Home |
Search |
Today's Posts |
#81
Posted to rec.audio.tubes
|
|||
|
|||
The KISS AMP: square-rooting the noise
My original long reply seem to have gone missing in the ether. Here's a
condensed recreation. John Stewart wrote: Andre Jute wrote: choke input ..... is traditionally built with two sections, meaning two chokes. Not traditionally built in two sections at all. RDH p1192ff. 1954 is traditional enough for me. Presumably you're now on the right circuit: http://members.lycos.co.uk/fiultra/T...trafi-crct.jpg That is a special case. Not at all. Rectifiers have spec limits on the value of capacitance that may be applied in different power filters. The only way to get the LC values desired is by using two sections. Many committed builders of SETs use several sections: LCLCLCLC is not uncommon. Don't throw a thrombie over the cost, John. If you're tonedeaf, before you start dancing you should ask what the tune is. This tune has buggerall to do with cost-engineering. If you think I will give up square-rooting the noise left after one LC section merely to save an additional LC section costing around a hundred bucks, you should raise your head from your cheap junkbox amps and smell the rich air of the 21st century. If you put one section in the -ve lead & the other in the +ve lead the common mode voltages if they are present will be stopped. I do it anyway, because I can. Everest; Hillary: "Because it is there." Ultrafi; KISS; Jute: "Screw the cost engineers." But then you need to be aware of other possible problems! List those "possible problems!", John, and then we'll compare them with the known disadvantages (!) of the standard engineering method (cheap single-coil choke on the positive lead) and with the extensive advantages (!) of my screw-the-cost-engineers method of a split choke with one section on the negative lead and one section on the positive lead. When it comes to protecting the rectifier(s) from excessive peak currents as seen in cap input filters a single section choke of at least the critical inductance will do just fine. No need at all for that 2nd winding. You haven't got it, pal. If the rectifier has to live dangerously and be sacrificed occasionally in the service of better sound, I'd gladly do. But in this amp the sonic imperatives dictate topologies that tend to conservative use and protection of the rectifier. The second choke, whatever its effect may be on the security of the rectifier, by intention has buggerall to do with protecting the rectifier. From the blank sheet, this amp was intended, cost no object, to drive noise into the basement. That is what the second choke and cap do. It also contributes to some other desirable features of the amp. Production is not contemplated; it is a DIY amp; cost is immaterial. You might ask what the design objectives are before you start pontificating from the burning bush: "This is The Great P'Eng. Second filter sections are an ostentatious abomination before my face. You will burn in Hell for double-coil chokes, double filter sections, and expensive tubes, because I am The P'Eng of The Cheap Junkbox." Sheesh, man, what did they teach you at engineering school about determining purpose as the first step? Or did you go only to cost-accounting school? JLS Thanks for your thoughts. Let's hear some more about the "possible problems!" of common mode rejection. Andre Jute Visit Jute on Amps at http://members.lycos.co.uk/fiultra/ "wonderfully well written and reasoned information for the tube audio constructor" John Broskie TubeCAD & GlassWare "an unbelievably comprehensive web site containing vital gems of wisdom" Stuart Perry Hi-Fi News & Record Review |
#82
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
Henry Pasternack wrote: "Patrick Turner" wrote in message ... Well, Si diodes and high value cap input filters are better regulated than any tubed choke input filter. Gone is the series resistance of choke and tube. The choke input allowed the tube rectifier to conduct throughout the cycle; there is no peak charging that had limits with tubes. It seems to be a strong concensus among subjective-minded hobbyists that choke-input filters sound better because a lot less RF hash is generated by diode switching. Even Morgan Jones, who's a pretty no-nonsense guy and not at all afraid of silicon, thinks so. The argument makes sense to me. I have built a lot of PS in my time even with 500V at 650mA for my 300 watt amps there is *no* problem with PS hash. The PS generates B+ using Si diodes, hundreds of uF, and a darn voltage doubler like McIntosh began to use all those years ago as soon as they could. if someone cannot prevent rectifier noise getting into the signal they are a power supply novice. I have never used tube rectifiers, and never had switching noise problems. The SS 2 x 300watt amp I built in '96 has two 100,000 uFcaps for the rails and directly charged by 35amp Si, bridge, and hum and noise is less than 0.25mV at the output. I also find that using fast recovery diodes isn't any quieter than plain slow old cheap Si diodes. But I am a careful with PS, prefering to have a mild steel sheet enclosure around the PS to stop magnetic fields and electrostatic fields interfering. I'd like to use a SMPS if I could lay my hands on a suitable design. Halcro do a SMPS in their amps and they sell for $50,000 a pair, and have a good rep, and no noise. This doesn't mean, though, that a given amplifier built with a cap-input power supply will sound bad. There are many counter-examples. Well if the choke is simply able to comply with RL / 1,100 where RL is the Edc / Idc and Idc is considered at 10% of that normally drawn, the choke is probably ok for choke input with a higher V ac HT winding to produce the same B+. There are a couple of concerns. The first is that the choke should have a high enough current rating that it won't saturate with the large AC currents flowing through it in this application. The initial few charge up cycles with an SI rectifier can cause slight saturation; the wave forms are weird, but really the chokes tend to limit inrush more than a cap input. With class B the choke current is always on the move with signal and slight mains voltage fluctuations, but because the swinger choke is designed to convey maximum dc at full power of a class B amp, usually current demands through the choke won't be any higher than for this condition unless someone connects a shorted output or a load well below rated. Manufacturers don't spec their chokes for choke-input service and don't publish detailed design information, so you have to make educated guesses. Well ya gotta know what you buy, or don't buy it; I wind all mine. The second concern is that the choke has to be made well enough that it can withstand the added stress due to magnetostriction and core heating. Hammond, which is the biggest supplier of affordable chokes in North America, specifically states that their chokes are not rated for use in choke-input filters. I don't know exactly what they mean by this. Well, you can work it all out for a given choke. RDH4 has Hanna's method, supposed to be good for filter chokes only with low ac across them. But if the DCR is low enough, Hanna works out OK with some choke inputs where high ac is across the chokeas well as the dc flow. The sum of the Tesla for dc and ac must not exceed about 1.5Tesla. Its the same issue with an SE OPT; you have large ac signal across the L, and dc flow... I have a pair of Hammond 10H @ 300mA, 60 Ohm, chokes that run absolutely cool and quiet as input chokes with a 1000VCT power transformer and 200mA DC load current. Nobody is perfectly competent. I do not post with purpose to point out mainly the incompetence of others, but more to offer an alternative which may or may not appear to be truer than what else may have been said, or to augment what has been said already. I found teachers who came at me to primarily kick my butt of ignorance to be rude. Sometimes there is no choice but to kick the ass of a disruptive student. I should keep my arse in a large steel bucket to prevent bruising eh.... Patrick Turner. -Henry |
#83
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
John Byrns wrote: In article .com, " wrote: Henry: Just ignore Mr. McCoy. He is no technician certainly and his design credentials are well-established. Peter, You are jumping into the middle of a blood feud, how can you expect Henry to ignore Andre? Henry's only interest in this group is attacking Andre. You may want to do some more reading in your psychology textbook. This is similar to the situation that went on for so long between Patrick and Phil A., where Patrick just couldn't ignore Phil's posts. Amazingly however Patrick seems to have been able to ignore Phil recently, although asking that of Henry may be asking too much. Phil has kept out of many threads where he could have been a pest. While that trend continues for awhile at least, I don't need to kick back. Patrick Turner. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#84
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
John Stewart wrote: Henry Pasternack wrote: "John Byrns" wrote in message ... This question is not meant as criticism of your general comments on "swinging chokes". Are you sure that it is true that "The reason swinging chokes are so rare nowadays is that the availability of cheap silicon rectifiers and large electrolytic capacitors makes capacitor-input filters more economical to build than choke- input filters"? My observation is that chokes of any sort, "swinging" or filter, were gone from Hi-Fi amps even before the use of silicon rectifiers became common, and also long before large electrolytic capacitors came on the scene, which appear to have come later than silicon rectifiers. You're the hi-fi historian, so I won't disagree. Most of the commercial hi-fi designs I'm familiar with used a tube rectifier, smallish electrolytic capacitors, and a small smoothing choke in the main B+ supply. Some, like the Marantz 8, used solid-state diodes but still retained the choke. Choke input filters, I think, were always costly. Regardless of the economics of power supply design during the interim period you refer to, in the present day the cost advantage of capacitor-input filters is clear. This ignores any ostensible performance advantages of the choke-input power supply. -Henry To a large degree the swinging choke was found in power supplies used with Class B, PP modulators as seen in the 30's & 40's. They were probably used in public address systems with Class B amplifiers as well. Fidelity was not a concern in those applications. Just lots of audio without too much demand on the power supply. Cheers, John Stewart I agree. Swinger chokes were used to get a lot out of a rectifier tube with with better regulation than a cap input filter. The whole idea of the swinger was to get less B+, variation, ie, better natural regulation, something very desirable in a class B modulator or class C RF amp. Fidelity with a class B amp can be quite fair. Klangfilm built 300 watt theatre amps using a sixpack of EL34; I once rebuilt a 400 watter with 8 EL34, and with pentode mode, I used only 12dB NFB; the guy used it for PA at a hall with 300 ppl, it was the most wondrous warm beautiful sound from "PA" that I had heard; it didn't measure like the best but it did the business ok. The screen voltage was well regged, that was the secret. Patrick Turner. |
#85
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
"Patrick Turner" wrote in message ...
I have built a lot of PS in my time even with 500V at 650mA for my 300 watt amps there is *no* problem with PS hash... if someone cannot prevent rectifier noise getting into the signal they are a power supply novice... I also find that using fast recovery diodes isn't any quieter than plain slow old cheap Si diodes... But I am a careful with PS, prefering to have a mild steel sheet enclosure around the PS to stop magnetic fields and electrostatic fields interfering. I know that diode switching artifacts are real and measurable, and I know there are ways to minimize them. This is even true with a choke input filter. Jones' book shows how to properly snub a filter choke and backs it up with actual scope traces. I think if you have the right measuring equipment it is pretty easy to prove that power supply switching noise makes it to the output of the signal circuitry. The noise is much harder to filter out than it is to prevent in the first place. There is a lot of objective as well as anecdotal evidence that proper use of common-mode chokes and effective bypassing can improve the performance of audio equipment. The results will vary a lot depending on the circuit, the wiring layout, chassis design, and of course the sensitivity of the listener and his state of mind. I think this is an area where you can legitimately probe deeper and deeper and deeper, and at each level open up a new realm of cost and complexity. If you are trying to build a neutrino detector, the cost is probably justified. The question is how good is good enough for consumer audio work? Well ya gotta know what you buy, or don't buy it; I wind all mine. It would be more cost effective for me to pay you to wind chokes for me than for me to wind them myself. But -- oh my god -- I would probably have Michael do the work, if only because he's closer. The second concern is that the choke has to be made well enough that it can withstand the added stress due to magnetostriction and core heating. Hammond, which is the biggest supplier of affordable chokes in North America, specifically states that their chokes are not rated for use in choke-input filters. I don't know exactly what they mean by this. Well, you can work it all out for a given choke. RDH4 has Hanna's method, supposed to be good for filter chokes only with low ac across them. That's fine, but it tells us nothing about the quality of the construction of off-the-shelf parts. If Hammond says their chokes aren't intended for choke-input service, could that mean they are cutting corners on mechanical robustness, or does it just mean that they don't want the hassle of dealing with hobbyists who don't properly derate the chokes for that application? I should keep my arse in a large steel bucket to prevent bruising eh.... If you intend to accuse me of child molestation, yes, you should plan to armor-plate your backside. -Henry |
#86
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
In article , "Henry Pasternack"
wrote: "John Byrns" wrote in message ... This question is not meant as criticism of your general comments on "swinging chokes". Are you sure that it is true that "The reason swinging chokes are so rare nowadays is that the availability of cheap silicon rectifiers and large electrolytic capacitors makes capacitor-input filters more economical to build than choke- input filters"? My observation is that chokes of any sort, "swinging" or filter, were gone from Hi-Fi amps even before the use of silicon rectifiers became common, and also long before large electrolytic capacitors came on the scene, which appear to have come later than silicon rectifiers. You're the hi-fi historian, so I won't disagree. Most of the commercial hi-fi designs I'm familiar with used a tube rectifier, smallish electrolytic capacitors, and a small smoothing choke in the main B+ supply. Some, like the Marantz 8, used solid-state diodes but still retained the choke. Choke input filters, I think, were always costly. Regardless of the economics of power supply design during the interim period you refer to, in the present day the cost advantage of capacitor-input filters is clear. This ignores any ostensible performance advantages of the choke-input power supply. Hi Henry, You are of course correct that some Hi-Fi amplifiers did use filter chokes as part of a capacitor input filter, among them as you point out was the Marantz 8B, and there were also others such McIntosh and some Dynaco models. I should have known this, and actually did know it, as I have owned the Marantz 8B and several Dynaco Stereo 70s. I am not sure how I ended up saying "that chokes of any sort, "swinging" or filter, were gone from Hi-Fi amps even before the use of silicon rectifiers became common", specifically how the word "filter" got in there. Well I sort of know how it got in there, I had originally typed the words "swinging or other" to indicate that the input choke might either be a so called "swinging" choke, or it might simply be an ordinary sort of choke. I remember changing the word "other", because it somehow didn't seem right, but what I might have been thinking to replace it with the word "filter" I haven't a clue. I have noticed before that sometimes when I go back and change a few words in isolation, to some that sound better, it is possible to grossly change the meaning of the whole paragraph unless I remember to go back and proof read the entire paragraph after making the change. Also while I don't believe that the coming of silicon rectifiers had anything to do with the disappearance of choke input filters from domestic Hi-Fi amplifiers, your basic thesis is probably correct in a broad sense. For example Mercury Vapor rectifiers, which were once common in higher power equipment, almost universally used choke input filters, as using capacitor input filters with mercury vapor rectifiers required including special protective circuits which probably weren't any less expensive than the choke they eliminated, and which also would introduce a new failure point reducing reliability. Silicon rectifiers made it possible to use capacitor input filters without special protective circuits. Personally the thing I like about choke input filters is that they reduce the stress on the rectifiers and power transformer, and so should result in a more reliable amplifier. When speaking of "large electrolytic capacitors" it is not clear that we are talking about the same thing, what do you consider to be a "large electrolytic capacitor"? I suspect that what you are referring to as a "large electrolytic capacitor" may be what I would call a "medium sized electrolytic capacitor". Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#87
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
In article .com,
" wrote: That he has made actionable accusations against Mr. Pasternack (and others, apparently) should make him glad that he is some thousands of miles away and has done nothing for which he might be extradited. That he sticks by them despite your attempt to show him otherwise demonstrates his irrationality as well as his lack of skill. It is possible that the reason Andre sticks by his story is because he may be talking about a different pornography incident than the one I know about, so even though the common thread was presumably an attempt to discredit Andre by sending out pornography under his name, the perpetrator(s) could have been different in the two cases. The case I am talking about occurred in late 1998 IIRC, and Andre appears to be talking about an incident that occurred in early 2001. I would also expect that you, as his friend, amanuensis and sycophant would hasten to help him come up with a proper series of designs rather than letting him spring his half-baked bunk (to quote Mr. Stewart) onto an unsuspecting public. Given your expertise, history and experience, that should be an easy thing for you, it would certainly have saved MANY threads of needless vituperation... and given your general self-effacing character, I am sure you would have been happy for Mr. McCoy to actually bask in adulation (for your design) for well-executed designs rather than having vultures picking over the dead corpse of his design(s). Certainly, Mr. McCoy would be happy to take your design whole-cloth... you may wait a considerable time before public acknowledgement, however.=20 I don't think this is a reasonable suggestion because Andre's and my design tastes and styles are completely different, although perhaps both a little offbeat. One of my designs just wouldn't look right under Andre's name, just as one of his designs wouldn't look right under my name. In the instant case I attribute much of the apparent weirdness of Andre's KISS amp to its heritage, that is that it is descended from his ill fated book attempt, and the fact that he has a large box of Lundahl transformers left over from that project that he wants to make use of. While the premise of his book was interesting, I thought that he was overreaching when he first brought it up back in 1998, and IIRC I told him so at the time. The idea of the book as I remember it was to build a series of amplifiers each of higher quality than the previous one, and always recycling the iron from the old amplifier into the new amplifier with the addition of additional transformers and chokes. This idea makes some sense for a series of two or possibly three different amplifiers, but I felt that Andre pushed the idea too far, IIRC to something like a half a dozen different amplifier designs, and the result was that some of them required a few very grotesque design details to be able to use the same iron as the rest of the series. I felt he would have done better to concentrate on maybe three amplifiers at the most, of increasing price and performance. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#88
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
"John Byrns" wrote in message ...
Personally the thing I like about choke input filters is that they reduce the stress on the rectifiers and power transformer, and so should result in a more reliable amplifier. I agree with that. When speaking of "large electrolytic capacitors" it is not clear that we are talking about the same thing, what do you consider to be a "large electrolytic capacitor"? I suspect that what you are referring to as a "large electrolytic capacitor" may be what I would call a "medium sized electrolytic capacitor". I mean "large" in the sense that you don't have to think twice about putting enough capacitance in to give your cap-input power supply stiff DC regulation. I don't mean, for instance, 1000uF or greater. -Henry |
#89
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
John Byrns wrote: I don't think this is a reasonable suggestion because Andre's and my design tastes and styles are completely different, although perhaps both a little offbeat. One of my designs just wouldn't look right under Andre's name, just as one of his designs wouldn't look right under my name. Much Snipped John, with all due respect, that is most likely true. Were Mr. McCoy to put forth a design that actually worked 'out of the box', we all would know instantly that he had nothing to do with it. And at the same time, were you to put out a half-baked bit of semi-deadly and unreliable bunk, we would also know that it was not yours... at least not yours when sober. You are playing your role to the hilt, however, and adding that of 'Apologist'... which is truly beneath you. Let Mr. McCoy manage his explanations. You would not get tarred with the same brush aiming for him.... something that you really do not deserve. Do you actually believe that any sane individual or group would let him design Bicycles!? That they would sell to the public? Did you note his artful implication without actually stating that he had such a commission? That is the bred-to-the-bone pretentious pathology that you are defending. A simple consideration of the necessities of Product Liability would expose his pretense for the lie that it is without further proof being necessary. Peter Wieck Wyncote, PA |
#90
Posted to rec.audio.tubes
|
|||
|
|||
The KISS AMP: square-rooting the noise
Andre Jute wrote: My original long reply seem to have gone missing in the ether. Here's a condensed recreation. John Stewart wrote: Andre Jute wrote: choke input ..... is traditionally built with two sections, meaning two chokes. Not traditionally built in two sections at all. RDH p1192ff. 1954 is traditional enough for me. Better take another look at your reference in RDH4. That is two seperate chokes, their cores in no way connected as they are in the Lundahl chokes. I will post a reference later, probably on the weekend in regard to choke caused problems when the choke is included in the PS return leg. BTW, our company today was part of an industry exhibit to the local customer base. I spent some time with Fred Hammond & some of his staff as well as other industry associates. Cheers, John Stewart |
#91
Posted to rec.audio.tubes
|
|||
|
|||
The KISS AMP: square-rooting the noise
Never let a fact get in the way of a good story...
You should know better, John! Peter Wieck Wyncote, PA |
#92
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
Mark S wrote: "Henry Pasternack" wrote in message ... "Patrick Turner" wrote in message ... Well, Si diodes and high value cap input filters are better regulated than any tubed choke input filter. Gone is the series resistance of choke and tube. The choke input allowed the tube rectifier to conduct throughout the cycle; there is no peak charging that had limits with tubes. It seems to be a strong concensus among subjective-minded hobbyists that choke-input filters sound better because a lot less RF hash is generated by diode switching. Even Morgan Jones, who's a pretty no-nonsense guy and not at all afraid of silicon, thinks so. The argument makes sense to me. This doesn't mean, though, that a given amplifier built with a cap-input power supply will sound bad. There are many counter-examples. Well if the choke is simply able to comply with RL / 1,100 where RL is the Edc / Idc and Idc is considered at 10% of that normally drawn, the choke is probably ok for choke input with a higher V ac HT winding to produce the same B+. There are a couple of concerns. The first is that the choke should have a high enough current rating that it won't saturate with the large AC currents flowing through it in this application. Manufacturers don't spec their chokes for choke-input service and don't publish detailed design information, so you have to make educated guesses. The second concern is that the choke has to be made well enough that it can withstand the added stress due to magnetostriction and core heating. Hammond, which is the biggest supplier of affordable chokes in North America, specifically states that their chokes are not rated for use in choke-input filters. I don't know exactly what they mean by this. I have a pair of Hammond 10H @ 300mA, 60 Ohm, chokes that run absolutely cool and quiet as input chokes with a 1000VCT power transformer and 200mA DC load current. Hi Henry, Interesting experience with the Hammond chokes. I tried their 5H 500 mA choke with a 1250 VCT trans @~250mA years ago in a choke input config and while it was a very stiff supply, the choke buzzed like nuts. I actually called Hammond about this and they said they would have to specially "dip" a choke for this app. 1,250V with CT means you'd have 550Vdc and at 50mA dc RL = 11k, and the L-critical required is 11k/1,100 = 10H, and you have only 5H. When max Idc occurs, RL = 1.1k, and L needn't be so much, so L is allowed to fall a bit, ie, swing, due to dc current, so the swing is fine, methinks you didn't have enough L, therefore causing weird wave forms with a switching content as the choke saturated partially and therefore hummed. Sure the chokes may also have needed another "dip", ie, it needed to be properly impregnated with varnish and your sample is a typical reason why I won;'t let anyone else wind anything for me. Having two chokes in series may have fixed the hum problem. This was about 1992 so things may have changed since then. I know the 1650 outputs went through a redesign since as my 100 w 1650 OPT's are significantly heavier / larger then the current offering. This was also when I learned about critical inductance in a choke input filter. Good thing I did have 1000v of caps in place when I first fired up that supply. Sowter also has a separate series of chokes for choke input filters. 625 Vrms will soar to +881V without a bleeder. A pair of 450V rated electros in series is barely enough rating. Patrick Turner. MarkS Nobody is perfectly competent. I do not post with purpose to point out mainly the incompetence of others, but more to offer an alternative which may or may not appear to be truer than what else may have been said, or to augment what has been said already. I found teachers who came at me to primarily kick my butt of ignorance to be rude. Sometimes there is no choice but to kick the ass of a disruptive student. -Henry |
#93
Posted to rec.audio.tubes
|
|||
|
|||
The KISS AMP: square-rooting the noise
Andre Jute wrote: My original long reply seem to have gone missing in the ether. Here's a condensed recreation. John Stewart wrote: Andre Jute wrote: choke input ..... is traditionally built with two sections, meaning two chokes. Not traditionally built in two sections at all. RDH p1192ff. 1954 is traditional enough for me. Presumably you're now on the right circuit: http://members.lycos.co.uk/fiultra/T...trafi-crct.jpg That is a special case. Not at all. Rectifiers have spec limits on the value of capacitance that may be applied in different power filters. The only way to get the LC values desired is by using two sections. Many committed builders of SETs use several sections: LCLCLCLC is not uncommon. Don't throw a thrombie over the cost, John. If you're tonedeaf, before you start dancing you should ask what the tune is. This tune has buggerall to do with cost-engineering. If you think I will give up square-rooting the noise left after one LC section merely to save an additional LC section costing around a hundred bucks, you should raise your head from your cheap junkbox amps and smell the rich air of the 21st century. I like to supply an OPT with dc at say 400V @ 150mA which has only mV of ripple. Same goes for any SE amp, where lack of ripple is far more important because the ripple is across the OPT and with SET the Ra is lower than RL so nearly all the ripple appears across the OPT primary. I explained in a previous lengthy post where I cited the reactance values needed for good ripple reduction. If you have low C values and a tube rectifier and high dc, chokes are necessary, and preferably CLCLC, or LCLC at least. Quad II got away with having a perfectly lousy power supply with 16uF to anchor down the CT and ripple was 17Vms. While in class A, a PP amp obstructs virtually all ripple current flow from the CT; the ripple appears at both ends of the OPT and same phase, so no ripple flow in the primary. In class AB the common mode rejection stops, and the hum becomes in series with the tube that is conducting. Most guitar amps have lousy filtering of their PS in the same way as Quad II and operate in low class A and almost classB. Musos like the growl in the background of loud notes due to hum. Quad used a 20H choke + 16uF to filter the screen voltage though. It only reduces the 17Vrms hum at the CT to 0.14vrms. 1H and 330uF would be just as good a filter, but because Idc for the screens is low you could get away with 470 ohms and 470uF. Quad II PS caused artifacts are about equal to the THD measurement. With a well done PS the PS artifacts are difficult to measure. The better idea for building good hi-fi amps as one off articles is to always exceed the lowest common denominator of the status quo mass market junk product. Patrick Turner. If you put one section in the -ve lead & the other in the +ve lead the common mode voltages if they are present will be stopped. I do it anyway, because I can. Everest; Hillary: "Because it is there." Ultrafi; KISS; Jute: "Screw the cost engineers." But then you need to be aware of other possible problems! List those "possible problems!", John, and then we'll compare them with the known disadvantages (!) of the standard engineering method (cheap single-coil choke on the positive lead) and with the extensive advantages (!) of my screw-the-cost-engineers method of a split choke with one section on the negative lead and one section on the positive lead. Good luck with dual winding chokes. I don't see the need for myself; my amps are complex and difficult to make already and I don't like adding to all that. I often just have CLC with the choke in the ground so there is no HV from winding to earthy core. Noise is never a problem in anything I build. When it comes to protecting the rectifier(s) from excessive peak currents as seen in cap input filters a single section choke of at least the critical inductance will do just fine. No need at all for that 2nd winding. You haven't got it, pal. If the rectifier has to live dangerously and be sacrificed occasionally in the service of better sound, I'd gladly do. But in this amp the sonic imperatives dictate topologies that tend to conservative use and protection of the rectifier. The second choke, whatever its effect may be on the security of the rectifier, by intention has buggerall to do with protecting the rectifier. From the blank sheet, this amp was intended, cost no object, to drive noise into the basement. That is what the second choke and cap do. It also contributes to some other desirable features of the amp. Production is not contemplated; it is a DIY amp; cost is immaterial. You might ask what the design objectives are before you start pontificating from the burning bush: "This is The Great P'Eng. Second filter sections are an ostentatious abomination before my face. You will burn in Hell for double-coil chokes, double filter sections, and expensive tubes, because I am The P'Eng of The Cheap Junkbox." Sheesh, man, what did they teach you at engineering school about determining purpose as the first step? Or did you go only to cost-accounting school? JLS Thanks for your thoughts. Let's hear some more about the "possible problems!" of common mode rejection. Andre Jute Visit Jute on Amps at http://members.lycos.co.uk/fiultra/ "wonderfully well written and reasoned information for the tube audio constructor" John Broskie TubeCAD & GlassWare "an unbelievably comprehensive web site containing vital gems of wisdom" Stuart Perry Hi-Fi News & Record Review |
#94
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
Henry Pasternack wrote: "Patrick Turner" wrote in message ... I have built a lot of PS in my time even with 500V at 650mA for my 300 watt amps there is *no* problem with PS hash... if someone cannot prevent rectifier noise getting into the signal they are a power supply novice... I also find that using fast recovery diodes isn't any quieter than plain slow old cheap Si diodes... But I am a careful with PS, prefering to have a mild steel sheet enclosure around the PS to stop magnetic fields and electrostatic fields interfering. I know that diode switching artifacts are real and measurable, and I know there are ways to minimize them. This is even true with a choke input filter. Jones' book shows how to properly snub a filter choke and backs it up with actual scope traces. Its all too easy to have an input lead or a feedback lead close to a mains or other power supply lead which has switching noise within it. I think if you have the right measuring equipment it is pretty easy to prove that power supply switching noise makes it to the output of the signal circuitry. The CRO can see it. If I set it for its mosrt sensitve range, and hums and switch spikes are less than 0.3mV, I am usually happy. The noise is much harder to filter out than it is to prevent in the first place. There is a lot of objective as well as anecdotal evidence that proper use of common-mode chokes and effective bypassing can improve the performance of audio equipment. common mode chokes are used to effect in IEC shielded/filered mains input sockets but I have not used any in a CLC supply and I don't see the need. There's no free lunch in a common mode choke in a LF CLC. Nothing much in RDH4 about them. The results will vary a lot depending on the circuit, the wiring layout, chassis design, and of course the sensitivity of the listener and his state of mind. I think this is an area where you can legitimately probe deeper and deeper and deeper, and at each level open up a new realm of cost and complexity. If you are trying to build a neutrino detector, the cost is probably justified. The question is how good is good enough for consumer audio work? Quad II got away with murder in their PS. I make mine with 60 dB less noise at the CT. This does not mean that listeners will gain a 60dB improvement in listeing pleasure. Just what does give such an improvement? What makes something 1,000 times better? But making a PS 1,000 times quieter than Quad II places thatv PS beyond blame, into the land of angelic power supplies, ie, like HV batteries. Angelic power supplies can be built at low cost. ( Wings are the most expensive parts, since not many spare feathers are available ). Well ya gotta know what you buy, or don't buy it; I wind all mine. It would be more cost effective for me to pay you to wind chokes for me than for me to wind them myself. No, I charge like then light brigade. Buy them from Hammond, they have costs because of mass production far cheaper than mine, and just use CLC. But -- oh my god -- I would probably have Michael do the work, if only because he's closer. Arch Angel Micheal should be able to provide angelic power supply components. Pray harder. The second concern is that the choke has to be made well enough that it can withstand the added stress due to magnetostriction and core heating. Hammond, which is the biggest supplier of affordable chokes in North America, specifically states that their chokes are not rated for use in choke-input filters. I don't know exactly what they mean by this. Well, you can work it all out for a given choke. RDH4 has Hanna's method, supposed to be good for filter chokes only with low ac across them. That's fine, but it tells us nothing about the quality of the construction of off-the-shelf parts. If Hammond says their chokes aren't intended for choke-input service, could that mean they are cutting corners on mechanical robustness, or does it just mean that they don't want the hassle of dealing with hobbyists who don't properly derate the chokes for that application? To really know, ya gotta buy a choke; they are dirt cheap, and a choke is cheaper than couple of big Macs with the lot. Big value caps are even cheaper. I should keep my arse in a large steel bucket to prevent bruising eh.... If you intend to accuse me of child molestation, yes, you should plan to armor-plate your backside. I have never accused anyone of such nefariousness. What folks do with other folks organs isn't my business. I'd even discuss tubes with George Bush if he was interested, and he's ****ing a whole country in the middle east. All I can say to George is that there is no accounting for taste. Patrick Turner. -Henry |
#95
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
John Byrns wrote: In article , "Henry Pasternack" wrote: "John Byrns" wrote in message ... This question is not meant as criticism of your general comments on "swinging chokes". Are you sure that it is true that "The reason swinging chokes are so rare nowadays is that the availability of cheap silicon rectifiers and large electrolytic capacitors makes capacitor-input filters more economical to build than choke- input filters"? My observation is that chokes of any sort, "swinging" or filter, were gone from Hi-Fi amps even before the use of silicon rectifiers became common, and also long before large electrolytic capacitors came on the scene, which appear to have come later than silicon rectifiers. You're the hi-fi historian, so I won't disagree. Most of the commercial hi-fi designs I'm familiar with used a tube rectifier, smallish electrolytic capacitors, and a small smoothing choke in the main B+ supply. Some, like the Marantz 8, used solid-state diodes but still retained the choke. Choke input filters, I think, were always costly. Regardless of the economics of power supply design during the interim period you refer to, in the present day the cost advantage of capacitor-input filters is clear. This ignores any ostensible performance advantages of the choke-input power supply. Hi Henry, You are of course correct that some Hi-Fi amplifiers did use filter chokes as part of a capacitor input filter, among them as you point out was the Marantz 8B, and there were also others such McIntosh and some Dynaco models. I should have known this, and actually did know it, as I have owned the Marantz 8B and several Dynaco Stereo 70s. I am not sure how I ended up saying "that chokes of any sort, "swinging" or filter, were gone from Hi-Fi amps even before the use of silicon rectifiers became common", specifically how the word "filter" got in there. Well I sort of know how it got in there, I had originally typed the words "swinging or other" to indicate that the input choke might either be a so called "swinging" choke, or it might simply be an ordinary sort of choke. I remember changing the word "other", because it somehow didn't seem right, but what I might have been thinking to replace it with the word "filter" I haven't a clue. I have noticed before that sometimes when I go back and change a few words in isolation, to some that sound better, it is possible to grossly change the meaning of the whole paragraph unless I remember to go back and proof read the entire paragraph after making the change. Also while I don't believe that the coming of silicon rectifiers had anything to do with the disappearance of choke input filters from domestic Hi-Fi amplifiers, your basic thesis is probably correct in a broad sense. For example Mercury Vapor rectifiers, which were once common in higher power equipment, almost universally used choke input filters, as using capacitor input filters with mercury vapor rectifiers required including special protective circuits which probably weren't any less expensive than the choke they eliminated, and which also would introduce a new failure point reducing reliability. Silicon rectifiers made it possible to use capacitor input filters without special protective circuits. Personally the thing I like about choke input filters is that they reduce the stress on the rectifiers and power transformer, and so should result in a more reliable amplifier. When speaking of "large electrolytic capacitors" it is not clear that we are talking about the same thing, what do you consider to be a "large electrolytic capacitor"? I suspect that what you are referring to as a "large electrolytic capacitor" may be what I would call a "medium sized electrolytic capacitor". You are right about all that John. Cap inputs are rare in industrial apps. They still are. A "large electro" is 470uF/450v rated or more as opposed to the puny values of 16uF used in the 1950s. I put 1,500uF/450V electros into a pair of 100 watt mono ampos I made, 150mm tall, 75mm dia, screw terminals, probably 20A ripple rating. They were taken out of little used university research equipment that was junked in the 1980s, so cost me only $25 each. Not much stinking ripple in those amps. Patrick Turner. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#96
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
On Wed, 5 Apr 2006 16:29:46 -0400, "Henry Pasternack"
wrote: I'd be interested in knowing what an RF vector network analyzer *was*. Been too far outa the loop lately, obviously. A device that that allows you to make very accurate complex impedance and transmission measurements (magnitude and phase) of RF networks over a wide range of frequencies. Info he http://www.n2pk.com A Smith chart generated by DSP. Now I've seen everything and can pass easily. Very cool, indeed; thanks. Makes me think I should do it with transistors and be done with it. Oh, sorry; you're using vacuum valves? Don't know why I'd assumed you'd gone over to the Dark Side (like I have, because I so often need to drive an A/D converter, and because I need to include gain trim controls and metering). Another design that is 180 degrees opposite from your approach, yet still very interesting, is the Pass Pearl phono preamp. Single- ended, zero feedback FETs. It's supposed to be pretty good. Doubtless true, but it's ag'in my religious beliefs. In the pure, no-extreme-loading world of vacuum-valves- only, you might be interested in a topology like Dick Burwen's from 1948, two triodes cascaded with the first stage run wide open and the second with EQ in the "anode follower" feedback. Still haven't found better, IMO, for use into very light loading. Thanks, as always, Chris Hornbeck |
#97
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
On Thu, 06 Apr 2006 06:57:21 +0100, Stewart Pinkerton
wrote: Here's an alternative, using *lots* of tiny transistors! :-) http://www.lurcher.org/ukra/stewart_p/stewart_p.html A very interesting topology, instrumentation input stage run flat, feeding 75 muS passive rolloff; then the lower frequency corners in the second stage's feedback. But why the third stage? Is the DC servo easier to implement? My only really cogent comment is: Great Photography! Really nice light. And don't blame the camera; if you get beautiful light, it's the photographer's fault. Thanks, as always, Chris Hornbeck |
#98
Posted to rec.audio.tubes
|
|||
|
|||
Swinging chokes
"Patrick Turner" wrote in message ... Mark S wrote: "Henry Pasternack" wrote in message ... "Patrick Turner" wrote in message ... Well, Si diodes and high value cap input filters are better regulated than any tubed choke input filter. Gone is the series resistance of choke and tube. The choke input allowed the tube rectifier to conduct throughout the cycle; there is no peak charging that had limits with tubes. It seems to be a strong concensus among subjective-minded hobbyists that choke-input filters sound better because a lot less RF hash is generated by diode switching. Even Morgan Jones, who's a pretty no-nonsense guy and not at all afraid of silicon, thinks so. The argument makes sense to me. This doesn't mean, though, that a given amplifier built with a cap-input power supply will sound bad. There are many counter-examples. Well if the choke is simply able to comply with RL / 1,100 where RL is the Edc / Idc and Idc is considered at 10% of that normally drawn, the choke is probably ok for choke input with a higher V ac HT winding to produce the same B+. There are a couple of concerns. The first is that the choke should have a high enough current rating that it won't saturate with the large AC currents flowing through it in this application. Manufacturers don't spec their chokes for choke-input service and don't publish detailed design information, so you have to make educated guesses. The second concern is that the choke has to be made well enough that it can withstand the added stress due to magnetostriction and core heating. Hammond, which is the biggest supplier of affordable chokes in North America, specifically states that their chokes are not rated for use in choke-input filters. I don't know exactly what they mean by this. I have a pair of Hammond 10H @ 300mA, 60 Ohm, chokes that run absolutely cool and quiet as input chokes with a 1000VCT power transformer and 200mA DC load current. Hi Henry, Interesting experience with the Hammond chokes. I tried their 5H 500 mA choke with a 1250 VCT trans @~250mA years ago in a choke input config and while it was a very stiff supply, the choke buzzed like nuts. I actually called Hammond about this and they said they would have to specially "dip" a choke for this app. 1,250V with CT means you'd have 550Vdc and at 50mA dc RL = 11k, and the L-critical required is 11k/1,100 = 10H, and you have only 5H. When max Idc occurs, RL = 1.1k, and L needn't be so much, so L is allowed to fall a bit, ie, swing, due to dc current, so the swing is fine, methinks you didn't have enough L, therefore causing weird wave forms with a switching content as the choke saturated partially and therefore hummed. Sure the chokes may also have needed another "dip", ie, it needed to be properly impregnated with varnish and your sample is a typical reason why I won;'t let anyone else wind anything for me. Having two chokes in series may have fixed the hum problem. Hi Patrick, Maybe the 10 H choke might have been a better choice but it was the 27 ohm DCR of this choke that caught my eye. Seems as though there are quite a few NOS potted chokes showing up on ebay these days. I replicated this supply with a potted UTC choke and that choke was dead silent.althought the DCR was a little higher, 50 ohms. This was about 1992 so things may have changed since then. I know the 1650 outputs went through a redesign since as my 100 w 1650 OPT's are significantly heavier / larger then the current offering. This was also when I learned about critical inductance in a choke input filter. Good thing I did have 1000v of caps in place when I first fired up that supply. Sowter also has a separate series of chokes for choke input filters. 625 Vrms will soar to +881V without a bleeder. A pair of 450V rated electros in series is barely enough rating. I had a pair of 500's in place. I determined the bleed resistor emperically as the critical inductance constants are a little conservative. 10K would knock the start voltage down to around 600v until the output tubes warmed up. I think I am going to revisit this type supply soon with the UTC choke and damper diodes for rectifiers. I'll post the results but I can't see getting to it til the fall. Best Regards, MarkS Patrick Turner. MarkS Nobody is perfectly competent. I do not post with purpose to point out mainly the incompetence of others, but more to offer an alternative which may or may not appear to be truer than what else may have been said, or to augment what has been said already. I found teachers who came at me to primarily kick my butt of ignorance to be rude. Sometimes there is no choice but to kick the ass of a disruptive student. -Henry |
#99
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
On Fri, 07 Apr 2006 05:21:09 GMT, Chris Hornbeck
wrote: On Thu, 06 Apr 2006 06:57:21 +0100, Stewart Pinkerton wrote: Here's an alternative, using *lots* of tiny transistors! :-) http://www.lurcher.org/ukra/stewart_p/stewart_p.html A very interesting topology, instrumentation input stage run flat, feeding 75 muS passive rolloff; then the lower frequency corners in the second stage's feedback. But why the third stage? Is the DC servo easier to implement? The third stage is simply there to ensure that the 20Hz IEC rolloff formed by C10-14 and R15 is not deranged by a low-impedance load, and hence to provide a controlled and very low output impedance. The phono stage sits within a foot of the arm base, so might have to drive quite a long interconnect. The passive treble rolloff avoids any possibility of scratch overload causing momentary latchup, and maintains the correct rolloff even after the opamp would level it off to unity gain in an active configuration. The nice thing about the input stage is of course that it's differential, so it rejects any common-mode hum and hash the cartridge and the arm leads might pick up. It's what Andre would call a very 'silent' amplifier. My only really cogent comment is: Great Photography! Really nice light. And don't blame the camera; if you get beautiful light, it's the photographer's fault. Thanks, but it's just the available daylight in the room, which has a pair of 6-foot windows spaced along each of the 25 foot long walls. Taken with a Sony DSC-S85, a great little camera now replaced by an absolutely superb Sony DSC-R1. I've done A2 sized prints off this new beauty and they are clean and crisp edge to edge, sharp as a tack with 200dpi resolution. I've even taken it up to A0 just for fun, and from a 3-foot viewing distance they appear flawless. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#100
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
"Patrick Turner" wrote in message ...
I'll send you the schematic of the amp, and if you like the look of it, I can send you the other 4 sheets. In each channel it uses one 2SK369, 1 x 6DJ8, 1 x 12AT7, 1 x 6CG7. These are paralleled so each tube socket has just one triode each. Thank you for sending me the schematic. It seems to be a defensible design. For my purposes, I would want to get a selected input FET with Idss suitable for running with the source resistor connected to ground. This is the way Allen Wright does it, and it avoids the need for the -6V bias supply. I'm also surprised you doubled up the tubes on the SRPP stage, but I guess you wanted to have a very robust output drive capability. The same for the 6DJ8, but if you hadn't paralleled it you'd have had an extra section lying around. I assume you don't believe in sharing tube halves between channels in a purist design. I still haven't decided what phono circuit to build, but I'm getting closer. One that intrigues me is this: http://www.pacifier.com/~gpimm/phono.htm The 6GK5 and 6ER5 are really hot tubes, but not very linear for large signals, I think. I received a new pulley for my turntable yesterday, to convert it from string drive to 1/2" tape drive. If I can find suitable splicing tape, this evening I will sacrifice one of my gay pornography videos to make a new drive belt. Hopefully there will be no more speed variations and that "hump, hump", I mean "thump, thump", every time the knot goes around will be gone. -Henry |
#101
Posted to rec.audio.tubes
|
|||
|
|||
ZFNB, Zweirdness in Memory Lane
John Byrns wrote: I attribute much of the apparent weirdness of Andre's KISS amp There's no weirdness to my KISS amps (plural, there are two of them); they are exceedingly conservative (except for a few details to trip up the clowns, of course!). Those two nice clean designs are he http://members.lycos.co.uk/fiultra/T...trafi-crct.jpg http://members.lycos.co.uk/fiultra/T44bis-'Populaire'-crct.jpg and are fully described he http://members.lycos.co.uk/fiultra/T...mp%20INDEX.htm Any "apparent weirdness" is the result of: a) the resident scum resolutely looking at the wrong circuit and screeching about it and b) a few noise-reducing measures, like split-wound chokes on the same cores used for common mode noise reduction, beyond the understanding of the cost-engineers. to its heritage, that is that it is descended from his ill fated book attempt, Oh, that baby was not only born but is a lusty, screaming infant. You may remember that the sociopath Michael LeFevre of Magnequest publicly threatened that if I published the book he would organize a campaign of denigration. I could not under those circumstances take the publisher's money, so I returned the advance and found other ways to take the money out of the skin of LeFevre and his gang of bullyboys. (I made a six-figure sum out of the self-immolation of Frank Deutschmann, for instance. In your face, scum, and in my bank account! LOL.) But the book isn't dead. It merely morphed into my KISS netsite, for which the Magnequest scum has paid far, far more handsomely than the original publisher's advance. and the fact that he has a large box of Lundahl transformers left over from that project I wish. No, those Lundahls have long since walked out of the door and been replaced, several times over. The truth is that the Lundahls are simply the most suitable at any price for my KISS amps; I don't see the point of paying more than Lundahls cost for transformers which are usually no better and very often inferior. There are no dropper resistors in the right circuits. As I said, the screeching scum whining about ballasts were (probably deliberately) looking at the wrong schematic, for a proofing knockup intended to last only a couple of days to prove the driver stage of the T39. (It was kept after that as a potato amp because it sounded so sweet... The circuit is here http://members.lycos.co.uk/fiultra/t...17acircuit.jpg for builders of my T39 KISS amp who want temporarily to experiment with a potato amp on the fullblown KISS experience.) While the premise of his book was interesting, I thought that he was overreaching when he first brought it up back in 1998, and IIRC I told him so at the time. I should have listened... Oh, well, like Edith Piaf sang, No time for regrets now. The idea of the book as I remember it was to build a series of amplifiers each of higher quality than the previous one, and always recycling the iron from the old amplifier into the new amplifier with the addition of additional transformers and chokes. This idea makes some sense for a series of two or possibly three different amplifiers, but I felt that Andre pushed the idea too far, IIRC to something like a half a dozen different amplifier designs, and the result was that some of them required a few very grotesque design details to be able to use the same iron as the rest of the series. I felt he would have done better to concentrate on maybe three amplifiers at the most, of increasing price and performance. Nah, the most interesting amps in my Modular series were the ones that were also most controversial. I fondly remember the absolute base amp which used photoflash caps, and which no one except me built because everyone else thought that if you were spending for 300B anyway photoflash caps were a whole mile of cheapies too far. And the one with the two power transformers operating each as a halfwave circuit to output fullwave rectified DC beyond the two tube rectifiers, while it is true that the usual scum screeched for months about it on RAT, earned me a lot of interest off-list from real engineers who wanted to know me -- precisely because I had the brains and the balls to resurrect that forgotten topology. Anyway, the second-cheapest amp in that Modular Series 300B, the T44, which you picked as your fave because of the silicon rectifier, was the one most people built, to my distress, because I thought then and still that a tube rectifier is A Good Idea. So I took a hint and morphed it into a stereo amp as the other KISS amp, the T44bis. An intelligent designer -- and intelligent artist, an intelligent human for that matter -- wastes nothing. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ I enjoyed that trip down memory lane. ZNFB. Zweirdness. Zregrets. Andre Jute Visit Jute on Amps at http://members.lycos.co.uk/fiultra/ "wonderfully well written and reasoned information for the tube audio constructor" John Broskie TubeCAD & GlassWare "an unbelievably comprehensive web site containing vital gems of wisdom" Stuart Perry Hi-Fi News & Record Review |
#102
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
Henry Pasternack wrote: "Patrick Turner" wrote in message ... I'll send you the schematic of the amp, and if you like the look of it, I can send you the other 4 sheets. In each channel it uses one 2SK369, 1 x 6DJ8, 1 x 12AT7, 1 x 6CG7. These are paralleled so each tube socket has just one triode each. Thank you for sending me the schematic. It seems to be a defensible design. For my purposes, I would want to get a selected input FET with Idss suitable for running with the source resistor connected to ground. This is the way Allen Wright does it, and it avoids the need for the -6V bias supply. The Allen Wright FVP just has the gate biased by 47k to 0V and an 18 ohm source R to 0V for source biasing just like having a cathode R to bias tube. Other versions of his amps have a CCS tail for a j-fet LTP which I tried but it was noiser than the SE version I sent you. The -6V supply is no problem in itself with the network of resistors which can be varied with a link to set the gain so that in the highest gain setting there is effectively 30 ohms from the source, R6&R7 with a link to R3, 10 ohms to 0V. The original FVP had no adjustment for gain. If you have a differential input pair of fets with CCS tail, there can be two source resistors and there is more scope for varying gain. But that's two more fets than I have.... I'm also surprised you doubled up the tubes on the SRPP stage, but I guess you wanted to have a very robust output drive capability. The same for the 6DJ8, but if you hadn't paralleled it you'd have had an extra section lying around. I assume you don't believe in sharing tube halves between channels in a purist design. I'd rather have only one function per bottle. In my own pre which is similar to the Rocket, I have a 6EJ7 in triode as the first triode; but almost anything will do, and you could have just one section of a 6DJ8 if you wanted to since its really unlikely there would be much cross talk betwen channels, so the tube count could be halved at the expense of slight dynamic range ability. I still haven't decided what phono circuit to build, but I'm getting closer. One that intrigues me is this: http://www.pacifier.com/~gpimm/phono.htm The 6GK5 and 6ER5 are really hot tubes, but not very linear for large signals, I think. 6GK5 has µ = 78, Ra = 5.4k, gm = 15mA/V all at Ia = 11.5mA. At 2.5 mA for ia as shown in the above schema the gm would be maybe only 1/3 but still OK, but about equal maybe to a parallel 6dJ8 at a total of 5mA. The above circuit is a little confusing since the load for the two triodes are twin circles of a CCS symbol but with output coming from a line between the circle which is normally the way a controlled CCS is indicated. So we are left to assume the guy really has the anode outputs from the anode and there are conventional CCS dc feeds to the triodes. Nothing wrong with CCS dc supplies with high ac impedance. There is 475 ohm unbypassed Rk for the 6GK5, so ra effective is about 10k plus 78 x 475 = 47k total. I doubt the guy factored in the 47k or actual Rout at the anode into his RIAA network because effectively he has 47k of Ra' + R4,22k in series with the C1, 0.1 & R5, 3k2 for the 3180 and 318 uS time constants. Well, 47k + 22k = 69k, so with 0.1uF, you have 6,900 uS, not 3,180 uS...l In my circuit the Ra' is extremely high for the cascode 6DJ8 triode anodes because the fet below it has a high Rd which is the cathode resistance for the triode, increasing Ra by µ x Rd, so the anode Ra is virtually a CCS. The Rout from the 6dJ8 is set by the 22k load resistance. I received a new pulley for my turntable yesterday, to convert it from string drive to 1/2" tape drive. If I can find suitable splicing tape, this evening I will sacrifice one of my gay pornography videos to make a new drive belt. Hopefully there will be no more speed variations and that "hump, hump", I mean "thump, thump", every time the knot goes around will be gone. Its sounding all a bit primitive there Henry. If that fails to work, hold the stylus carefully and run around the record on a table at 33.3RPM. One guy here mounted his house on an old merry-go-round with record on a shaft up through the centre to sit the record on. His wife got dizzy... Patrick Turner. -Henry |
#103
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
I still haven't decided what phono circuit to build, but I'm getting closer. One that intrigues me is this: http://www.pacifier.com/~gpimm/phono.htm The 6GK5 and 6ER5 are really hot tubes, but not very linear for large signals, I think. Oh, and BTW, phono signals are tiny, so linearity with *any* triode will be OK at less than 0.5Vrms output. Patrick Turner. |
#104
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
"Patrick Turner" wrote in message ...
The Allen Wright FVP just has the gate biased by 47k to 0V and an 18 ohm source R to 0V for source biasing just like having a cathode R to bias tube. Other versions of his amps have a CCS tail for a j-fet LTP which I tried but it was noiser than the SE version I sent you. It should be 3dB noisier due to the two tubes' noise adding (but not 6dB, evidently). Maybe Andre can explain why. The -6V supply is no problem in itself with the network of resistors which can be varied with a link to set the gain... The original FVP had no adjustment for gain. If you have a differential input pair of fets with CCS tail, there can be two source resistors and there is more scope for varying gain. But that's two more fets than I have.... There are several theoretical advantages to the differential design. Allen swears it's the bee's knees. The above circuit is a little confusing since the load for the two triodes are twin circles of a CCS symbol but with output coming from a line between the circle which is normally the way a controlled CCS is indicated. You have to read the documentation for the current sources. The middle connection is what he calls the "mu" output. Basically, it turns the tube and the current source into a hybrid SRPP. Around 400 Ohms output impedance. Supposed to be sonically transparent, and these guys are *very* particular. I doubt the guy factored in the 47k or actual Rout at the anode into his RIAA network... He's smarter than you think. I received a new pulley for my turntable yesterday... Its sounding all a bit primitive there Henry. I whipped a belt together quickly. It's not as nice as I'd like, but there's time to learn to make them better. There's no noise when the splice goes over the pulley. Here's a picture. http://users.rcn.com/hpasternack/teres/belt.jpg -Henry |
#105
Posted to rec.audio.tubes
|
|||
|
|||
The KISS AMP: square-rooting the noise
Go over to alt.binaries.pictures.radio to read of a problem arising when a choke input
filter is connected in the return leg of a rectifier circuit. The page is taken from the Electrical Engineering Series, John D Ryder, Author. Cheers, John Stewart |
#106
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
Henry Pasternack wrote: "Patrick Turner" wrote in message ... The Allen Wright FVP just has the gate biased by 47k to 0V and an 18 ohm source R to 0V for source biasing just like having a cathode R to bias tube. Other versions of his amps have a CCS tail for a j-fet LTP which I tried but it was noiser than the SE version I sent you. It should be 3dB noisier due to the two tubes' noise adding (but not 6dB, evidently). Maybe Andre can explain why. The noise problem wasn't the fet noise; it was mainly hum, but after examining all the figures and taking into account the fact I only wanted an unbalanced output there was no point in having balanced input and following balanced gain stage with all the extra bits. Simplicity won, complexity lost, music gained.... The -6V supply is no problem in itself with the network of resistors which can be varied with a link to set the gain... The original FVP had no adjustment for gain. If you have a differential input pair of fets with CCS tail, there can be two source resistors and there is more scope for varying gain. But that's two more fets than I have.... There are several theoretical advantages to the differential design. Allen swears it's the bee's knees. I know, he would, most folks do, but I got excellent low noise with the circuit i gave you although it is hum prone near magnetic fields when using MC carts. If you have 0.4mV at 1kHz to make the full output, low level hum signals can cause problems because bass signal sensitivity is higher. The above circuit is a little confusing since the load for the two triodes are twin circles of a CCS symbol but with output coming from a line between the circle which is normally the way a controlled CCS is indicated. You have to read the documentation for the current sources. The middle connection is what he calls the "mu" output. Basically, it turns the tube and the current source into a hybrid SRPP. Around 400 Ohms output impedance. Supposed to be sonically transparent, and these guys are *very* particular. Yes you are right, the there is a mu follower output which has low Rout see http://www.pacifier.com/~gpimm/Activ...t_control.html I'd prefer a fully tubed µ follower... A tubed µ follower with 6DJ8 can have a low Rout, and the bottom gain tube sees a high RL. A fet can be still used in cascode with the bottom gain triode. The overall gain of the fet plus gain tube + load in the anode circuit of the tube is simply fet gm x tube RL, so in my circuit gain without the source resistors under the fet the gain = gm x RL = 0.04 x 22k = 880. Now if the 22k RL was replaced with a bootstrapped follower with say 10k between bottom triode anode and top triode cathode then the load the gain triode sees is about 300k, so the overall gain would be huge, maybe 12,000, and way too much and unmanagable, and most likely to oscillate at RF badly. I doubt the guy factored in the 47k or actual Rout at the anode into his RIAA network... He's smarter than you think. I missed the µ follower bit. His tube schematic does not show exactly what is there. With Rout very low, the 0.1 and 22k give 2,200 uS ??? I received a new pulley for my turntable yesterday... Its sounding all a bit primitive there Henry. I whipped a belt together quickly. It's not as nice as I'd like, but there's time to learn to make them better. There's no noise when the splice goes over the pulley. Here's a picture. http://users.rcn.com/hpasternack/teres/belt.jpg -Henry Splice over pulley noise shouldn't be audible because the inertia of the rotating masses would limit any miniscule platter speed changes. Maybe the tape won't last forever. Duzzen madder, just rejoin another bit of tape. But if there's porno on the tapes maybe the music will sound all hot and sweaty... Patrick Turner. |
#107
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
"Patrick Turner" wrote in message ...
I know, he would, most folks do, but I got excellent low noise with the circuit i gave you although it is hum prone near magnetic fields when using MC carts. If you have 0.4mV at 1kHz to make the full output, low level hum signals can cause problems because bass signal sensitivity is higher. My tonearn is unsuitable for MC cartridges, so for better or worse I don't have to solve that problem. I'm really bummed that Shure has canceled production of the V15, though. Hopefully they will come out with a replacement model one of these days. I'd prefer a fully tubed µ follower... Gary's current source has extremely high impedance and low noise. It's a nice example of a design that's been optimized over successive iterations. A lot of people are reporting very good results using these things. Now if the 22k RL was replaced with a bootstrapped follower with say 10k between bottom triode anode and top triode cathode then the load the gain triode sees is about 300k, so the overall gain would be huge, maybe 12,000, and way too much and unmanagable, and most likely to oscillate at RF badly. Obviously you don't want to load a cascode with a current source... With Rout very low, the 0.1 and 22k give 2,200 uS I agree the time constant looks to be off by 50%. Maybe I'll ask him about it. -Henry |
#108
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
On Fri, 7 Apr 2006 12:49:12 -0400, "Henry Pasternack"
wrote: http://www.pacifier.com/~gpimm/phono.htm Note that C5 and the output transformer are series resonant at about 50 Hz. Good fortune, Chris Hornbeck |
#109
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
On Fri, 07 Apr 2006 23:28:59 GMT, Patrick Turner
wrote: Oh, and BTW, phono signals are tiny, so linearity with *any* triode will be OK at less than 0.5Vrms output. I strongly disagree. FWIW. Good fortune, Chris Hornbeck |
#110
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
"Patrick Turner" wrote in message ...
I missed the µ follower bit. His tube schematic does not show exactly what is there. With Rout very low, the 0.1 and 22k give 2,200 uS ??? There's an RIAA calculator he http://www.kabusa.com/frameset.htm?/phonpre.htm It confirms the values in the Pimm schematic. Supposedly, the calculator is based on math by Stanley Lip****z. I haven't had a chance to check the calculations, but I suspect it's correct. I guess there's interaction between the sections that accounts for the seeming discrepancy, although it isn't intuitively obvious to me why right at this moment. Gotta get dressed and take the kids to piano lessons. -Henry |
#111
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
"Henry Pasternack" wrote in message ... "Patrick Turner" wrote in message ... I guess there's interaction between the sections that accounts for the seeming discrepancy, although it isn't intuitively obvious to me why right at this moment. The answer is you need to put the two capacitors in parallel when figuring the 50Hz cutoff. -Henry |
#112
Posted to rec.audio.tubes
|
|||
|
|||
The KISS AMP: square-rooting the noise
This particular thread is now so contaminated with irrelevant noise,
some of it from transistors and child molesters, that I have opened a new thread to discuss John Stewart's contribution. It is called "The KISS AMP: square-rooting the noise: the real work starts" -- AJ John Stewart wrote: Go over to alt.binaries.pictures.radio to read of a problem arising when a choke input filter is connected in the return leg of a rectifier circuit. The page is taken from the Electrical Engineering Series, John D Ryder, Author. Cheers, John Stewart |
#113
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
Chris Hornbeck wrote: On Fri, 07 Apr 2006 23:28:59 GMT, Patrick Turner wrote: Oh, and BTW, phono signals are tiny, so linearity with *any* triode will be OK at less than 0.5Vrms output. I strongly disagree. FWIW. Good fortune, Chris Hornbeck Hang on a second please. If you had a fairly non linear triode like a 6J6, 6GK5 whatever, most are capable of less than 1% at 10Vrms, resistance loaded. Expect 5% at 50Vrms output. So at an output of 100mV, expect THD = 0.01%. Isn't that linear enough? Now many relatively non linear triodes such as the high gm RF triodes are much more linear when set up with a CCS load to buffer them from any R loading effect. The steeper the RL slope of the RL loadline across the curves, the worse the distortion. If the load line is virtually horizontal, distorion is often only 0.1% for 10vrms output. i get that with 12AT7 in µ-follower. So at 0.1Vrms, thd is rather low! The RCA book curves for 6GK5 show rather non linear curve spacing at 2.5mA of anode current but its over a possible 200V p-p V swing. Over 0.1Vrms, the distortion is the least of our concerns. I have not seen 6ER8 curves but even if as poor as 6GK5, I suspect thd 0.1% at 500mV output. I'd never lumber a customer with tubes not being produced now except perhaps the occasional13E1, where I have sold a spare to the customer and keep a few more myself. Those guys will die before i do probably, and the circuit is alterable to take multiples of 6L6 if the 13E1 is no longer avaliable. I have dozens of 6EJ7. It is a fabulous triode. I don't believe all the curves people have drawn for data curves over the last 60 years. Many are quite ficticious. In a recent test of GE6550A in triode I got 1.5% thd with 235Vrms at the anode with a high value load of 100k with a choke. GE6550A curves show about 3% THD is likely along a horizontal load line. at 1Vrms such power tubes have THD noise. Nearly all the little signal triodes measure better than curves predict; errors were routinely included with inaccurate test gear in the 1950s and 1960s when the analog gear was tubed and had worse thd than the bloomin tubes they were attempting to measure. Recent tests of EH6550 and KT90EH show such tubes in triode to be expemplary and every bit as linear as a 300B, yet the curves are only an approximation of the linearity, often incorrect by 2%. 2% is SFA, and not a big deal because from the curves we can still plot valid load lines even if they were 5% in error. A 2SK369 with a 2k RL with Ia = 5mA and with bypassed source has a gain of 80, but you can only get about 6vrms output and THD = about 7%. Its like a pentode at 60Vrms output. nevertheless, in cascode with a triode, and with 6dB current FB in the source circuit, it has a gain of say 15, and if Vin = 5mV, output = 75mV and thd = about 0.05% max. with MC and .4mV input, thd falls to 0.01%, and the following tubes don't add anything extra. I have much more source degenerative current FB in my fet input stage than Allen Wright uses in his FVP. Allen's phono stage and mine are quite linear enough and quiet enough with no loop NFB, just passive RIAA eq. A nice phono stage can be built with all fets, take a leaf from the work by Erno Borbelli... But triodes are pretty linear at low levels. The 10 tube preamp I built with a two stage MM phono amp with a pair of µ-followers but without a fet works great with cart inputs above 2mV, quiet enough, and certainly linear enough, and all passive RIAA eq. My next edition of my website will have all the grimy details of phono amp developments over the last few years. Patrick Turner. |
#114
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
Henry Pasternack wrote: "Patrick Turner" wrote in message ... I missed the µ follower bit. His tube schematic does not show exactly what is there. With Rout very low, the 0.1 and 22k give 2,200 uS ??? There's an RIAA calculator he http://www.kabusa.com/frameset.htm?/phonpre.htm It confirms the values in the Pimm schematic. Supposedly, the calculator is based on math by Stanley Lip****z. I tried using 22k and 47k for R1 in the RIAA calculator and got very different time constant R x C products. Something is drastically wrong with this online calculator. Assume it is BS unless you can proove otherwise. I haven't had a chance to check the calculations, but I suspect it's correct. I guess there's interaction between the sections that accounts for the seeming discrepancy, although it isn't intuitively obvious to me why right at this moment. Add C1 to C2, and the total should act with R1 to form the 3180uS time constant. I got strange results. I prefer my own tested trimmed values for RIAA using a reverse RIAA filter and flat input signal. You have all the details I sent you. My RIAA filter has the 75uS filter with a pole added for 50kHz, and has build out R rather than have the 75uS cap straight across the RC used for 3180 and 318 uS. There is better isolation and freedom from interaction between the 3 time constants I believe. Perhaps you'd like to simulate the circuit i sent; and see what BS a program comes up with. You cannot assume a program is telling you the truth unless you confirm the outcome by measurement. But the reverse RIAA filter I have has been used to test CJ and ARC and FVP preamps which appeared to have virtually near perfect responses so I guess my RIAA reverse eq filter values are also perfect enough. Gotta get dressed and take the kids to piano lessons. Well, Naked dads delivering kiddies to piano lessons are frowned upon here. Patrick Turner. -Henry |
#115
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
Henry Pasternack wrote: "Henry Pasternack" wrote in message ... "Patrick Turner" wrote in message ... I guess there's interaction between the sections that accounts for the seeming discrepancy, although it isn't intuitively obvious to me why right at this moment. The answer is you need to put the two capacitors in parallel when figuring the 50Hz cutoff. See my other post. Doing the above I still get queer answers. Patrick Turner. -Henry |
#116
Posted to rec.audio.tubes
|
|||
|
|||
The KISS AMP: square-rooting the noise
Andre Jute wrote: This particular thread is now so contaminated with irrelevant noise, some of it from transistors and child molesters, that I have opened a new thread to discuss John Stewart's contribution. It is called "The KISS AMP: square-rooting the noise: the real work starts" -- AJ John Stewart wrote: Go over to alt.binaries.pictures.radio to read of a problem arising when a choke input filter is connected in the return leg of a rectifier circuit. The page is taken from the Electrical Engineering Series, John D Ryder, Author. Cheers, John Stewart Ya gotta watch out from Irrelevanlty Noisy Transelvanian Sisters ( transistors ). Take garlic in each pocket wherever you go. Be out watching when any said sister tries to square root you in the misty dawns of the far off mountaines, and if in doubt take a priest with stake and mallet in the coach with thee..... Patrick Turner. |
#117
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
"Patrick Turner" wrote in message ...
See my other post. Doing the above I still get queer answers. If you put 22,550 Ohms into the calculator (taking into account the 450 Ohm output impedance of the "mu follower"), you get 3279 Ohms, 0.097uF, and 0.0333uF. So it's evident where Gary Pimm got his component values. At low frequencies, the reactance of the capacitors dominates R5 and R6. The two capacitors are effectively in parallel. If you figure the corner frequency of 22,500 Ohms in parallel with 0.133uF, you get 53.2 Hz. That's close enough to 50Hz to serve as a starting point for trimming. Morgan Jones has the formulas in his book. He says the numbers are "exact". If R1 is the input resistor, C1 is the first capacitor, R2 is the resistor in series with C1, and C2 is the second capacitor, then: R1C1 = 2187uS R1C2 = 750uS R2C1 = 318uS C1/C2 = 2.916 With a low-impedance network like the one in Pimm's design, afforded by the low output impedance of the first stage, the effect of stray and interelectrode capacitance is greatly reduced. I think the design is reasonable and I disagree with you that the KAB calulator is BS. -Henry |
#118
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
Henry Pasternack wrote: "Patrick Turner" wrote in message ... See my other post. Doing the above I still get queer answers. If you put 22,550 Ohms into the calculator (taking into account the 450 Ohm output impedance of the "mu follower"), you get 3279 Ohms, 0.097uF, and 0.0333uF. So it's evident where Gary Pimm got his component values. alright, but with 22.55k and 0.13 uf, the product is 2,931 uS, not 3,180 uS. so -3db is 54.2 Hz. The two caps together are supposed to begin pulling down the voltage at the output of the 22k so -3dB is supposed to be 50Hz. I would have though total C should be 0.141uF. 0.141 uF plus 22.55k have -3db at 50Hz and have a product of 3180 uS. if C2 wasn't there, the attenuation at R1 output should be negligible at 10Hz, but by 10kHz should be exactly -20dB; you should have poles at 50hz and 500Hz. so R1 = 9R2, so if R2 = 3.2k, R1 = 28.8k. can you show by first principle reasoning that I am wrong? I ain't very good with equations with reactive j quanties, vector analysis etc..... I just measure and trim; its quicker than calculations. When C2 is added, the response is -20dB down at 1kHz, and -40db down at about 22 kHz. Everywhere i look around the web there are different relevant values of R&C which are supposed to give the same error-less outcome, but they are just not all right; and especially RIAA feedback networks. They all blather on with schematics and calculations, but few measurements. At low frequencies, the reactance of the capacitors dominates R5 and R6. The two capacitors are effectively in parallel. If you figure the corner frequency of 22,500 Ohms in parallel with 0.133uF, you get 53.2 Hz. That's close enough to 50Hz to serve as a starting point for trimming. Well that is about true. But trimming should be not necessary when a computer is trying to see us right; we employ computers to be accurate within 0.1%. Morgan Jones has the formulas in his book. He says the numbers are "exact". If R1 is the input resistor, C1 is the first capacitor, R2 is the resistor in series with C1, and C2 is the second capacitor, then: R1C1 = 2187uS R1C2 = 750uS R2C1 = 318uS C1/C2 = 2.916 With a low-impedance network like the one in Pimm's design, afforded by the low output impedance of the first stage, the effect of stray and interelectrode capacitance is greatly reduced. I think the design is reasonable and I disagree with you that the KAB calulator is BS. Well, BS-ishy a bit imho. I never trust these things unless they measure right after a good test. The noise generated by the network after the first gain stage is quite negligible compared to the noise already in the amp from V1 grid input noise and flicker noise and shot noise. You could have R1 = 100k, and still the noise and stray C effects are negligible, and especially since you have a high output MM cart of 2mV input With my circuit, noise is also very low since stage 1 cascode has a lot of gain, and RC filter components are low enough. Problems with filter stray C and noise only arrises if you have R1 = 1M like they used to use with V1 as say 1/2 a 6SL7, or a damn pentode, a real BS solution imho. -Henry |
#119
Posted to rec.audio.tubes
|
|||
|
|||
All right, Patrick
"Patrick Turner" wrote in message ...
alright, but with 22.55k and 0.13 uf, the product is 2,931 uS, not 3,180 uS. so -3db is 54.2 Hz. I neglected to consider the effect of the 3.2K resistor, which adds to the total series resistance seen by C1. This lowers the corner frequency. if C2 wasn't there, the attenuation at R1 output should be negligible at 10Hz, but by 10kHz should be exactly -20dB; you should have poles at 50hz and 500Hz. so R1 = 9R2, so if R2 = 3.2k, R1 = 28.8k can you show by first principle reasoning that I am wrong? But C2 is there, and the network never hits a flat response of -20dB. At midband (1000Hz), where RIAA is specified to be -19.911dB, the reactance of C2 is a non-negligible 4.8K Ohms. I ain't very good with equations with reactive j quanties, vector analysis etc. I just measure and trim; its quicker than calculations. It's certainly quicker if you're no good with equations. Everywhere i look around the web there are different relevant values of R&C which are supposed to give the same error-less outcome, but they are just not all right; and especially RIAA feedback networks. They all blather on with schematics and calculations, but few measurements. If you account for all strays and second-order effects, calculations will be exact. It's easier to use a reasonable approximation to get within, say, 0.5dB and trim the network on the bench to get it spot-on. Well that is about true. But trimming should be not necessary when a computer is trying to see us right; we employ computers to be accurate within 0.1%. We would have no RIAA curve in the first place without calculations. I don't deny that you can get good results without math, but in this case I think your lack of math background causes you to be unduly suspicious of computers and calculations The noise generated by the network after the first gain stage is quite negligible compared to the noise already in the amp from V1 grid input noise and flicker noise and shot noise. The 20dB loss introduced by the network adds directly to the noise figure measured at the input to the second stage. If the first stage had only 20dB of gain, at midband the signal level at the input to the second stage would be the same as at the first stage, and so the second stage noise would be as significant as the first's. There doesn't seem to be any advantage to degeneration in the cathode (or source) of the first stage except as required to set the bias current and to insure sufficient input overload margin. Every dB of gain in the first stage lessens the effective noise contribution of the second stage. -Henry |
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 100 4 December 2004 | Vacuum Tubes | |||
What are they Teaching | Audio Opinions |