linearity were my only concern, I would have chosen the 6SN7 which itself sounds uncommonly fine.) So we have, -high current; from experience 20mA is pretty linear though the hardliners prefer 24mA -I need high current in any event because there will as a matter of principle be zero negative feedback For sound sonic (hee-hee) reasons, I want battery bias on a 417A used as a driver. Remember, we are coming to this KISS ultrafi amp from the KISS standard reference of 6SN76SN7300Bhigh impedance, an ultra-silent amp. While we are satisfied that we can back away from that perhaps overhigh standard a little, there is no call recklessly to embrace noise. I despise regulated power on the filaments of high-class amps for the bleached out, lowest common denominator sound it produces. (One of the reasons those big American amps sound like silicon is because they use beautifully designed regulated power supplies.) Battery bias is one way of avoiding noise. It eliminates a major sound-shaping element, the cathode bypass cap, leaving us enjoy the sound of the 417A unadorned. The bass boost of fixed bias can be used by the selection of other elements to shelve and tilt the driver-output tube transfer curve just so. -battery bias fixes the quiescent or zero-signal operating point at 2.4V because that is what we can get ni-cad batteries for, 1.2V being too little and 3.6V taking us outside the tube's permitted parameters (too much voltage on the plate-check the Eb-Ia-Eg curves) At this point we thus have a tube operated at 20mA (arbitrary) and -2.4V (battery availability) and 175V on the plate (read off the transfer curves). 6K8 is conveniently close to our preferred multiple of four of the 417A's plate resistance of about 1750--1800 ohms. All that remains is to calculate that a 6K8 load resistor will drop 136V at 20mA and now we know the power supply must deliver 311V at the takeoff for the driver. The grid leak resistor of the tube following, the 300B, which forms part of the load on the 417A plate, thereby becomes a response-shaping element in precisely the same way as its associated coupling capacitor (with which it forms an HF filter). For the time being I shall merely tell you the grid resistor is 47Kohm and the coupling capacitor 0=2E1uF because we want matching time constants on the RC filters to the grid of the 300B and to its cathode. In a later chapter I will show how I deliberately used these two elements to shift and tilt the transfer function of the 417A/300B combination, what the pretentious call distortion cancellation; they're in for a shock in just a moment... Remember what I said about the relationship of wu to the swan? The way all this falls together so tidily has nothing to do with luck and a great deal with years of hard work on drivers for 300B over more than a decade. The only remaining design decision is the value of the stepped attenuator across the input. The QUAD 67 CD player I favour demands to see at least 10K from the next device in the chain. But the 417A, which never stops reminding the clumsy that it is an RF tube, doesn't like more than 10Kohm attached to any part of it. The compromise I choose is a 20K DACT stepped attenuator with very short tracks (it is built with surface mount resistors) plus a 220 ohm grid stopper soldered right to the socket pins of each of the four grid inlets of the 417A. (Use the centre pin of the socket for a common input end.) Now we have a complete signal section for The KISS Amp 300B "Ultrafi" from volume control through driver, power tube and output transformer to loudspeakers. By now the power supply is essentially designed by reference to the parameters guiding the signal section; we shall discuss the general principles of the power supply in another chapter. *** At this stage you may compare what I have done with the transfer function in the audio band with the standard reference amp of 2x 6SN7 working with 300B to what I have done with 417A in the Ultrafi working very substantially against the linearity of the 300B. When it was put to me by someone, who hadn't yet seen the schematic for the circuit, that I would of course use the 417A for harmonic cancellation on the 300B, I burst out laughing. A 300B that requires linearization is incompetently implemented. Make no mistake. I deliberately used the superb warmth of the 417A to subvert the blameless linearity of the 300B. I did it only for a little way, and very selectively, of course, so you might say I did a reverse harmonic cancellation job. There are several good reasons for what I did. First, I wanted highish and very quick rolloff below 36Hz to protect my expensive Lowther drivers fitted to horns because horns simply don't load the driver below the fundamental resonance. This in turn requires that the top end be very carefully limited perceptually to balance the response around 700Hz or 1000Hz; this is a psychoacoustic matter with no electronic justification. Then I went one step further and tilted the system response so that, instead of a sharp peak in low presence range like most SE amps (the source of the silly myth that SE amps must necessarily possess "added euphonics), there is a shelf that runs from 36Hz to 17kHz, that is from the fundamental resonance of my Lowther loudspeakers to the fourth (natural) harmonic of the highest notes on a piano. The purpose of this is to boost the frequencies which carry most of the energy in the sound (which is not, repeat not, the fundamental) especially since they fall neatly in the specially pleasing territory of my horns; thus, without endangering my horns, I have boosted the frequencies which most strongly by subsonics suggest to the ears that fundamental which nature has made weak or which by electronic legerdemain I (or more usually the recording engineer before me) have written out of the script. These are artificial means to attain a sound professional chamber musicians will recognize as "more natural" than the sound made by amps which measure better. It is a key example of electronics serving taste and culture. Of course, before one can demand that electronics serve taste and culture, one must first earn taste and culture. In a later chapter I will show how all this is based on very hard science, a nasty wodge of tricky math straight out of the RCA Institutes by hand of Mr Julian L Bernstein. Mr Bernstein was Associate Dean of the RCA Day School. Here Andre Jute on the left receives The Compleat Tranfer Formula directly from his hand. Muscle tone like that requires a lot of soldering! Illustration by Michael O'Dwyer after Michelangelo I shall also present the spreadsheet in which by optimization search I discovered the correct component values to achieve this advantageous transfer function; trying to handle a complicated complete transfer function formula on a slide rule or calculator will drive the most patient of men nuts. As it was, this driver stage took me many months to design, test and optimize. Without the help of a computer it is the sort of development only a big team can do, beyond the resources of a hobbyist to handle manually. Of course the Ultrafi 300B sounds warmer than the 6SN7 driven amp. I designed it to do precisely that while casually waving a fig leaf of engineering respectability ("within 3dB over the entire audio band covered by the intended speakers") in the approximate vicinity of its impressively big-- er-- sound every time puritan silicon curtain-twitchers approach primly. It is a musical hedonist's amp and sound like it. JUTE ON AMPS =B7 KISSmain =B7 KISS190index HOME =B7 JUTE ON AMPS =B7 CLASSICAL JUKEBOX THE WRITER'S HOUSE =B7 THE TRUTH =B7 OTHER MATTERS ARISING THE VOLTAGES IN THIS AMP WILL KILL YOU. GET EXPERIENCED SUPERVISION IF IT IS YOUR FIRST TUBE AMP All text and illustration is Copyright =A9 Andre Jute 1996, 2005 and may not be reproduced except in the thread KISS xxx on rec.audio.tubes |
#3
|
|||
|
|||
Dear Mr. Jute, I am sure you are correct that the Populaire should be built first to provide a reference point for the Ultrafi, but Frank's questions raise another point of interest to me. I generally find the sound of cascade drivers, like that used in the Populaire, to be somewhat congested. On the other hand I find the sound of tubes like the 417A to be, well too "Ultrafi" for my taste. Would it be reasonable to construct a version of your 300B KISS Amp using a triode connected 6AU6 as the driver? The 6AU6 in this connection has a plate resistance similar to the 6SN7, and an amplification factor roughly equal to the 417A, so I am hoping that it might offer a happy compromise between your Populaire and Ultrafi KISS Amps. Regards, John Byrns In article .com, " wrote: The end result of your KISS Amp 300B project is not one amp but two amps, right? Right. The T44bis Populaire is a standard good, indeed excellent, 300B design with a two-stage 6SN7 front end. The T39 Mk VI Ultrafi is what it says, a 300B with a 417A front end designed to the tastes of the minority SE niche I christened the Ultrafidelista about a dozen years ago. How should the person who builds his first 300B amp choose between the two designs, one with a "cascade" of two 6SN7 and the other with a single 417A? The builder of a *first* 300B amp should not choose between the two. He should first build the Populaire, otherwise he will have no reference. That is why I kept calling the Populaire the *standard good*. This is an amp that can compete on level ground with silicon numbers but that sounds better. Once the 300B virgin has built this elevated quality reference, he can build the Ultrafi and understand what I am talking about when I say it lifts your music onto a different level. If you don't do it stepwise, you will not understand that the Ultrafi is as much an engineering marvel as a marvel of taste. I am not at all keen to encourage the mysticism which has conquered a significant part of audiophilia, nor am I keen on people doing what I say because I say so; I'd rather they took the journey with me and decided at each stage that I am right and therefore they will take the next step with me. Comparing the two schematics you published and the related frequency response graphs for the two amps, they are tuned differently, right? Right. That is the only reason for building two amps. (Boredom is not a good reason for working with high voltage!) I assume from the graphs on your netsite that the flat amp with 6SN7 you call "Popular" is intended to work with any relatively sensitive speakers and the "Ultrafi" with 417A is tuned to your preferred Lowther horns. But can they be tuned the same? Both will drive my Lowther horns well, though the Populaire (not Popular as you have it) will require rolling off higher in the LF to protect the speakers. Whether they *can* be tuned the same is one of those interesting but ultimately useless questions that will run and run on RAT. The short answer is yes, they *can* be tuned the same. As always in tubes, the short answer hides more than it illuminates. The two amps are *deliberately* differently tuned, the Populaire to be flat (near enough DC to daylight) and have the lowest possible THD within the constraints of the available transformers and of course absent NFB; it is a precision instrument. The 417A was chosen for the Ultrafi for the quality of its sound regardless of the fact that it is less linear than the 6SN7 (the 417A is still a very linear tube). To make the 6SN7 work like a 417A is to pervert its character, though it would be easier than the other way round. To make a 417A work like a 6SN7 is simply an engineering waste, considering how much cheaper a good 6SN7 is; it would take an enormous amount of development work. Surf my web pages at, http://users.rcn.com/jbyrns/ |
#4
|
|||
|
|||
|
#6
|
|||
|
|||
On 29 Apr 2005 12:19:18 -0700, "
wrote: The builder of a *first* 300B amp should not choose between the two. He should first build the Populaire, otherwise he will have no reference. That is why I kept calling the Populaire the *standard good*. This is an amp that can compete on level ground with silicon numbers but that sounds better. BWAHAHAHA! If it competed in any way with a well-designed 8-watt silicon amp (say for instance the original 1969 Linsley Hood design), then it would of course not *have* any sound of its own, and therefore could not sound 'better' than the SS amp. Once the 300B virgin has built this elevated quality reference, he can build the Ultrafi and understand what I am talking about when I say it lifts your music onto a different level. If you don't do it stepwise, you will not understand that the Ultrafi is as much an engineering marvel as a marvel of taste. BWAHAHAHAHA! No, Jute, it's not an 'engineering marvel', it's a 'back of a fag packet' amp that you are cobbling up from what's in your parts bin, in the demented belief that the use of WE tubes somehow connects it to the classic (and *utterly* different) WE 91A movie theater amp. I am not at all keen to encourage the mysticism which has conquered a significant part of audiophilia, nor am I keen on people doing what I say because I say so; BWAHAHAHAHA! I'd rather they took the journey with me and decided at each stage that I am right and therefore they will take the next step with me. Any intelligent reader will note that at each stage you just add a fudge factor to get back either to what's in your parts bin, or to what everyone else does. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#7
|
|||
|
|||
John Byrns wrote:
Dear Mr. Jute, Let's not fall into any bad habits on our nice friendly newsgroup now, Mr Byrns. I am sure you are correct that the Populaire should be built first to provide a reference point for the Ultrafi, but Frank's questions raise another point of interest to me. I generally find the sound of cascade drivers, like that used in the Populaire, to be somewhat congested. I agree with you that in general a cascade is almost never as good as a single really suitable tube; in particular some practices thoughtlessly carried forward from pre-amps lead to pretty dire cascades in power amps. But if you find two stages of 6SN7 congested, your ear must be ultra-refined. The complaint with 6SN7 is generally that people don't really want that much clarity and separation. I chose the 6SN7 for precisely that reason, and my choice has been validated by others. It is the 6SN7 that makes the Populaire a reference 300B. On the other hand I find the sound of tubes like the 417A to be, well too "Ultrafi" for my taste. Well now, that's a different matter precisely because the 417A *is* a matter of taste. There will always be some people who prefer the precision of the Populaire. Would it be reasonable to construct a version of your 300B KISS Amp using a triode connected 6AU6 as the driver? The 6AU6 in this connection has a plate resistance similar to the 6SN7, and an amplification factor roughly equal to the 417A, so I am hoping that it might offer a happy compromise between your Populaire and Ultrafi KISS Amps. I don't see why not. I don't have any 6AU6 experience but the spec seems good at the recommended 12mA operating point. Others to consider is the 6SJ7 which is proven as a driver for 300B, and the 310A and B which also have credibility (but are expensive). I also liked a 6SL7 SRPP, but that is not really KISS (and a mu stage is even less KISS). If you like SRPP, you might check out a big page of driver topologies with distortion measurements on Steve Bench's site, where he also shows a mixed 6SL7/6SN7 SRPP which is almost as silent as 417A SRPP with ground lift. Further to Frank B's question about tuning the two KISS amps the same, SRPP is another way approach the silence and bandwidth of 6SN7 with 417A, with additional benefits in impedance matching. Regards, John Byrns The whole point of a project like KISS is the thought process, not the precise circuits I publish. If DIYers just build my circuits slavishly, they will have nothing to share with me. I'm far too slack to want to do all the work myself. I look forward keenly to the variations and reports here on RAT. HTH. Andre Jute In article .com, " wrote: The end result of your KISS Amp 300B project is not one amp but two amps, right? Right. The T44bis Populaire is a standard good, indeed excellent, 300B design with a two-stage 6SN7 front end. The T39 Mk VI Ultrafi is what it says, a 300B with a 417A front end designed to the tastes of the minority SE niche I christened the Ultrafidelista about a dozen years ago. How should the person who builds his first 300B amp choose between the two designs, one with a "cascade" of two 6SN7 and the other with a single 417A? The builder of a *first* 300B amp should not choose between the two. He should first build the Populaire, otherwise he will have no reference. That is why I kept calling the Populaire the *standard good*. This is an amp that can compete on level ground with silicon numbers but that sounds better. Once the 300B virgin has built this elevated quality reference, he can build the Ultrafi and understand what I am talking about when I say it lifts your music onto a different level. If you don't do it stepwise, you will not understand that the Ultrafi is as much an engineering marvel as a marvel of taste. I am not at all keen to encourage the mysticism which has conquered a significant part of audiophilia, nor am I keen on people doing what I say because I say so; I'd rather they took the journey with me and decided at each stage that I am right and therefore they will take the next step with me. Comparing the two schematics you published and the related frequency response graphs for the two amps, they are tuned differently, right? Right. That is the only reason for building two amps. (Boredom is not a good reason for working with high voltage!) I assume from the graphs on your netsite that the flat amp with 6SN7 you call "Popular" is intended to work with any relatively sensitive speakers and the "Ultrafi" with 417A is tuned to your preferred Lowther horns. But can they be tuned the same? Both will drive my Lowther horns well, though the Populaire (not Popular as you have it) will require rolling off higher in the LF to protect the speakers. Whether they *can* be tuned the same is one of those interesting but ultimately useless questions that will run and run on RAT. The short answer is yes, they *can* be tuned the same. As always in tubes, the short answer hides more than it illuminates. The two amps are *deliberately* differently tuned, the Populaire to be flat (near enough DC to daylight) and have the lowest possible THD within the constraints of the available transformers and of course absent NFB; it is a precision instrument. The 417A was chosen for the Ultrafi for the quality of its sound regardless of the fact that it is less linear than the 6SN7 (the 417A is still a very linear tube). To make the 6SN7 work like a 417A is to pervert its character, though it would be easier than the other way round. To make a 417A work like a 6SN7 is simply an engineering waste, considering how much cheaper a good 6SN7 is; it would take an enormous amount of development work. Surf my web pages at, http://users.rcn.com/jbyrns/ |
#8
|
|||
|
|||
" wrote: John Byrns wrote: Dear Mr. Jute, Let's not fall into any bad habits on our nice friendly newsgroup now, Mr Byrns. I am sure you are correct that the Populaire should be built first to provide a reference point for the Ultrafi, but Frank's questions raise another point of interest to me. I generally find the sound of cascade drivers, like that used in the Populaire, to be somewhat congested. I agree with you that in general a cascade is almost never as good as a single really suitable tube; in particular some practices thoughtlessly carried forward from pre-amps lead to pretty dire cascades in power amps. But if you find two stages of 6SN7 congested, your ear must be ultra-refined. The complaint with 6SN7 is generally that people don't really want that much clarity and separation. I chose the 6SN7 for precisely that reason, and my choice has been validated by others. It is the 6SN7 that makes the Populaire a reference 300B. I have used the 6SL7 with a metal screen around it as an input tube, followed by an big bottle ECC32, each with both halves in parallel for the inputs of the 13E1 amps I made, and no complaints about detail, instrumnet separation, air, bloom etc, even with the mild global NFB applied, sensible due to the high Ro of the output tube in UL. The reason for UL was to get the same spectra one sees in a triode. We could ague all day about NFB and UL and using large glass beam tetrodes, but the point is that the SL7 is also a nice tube in cascade with a low µ triode. The warmth of such tubes is the natural warmth of the musicians naturally resonating their instruments together with each other; years of practice get them there. Several recent live concerts here have reminded my poor mold ears what hi-fi is really about. Give me clarity, give me separation, and I have music! Give me dynamics unspoilt by a buzz riding on massed voices or brass, let me hear the oomph of that brass band without a skerick of compression.... Tis unfortunate that much music is fiddled with before we get a disc of it. "Bright" microphones have a lot to answer for. On the other hand I find the sound of tubes like the 417A to be, well too "Ultrafi" for my taste. Well now, that's a different matter precisely because the 417A *is* a matter of taste. There will always be some people who prefer the precision of the Populaire. Would it be reasonable to construct a version of your 300B KISS Amp using a triode connected 6AU6 as the driver? The 6AU6 in this connection has a plate resistance similar to the 6SN7, and an amplification factor roughly equal to the 417A, so I am hoping that it might offer a happy compromise between your Populaire and Ultrafi KISS Amps. I don't see why not. I don't have any 6AU6 experience but the spec seems good at the recommended 12mA operating point. Others to consider is the 6SJ7 which is proven as a driver for 300B, and the 310A and B which also have credibility (but are expensive). You could do worse than try a trioded 6EJ7/EF184, frame grid pentode, Ra 10k approx, µ = 60 approx. Its much more gutsy cousin is the frame grid pentode, E280F. In triode Ra = 1.8k and µ = 60. BTW, not too many E280F are lying around. Then there is the humble 6BX6/EF80, also with higher merit than a little 6AU6, which isn't bad at all as a triode. I also liked a 6SL7 SRPP, but that is not really KISS (and a mu stage is even less KISS). If you like SRPP, you might check out a big page of driver topologies with distortion measurements on Steve Bench's site, where he also shows a mixed 6SL7/6SN7 SRPP which is almost as silent as 417A SRPP with ground lift. I have swung to µ follower preamp stages because the topology yields the clearest sound I have heard from triodes, and there is less 3H than SRPP, which has to be carefully set up to actually get the PP action. The 2H is minimized with µ follower because the gain tube is operating with an RL much higher than SRPP, and tending towards CCS, and thus more linearly, and Ro is very low, and 0.1% 2H is routine at 10rms output which is nice for a preamp which only has to make a volt. I find that when allowed act with minimal current change in gain stages triodes sound pure, clear, and without blame. There is in effect a cathode follower involved, but with a good amount of idle current and by staying well away from the dreaded cathode follower cut off when any cap coupled output load is too low, then triode followers don't degrade the sound, as some would have me believe. Further to Frank B's question about tuning the two KISS amps the same, SRPP is another way approach the silence and bandwidth of 6SN7 with 417A, with additional benefits in impedance matching. Regards, John Byrns The whole point of a project like KISS is the thought process, not the precise circuits I publish. If DIYers just build my circuits slavishly, they will have nothing to share with me. I'm far too slack to want to do all the work myself. I look forward keenly to the variations and reports here on RAT. I currently have a trioded EL84 driving a quad of SE 6CA7 and I don't hear too much wrong..... Patrick Turner. HTH. Andre Jute In article .com, " wrote: The end result of your KISS Amp 300B project is not one amp but two amps, right? Right. The T44bis Populaire is a standard good, indeed excellent, 300B design with a two-stage 6SN7 front end. The T39 Mk VI Ultrafi is what it says, a 300B with a 417A front end designed to the tastes of the minority SE niche I christened the Ultrafidelista about a dozen years ago. How should the person who builds his first 300B amp choose between the two designs, one with a "cascade" of two 6SN7 and the other with a single 417A? The builder of a *first* 300B amp should not choose between the two. He should first build the Populaire, otherwise he will have no reference. That is why I kept calling the Populaire the *standard good*. This is an amp that can compete on level ground with silicon numbers but that sounds better. Once the 300B virgin has built this elevated quality reference, he can build the Ultrafi and understand what I am talking about when I say it lifts your music onto a different level. If you don't do it stepwise, you will not understand that the Ultrafi is as much an engineering marvel as a marvel of taste. I am not at all keen to encourage the mysticism which has conquered a significant part of audiophilia, nor am I keen on people doing what I say because I say so; I'd rather they took the journey with me and decided at each stage that I am right and therefore they will take the next step with me. Comparing the two schematics you published and the related frequency response graphs for the two amps, they are tuned differently, right? Right. That is the only reason for building two amps. (Boredom is not a good reason for working with high voltage!) I assume from the graphs on your netsite that the flat amp with 6SN7 you call "Popular" is intended to work with any relatively sensitive speakers and the "Ultrafi" with 417A is tuned to your preferred Lowther horns. But can they be tuned the same? Both will drive my Lowther horns well, though the Populaire (not Popular as you have it) will require rolling off higher in the LF to protect the speakers. Whether they *can* be tuned the same is one of those interesting but ultimately useless questions that will run and run on RAT. The short answer is yes, they *can* be tuned the same. As always in tubes, the short answer hides more than it illuminates. The two amps are *deliberately* differently tuned, the Populaire to be flat (near enough DC to daylight) and have the lowest possible THD within the constraints of the available transformers and of course absent NFB; it is a precision instrument. The 417A was chosen for the Ultrafi for the quality of its sound regardless of the fact that it is less linear than the 6SN7 (the 417A is still a very linear tube). To make the 6SN7 work like a 417A is to pervert its character, though it would be easier than the other way round. To make a 417A work like a 6SN7 is simply an engineering waste, considering how much cheaper a good 6SN7 is; it would take an enormous amount of development work. Surf my web pages at, http://users.rcn.com/jbyrns/ |
#9
|
|||
|
|||
It is always worth pointing out, at least in rational society like RAT
sometimes manages to be (1), that an SRPP is not a CCS nor strictly speaking an SE stage. It should be thought of more as a PP stage with the "upper" tube arranged as a cathode follower. There is a heavy-duty mathematical treatment in Valley and Wallman, MIT Radiation Lab Series Vol 18. About ten years ago I published a spreadsheet for calculating and tuning output impedance in SRPP by choosing dissimilar resistors for the upper and lower positions. For the mixed 6SL7/xxx combo SRPP Patrick refers to, I first heard about mixed combos from Steve Bench about ten years ago. His site is always worth a visit. For another project I lifted an ultrarefined 417A SRPP with ground lift straight from Steve. On test it performed precisely as he said it would. Steve's page on driver topologies also has tested circuits for inductor loads and pseudo inductors made with transistors. These two netpages are the Book of Genesis for those interested in driver stages for blameless amps... http://members.aol.com/sbench101/TubeMisc/testsch.txt http://members.aol.com/sbench101/TubeMisc/testsch.gif Andre Jute (1) Among the more fanatical SE groupies you can get knackered for saying less than I say in the opening par of my letter. None so deaf as those who do not want to hear. Patrick Turner wrote: " wrote: John Byrns wrote: Dear Mr. Jute, Let's not fall into any bad habits on our nice friendly newsgroup now, Mr Byrns. I am sure you are correct that the Populaire should be built first to provide a reference point for the Ultrafi, but Frank's questions raise another point of interest to me. I generally find the sound of cascade drivers, like that used in the Populaire, to be somewhat congested. I agree with you that in general a cascade is almost never as good as a single really suitable tube; in particular some practices thoughtlessly carried forward from pre-amps lead to pretty dire cascades in power amps. But if you find two stages of 6SN7 congested, your ear must be ultra-refined. The complaint with 6SN7 is generally that people don't really want that much clarity and separation. I chose the 6SN7 for precisely that reason, and my choice has been validated by others. It is the 6SN7 that makes the Populaire a reference 300B. I have used the 6SL7 with a metal screen around it as an input tube, followed by an big bottle ECC32, each with both halves in parallel for the inputs of the 13E1 amps I made, and no complaints about detail, instrumnet separation, air, bloom etc, even with the mild global NFB applied, sensible due to the high Ro of the output tube in UL. The reason for UL was to get the same spectra one sees in a triode. We could ague all day about NFB and UL and using large glass beam tetrodes, but the point is that the SL7 is also a nice tube in cascade with a low =B5 triode. The warmth of such tubes is the natural warmth of the musicians naturally resonating their instruments together with each other; years of practice get them there. Several recent live concerts here have reminded my poor mold ears what hi-fi is really about. Give me clarity, give me separation, and I have music! Give me dynamics unspoilt by a buzz riding on massed voices or brass, let me hear the oomph of that brass band without a skerick of compression.... Tis unfortunate that much music is fiddled with before we get a disc of it. "Bright" microphones have a lot to answer for. On the other hand I find the sound of tubes like the 417A to be, well too "Ultrafi" for my taste. Well now, that's a different matter precisely because the 417A *is* a matter of taste. There will always be some people who prefer the precision of the Populaire. Would it be reasonable to construct a version of your 300B KISS Amp using a triode connected 6AU6 as the driver? The 6AU6 in this connection has a plate resistance similar to the 6SN7, and an amplification factor roughly equal to the 417A, so I am hoping that it might offer a happy compromise between your Populaire and Ultrafi KISS Amps. I don't see why not. I don't have any 6AU6 experience but the spec seems good at the recommended 12mA operating point. Others to consider is the 6SJ7 which is proven as a driver for 300B, and the 310A and B which also have credibility (but are expensive). You could do worse than try a trioded 6EJ7/EF184, frame grid pentode, Ra 10k approx, =B5 =3D 60 approx. Its much more gutsy cousin is the frame grid pentode, E280F. In triode Ra =3D 1.8k and =B5 =3D 60. BTW, not too many E280F are lying around. Then there is the humble 6BX6/EF80, also with higher merit than a little 6AU6, which isn't bad at all as a triode. I also liked a 6SL7 SRPP, but that is not really KISS (and a mu stage is even less KISS). If you like SRPP, you might check out a big page of driver topologies with distortion measurements on Steve Bench's site, where he also shows a mixed 6SL7/6SN7 SRPP which is almost as silent as 417A SRPP with ground lift. I have swung to =B5 follower preamp stages because the topology yields the clearest sound I have heard from triodes, and there is less 3H than SRPP, which has to be carefully set up to actually get the PP action. The 2H is minimized with =B5 follower because the gain tube is operating with an RL much higher than SRPP, and tending towards CCS, and thus more linearly, and Ro is very low, and 0.1% 2H is routine at 10rms output which is nice for a preamp which only has to make a volt. I find that when allowed act with minimal current change in gain stages triodes sound pure, clear, and without blame. There is in effect a cathode follower involved, but with a good amount of idle current and by staying well away from the dreaded cathode follower cut off when any cap coupled output load is too low, then triode followers don't degrade the sound, as some would have me believe. Further to Frank B's question about tuning the two KISS amps the same, SRPP is another way approach the silence and bandwidth of 6SN7 with 417A, with additional benefits in impedance matching. Regards, John Byrns The whole point of a project like KISS is the thought process, not the precise circuits I publish. If DIYers just build my circuits slavishly, they will have nothing to share with me. I'm far too slack to want to do all the work myself. I look forward keenly to the variations and reports here on RAT. I currently have a trioded EL84 driving a quad of SE 6CA7 and I don't hear too much wrong..... Patrick Turner. HTH. Andre Jute In article .com, " wrote: The end result of your KISS Amp 300B project is not one amp but two amps, right? Right. The T44bis Populaire is a standard good, indeed excellent, 300B design with a two-stage 6SN7 front end. The T39 Mk VI Ultrafi is what it says, a 300B with a 417A front end designed to the tastes of the minority SE niche I christened the Ultrafidelista about a dozen years ago. How should the person who builds his first 300B amp choose between the two designs, one with a "cascade" of two 6SN7 and the other with a single 417A? The builder of a *first* 300B amp should not choose between the two. He should first build the Populaire, otherwise he will have no reference. That is why I kept calling the Populaire the *standard good*. This is an amp that can compete on level ground with silicon numbers but that sounds better. Once the 300B virgin has built this elevated quality reference, he can build the Ultrafi and understand what I am talking about when I say it lifts your music onto a different level. If you don't do it stepwise, you will not understand that the Ultrafi is as much an engineering marvel as a marvel of taste. I am not at all keen to encourage the mysticism which has conquered a significant part of audiophilia, nor am I keen on people doing what I say because I say so; I'd rather they took the journey with me and decided at each stage that I am right and therefore they will take the next step with me. Comparing the two schematics you published and the related frequency response graphs for the two amps, they are tuned differently, right? Right. That is the only reason for building two amps. (Boredom is not a good reason for working with high voltage!) I assume from the graphs on your netsite that the flat amp with 6SN7 you call "Popular" is intended to work with any relatively sensitive speakers and the "Ultrafi" with 417A is tuned to your preferred Lowther horns. But can they be tuned the same? Both will drive my Lowther horns well, though the Populaire (not Popular as you have it) will require rolling off higher in the LF to protect the speakers. Whether they *can* be tuned the same is one of those interesting but ultimately useless questions that will run and run on RAT. The short answer is yes, they *can* be tuned the same. As always in tubes, the short answer hides more than it illuminates. The two amps are *deliberately* differently tuned, the Populaire to be flat (near enough DC to daylight) and have the lowest possible THD within the constraints of the available transformers and of course absent NFB; it is a precision instrument. The 417A was chosen for the Ultrafi for the quality of its sound regardless of the fact that it is less linear than the 6SN7 (the 417A is still a very linear tube). To make the 6SN7 work like a 417A is to pervert its character, though it would be easier than the other way round. To make a 417A work like a 6SN7 is simply an engineering waste, considering how much cheaper a good 6SN7 is; it would take an enormous amount of development work. Surf my web pages at, http://users.rcn.com/jbyrns/ |
#10
|
|||
|
|||
" wrote: It is always worth pointing out, at least in rational society like RAT sometimes manages to be (1), that an SRPP is not a CCS nor strictly speaking an SE stage. It should be thought of more as a PP stage with the "upper" tube arranged as a cathode follower. There is a heavy-duty mathematical treatment in Valley and Wallman, MIT Radiation Lab Series Vol 18. About ten years ago I published a spreadsheet for calculating and tuning output impedance in SRPP by choosing dissimilar resistors for the upper and lower positions. If anyone should dig through the group archives, they will find a formula I developed a couple of years back to determine the cap coupled RL value, usually a volume control pot, to get the current change in the top and bottom tubes to be nearly equal, and thus have top and bottom tubes see the most nearly equal loads, ie, approximately twice the cap coupled load on the output. Maybe my formula was a re-invention or repeat of what somebody else has done before, but anyone with a brain can work out what load and Rk is needed to get the condition for maximum PP action. Its impossible to get each to see exactly twice the cap coupled RL, but the higher the tube µ the more complete the PP action will be, and PP action only occurs with the right load value. The alleged benefits of PP action and cancellation of 2H are best realised with careful choice of the resistor between tubes if one starts off with a fixed load value. With a very high RL value, the action of the tubes doesn't include much PP action, and you'd think it similar to common cathode. Where cap coupled RL is say 500k, the load seen by the bottom tube approaches Ra + ( µ + 1 )x Rk, so that if Rk of the top tube is 1.8k and µ = 20, then the bottom 6SN7 triode sees a load of only say 10k + 21 x 1.8k = 47.8k, barely any different to a common cathode triode loaded by 47k. Using the two halves in parallel as a common cathode stage with 22k RL would be almost the same as SRPP. For the mixed 6SL7/xxx combo SRPP Patrick refers to, I first heard about mixed combos from Steve Bench about ten years ago. His site is always worth a visit. For another project I lifted an ultrarefined 417A SRPP with ground lift straight from Steve. On test it performed precisely as he said it would. Item 7 in Steve's site with 417A gives an astonishly low thd of 0.06% at 20v output.... But all the SRPP circuits have 2H/3H which varies with the following cap coupled load. Steve's page on driver topologies also has tested circuits for inductor loads and pseudo inductors made with transistors. These two netpages are the Book of Genesis for those interested in driver stages for blameless amps... http://members.aol.com/sbench101/TubeMisc/testsch.txt http://members.aol.com/sbench101/TubeMisc/testsch.gif There is a wealth of info there. Only schematic no 8 with an additional choke between top and bottom tubes of a 6SN7 acts differently to all the rest, which are plain common cathode or SRPP schemos. The artificial choke, or gyrator, as it is called can have its 22uF and 470k replaced by a suitable R divider, maybe 50k total value, and a 1k as the Re on the mpsa42, thus converting it to a better F range CCS bypassed with the 50k divider. The 1.8k Rk can be left out, but the resistor between base and + v should be well bypassed to stop the bjt from being an active current varying device. The DC current can be adjusted with the values in the 50k divider. With such a high R = approx 50k between top and bottom triodes, the bottom triode is loaded only by A x 50k, where A is the open loop gain of the top triode, and assuming the top triode does nearly all the current delivery to the outside world load which could be 50k, its gain is around 16, so the bottom tube sees a load = 16 x 50k = 800k, and thd will be around what Steve claims, a mere 0.16% at 20vo. But the effort needed to build the CCS between tubes isn't really needed, and a 20k resistor will do if one can accept the 100v drop across it if Iaq = 5 mA. Since the bottom tube will see A x 20k, or 16 x 20 = 320k, then you will still get 0.2% at 20v; I routinely get 0.1% at 10vo, and declining to 0% at 0.0vo approximately linearly, so that at 0.1vo, thd = 0.001%. Noise can spoil any measurement of such low signals, but reliable thd measures of 0.01% nearly all 2H are possible at a volt of output. The use of the µ follower reduced thd by 20dB compared to the ordinary SRPP which is only marginally better than a common cathode stage for thd. Its all because the loadings for the triodes are raised well beyond the rule that RL should be approximately 4 x Ra; thd plummets if RL = 40 Ra. But it doesn't continue to reduce much beyond 40Ra; 400Ra is a virtual CCS, and nice if its done so easily with a bjt CCS, but if there are simpler ways than CCS which give substantial benefits. I have never known a triode loaded with 40Ra to ever sound bad. In the case where 417A is needed to drive a 300B, then the thd issue isn't a big deal since most of the 417A thd is 2H, and will cancel to some useful extent the thd of the 300B. I think the secret is to set up the 417A with optimal load values if common cathode is used, and not try to reduce its RL to get more 2H to cancel the 2H of the 300B. That you do get harmonic voltage cancelation is an electronic freebie, rare with electronics. Bypassing the 417A cathode will keep its gain at maximum, keep its effective Ra as low as it should be, and also keep the amp output thd lowest. Andre Jute (1) Among the more fanatical SE groupies you can get knackered for saying less than I say in the opening par of my letter. None so deaf as those who do not want to hear. I get complaints about my long explanations at times, and especially when I am attempting to unravel the validity of some pet theory by somebody. Not all the people can be pleased all the time. Patrick Turner. Patrick Turner wrote: " wrote: John Byrns wrote: Dear Mr. Jute, Let's not fall into any bad habits on our nice friendly newsgroup now, Mr Byrns. I am sure you are correct that the Populaire should be built first to provide a reference point for the Ultrafi, but Frank's questions raise another point of interest to me. I generally find the sound of cascade drivers, like that used in the Populaire, to be somewhat congested. I agree with you that in general a cascade is almost never as good as a single really suitable tube; in particular some practices thoughtlessly carried forward from pre-amps lead to pretty dire cascades in power amps. But if you find two stages of 6SN7 congested, your ear must be ultra-refined. The complaint with 6SN7 is generally that people don't really want that much clarity and separation. I chose the 6SN7 for precisely that reason, and my choice has been validated by others. It is the 6SN7 that makes the Populaire a reference 300B. I have used the 6SL7 with a metal screen around it as an input tube, followed by an big bottle ECC32, each with both halves in parallel for the inputs of the 13E1 amps I made, and no complaints about detail, instrumnet separation, air, bloom etc, even with the mild global NFB applied, sensible due to the high Ro of the output tube in UL. The reason for UL was to get the same spectra one sees in a triode. We could ague all day about NFB and UL and using large glass beam tetrodes, but the point is that the SL7 is also a nice tube in cascade with a low µ triode. The warmth of such tubes is the natural warmth of the musicians naturally resonating their instruments together with each other; years of practice get them there. Several recent live concerts here have reminded my poor mold ears what hi-fi is really about. Give me clarity, give me separation, and I have music! Give me dynamics unspoilt by a buzz riding on massed voices or brass, let me hear the oomph of that brass band without a skerick of compression.... Tis unfortunate that much music is fiddled with before we get a disc of it. "Bright" microphones have a lot to answer for. On the other hand I find the sound of tubes like the 417A to be, well too "Ultrafi" for my taste. Well now, that's a different matter precisely because the 417A *is* a matter of taste. There will always be some people who prefer the precision of the Populaire. Would it be reasonable to construct a version of your 300B KISS Amp using a triode connected 6AU6 as the driver? The 6AU6 in this connection has a plate resistance similar to the 6SN7, and an amplification factor roughly equal to the 417A, so I am hoping that it might offer a happy compromise between your Populaire and Ultrafi KISS Amps. I don't see why not. I don't have any 6AU6 experience but the spec seems good at the recommended 12mA operating point. Others to consider is the 6SJ7 which is proven as a driver for 300B, and the 310A and B which also have credibility (but are expensive). You could do worse than try a trioded 6EJ7/EF184, frame grid pentode, Ra 10k approx, µ = 60 approx. Its much more gutsy cousin is the frame grid pentode, E280F. In triode Ra = 1.8k and µ = 60. BTW, not too many E280F are lying around. Then there is the humble 6BX6/EF80, also with higher merit than a little 6AU6, which isn't bad at all as a triode. I also liked a 6SL7 SRPP, but that is not really KISS (and a mu stage is even less KISS). If you like SRPP, you might check out a big page of driver topologies with distortion measurements on Steve Bench's site, where he also shows a mixed 6SL7/6SN7 SRPP which is almost as silent as 417A SRPP with ground lift. I have swung to µ follower preamp stages because the topology yields the clearest sound I have heard from triodes, and there is less 3H than SRPP, which has to be carefully set up to actually get the PP action. The 2H is minimized with µ follower because the gain tube is operating with an RL much higher than SRPP, and tending towards CCS, and thus more linearly, and Ro is very low, and 0.1% 2H is routine at 10rms output which is nice for a preamp which only has to make a volt. I find that when allowed act with minimal current change in gain stages triodes sound pure, clear, and without blame. There is in effect a cathode follower involved, but with a good amount of idle current and by staying well away from the dreaded cathode follower cut off when any cap coupled output load is too low, then triode followers don't degrade the sound, as some would have me believe. Further to Frank B's question about tuning the two KISS amps the same, SRPP is another way approach the silence and bandwidth of 6SN7 with 417A, with additional benefits in impedance matching. Regards, John Byrns The whole point of a project like KISS is the thought process, not the precise circuits I publish. If DIYers just build my circuits slavishly, they will have nothing to share with me. I'm far too slack to want to do all the work myself. I look forward keenly to the variations and reports here on RAT. I currently have a trioded EL84 driving a quad of SE 6CA7 and I don't hear too much wrong..... Patrick Turner. HTH. Andre Jute In article .com, " wrote: The end result of your KISS Amp 300B project is not one amp but two amps, right? Right. The T44bis Populaire is a standard good, indeed excellent, 300B design with a two-stage 6SN7 front end. The T39 Mk VI Ultrafi is what it says, a 300B with a 417A front end designed to the tastes of the minority SE niche I christened the Ultrafidelista about a dozen years ago. How should the person who builds his first 300B amp choose between the two designs, one with a "cascade" of two 6SN7 and the other with a single 417A? The builder of a *first* 300B amp should not choose between the two. He should first build the Populaire, otherwise he will have no reference. That is why I kept calling the Populaire the *standard good*. This is an amp that can compete on level ground with silicon numbers but that sounds better. Once the 300B virgin has built this elevated quality reference, he can build the Ultrafi and understand what I am talking about when I say it lifts your music onto a different level. If you don't do it stepwise, you will not understand that the Ultrafi is as much an engineering marvel as a marvel of taste. I am not at all keen to encourage the mysticism which has conquered a significant part of audiophilia, nor am I keen on people doing what I say because I say so; I'd rather they took the journey with me and decided at each stage that I am right and therefore they will take the next step with me. Comparing the two schematics you published and the related frequency response graphs for the two amps, they are tuned differently, right? Right. That is the only reason for building two amps. (Boredom is not a good reason for working with high voltage!) I assume from the graphs on your netsite that the flat amp with 6SN7 you call "Popular" is intended to work with any relatively sensitive speakers and the "Ultrafi" with 417A is tuned to your preferred Lowther horns. But can they be tuned the same? Both will drive my Lowther horns well, though the Populaire (not Popular as you have it) will require rolling off higher in the LF to protect the speakers. Whether they *can* be tuned the same is one of those interesting but ultimately useless questions that will run and run on RAT. The short answer is yes, they *can* be tuned the same. As always in tubes, the short answer hides more than it illuminates. The two amps are *deliberately* differently tuned, the Populaire to be flat (near enough DC to daylight) and have the lowest possible THD within the constraints of the available transformers and of course absent NFB; it is a precision instrument. The 417A was chosen for the Ultrafi for the quality of its sound regardless of the fact that it is less linear than the 6SN7 (the 417A is still a very linear tube). To make the 6SN7 work like a 417A is to pervert its character, though it would be easier than the other way round. To make a 417A work like a 6SN7 is simply an engineering waste, considering how much cheaper a good 6SN7 is; it would take an enormous amount of development work. Surf my web pages at, http://users.rcn.com/jbyrns/ |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
KISS 100 by Andre Jute at 31 March 2004 -- The KISS Amp INDEX | Vacuum Tubes | |||
KISS 123 by Andre Jute: Why the KISS 300B is ZNFB | Vacuum Tubes | |||
Re KISS 123 by Andre Jute | Vacuum Tubes | |||
KISS 101B by Andre Jute | Vacuum Tubes | |||
KISS 100 4 December 2004 | Vacuum Tubes |