Home |
Search |
Today's Posts |
#1
|
|||
|
|||
KISS 117 by Andre Jute
KISS 117 by Andre Jute
This text is copyright Andre Jute 2004 and may not be reproduced except in the thread KISS xxx on rec.audio.tubes THE VOLTAGES IN THIS AMP WILL KILL YOU. GET EXPERIENCED SUPERVISION IF IT IS YOUR FIRST TUBE AMP. There are fewer than a handful of books that no tube audiophile can do without. One of them, Transformers and Tubes in Power Amplifiers by Menno van der Veen, lies at my right hand with Langford-Smith's Radio Designer's Handbook. Many tube audiofiles who may think they don't know Menno have already met him because they own Plitron or Amplimo transformers which he designed. Below is an article I asked Menno to write on selecting a transformer, followed by more fruits from his fertile mind in rare and wonderful topologies and the transformers for them. Enjoy. - - Andre Jute Copyright text and figures ©1998 Menno van der Veen The secret of selecting a good output transformer by Menno van der Veen President Ir. buro Vanderveen Plitron/Amplimo/OPT-R&D The article is in two parts: first what everyone should know, then remarks about coupling schemes, including a couple Menno has just invented. Here's Menno: "This article explains how to select an optimal output transformer (OPT) for your special amplifier. I shall introduce the challenging possibilities of the new "Specialist" toroidal OPTs. The only restriction Andre gave me for this article was: "please, no formulae--we can't control how they appear on the net". Therefore I will explain with words and point to my book, published lab reports and AES preprints for those who love doing complex calculations. Here we go." Part 1: What every audiophile should know SE AND PUSH-PULL OUTPUT TRANSFORMERS In tube amplifiers you need an OPT because the voltages in the tube amplifier are too large for your loudspeaker, while the current capability of the tubes is too small to drive your speaker correctly. There exist fabulous Output Transformerless amplifier designs (OTL) but most tube amps use output transformers. The function of an OPT is to lower the high voltage to safe values and to multiply the weak tube currents into larger values. This action is performed by winding different amounts of turns on the incoming (primary) and outgoing (secondary) side of the OPT. The turns ratio between primary and secondary is the major tool performing this job. All output transformers can be divided into two groups. Transformers for Single Ended amplifiers and transformers for push-pull amplifiers. The major difference between these two is that in SE OPTs the quiescent current of the power triode is not compensated in the transformer core, while in push-pull transformers the quiescent currents of the two push-pull power tubes cancel each other out in the core of the OPT. This means that an SE transformer must be constructed differently from a push-pull type. In general one can say that an SE-transformer includes a gap in the core to deal with the quiescent current while the push-pull version has a closed core with almost no gap in it. This means: even when you own a very good push-pull OPT you can't use it in a SE-amplifier! IMPEDANCES Suppose you want to select an OPT for a special design. Suppose that we are dealing with a push-pull amplifier. Somewhere in the tube spec or design notes you should find the primary impedance (Zaa) for optimal loading of the power tubes. Let's imagine a design with a primary impedance of 3300 Ohms. On the secondary side you wish to connect a 4 or an 8 Ohms loudspeaker. I have standardized my toroidal designs to a 5 Ohms secondary, but very often 4 and 8 Ohms connections are found. Suppose for now that you have found a transformer with a primary impedance of 4000 Ohms and secondaries at 4 and 8 Ohms. Can you use this transformer for your special design where Zaa should equal 3300 Ohms? The answer is yes, you only have to perform a minor calculation to see how. In your transformer you have an impedance ratio of 4000/4 = 1000. Now suppose that you don't apply a 4 Ohms loudspeaker but a 3.3 Ohms version, then with this impedance ratio of 1000 you get a primary impedance of 3300 Ohms. When you use the 8 Ohms secondary connection, your impedance ratio is 4000/8 = 500. To get a primary impedance of 3300 Ohm you should apply a loudspeaker with an impedance of 3300/500 = 6.6 Ohms. These examples demonstrate the following important rule: "The impedance ratio of the OPT combined with the impedance of the loudspeaker delivers the primary impedance." Another example: suppose you have an SE-OPT with a primary impedance Za = 2500 Ohms. On the secondary you have a 4 Ohms connection. The impedance ratio is 2500/4 = 625. Now suppose that you wish to build a 300B SE-amplifier with a primary impedance of 3500 Ohms. What speaker should you connect? The answer is: 3500/625 = 5.6 Ohms. Every one knows that the impedance of a loudspeaker is frequency dependent, meaning that at each frequency the impedance has a different value. Therefore loudspeaker manufacturers give a mean impedance value. The consequence of this is that you never can calculate exactly the value of the primary impedance. Just try to be close to the primary and secondary impedances intended in the design, but don't worry about deviations up to about 20%. This criterium will make your selection of an output transformer much easier. POWER CAPABILITY The output transformer must be able to handle the output power without major losses and distortions, over the intended frequency range. This is a rather difficult topic because not all manufacturers deliver all the information you need to judge whether the OPT is applicable or not. What you need is the following: "which is the lowest frequency at which the transformer can handle its nominal power"? Take an example: suppose you select an OPT which can handle 50 Watts at 30 Hz. Then this transformer can handle 50/2 = 25 Watt at 30/1.414 = 21.2 Hz. Or the transformer can handle 50*2 = 100 Watt at 30*1.414 = 42.4 Hz. The rule behind all this is: "The power capability doubles when the frequency is a factor 1.414 larger. The power capability halves when the lowest frequency is devided by 1.414 (squareroot 2)." But what to do when the manufacturer only tells you that you are buying a 100 Watt transformer without mentioning the lowest nominal power frequency? To be honest: the lack of information makes you 'blind' and you don't know how the behaviour of this transformer will be at low frequencies. The lower the frequency of the input, the more the core of the OPT gets saturated and you only can guess at which frequency severe distortions will start. The only thing you should count on is the good name of the manufacturer, expecting that he is knowing what he is doing. I plead, however, for OPT power specifications to be specified with the lowest frequency clearly stated. That would help you in selecting the optimal OPT for your application. LOSSES All the magnet wire turns of the OPT have a resistance. The currents of the tubes are partly converted into heat in this internal resistance and therefore you lose power. This is expressed in the "Insertion Loss" which you find in the specification of the OPT. Let me give an example: suppose an I-loss of 0.3 dB, how much power is lost in heat in the transformer? Now take your calculator and calculate: 0.3/10 = 0.03, calculate -0.03 with the inverse log-function (10 to the power of x) resulting in 0.933. This results means that 93.3 % of the output power is converted into music while 100-93.3 = 6.7 % is converted into heat. This knowledge does not enable you to fry an egg on the transformer, so a more general rule will give better information: "Insertion losses smaller than 0.3 dB in an OPT indicate an acceptable heat loss without causing major difficulties." LOW FREQUENCY RANGE and DC-IMBALANCE The calculation of the frequency range of an OPT is very complex. You find all the information and details in my AES preprint 3887: "Theory and Practice of Wide Bandwidth Toroidal Output Transformers", which can be ordered from the AES. The most important quantity determining the low frequency range of an OPT is the primary inductance Lp (its value is given in H = Henry). The larger Lp is, the better the low frequency response of the transformer. To make Lp large you need a lot of magnet wire turns around the core and you need to use a large core. A second factor determing the low frequency range is the primary impedance of the OPT paralleled with the plate resistances of the output tubes. The smaller the plate resistances and primary impedance the wider the frequency range at the low frequency side. Select power tubes with a low plate resistance (like triodes) for a good bass response with very little distortion combined with OPT's with a large Lp value. (For a more detailed study see my article in Glass Audio 5/97: "Measuring Output Transformer Performance", p20ff.) However, the larger you make Lp, the more sensitive the OPT becomes to an imbalance of the quiescent currents of the power tubes in a push-pull amplifier design. In practice this means: when you use high quality OPTs with good bass response and a large primary inductance, you should pay special attention to carefully balancing the quiescent currents of the power tubes. Whether you use my toroidal designs or EI-designs or C-core designs, this is a general rule for large primary inductance OPT designs. If you don't balance your quiescent currents carefully, your maximum power capability at low frequencies gets less and the distortions become larger. HIGH FREQUENCY RANGE At the high frequency side, two internal quantities of the transformer limit the high frequencies. These a the effective internal capacitance between the windings (Cip) and the leakage inductance of the transformer (Lsp). The leakage is caused by the simple fact that not all the magnetic fieldlines are captured in the core. Some leave the core and are outside the transformer. In this aspect the toroidal transformers show very good specifications, because the round shaped core captures almost all the fieldlines, resulting in very small leakage inductances. The smaller the leakage, the wider the frequency range. The influence of the internal capacitance is the same: the smaller the internal capacitance, the wider the high frequency range. A transformer designer therefore has to find an optimal balance between the leakage, the capacitance and the tubes and impedances used to create an optimal frequency range. I discuss this in detail in my 3887 AES preprint. Now some rules: "The smaller the plate resistances of the tubes, the wider the frequency range." "When the balance between Lsp and Cip is not correct, square wave overshoot will occur (incorrect Q-factor)." "When both Lsp and Cip are large, the high frequency range becomes limited and this will result in differential phase distortions" (meaning that the frequency components of a tone, or of the music, will be time-shifted towards each other, resulting in a distorted tone envelope, detectable by the ear due to its a-linear behaviour). Let me summarise this as follows: it is up to the transformer designer not only to create a wide frequency range, but to tune the high frequency behaviour with the correct Q-factor (somewhere between 0.5 and 0.7). In that case no square wave overshoot will occur and the differential phase distortion will be minimal. Look into the specifications of the manufacturer to find more details about the high frequency tuning of his designs. I have optimized my toroidal designs for a very wide frequency range, in order to be prepared for the new digital developments with sampling rates now up to 194 kHz--and who knows what the near future will bring? Part 2 of the article by Menno van der Veen: At the leading edge SPECIAL CONFIGURATIONS & NEW ADVENTURES Recently I finalised a study and research about special coupling techniques between an OPT and power tubes. The results of this research can be found in my AES preprint 4643: "Modelling Power Tubes and their Interaction with Output Transformers", obtainable from the AES. My basic question was: "How can I couple power tubes optimally to an output transformer?" In order to answer this simple question I first had to design a mathematical model describing the behaviour of power tubes. Fortunately many others (like for instance the pioneers Scott Reynolds and Norman Koren) already had studied this subject and I only had to add a very small extension to their models to be able to model pentode power tubes rather accurately. My next important step was the understanding that there are many ways to connect power tubes to output transformers. I only mention a few possiblities: pentode push-pull, ultra linear, triode push-pull, cathode feedback, cathode out, and so on. I discovered that it is possible to bring all these various coupling techniques into one general model by means of the introduction of the screen grid feedback ratio X and the introduction of the cathode feedback ratio. The figure shows eight different coupling methods between the push-pull power tubes and the OPT. To investigate all these amplifiers, I designed new toroidal output transformers: the "Specialist Range". These new transformers contain very special windings for the application of selectable cathode and screen grid feedback. For more information see the web pages of Plitron and Amplimo. The major time consuming element in the designing of these new toroidal transformers was the demand that the amplifiers should be absolutely stable, not oscilating, and optimized in power, frequency range and damping factor behaviour. During my research and the development of the new OPTs I discovered two brand new circuits (numbers 5 and 7) which are under my registration and copyright. For trade use and/or manufacture please contact me for licensing. I will not deal now with all the details of this new research. The technically inclined can order the 4643 AES preprint. The circuits 5 (Super Pentode) and 7 (Super Triode) both show very special qualities not seen before by me in push-pull amps. For instance, circuit 5 delivers amazingly large output powers (80 Watt with 2 x EL34 at 450-470 V supply), while circuit 7 delivers extremely small distortions (harmonic as well as intermodulation) and a damping factor surpassing triode amplifiers. In all this I noticed (and calculated) that especially the quiescent current per power tube is a major tool in creating minimal distortions (while hardly decreasing the maximum output power). The new toroidal "specialist" transformers are available for any one to perform his/her private tests with these new amplifier possibilities. See Plitron's and Amplimo's web-sites. Their internal research reports can be ordered; they deal with these new technologies, giving a background information and important references to the relevant literature . SUMMARY I did not talk in length about SE-transformers, their selection and optimal application. All this information can be found in my new book which I hope to finish soon. I paid attention to impedances, powers and losses, the frequency range, distortions and new coupling techniques. For those who wish to study these subjects in depth, I attach a bibliography of only my own writing, which in turn contain bibliographies of the relevant references. If comments, reactions and advice should appear in my email, I would be a very happy man. LITERATURE 1) Menno van der Veen; "Transformers and Tubes"; published by Plitron; www.plitron.com 2) Menno van der Veen; "Het Vanderveen Buizenbouwboek"; published by Amplimo; www.amplimo.nl 3) Menno van der Veen; "Theory and Practice of Wide Bandwidth Toroidal Output Transformers"; AES preprint 3887 4) Menno van der Veen; "Modelling Power Tubes and their Interaction with Output Transformers"; AES preprint 4643 5) Menno van der Veen; "Measuring Output Transformer Performance"; Glass Audio 5/97 6) Menno van der Veen; "Lab Report Specialist Range Toroidal Output Transformers"; published by Plitron; www.plitron.com 7) Menno van der Veen; "Specialist Ringkern Uitgangstransformatoren, de Super Pentode Schakeling"; published by Amplimo; www.amplimo.nl All the above literature contains a wealth of references to other authors. Super Pentode and Super Triode: Names and principles are registered by the author and are subject to European Union and International Copyright Laws. Licensing enquires for reproduction of and manufacture for trade sale should be directed to Menno van der Veen Menno van der Veen (b1949) graduated in physics and electronics. He taught physics at the physics department of a teachers' college until he founded his research and consultancy company in 1986. He is a board member of the Dutch Section of the Audio Engineering Society and of the Elpec (Dutch Electronic Press Association). He is a member of the Dutch Acoustic Society (NAG). He is the designer of special toroidal output transformers for tube amplifiers in close cooperation with Plitron and Amplimo. Results of his output transformer research were published at AES conventions in 1994 and 1998. He wrote over 360 articles for various Dutch high end audio magazines and is the author of the book "Transformers and Tubes". Designing tube amplifiers is his vocation--combined with playing the Jazz guitar. Currently he is writing two new books on tube amplifiers. Copyright text and figures ©1998 Menno van der Veen |
#2
|
|||
|
|||
"Andre Jute" wrote in message
om... KISS 117 by Andre Jute Hmmmm, slipped outa my trashbin. Whelp, back in ya go... Tim -- "I've got more trophies than Wayne Gretsky and the Pope combined!" - Homer Simpson Website @ http://webpages.charter.net/dawill/tmoranwms |
#3
|
|||
|
|||
Andre,
I worry about that power rating scheme. A certain diameter wire will only handle a certain amount of DC current. This is a factor in the power handling capabilities of the output transformer as well as it's AC power rating for the actual audio signal. You need to know both the DC current rating and what audio power the transformer is rated for. If I have a push pull amp using a pair of EL34's which is rated at 30 watts, you might think a 35 watt output transformer should be fine. But when you see the full specifications of 35 watts music power with a 100 ma DC current rating, it is all too obvious that the transformer is inadequate since the EL34's will pull 120 ma at idle through the output transformer. So, in many cases, although the amp puts out only 30 - 35 watts, the average 50 watt output transformer is what you need. Even if you use the frequency formula and limit your low frequencies to 30 hz as opposed to the transformer's 20hz low end rating, you still have a primary winding not up to the task of handling the 120ma idle current and 145 ma draw at full power. Another headache with choosing a transformer is the frequency bandwidth. Some companies rate their frequency response at full power. So, their 50 watt output delivers a full 20 - 20Khz at 50 watts. Others claim a 50 watt output, but rate their frequency response at only 1 watt to give a much inflated bandwidth. When you test the transformer at 50 watts, it falls far short of the 20 - 20Khz bandwidth. The ultimate test of any transformer is to have it in a circuit and do both square wave and sine wave tests to see how well the wave form is reproduced at the different frequencies as well as at both low and high power. This is of course assuming that the amplifier's circuitry is up to the task and not creating problems of it's own. Everything from the way you wind the transformer to the core material and size have an impact on the output transformer's performance. The text books and spice models are nice, but there is no substitute for actual experimentation with real functioning circuits. An actual amplifier is the best test bed and will present the transformer in question with all the dynamics of actual use. Although I have the Radiotron Designer's handbook and use it as well as tube manuals and other resources, I find that they will just get me in the ball park. They do not afford me the answers to end all debate. Thus the reason for my prototypes. There is another variable with respect to both power and output transformers. Heat. How hot do you want your amp to get? I could use a 175 ma power transformer and a 35 watt output in one amp I designed and it will work, and would probably last a reasonably long time. But the transformers get real hot. They work, but they feel too hot. Moving up to a 50 watt output and a 200 ma power transformer nets me a much cooler running amp with a total increase in transformer cost of around $25.00. Bill B. |
#4
|
|||
|
|||
Wbittle wrote: Andre, I worry about that power rating scheme. A certain diameter wire will only handle a certain amount of DC current. This is a factor in the power handling capabilities of the output transformer as well as it's AC power rating for the actual audio signal. You need to know both the DC current rating and what audio power the transformer is rated for. If I have a push pull amp using a pair of EL34's which is rated at 30 watts, you might think a 35 watt output transformer should be fine. But when you see the full specifications of 35 watts music power with a 100 ma DC current rating, it is all too obvious that the transformer is inadequate since the EL34's will pull 120 ma at idle through the output transformer. So, in many cases, although the amp puts out only 30 - 35 watts, the average 50 watt output transformer is what you need. Even if you use the frequency formula and limit your low frequencies to 30 hz as opposed to the transformer's 20hz low end rating, you still have a primary winding not up to the task of handling the 120ma idle current and 145 ma draw at full power. Another headache with choosing a transformer is the frequency bandwidth. Some companies rate their frequency response at full power. So, their 50 watt output delivers a full 20 - 20Khz at 50 watts. Others claim a 50 watt output, but rate their frequency response at only 1 watt to give a much inflated bandwidth. When you test the transformer at 50 watts, it falls far short of the 20 - 20Khz bandwidth. The ultimate test of any transformer is to have it in a circuit and do both square wave and sine wave tests to see how well the wave form is reproduced at the different frequencies as well as at both low and high power. This is of course assuming that the amplifier's circuitry is up to the task and not creating problems of it's own. Everything from the way you wind the transformer to the core material and size have an impact on the output transformer's performance. The text books and spice models are nice, but there is no substitute for actual experimentation with real functioning circuits. An actual amplifier is the best test bed and will present the transformer in question with all the dynamics of actual use. Although I have the Radiotron Designer's handbook and use it as well as tube manuals and other resources, I find that they will just get me in the ball park. They do not afford me the answers to end all debate. Thus the reason for my prototypes. There is another variable with respect to both power and output transformers. Heat. How hot do you want your amp to get? I could use a 175 ma power transformer and a 35 watt output in one amp I designed and it will work, and would probably last a reasonably long time. But the transformers get real hot. They work, but they feel too hot. Moving up to a 50 watt output and a 200 ma power transformer nets me a much cooler running amp with a total increase in transformer cost of around $25.00. Bill B. I know your question was aimed at Andre, who may have a lot more to say, but since its late here, I have a simple comment. For all OPTs, the power loss should be less than 10% at the worst, ie, the lowest load value, say 3 ohms. This means for 6 ohms, the losses will be 5%. For this ideal condition to apply, the winding resistance of the primary end to end should be 2.5% of the load resistance a-a seen by the output tubes. So if you have 10k a-a RL, the Rwp should be les than 250 ohms. The winding resistance of the secondary if its to suit 6 ohms should be also 2.5% of 6 ohms which is 0.15 ohms. As long as these conditions are met, you have a decent OPT with regard for losses. you will generally find that 3 amps per sq.mm is the current density often used allowable for design, but 2 amps/sq.mm is better, more likely to give the magic 5% winding losses. Manufacturers are often dominated by bean counters who tell the designers to reduce iron size, and reduce wire dia, and the 20% losses won't be noticed by the listeners. Patrick Turner. |
#5
|
|||
|
|||
Bill
Super to hear from someone with a relevant opinion and a courteous manner of expressing it. I see Patrick has already come up with some technical answers. I published that article by Menno van der Veen, the designer of the Plitron and Amplimo toroidal transformers, to give newbies a feeling for handling the numbers. Of course safety margins must be built in. I believe in generous margins, myself. You can see Menno's transformers recommended for various tubes, for which designs are in his book referenced in the article, at the Plitron site. His documentation for commercial designs is a thing of beauty and when I first saw the documentation for a custom design I nearly nearly threw a thrombie as I wondered who would pay for the days taken by such thorough tests. You will see that for The KISS Amp, the 300B I'm designing in the series to which KISS 117 is an informational supplement, the choice of output transformer, the Lundahl LL1623SE 90mA, on the 5K6 primary impedance linkup is rated 13W, more than adequate for the 3.8W the amp will actually produce. The Lundahls are particularly well-specified for low end power but in this instance, where the amp is being designed for a horn, we really don't want any power on the speaker under about 32Hz or even 36Hz if we are being paranoid. (As I am sure you know, a horn driver below its free air resonance is unloaded.) But I'm recommending these trannies all the same because they are so versatile and so well-priced for the quality. Same with the power supply, as you will see when the series reaches there. Essentially, including all bleeds, the amp requires 180mA but I'll be recommending the overspecced Lundahl 500mA LL1651 for the really rather good reason that one carriage charge for transformers is cheaper than two carriage charges from different suppliers. Again, it is a good tranny for its versatility (250-0-250V 500mA, 4x 4A 6.3V filament supplies), quality and price. It shouldn't overheat g. The versatile double chokes will be from Lundahl too. Actually, I know damn well that these trannies will handle the power way down beyond what I want and won't overheat because I've been using them for years. While I appreciate what you say about bench work I am, nominally at least, writing here for relative newbies. We don't want them to stick their fingers into a 500V amp more than is absolutely necessary. In fact, we don't want anyone to stick his fingers into HV gear more often than is absolutely necessary. I was horrified the other day to see on a kit sent for review, to be offered to Joe Public, that the bias adjustment and the humbusting are lid-off operations. How much could it have cost to bring the pot shafts outside the case? Andre Jute Wbittle wrote in message ... Andre, I worry about that power rating scheme. A certain diameter wire will only handle a certain amount of DC current. This is a factor in the power handling capabilities of the output transformer as well as it's AC power rating for the actual audio signal. You need to know both the DC current rating and what audio power the transformer is rated for. If I have a push pull amp using a pair of EL34's which is rated at 30 watts, you might think a 35 watt output transformer should be fine. But when you see the full specifications of 35 watts music power with a 100 ma DC current rating, it is all too obvious that the transformer is inadequate since the EL34's will pull 120 ma at idle through the output transformer. So, in many cases, although the amp puts out only 30 - 35 watts, the average 50 watt output transformer is what you need. Even if you use the frequency formula and limit your low frequencies to 30 hz as opposed to the transformer's 20hz low end rating, you still have a primary winding not up to the task of handling the 120ma idle current and 145 ma draw at full power. Another headache with choosing a transformer is the frequency bandwidth. Some companies rate their frequency response at full power. So, their 50 watt output delivers a full 20 - 20Khz at 50 watts. Others claim a 50 watt output, but rate their frequency response at only 1 watt to give a much inflated bandwidth. When you test the transformer at 50 watts, it falls far short of the 20 - 20Khz bandwidth. The ultimate test of any transformer is to have it in a circuit and do both square wave and sine wave tests to see how well the wave form is reproduced at the different frequencies as well as at both low and high power. This is of course assuming that the amplifier's circuitry is up to the task and not creating problems of it's own. Everything from the way you wind the transformer to the core material and size have an impact on the output transformer's performance. The text books and spice models are nice, but there is no substitute for actual experimentation with real functioning circuits. An actual amplifier is the best test bed and will present the transformer in question with all the dynamics of actual use. Although I have the Radiotron Designer's handbook and use it as well as tube manuals and other resources, I find that they will just get me in the ball park. They do not afford me the answers to end all debate. Thus the reason for my prototypes. There is another variable with respect to both power and output transformers. Heat. How hot do you want your amp to get? I could use a 175 ma power transformer and a 35 watt output in one amp I designed and it will work, and would probably last a reasonably long time. But the transformers get real hot. They work, but they feel too hot. Moving up to a 50 watt output and a 200 ma power transformer nets me a much cooler running amp with a total increase in transformer cost of around $25.00. Bill B. |
#6
|
|||
|
|||
Hi Andre, Your series of articles on the "KISS" ultra-fu amplifier may inspire me to see if I can find enough parts in my parts bin to build something along the lines you are describing. I am patiently waiting with some anticipation for the article on the power supply design, when will that be posted? I am also curious to see what you have to say about the driver stage. I hope I haven't accidentally missed these parts as I have been skimming some of the recent articles that seem to be repeating some material from earlier articles in the series? I am curious why you have dismissed push pull operation as more complex? I would think you could build a nice ultra-fi amplifier using a pair of class A1 push pull 2A3s, which wouldn't be anymore complicated than a 300B SE design. You could keep it simple by using a single ended driver with a transformer for phase splitting, and the driver parameters could be adjusted to provide the desired proportions of second and higher order harmonics in the output. Offsetting the added cost of the driver transformer would be lower costs for the power supply and output transformer in a push pull ultra-fi amplifier. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#7
|
|||
|
|||
John Byrns wrote:
Hi Andre, Yo, mon, I thought you'd gone, given up on RAT. Welcome back! Your series of articles on the "KISS" ultra-fu amplifier Ultra-kung-fu high voltage amplifier for dojo! Three falls loser get shocko! may inspire me to see if I can find enough parts in my parts bin to build something along the lines you are describing. I am patiently waiting with some anticipation for the article on the power supply design, when will that be posted? First I need to deal with the driver stage, then I come to the power stage, nasty little distractions permitting (see below). I am also curious to see what you have to say about the driver stage. I hope I haven't accidentally missed these parts You haven't missed anything. I am just waiting out a diplomaed quarterwit who came to RAT to set himself up as a sub-Plodnickian spoiler. When he tires of lunging about usurping my threads with negative messages and screeching about how his opinions are ‘facts', off I go again with The KISS Amp. Meanwhile a few of the guys thought it amusing in return for a little courtesy (which we haven't seen yet) to invite this unappetizing fellow to submit a silicon design tuned to sound like tubes. If and when it arrives we shall be able to judge if on his own patch the man knows anything we don't already know or can't find out from someone with better manners. You may remember that about ten years ago I published a pair of articles on the subject of making transistors sound like tubes (loud screeching from the usual useless monkeys in the peanut gallery), and then published the circuit from those articles in a gainclone barney some time later to demonstrate what an irreducible parts count really looks like. as I have been skimming some of the recent articles that seem to be repeating some material from earlier articles in the series? As a teenager I made a film with an old touring vaudevillian in a part I wrote especially for him. He taught me always to tell ‘em I'm gonna tell ‘em a joke, to tell ‘em the joke, then to tell ‘em I told ‘em the joke. Valuable advice in any branch of communications. But I'm not repeating material, I'm layering more complete and therefore more complicated versions, or deeper information, onto the earliest simplifications. I must say I honestly doubt that I know anything about power supply design that you don't. I do intend publishing in the power supply series of articles one about the math required to specify a power transformer but I don't intend to go into transformer design at all; I will treat it as black box, as I treated output transformers in the earlier part of the series, for fear of confusing some in my audience who, like me, actually trust the better winders to know their business. Perhaps that article when it arrives might be a hook for Patrick to tell us something deeper. I am curious why you have dismissed push pull operation as more complex? I would think you could build a nice ultra-fi amplifier using a pair of class A1 push pull 2A3s, which wouldn't be anymore complicated than a 300B SE design. I picked on the 300B for the perfect reason that it is the amp I am building right now and through the holidays, and because I have a lot of material in various articles I have written about 300B over the years. The reason I wrote so much about SE 300B was that there were always enough guys to write about PP amps. The 300B is the sensible hedonist's SE choice: everything bigger requires a kilovolt, anything smaller may be compromised by less sensitive speakers. For the 300B there is a range of speakers available at all prices. Anything smaller limits you to rare or very expensive speakers. The 300B is also the paradigm of the aspirational amp: everyone wants one at least once in his life. But it's the principles and examples that matter, not the particular tube. Perhaps when I finish this series I will describe my own favourite among my designs, my 25W T113 Class A PP triode linked EL34, or at least update the schematic for re-publication. You could keep it simple by using a single ended driver with a transformer for phase splitting, and the driver parameters could be adjusted to provide the desired proportions of second and higher order harmonics in the output. It's possible, John. (A nice little discussion and accompanying flame war on harmonic cancellation is just what we need to keep us giggling at Christmas!) But you know, my series would be twice as long if I wrote about PP before I wrote about SE. PP is fundamentally a more complicated concept than SE. And the correspondence about paraphase stages would take forever! And the flame war about NFB would take twice as long as forever! As for transformer coupling, I'm a longtime fan, but I'm trying to keep it really basic and simple. Back to my roots, I almost said, but actually my roots are in the Quad II and first series Quad ESL I bought as a student at the Rand Agricultural show lo those years ago. More recently (c1990) I was Born Again as a Class A1 ZNFB triode tubie. There is absolutely no reason that you should not design and build such an amplifier and describe it stage by stage here in a series of articles that could build into an instruction set. I for one would be very interested in reading you on the subject. Offsetting the added cost of the driver transformer would be lower costs for the power supply and output transformer in a push pull ultra-fi amplifier. I just don't see that, John. A good splitter tranny is a pricy item. The Lundahls I like don't have much price difference between SE and PP output transformers, and I always use their matching power supply too because it is beefy and has four equally beefy filament supplies and very handy segmentation. What you can actually save with push-pull operation is one choke because you may argue that PP doesn't need the choke input filter I put on SE amps. That one choke, 70 or 80 bucks even for a good dual-coil swinging choke 10H 200mA, won't pay for an interstage transformer of any quality, never mind for two, or to double up on the power tubes. Mind you, with the 15W one can get from a PP 2A3 pair one would have another 3dB over 300B, quite enough to drive more common, less sensitive speakers. Audio Innovations, referring now to the first valve incarnation of the company which gave birth to Audio Note UK, used to make a very sweet PP 2A3 amp, and I seem to remember that they publicly declared it their favourite amp by calling it First. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ Nice to hear from you. Andre Jute |
#8
|
|||
|
|||
|
#9
|
|||
|
|||
|
#11
|
|||
|
|||
John Byrns wrote: In article , (Andre Jute) wrote: John Byrns wrote: Hi Andre, Yo, mon, I thought you'd gone, given up on RAT. Welcome back! Not gone, just keeping a low profile because the discussion here often tends to be less than inspiring. Your series of articles on the "KISS" ultra-fu amplifier Ultra-kung-fu high voltage amplifier for dojo! Three falls loser get shocko! Just a typo, no shocko. may inspire me to see if I can find enough parts in my parts bin to build something along the lines you are describing. I am patiently waiting with some anticipation for the article on the power supply design, when will that be posted? First I need to deal with the driver stage, then I come to the power stage, nasty little distractions permitting (see below). I am also curious to see what you have to say about the driver stage. I hope I haven't accidentally missed these parts You haven't missed anything. I am just waiting out a diplomaed quarterwit who came to RAT to set himself up as a sub-Plodnickian spoiler. When he tires of lunging about usurping my threads with negative messages and screeching about how his opinions are ‘facts', off I go again with The KISS Amp. I have skipped reading the sub threads created by the diplomaed quarterwits. Meanwhile a few of the guys thought it amusing in return for a little courtesy (which we haven't seen yet) to invite this unappetizing fellow to submit a silicon design tuned to sound like tubes. If and when it arrives we shall be able to judge if on his own patch the man knows anything we don't already know or can't find out from someone with better manners. A few years back I toyed with the idea of a "KISS" transistor amp, it wasn't intended to sound like tubes though, and I never built it. You may remember that about ten years ago I published a pair of articles on the subject of making transistors sound like tubes (loud screeching from the usual useless monkeys in the peanut gallery), and then published the circuit from those articles in a gainclone barney some time later to demonstrate what an irreducible parts count really looks like. I can't say I remember that pair of articles, where did you publish them? I am curious why you have dismissed push pull operation as more complex? I would think you could build a nice ultra-fi amplifier using a pair of class A1 push pull 2A3s, which wouldn't be anymore complicated than a 300B SE design. I picked on the 300B for the perfect reason that it is the amp I am building right now and through the holidays, and because I have a lot of material in various articles I have written about 300B over the years. The reason I wrote so much about SE 300B was that there were always enough guys to write about PP amps. The 300B is the sensible hedonist's SE choice: everything bigger requires a kilovolt, anything smaller may be compromised by less sensitive speakers. For the 300B there is a range of speakers available at all prices. Anything smaller limits you to rare or very expensive speakers. The 300B is also the paradigm of the aspirational amp: everyone wants one at least once in his life. But it's the principles and examples that matter, not the particular tube. Perhaps when I finish this series I will describe my own favourite among my designs, my 25W T113 Class A PP triode linked EL34, or at least update the schematic for re-publication. You could keep it simple by using a single ended driver with a transformer for phase splitting, and the driver parameters could be adjusted to provide the desired proportions of second and higher order harmonics in the output. It's possible, John. (A nice little discussion and accompanying flame war on harmonic cancellation is just what we need to keep us giggling at Christmas!) But you know, my series would be twice as long if I wrote about PP before I wrote about SE. PP is fundamentally a more complicated concept than SE. It isn't clear to me that class A1 PP is actually anymore complicated than SE when you consider all the subtleties of SE design. IMHO, it is. You need more elements of gain. There are always two output tubes, not one. However, the phase splitter can be done away with completely by using a CCS tail to the class A PP pair of output tubes, therefore preventing any hint of class AB. The input drive can be with just one SET driver tube to one grid of the PP opvs. The other grid is grounded. This way the currents in the PP pair are always equalised, or balanced with equal but opposite phased currents. You get lower thd, still predominantly 2H from the driver triode, and you get twice the po of SE. And the correspondence about paraphase stages would take forever! And the flame war about NFB would take twice as long as forever! Who said anything about "NFB"? Why is it always assumed that PP involves "NFB"? I'm talking about a simple class A1 PP triode amp which has less need for "NFB" than does an SE design. The issue of NFB will stir up controversy. However, for those who don't wish to use it in a powe amp and who also want low Ro for their triode amps, then the only recourse is to high output transformer P-S turn ratios which gi9ve high impedance ratios, and thus the plate resistance is transformed down, when looking into the secondary of the OPT. The same equally applies to PP or SE amps, and imho, there is no more need for NFB in SE amps compared to PP. As for transformer coupling, I'm a longtime fan, but I'm trying to keep it really basic and simple. Back to my roots, I almost said, but actually my roots are in the Quad II and first series Quad ESL I bought as a student at the Rand Agricultural show lo those years ago. More recently (c1990) I was Born Again as a Class A1 ZNFB triode tubie. There is absolutely no reason that you should not design and build such an amplifier and describe it stage by stage here in a series of articles that could build into an instruction set. I for one would be very interested in reading you on the subject. Offsetting the added cost of the driver transformer would be lower costs for the power supply and output transformer in a push pull ultra-fi amplifier. I just don't see that, John. A good splitter tranny is a pricy item. The Lundahls I like don't have much price difference between SE and PP output transformers, and I always use their matching power supply too because it is beefy and has four equally beefy filament supplies and very handy segmentation. What you can actually save with push-pull operation is one choke because you may argue that PP doesn't need the choke input filter I put on SE amps. That one choke, 70 or 80 bucks even for a good dual-coil swinging choke 10H 200mA, won't pay for an interstage transformer of any quality, never mind for two, or to double up on the power tubes. With PP, in addition to saving at least one choke in the power supply, the output transformers can achieve the same performance with less iron and copper, making them less expensive. Imho, a choke in the PS is good practice for either PP or SE amps. The PP amp needs less smoothing, but I jave never built an amp without a smooth B+ supply to the CT. It makes for easier less lossy B+ voltages for the driver stages which can rely on RC filtering. The doubled up power tubes in PP are smaller and hence cost less. This would be true if one said we'll use a pair of 2A3 instead of one 300B. But 2A3 are not all that cheap either. If the cost of a splitter transformer bothers you, you can always add a second triode to the front end to form an electronic splitter at minimal cost. I suggested the transformer splitter in the spirit of "KISS". See above, and have neither phase splitter, transformer, or two input tubes. But a CCS is needed. That can be a solid state item, as a slave to the current whims of the tubes. ( I know any talk of transistorized CCS will have lotsa folks vomiting all over me...) But a choke can also be used in place of the CCS. As I have been reading through your "KISS" series I noticed that in the "KISS 113" article you make the following statement: "The Lundahl 1623-SE can be wired on the secondary to reflect impedances of 1.6-3.0-5.6Kohm onto the primary. (This is the correct way to make a multi-use output transformer; actually tapping the primary is an egregious practice for reasons we don't have the space to go into here.) The two useful ones for SE 300B are 3.0Kohm, which is pretty much a default 300B load, and 5.6Kohm, which is near enough the upper limit of 8x plate resistance. It has different power ratings at these impedances ranging from 13W to 50W, so our desired 3W will never drive it into saturation." I noted with particular interest the part of your statement where you say "tapping the primary is an egregious practice". Didn't you once write an editorial for Glass Audio magazine which if not advocating tapping the primary, at least questioned why it wasn't done more often? It seems to me that if it is necessary to provide multiple primary impedances, tapping the primary is less egregious than Lundahl's scheme of changing the impedance level of the whole transformer by reconfiguring the secondary, which results in higher copper losses than simply tapping the primary to provide a lower impedance. I know some people find the idea of the loose end of the primary winding flapping in the breeze, as it were, to be somewhat unsettling. I don't see that as a great problem with proper design of the transformer and its application in the amplifier circuit. Its so much better practice to use all the windings at all times and have the same current densities in all multi section secondaries at all configurations to achieve impedance matching. However, if a designer set out to make his primary winding tapable, without increasing leakage inductance, and without reducing the **ratio** between the load and the nominal RL seen by the tube, then reducing primary turns would be OK. In other words, we may like to see 24 Henrys of inductance with a 3k : 6 ohm OPT. If we had a tap at 0.7 of the turns, then the match would be 1.5k to 6 ohms, and the inductance could be 1/2 the 24 H, and indeed it would be because the inductance varies as the square or the square root of the turn ratio change. The leakage also reduces with reducing P turns, providing the PS interface geometry is the same. So reducing the load match from 3k to 1.5k, two output tubes could be used in parallel, if the OPT is designed for that. But its better to design for 2 tubes to begin with, and when one tube is used, the P inductance is still 24 H, and we **adjust the windings of the secondary** to get the load match, and without wasting sections or having unequal current density. The arrangement of the OPT No 1 details at my website has the winding details that achieve this. AFAIK, that design of mine could be used for an SE 300B amp, the core would be gapped, and the DC fed from one end to the other of the promary. I wouldn't like to see anyone tap the primary down. The old UTC here in the US offered several output transformers in their LS series that provided two different primary impedances by means of primary taps. In my opinion the best way to accomplish multiple primary impedances is the way Sowter does it in their series of experimenters transformers, IIRC. What Sowter did is bring out separate leads for the multiple primary sections so that they could be connected in various series/parallel combinations to achieve the desired primary impedance with a fixed secondary impedance. But this has serious limitations with regard to current density. using 0.7 of the P turns means that 0.3 of the P turns are wasted, or then connected across 0.3 of the turns within the winding of 0.7 of the turns. Its not quite right. With this method the magnetic operating point of the transformer does not change as it does when the secondary is reconfigured. I am not so sure. If you have an OPT to suit 3k and with 80 mA, then one reduces the turns by 0.7 times to suit 1.5k, and you want the same power, then the DC has to increase. But I guess is that if the DC magnetisation stays constant, its OK, ie, the turns x DC current, or amp-turns, is constant. Besides eliminating the loose primary end, losses are also minimized because all the copper is used in a way that minimizes the winding resistance, unlike what happens when the secondary is reconfigured. If the turns were reduced by a factor of 0.7, for halving the P load, there will be at least 0.4 times the total turns that are in just one winding, so the primary Rw will not decline by half to get the same low% winding losses. As I said in another post, having winding losses below 10% is not too bad for any SE OPT. I prefer to have an OPT wound to suit the intended primary load on the tubes. If you have an OPT designed for 1.5k to 6 ohms, then it may be suitable for a 3k to 12 ohm situation, but then the P inductance will be too low, although the winding losses are lower, and HF is better, but the LF is poorer. Using a 3k to 6 ohm tranny for connection to 1.5k to 3ohms will double the winding % losses, and reduce HF, but extend the LF. But the tranny may be over stressed by the DC. Any discussion about OPT design is a minefield for anyone not prepared to consider six variables at the one time. Patrick Turner. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#12
|
|||
|
|||
|
#13
|
|||
|
|||
You promised a circuit. You have sent hundreds of negative and abusive
messages, including racist slurs, but no circuit. Mr Jute ignores you with contempt. Why should I not conclude that he is right? Frank B Stewart Pinkerton wrote in message . .. On 4 Dec 2004 21:09:04 -0800, (Frank B) wrote: I too thought we were waiting for Pinkerton to deliver a transistor amplifier. But now it appears we were wasting our time. No one seems to be clear whether this is supposed to be a tranny equivalent of Andre's design, an 'ultra-fidelity' SE SS amp, or an amp designed specifically to mimic the sound of a SE 300B design. My own inclination would be to make it all BJT, with a CCS in the supply rail and a CF input buffer, and employing a custom-wound OPT. That's about as KISS as it gets, IMHO. Whichever, I have yet to dust down my slide rule on such a venture, since it would for me be only a bit of fun, not any saerious attempt to produce a truly 'ultra-fidelity' amplifier, which requires solid-state *and* push-pull for success. |
#14
|
|||
|
|||
(Frank B) wrote in message m...
Dear Mr Jute, (Andre Jute) wrote in message . com... John Byrns wrote: I am also curious to see what you have to say about the driver stage. I hope I haven't accidentally missed these parts You haven't missed anything. I am just waiting out a diplomaed quarterwit who came to RAT to set himself up as a sub-Plodnickian spoiler. When he tires of lunging about usurping my threads with negative messages and screeching about how his opinions are ?facts', off I go again with The KISS Amp. Meanwhile a few of the guys thought it amusing in return for a little courtesy (which we haven't seen yet) to invite this unappetizing fellow to submit a silicon design tuned to sound like tubes. If and when it arrives we shall be able to judge if on his own patch the man knows anything we don't already know or can't find out from someone with better manners. You may remember that about ten years ago I published a pair of articles on the subject of making transistors sound like tubes (loud screeching from the usual useless monkeys in the peanut gallery), and then published the circuit from those articles in a gainclone barney some time later to demonstrate what an irreducible parts count really looks like. I too thought we were waiting for Pinkerton to deliver a transistor amplifier. But now it appears we were wasting our time. Is there any chance we can see your articles on tube sound from transistors while we wait for the rest of the KISS amp? A sincere thanks in advance for all your hard work. Frank B. I've posted those articles as KISS 191A and 191B. The schemo, ps and layout are on a netsite I have set up for The Kiss Amp at http://members.lycos.co.uk/fiultra/ . Enjoy! Pinkerton isn't interested in doing useful work but in posing as an expert. He isn't, he's a gossip monger. I stopped reading his posts the moment I discovered he knows nothing I don't. You could grow old waiting for his sort of tenth-rate "engineer" to do something useful or even mildly entertaining. Pinky is a waste of good oxygen. Andre Jute |
#15
|
|||
|
|||
On 5 Dec 2004 11:58:40 -0800, (Frank B) wrote:
You promised a circuit. You have sent hundreds of negative and abusive messages, including racist slurs, but no circuit. Mr Jute ignores you with contempt. Why should I not conclude that he is right? Excuse me? I made no 'promise' whatever, I simply suggested that it might be interesting to devise such a circuit. Furthermore, don't take Andre's word for content, he's a known liar and pathological abuser. Remember, *he* is the one who started all these pure attack threads, not me. One thing Andre is *not* doing, is ignoring me, in fact he has gone out of his way to start numerous attack threads, but has *never* refuted my correction of his mistakes. The man is clearly an egomaniacal old windbag who cannot stand his factual errors being exposed. Stewart Pinkerton wrote in message . .. On 4 Dec 2004 21:09:04 -0800, (Frank B) wrote: I too thought we were waiting for Pinkerton to deliver a transistor amplifier. But now it appears we were wasting our time. No one seems to be clear whether this is supposed to be a tranny equivalent of Andre's design, an 'ultra-fidelity' SE SS amp, or an amp designed specifically to mimic the sound of a SE 300B design. My own inclination would be to make it all BJT, with a CCS in the supply rail and a CF input buffer, and employing a custom-wound OPT. That's about as KISS as it gets, IMHO. Whichever, I have yet to dust down my slide rule on such a venture, since it would for me be only a bit of fun, not any serious attempt to produce a truly 'ultra-fidelity' amplifier, which requires solid-state *and* push-pull for success. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#16
|
|||
|
|||
In article , Patrick Turner
wrote: John Byrns wrote: As I have been reading through your "KISS" series I noticed that in the "KISS 113" article you make the following statement: "The Lundahl 1623-SE can be wired on the secondary to reflect impedances of 1.6-3.0-5.6Kohm onto the primary. (This is the correct way to make a multi-use output transformer; actually tapping the primary is an egregious practice for reasons we don't have the space to go into here.) The two useful ones for SE 300B are 3.0Kohm, which is pretty much a default 300B load, and 5.6Kohm, which is near enough the upper limit of 8x plate resistance. It has different power ratings at these impedances ranging from 13W to 50W, so our desired 3W will never drive it into saturation." I noted with particular interest the part of your statement where you say "tapping the primary is an egregious practice". Didn't you once write an editorial for Glass Audio magazine which if not advocating tapping the primary, at least questioned why it wasn't done more often? It seems to me that if it is necessary to provide multiple primary impedances, tapping the primary is less egregious than Lundahl's scheme of changing the impedance level of the whole transformer by reconfiguring the secondary, which results in higher copper losses than simply tapping the primary to provide a lower impedance. I know some people find the idea of the loose end of the primary winding flapping in the breeze, as it were, to be somewhat unsettling. I don't see that as a great problem with proper design of the transformer and its application in the amplifier circuit. Its so much better practice to use all the windings at all times and have the same current densities in all multi section secondaries at all configurations to achieve impedance matching. However, if a designer set out to make his primary winding tapable, without increasing leakage inductance, and without reducing the **ratio** between the load and the nominal RL seen by the tube, then reducing primary turns would be OK. In other words, we may like to see 24 Henrys of inductance with a 3k : 6 ohm OPT. If we had a tap at 0.7 of the turns, then the match would be 1.5k to 6 ohms, and the inductance could be 1/2 the 24 H, and indeed it would be because the inductance varies as the square or the square root of the turn ratio change. The leakage also reduces with reducing P turns, providing the PS interface geometry is the same. So reducing the load match from 3k to 1.5k, two output tubes could be used in parallel, if the OPT is designed for that. But its better to design for 2 tubes to begin with, and when one tube is used, the P inductance is still 24 H, and we **adjust the windings of the secondary** to get the load match, and without wasting sections or having unequal current density. The arrangement of the OPT No 1 details at my website has the winding details that achieve this. AFAIK, that design of mine could be used for an SE 300B amp, the core would be gapped, and the DC fed from one end to the other of the promary. I wouldn't like to see anyone tap the primary down. The old UTC here in the US offered several output transformers in their LS series that provided two different primary impedances by means of primary taps. In my opinion the best way to accomplish multiple primary impedances is the way Sowter does it in their series of experimenters transformers, IIRC. What Sowter did is bring out separate leads for the multiple primary sections so that they could be connected in various series/parallel combinations to achieve the desired primary impedance with a fixed secondary impedance. But this has serious limitations with regard to current density. using 0.7 of the P turns means that 0.3 of the P turns are wasted, or then connected across 0.3 of the turns within the winding of 0.7 of the turns. Its not quite right. With this method the magnetic operating point of the transformer does not change as it does when the secondary is reconfigured. I am not so sure. If you have an OPT to suit 3k and with 80 mA, then one reduces the turns by 0.7 times to suit 1.5k, and you want the same power, then the DC has to increase. But I guess is that if the DC magnetisation stays constant, its OK, ie, the turns x DC current, or amp-turns, is constant. Besides eliminating the loose primary end, losses are also minimized because all the copper is used in a way that minimizes the winding resistance, unlike what happens when the secondary is reconfigured. If the turns were reduced by a factor of 0.7, for halving the P load, there will be at least 0.4 times the total turns that are in just one so winding, the primary Rw will not decline by half to get the same low% winding losses. As I said in another post, having winding losses below 10% is not too bad for any SE OPT. I prefer to have an OPT wound to suit the intended primary load on the tubes. If you have an OPT designed for 1.5k to 6 ohms, then it may be suitable for a 3k to 12 ohm situation, but then the P inductance will be too low, although the winding losses are lower, and HF is better, but the LF is poorer. Using a 3k to 6 ohm tranny for connection to 1.5k to 3ohms will double the winding % losses, and reduce HF, but extend the LF. But the tranny may be over stressed by the DC. Any discussion about OPT design is a minefield for anyone not prepared to consider six variables at the one time. Yes, it is indeed a minefield for mere mortals who typically can keep track of only five variables at one time, perhaps that explains your surprising endorsement of building transformers with multiple primary impedances by changing only the configuration of the secondary. The only virtue I can see in changing the primary impedance by reconfiguring the secondary is that the secondary is generally already designed to be reconfigured to accommodate various load impedances, so using the same facility to change the primary impedance simplifies the transformer by reducing the number of lead outs required. Without getting into all the variables involved, changing the impedance level of a fixed primary winding from say 1.5k to 3.0k, as in your example, by reconfiguring the connections of secondary sections will result in a loss of low frequency power bandwidth at the higher impedance, due to core saturation effects. The transformer will also have greater copper losses at the lower primary impedance because the copper will not be used as efficiently as at the higher impedance. Also if it is a single ended transformer, for a given power output, the DC current in the primary will be greater at the lower impedance meaning more ampere turns which will require more iron than would otherwise be required to avoid saturation. Other than DC heating effects, your objection to non uniform current densities in the primary doesn't make any sense because you are simply transferring the problem of non uniform current density to the secondary when you reconfigure the secondary. By allowing for the reconfiguration of the connections of the primary sections these problems are eliminated. The low frequency power bandwidth remains independent of the primary impedance, copper losses are minimized when the transformer is configured for the low primary impedance, and extra iron is not required to accommodate the increased DC necessary for a given output power level in an SE transformer. The bottom line is that just as the best way to accommodate various loads is by providing for the reconfiguration of the secondary, the best way to provide for various primary impedances is to provide for a reconfigurable primary. It's sort of like making sausage, you make use of all the parts. In fact didn't you have some transformers on your web page that were built just this way, making your endorsement of changing the primary impedance by reconfiguring the secondary even more surprising? Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#17
|
|||
|
|||
Stewart Pinkerton said:
My own inclination would be to make it all BJT, with a CCS in the supply rail and a CF input buffer, and employing a custom-wound OPT. That's about as KISS as it gets, IMHO. I, for one, would applaud any attempt in this direction. Especially the transformer-coupling interests me! -- Sander de Waal " SOA of a KT88? Sufficient. " |
#18
|
|||
|
|||
Patrick Turner said:
But a CCS is needed. That can be a solid state item, as a slave to the current whims of the tubes. ( I know any talk of transistorized CCS will have lotsa folks vomiting all over me...) Why? Makes perfect sense to me. -- Sander de Waal " SOA of a KT88? Sufficient. " |
#19
|
|||
|
|||
On Mon, 06 Dec 2004 19:48:25 +0100, Sander deWaal
wrote: Stewart Pinkerton said: My own inclination would be to make it all BJT, with a CCS in the supply rail and a CF input buffer, and employing a custom-wound OPT. That's about as KISS as it gets, IMHO. I, for one, would applaud any attempt in this direction. Especially the transformer-coupling interests me! Ahhh. I've been thinking, and I've kinda gone off the OPT. Very expensive to get a really good one (Lundahl or Sowter custom-wound), and not really necessary for a SS SE amp. One might argue that it would more closely mimic the valve equivalent, but in terms of the common claims of super- linearity for the SET, the OPT is an obvious weak spot. In view of the possiblity that low-level linearity is indeed critical, I have instead determined to go down the path of a *truly* maximised linearity, while eschewing the easy options of global NFB and PP operation. It will be interesting to see where this path leads............ -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#20
|
|||
|
|||
Stewart Pinkerton said:
My own inclination would be to make it all BJT, with a CCS in the supply rail and a CF input buffer, and employing a custom-wound OPT. That's about as KISS as it gets, IMHO. I, for one, would applaud any attempt in this direction. Especially the transformer-coupling interests me! Ahhh. I've been thinking, and I've kinda gone off the OPT. Very expensive to get a really good one (Lundahl or Sowter custom-wound), and not really necessary for a SS SE amp. One might argue that it would more closely mimic the valve equivalent, but in terms of the common claims of super- linearity for the SET, the OPT is an obvious weak spot. In view of the possiblity that low-level linearity is indeed critical, I have instead determined to go down the path of a *truly* maximised linearity, while eschewing the easy options of global NFB and PP operation. It will be interesting to see where this path leads............ OTOH, McIntosh used iron in some SS amps with good results. Analogous to their tube amps, you could use one primary winding in the drain, and one in the source path. Feedback, anyone? ;-) With the lower impedances involved, a simple transformer shouldn't have to be the bottleneck. Miller's the one to beat! -- Sander de Waal " SOA of a KT88? Sufficient. " |
#21
|
|||
|
|||
On Mon, 06 Dec 2004 23:23:13 +0100, Sander deWaal
wrote: Stewart Pinkerton said: My own inclination would be to make it all BJT, with a CCS in the supply rail and a CF input buffer, and employing a custom-wound OPT. That's about as KISS as it gets, IMHO. I, for one, would applaud any attempt in this direction. Especially the transformer-coupling interests me! Ahhh. I've been thinking, and I've kinda gone off the OPT. Very expensive to get a really good one (Lundahl or Sowter custom-wound), and not really necessary for a SS SE amp. One might argue that it would more closely mimic the valve equivalent, but in terms of the common claims of super- linearity for the SET, the OPT is an obvious weak spot. In view of the possiblity that low-level linearity is indeed critical, I have instead determined to go down the path of a *truly* maximised linearity, while eschewing the easy options of global NFB and PP operation. It will be interesting to see where this path leads............ OTOH, McIntosh used iron in some SS amps with good results. You might argue that it made their SS amps sound like tube amps - not necessarily a good thing? :-) Analogous to their tube amps, you could use one primary winding in the drain, and one in the source path. Feedback, anyone? ;-) With the lower impedances involved, a simple transformer shouldn't have to be the bottleneck. Agreed it's a toss-up between putting iron or a large electrolytic in the signal path. I don't want to complicate the design with split supplies and a DC servo, that seems to miss the point. Miller's the one to beat! Grolsch beats it every time! :-) -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#22
|
|||
|
|||
John Byrns wrote: In article , Patrick Turner wrote: John Byrns wrote: As I have been reading through your "KISS" series I noticed that in the "KISS 113" article you make the following statement: "The Lundahl 1623-SE can be wired on the secondary to reflect impedances of 1.6-3.0-5.6Kohm onto the primary. (This is the correct way to make a multi-use output transformer; actually tapping the primary is an egregious practice for reasons we don't have the space to go into here.) The two useful ones for SE 300B are 3.0Kohm, which is pretty much a default 300B load, and 5.6Kohm, which is near enough the upper limit of 8x plate resistance. It has different power ratings at these impedances ranging from 13W to 50W, so our desired 3W will never drive it into saturation." I noted with particular interest the part of your statement where you say "tapping the primary is an egregious practice". Didn't you once write an editorial for Glass Audio magazine which if not advocating tapping the primary, at least questioned why it wasn't done more often? It seems to me that if it is necessary to provide multiple primary impedances, tapping the primary is less egregious than Lundahl's scheme of changing the impedance level of the whole transformer by reconfiguring the secondary, which results in higher copper losses than simply tapping the primary to provide a lower impedance. I know some people find the idea of the loose end of the primary winding flapping in the breeze, as it were, to be somewhat unsettling. I don't see that as a great problem with proper design of the transformer and its application in the amplifier circuit. Its so much better practice to use all the windings at all times and have the same current densities in all multi section secondaries at all configurations to achieve impedance matching. However, if a designer set out to make his primary winding tapable, without increasing leakage inductance, and without reducing the **ratio** between the load and the nominal RL seen by the tube, then reducing primary turns would be OK. In other words, we may like to see 24 Henrys of inductance with a 3k : 6 ohm OPT. If we had a tap at 0.7 of the turns, then the match would be 1.5k to 6 ohms, and the inductance could be 1/2 the 24 H, and indeed it would be because the inductance varies as the square or the square root of the turn ratio change. The leakage also reduces with reducing P turns, providing the PS interface geometry is the same. So reducing the load match from 3k to 1.5k, two output tubes could be used in parallel, if the OPT is designed for that. But its better to design for 2 tubes to begin with, and when one tube is used, the P inductance is still 24 H, and we **adjust the windings of the secondary** to get the load match, and without wasting sections or having unequal current density. The arrangement of the OPT No 1 details at my website has the winding details that achieve this. AFAIK, that design of mine could be used for an SE 300B amp, the core would be gapped, and the DC fed from one end to the other of the promary. I wouldn't like to see anyone tap the primary down. The old UTC here in the US offered several output transformers in their LS series that provided two different primary impedances by means of primary taps. In my opinion the best way to accomplish multiple primary impedances is the way Sowter does it in their series of experimenters transformers, IIRC. What Sowter did is bring out separate leads for the multiple primary sections so that they could be connected in various series/parallel combinations to achieve the desired primary impedance with a fixed secondary impedance. But this has serious limitations with regard to current density. using 0.7 of the P turns means that 0.3 of the P turns are wasted, or then connected across 0.3 of the turns within the winding of 0.7 of the turns. Its not quite right. With this method the magnetic operating point of the transformer does not change as it does when the secondary is reconfigured. I am not so sure. If you have an OPT to suit 3k and with 80 mA, then one reduces the turns by 0.7 times to suit 1.5k, and you want the same power, then the DC has to increase. But I guess is that if the DC magnetisation stays constant, its OK, ie, the turns x DC current, or amp-turns, is constant. Besides eliminating the loose primary end, losses are also minimized because all the copper is used in a way that minimizes the winding resistance, unlike what happens when the secondary is reconfigured. If the turns were reduced by a factor of 0.7, for halving the P load, there will be at least 0.4 times the total turns that are in just one so winding, the primary Rw will not decline by half to get the same low% winding losses. As I said in another post, having winding losses below 10% is not too bad for any SE OPT. I prefer to have an OPT wound to suit the intended primary load on the tubes. If you have an OPT designed for 1.5k to 6 ohms, then it may be suitable for a 3k to 12 ohm situation, but then the P inductance will be too low, although the winding losses are lower, and HF is better, but the LF is poorer. Using a 3k to 6 ohm tranny for connection to 1.5k to 3ohms will double the winding % losses, and reduce HF, but extend the LF. But the tranny may be over stressed by the DC. Any discussion about OPT design is a minefield for anyone not prepared to consider six variables at the one time. Yes, it is indeed a minefield for mere mortals who typically can keep track of only five variables at one time, perhaps that explains your surprising endorsement of building transformers with multiple primary impedances by changing only the configuration of the secondary. The only virtue I can see in changing the primary impedance by reconfiguring the secondary is that the secondary is generally already designed to be reconfigured to accommodate various load impedances, so using the same facility to change the primary impedance simplifies the transformer by reducing the number of lead outs required. Without getting into all the variables involved, changing the impedance level of a fixed primary winding from say 1.5k to 3.0k, as in your example, by reconfiguring the connections of secondary sections will result in a loss of low frequency power bandwidth at the higher impedance, due to core saturation effects. The transformer has to be designed for worst case use if it is also to be allowed to be used various primary impedance matches. The transformer will also have greater copper losses at the lower primary impedance because the copper will not be used as efficiently as at the higher impedance. Also if it is a single ended transformer, for a given power output, the DC current in the primary will be greater at the lower impedance meaning more ampere turns which will require more iron than would otherwise be required to avoid saturation. Other than DC heating effects, your objection to non uniform current densities in the primary doesn't make any sense because you are simply transferring the problem of non uniform current density to the secondary when you reconfigure the secondary. All the transformers I wind have variable impedance matching and employ secondaries which can be re-arranged in various ways to match loads so that the primary load seen by a pair of tubes is the same, so if you want 50 watts into 5k a-a in a PP amp, it matters not whether one uses 8 x EL84, 4 x EL34, or 2 x KT90, the load is 5k, and max allowable po = 50 watts for the saturation F to remain as it was designed to be. 50 watts into 5k is 500vrms. Using such a transformer for 100 watts into 10k would mean Va-a = 1,000v. The problem would be that Fsat would be an octave higher. The copper losses will be lower, and HF roll off at a higher F, ( but still depending on the R source.) There are no dissimilar current densities in each strand of secondary wire once the OPT has been set up for chosen load match. I would have thought I made all this perfectly clear at my website where considerable info exists about OPT design. By allowing for the reconfiguration of the connections of the primary sections these problems are eliminated. The low frequency power bandwidth remains independent of the primary impedance, copper losses are minimized when the transformer is configured for the low primary impedance, and extra iron is not required to accommodate the increased DC necessary for a given output power level in an SE transformer. You echo exactly what I said about OPTs with regard to designing with primary impedance changing in mind. I still think "universal transformers" have limitations. I find that designing an OPT for an optimum nominal primay load will yield the best design, and once you start to reduce the number of turns used on a primary, the fill factor declines, and the benefits of having a full window full of wire carrying current is lost, and its a compromise I won't make. My recomendation is to use all the primary of a given OPT, and always use enough devices to suit the transformer, rather than use primary taps Consider a 300B used with an OPT for 3k to 6 ohms for 8 watts. 154 vrms is the voltage across the primary. Now say we used 2 x 300B, so we'd get 16 watts. We'd need to have the load dropped to 1.5k at the primary, so we'd need to have the secondary arranged with 1.414 more turns than for 6 ohms, so that it then gives 1.5k to 6 ohms. The voltage across the primary will be 154 vrms for the two tubes. The transformer would have twice the winding losses for the 16 watt version. The inductance and gapping should have been designed for the worser case of the 16 watt amp, so that with increased DC, the inductance is still high enough and the permeability hasn't been lessened to allow saturation at a higher F. The actual inductance required for the primary is halved if the load is halved The bottom line is that just as the best way to accommodate various loads is by providing for the reconfiguration of the secondary, the best way to provide for various primary impedances is to provide for a reconfigurable primary. It's sort of like making sausage, you make use of all the parts. But not if you waste unused portions of windings. I make OPTs, not sausages, and only one at a time comes out of the system, and but they do tend to fit a reasonable range of load matching. OPT No1 is a fairly versatile design, not too hard to wind, good leakage figures, good inductance figures, and with low enough losses and high enough Lp to suit the range of loads intended. I think it would also suit an SE design OK for about 1/2 the PP power possible. The core needs gapping, instead of the usual interleaved core. Somebody could divide the primary up into sections if they wanted to but any advantage in using a potion of the primary as-is is not great, and usually involves a disadvantage, unless the halves of the primary are paralleled, to provide a load equal to 1/4 the full primary winding. Then you find the voltage at the P should be 1/2 that for the whole winding, and since RL is 1/4 of that for the full winding, then the same power output is all that is possible, but losses are the same. In fact didn't you have some transformers on your web page that were built just this way, making your endorsement of changing the primary impedance by reconfiguring the secondary even more surprising? There are no recomendations at my site for reducing the primary turns except by way of paralleling 1/2 primaries. Its very unlikely that someone would ever need to parallel the two halves of the OPT No 1 to get a load match to 2k instead of 8k. It only allows the same power output. It would mean that although OPT No1 could normally be used for 2 x KT88 with an 8 k load, the OPT could also be used with a octet pack of EL84, so that each pair would be seeing 8k. The 2k load in lieu of the 8k load may also suit 6AS7 instead of KT88 etc. Patrick Turner. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#23
|
|||
|
|||
Sander deWaal wrote: Patrick Turner said: But a CCS is needed. That can be a solid state item, as a slave to the current whims of the tubes. ( I know any talk of transistorized CCS will have lotsa folks vomiting all over me...) Why? Makes perfect sense to me. And I thought you liked me. Oh well, if they wanna spew, I won't stop em, but when I come over soon, you won't like the pong... Patrick Turner. -- Sander de Waal " SOA of a KT88? Sufficient. " |
#24
|
|||
|
|||
"Stewart Pinkerton" wrote in message ... On 4 Dec 2004 21:09:04 -0800, (Frank B) wrote: I too thought we were waiting for Pinkerton to deliver a transistor amplifier. But now it appears we were wasting our time. No one seems to be clear whether this is supposed to be a tranny equivalent of Andre's design, an 'ultra-fidelity' SE SS amp, or an amp designed specifically to mimic the sound of a SE 300B design. My own inclination would be to make it all BJT, with a CCS in the supply rail and a CF input buffer, and employing a custom-wound OPT. That's about as KISS as it gets, IMHO. Whichever, I have yet to dust down my slide rule on such a venture, since it would for me be only a bit of fun, not any saerious attempt to produce a truly 'ultra-fidelity' amplifier, which requires solid-state *and* push-pull for success. -- Stewart Pinkerton | Music is Art - Audio is Engineering Hello Stewart, It was my suggestion that you could run a parallel thread to Andre's excellent KISS amp thread, and take us through the design of a low power SS amplifier, so that those of us who are interested could build both and compare them. This too will lead to an interesting new thread. We must be careful here to compare apples with apples, so your brief should be to design an amplifier with a similar power output to Andre's SE valve amp, with a similar amount of feedback (if any). No more. No less. For a man of your capabilility, it should present no problems:-) Iain |
#25
|
|||
|
|||
Patrick Turner said:
( I know any talk of transistorized CCS will have lotsa folks vomiting all over me...) Why? Makes perfect sense to me. And I thought you liked me. Oh well, if they wanna spew, I won't stop em, but when I come over soon, you won't like the pong... Umm.......I meant the SS CCS ;-) -- Sander de Waal " SOA of a KT88? Sufficient. " |
#26
|
|||
|
|||
On Tue, 7 Dec 2004 18:28:07 +0200, "Iain M Churches"
wrote: "Stewart Pinkerton" wrote in message .. . On 4 Dec 2004 21:09:04 -0800, (Frank B) wrote: I too thought we were waiting for Pinkerton to deliver a transistor amplifier. But now it appears we were wasting our time. No one seems to be clear whether this is supposed to be a tranny equivalent of Andre's design, an 'ultra-fidelity' SE SS amp, or an amp designed specifically to mimic the sound of a SE 300B design. My own inclination would be to make it all BJT, with a CCS in the supply rail and a CF input buffer, and employing a custom-wound OPT. That's about as KISS as it gets, IMHO. Whichever, I have yet to dust down my slide rule on such a venture, since it would for me be only a bit of fun, not any saerious attempt to produce a truly 'ultra-fidelity' amplifier, which requires solid-state *and* push-pull for success. -- Stewart Pinkerton | Music is Art - Audio is Engineering Hello Stewart, It was my suggestion that you could run a parallel thread to Andre's excellent KISS amp thread, and take us through the design of a low power SS amplifier, so that those of us who are interested could build both and compare them. This too will lead to an interesting new thread. We must be careful here to compare apples with apples, so your brief should be to design an amplifier with a similar power output to Andre's SE valve amp, with a similar amount of feedback (if any). No more. No less. For a man of your capabilility, it should present no problems:-) That's the plan, and I'm even going to use the hated BJT rather than the 'valve-like' MOSFET. Targeting 12 watts into 4 ohms, which should give similar max output voltage to Andre's amp, but with a bit more drive capacity for typical modern low-impedance speakers. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#27
|
|||
|
|||
Sander deWaal wrote: Patrick Turner said: ( I know any talk of transistorized CCS will have lotsa folks vomiting all over me...) Why? Makes perfect sense to me. And I thought you liked me. Oh well, if they wanna spew, I won't stop em, but when I come over soon, you won't like the pong... Umm.......I meant the SS CCS ;-) I, I, I Knew what yoooo meant young man, and its all too much.... All those Silly Stupid cute current sauces are all a bit much after awhile.... Makes them sick they do. Patrick Turner. -- Sander de Waal " SOA of a KT88? Sufficient. " |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
KISS 115 by Andre Jute | Vacuum Tubes | |||
KISS 114 by Andre Jute | Vacuum Tubes | |||
KISS 113 by Andre Jute | Vacuum Tubes | |||
KISS 111 by Andre Jute | Vacuum Tubes | |||
KISS 102 by Andre Jute | Vacuum Tubes |