Home |
Search |
Today's Posts |
#1
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-P amp
Hi RATs!
Took me long enough. I have now installed garters in three types of amp. All much improved in bass and clarity. This was the first fixed bias converted to balanced cathode bias. Another one bites the dust. www.tubecad.com May 2005 Happy Ears! Al |
#2
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
Hi RATs!
Took me three and a half years to try eight resistors and four capacitors Perhaps, in the fullness of time. Perhaps. Happy Ears! Al On Oct 9, 1:29�am, flipper wrote: Glad to hear it was a success and I, too, think current balancing makes a difference. Why not also try the current mirror approach? |
#3
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
tubegarden wrote: Hi RATs! Took me long enough. I have now installed garters in three types of amp. All much improved in bass and clarity. This was the first fixed bias converted to balanced cathode bias. Another one bites the dust. www.tubecad.com May 2005 Happy Ears! Al Broskie tries to explain the Blumlien Garter circuit but I don't like it much. The original Garter has 2 x 200 ohms in series to OV for each Rk of two output triodes with Ek at +60V in each. Each cathode is bypassed to 0V. The junction of one tube's 200 ohm Rks is taken to the grid bias R of the other tube. Suppose the gm of one tube varies so it tries to set itself with say Ek = +70V. This is +10Vdc above the wanted +60V. Then 1/2 this +10V rise is fed to the other tube grid, so its Ek will rise, and then half that rise is fed back to the first tube grid so the tube bias currents cause rising grid voltages which try to rise and tend to balance but the amount of balance you get isn't really worth the trouble imho. Broskie goes on to show a range of methods using SS circutry to maintain nearly identical Ia dc in each output tube. I've tried similar ideas in a "fixed bias" amp using an LTP with MJE350 so the collectors control the negative grid bias ( via the attached Rg ) and balance Ia bias from current sensing resistors from each cathode to 0V, say 10 ohms. The sensing R ac is filtered out with RC, and the Ik dc is fed to the bases of the LTP. Its very simple and leaves the Blumlien circuit in the dust except that when you apply GNFB, the whole thing tends to oscillate badly at LF beacuse with so many RC couplings the circuit works like a decent phase shift oscillator, and adding more C to ever lengthen the time constants merely moves the oscillation frequency ever lower, but oscillate it will. Trying to balance the Ia dc of PP tubes when pushed into class AB is like catching a horse that has bolted. There is always a delay of bias correction. Usually the use of individual bias resistors and large bypass caps over 470uF in a pair of OPT tubes gives very adequate bias equality unless the tubes are nearly worn out. Another trouble with providing self balancing in what are cathode biasing networks with large R&C values is that when the amps move into class AB operation then BOTH tubes will have a much rising Ek and thus become wrongly biased and producing high crossover THD as a result from the rectifier action because of very distorted cathode currents which charge up Ck like in a power rectifier. The garter only works a bit with class A circuits. For something that gives the nice self regulation of cathode bias and the low THD of fixd bias even with class AB operation, people could do a lot worse than use my Dynamic Bias Stabilization method of bias control, see http://www.turneraudio.com.au/schem-...tabilizer.html This is slightly more complex than Blumlein who didn't have transistors available for use as happy slaves to manage tube cathode currents. Its less complex than much of Broskie, but will work on any cathode bias amp if implemented properly. There are full notes for everyone at my site. The circuit of mine only begins to work when any signal produces signal peak currents more than twice the idle current of an OP tube, ie, once the tubes begin to work in class AB after the limit for pure class A. Most ppl's class AB amps don't move out of class A often, and then maybe only on transients like drum beats and my system automatically bypasses these transient currents *as they are produced* through a transistor and thus the voltage bias at the cathode remains very stable and doesn't move around like it does on all AB amps when pushed a bit hard. When pressed very hard with a sine wave well into class AB, there will be a small square wave at the low value current sensing R. The amount of THD fed back into the OP tube cathode is instantly much less than the THD if there was no active bypassing of excessive Ik signals. One could possibly use 1 ohm current sensing in my scheme to reduce series distortion signals by 20dB over what is found with 10 ohms. but you'd need to maybe amplifiy the signal from the current sensor R, and that complicates matters unecessarily. Patrick Turner. |
#4
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-P amp
In article ,
tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Doesn't meet the KISS principle, why use "eight resistors and four capacitors" when only one resistor and one capacitor will do the job? -- Regards, John Byrns Surf my web pages at, http://fmamradios.com/ |
#5
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
On Oct 9, 8:17�am, John Byrns wrote:
Doesn't meet the KISS principle, why use "eight resistors and four capacitors" when only one resistor and one capacitor will do the job? -- Regards, John Byrns Hi RATs! "Keep everything as simple as possible, but NO SIMPLER!" - G.W. Bush or A. E. Newman or A. Einstein, whoever ... If you are going to listen to music, the current imbalance in the output transformer gobbles bandwidth and sensativity. Blumlein built the circuit before WW2. It really works, despite Patrick's musings. Early bean counters were every bit as stupid as the spreadsheet readers of today. Saying a circuit works is not the same as saying it sounds wonderful, but, a penny saved is a world lost. I am glad I do not rely on anyone to tell me what to try. Even Broskie ran past it on his way to a Real Solution Not much progress in the tube amp game. Can't imagine why "when only one resistor and one capacitor will do the job?" Because that noble little pair sound like ****, relatively. And I am using cheap, but carefully matched parts. Matching parts is fun, and free, if your time is not too valuable Happy Ears! Al PS Patrick, the correction signal for each tube's grid is based on the other tubes current at each moment, and drives both toward the middle, away from any imbalance. Build it and measure all you like Quit spouting vapid theory into the eyes of the world. You are a fine builder. Criticism is a sucker's game. |
#6
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
flipper wrote: On Thu, 9 Oct 2008 03:44:21 -0700 (PDT), tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Hehe. I can appreciate that. One advantage the current mirror has is it doesn't alter the rest of the amp. By that I mean, it 'fits' inside the existing bias voltage so the rest of the amp is unaware there's been a 'change'. Of course, it's also more accurate. I can also see how some might want a 'no sand', 'purist', amp too, though. The current mirror is unecessary. It maybe easier and just as good to have two separate ss current sinks, one for each cathode and each can be arranged to get less than 1mA of current. But the trouble is when the amp works in class AB. Then you get an even larger rise in Ek than with plain resistors used for cathode current, and the Ek rise occurs on each cathode, and the amp effectively becomes biased for class B, or even class C, and with a dramatic increase in odd order distortion, ie, crossover distortion. Between class A and B the Ek of each OP tube lurches around. I have spelt out a way of dynamic bias stabilization to overcome this Ek variation when cathode biasing is used see my last post. The only other thing that *IS SIMPLE* that works against Ek rise in class AB is fixed bias, and then the use of an active balancing circuit is OK, either with an LTP made using a pair of MJE350 to balance the grid bias voltage, or else have a 12AX7 set up as an LTP with a -300V supply to so the same thing. BUT these arrangements are prone to make anyone's amp into a LF phase shift oscillator, especially with GNFB. I found that simply having fixed bias and with a pair of LED to indicate the bias balance worked just fine. A small pot on top of the chassis and workable with a fingernail for bias balance adjustment gives all the active Ia dc balancing anyone would want for hi-fi becaue if theamp make a max of say 40W class AB, and 20W in class A before moving to AB, then that's enough class A power for 99% of listeners and the amp just does not de-balance because there are no RC networks on each cathode. Bias unbalance could come from serious clipping when the coupling caps charge up due to grid current, but by that point most listeners are being deafened, or there is a serious amp problem. The unblance will register in the LEDs used to monitor the balance. So a shorted speaker lead is easily spotted, or bias failure. See the ideas and design detials at http://www.turneraudio.com.au/5050Integrated.html My ideas on this have not been improved by anyone, or used commercially by anyone because most designers of commercial amps are dumbasses, stubborn, and in love with absurd complexity which are monuments to their ego but not to fine music, and they don't care if their products blow up and cause much regret amoung the buying public, and bring a bad name to the use of tubes. But with my designs you always get superlative sound and reliablity and ease of bias management and you keep an eye on the bias whenever the amps are on. And if you go out to buy some beer and leave the amps turned on, that will always be the time when a tube decides to fail to a short, ot just conduct far too much Idc, then my amps all just turn themselves off politely instead of burning a house down. Don't ask me if the margin of LF stability is improved with the Blumlein circuit, but probably not. You guys with simulators might be able to answer that; whether you'd be right needs a check with the breadboard - of course!. Patrick Turner. Perhaps, in the fullness of time. Perhaps. Happy Ears! Al On Oct 9, 1:29?am, flipper wrote: Glad to hear it was a success and I, too, think current balancing makes a difference. Why not also try the current mirror approach? |
#7
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
Hi RATs!
Seems to be some political noise ... If you have a P-P amp: get 8 resistors and 4 capacitors and put them in and listen... if you ain't impressed, I apologize. (you will be very impressed) Happy Ears! Al |
#8
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34P-P amp
Hi RATs,
Al's right. I'm building a PP 71A for my office and I was disappointed with the bass response I was getting on the breadboard amp. I tried the garters and it was a pleasant surprise. Much better bass with only about 2 watts. I've disassembled the breadboard and am currently building a chassis. Slow going... Best regards to all, Raymond tubegarden wrote: Hi RATs! Seems to be some political noise ... If you have a P-P amp: get 8 resistors and 4 capacitors and put them in and listen... if you ain't impressed, I apologize. (you will be very impressed) Happy Ears! Al |
#9
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
tubegarden wrote: Hi RATs! Seems to be some political noise ... If you have a P-P amp: get 8 resistors and 4 capacitors and put them in and listen... if you ain't impressed, I apologize. (you will be very impressed) I could say the same thing politically about all my methods of doing things. But I won't assume everyone will be impressed. And I won't apologize. Like you I ain't done nothin wrong. I may not be able to help those who are not impressed. Keep smilin, Patrick Turner. Happy Ears! Al |
#10
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
Patrick Turner wrote: I could say the same thing politically about all my methods of doing things. But I won't assume everyone will be impressed. And I won't apologize. Like you I ain't done nothin wrong. I may not be able to help those who are not impressed. Keep smilin, Patrick Turner. Hi RATS! Yes, we ain't doin nothing wrong, just different The world is headed for a lesson in money and power. Good time to seek close to free entertainment ... I am happy my son had room for us. Some folks are not so lucky. And it will not be brighter soon. My pile of old stereo junk works fine 8^D The other 95% was just for the joy of seeing it. Music is the best part. No matter what we think or see around us. Happy Ears! Al |
#11
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-P amp
In article ,
flipper wrote: On Thu, 09 Oct 2008 08:17:24 -0500, John Byrns wrote: In article , tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Doesn't meet the KISS principle, why use "eight resistors and four capacitors" when only one resistor and one capacitor will do the job? The first task is, or should be, to define 'the job' and if current balance is a part of it then the "one resistor and one capacitor" doesn't do it. Al's definition of the job seems to be first preventing "current imbalance in the output transformer", because it "gobbles bandwidth and sensativity", and secondly insuring that "it sounds wonderful". Presumably Al's point is that if "current imbalance in the output transformer gobbles bandwidth" then it won't sound so wonderful. While that seems reasonable, I'm not sure how gobbling "sensativity", what ever that is, will keep an amplifier from sounding wonderful? My requirements for the job are a little different than Al's. I start with the same requirement that the amplifier sound wonderful, however I don't feel that current balance in the output transformer is a necessary condition for an amplifier to sound wonderful, so I don't make current balance a requirement of the job. Secondly for numerous reasons I prefer simplicity, so I don't like solutions that take extra parts to implement, KISS. Presumably the problem with current imbalance in the output transformer is that it can cause the transformer to saturate under some conditions, producing distortion and restricting the power capability at low frequencies. An alternate approach to the problem of current imbalance in the output transformer is to use a simple circuit and deal with the current imbalance by over specifying the output transformer so that saturation won't occur with a reasonably balanced pair of output tubes. While this approach may cost a bit more, it is only a hobby, and is harmonious with Al's maxim that "a penny saved is a world lost". -- Regards, John Byrns Surf my web pages at, http://fmamradios.com/ |
#12
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
John Byrns wrote: In article , flipper wrote: On Thu, 09 Oct 2008 08:17:24 -0500, John Byrns wrote: In article , tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Doesn't meet the KISS principle, why use "eight resistors and four capacitors" when only one resistor and one capacitor will do the job? The first task is, or should be, to define 'the job' and if current balance is a part of it then the "one resistor and one capacitor" doesn't do it. Al's definition of the job seems to be first preventing "current imbalance in the output transformer", because it "gobbles bandwidth and sensativity", and secondly insuring that "it sounds wonderful". Presumably Al's point is that if "current imbalance in the output transformer gobbles bandwidth" then it won't sound so wonderful. While that seems reasonable, I'm not sure how gobbling "sensativity", what ever that is, will keep an amplifier from sounding wonderful? My requirements for the job are a little different than Al's. I start with the same requirement that the amplifier sound wonderful, however I don't feel that current balance in the output transformer is a necessary condition for an amplifier to sound wonderful, so I don't make current balance a requirement of the job. Secondly for numerous reasons I prefer simplicity, so I don't like solutions that take extra parts to implement, KISS. Presumably the problem with current imbalance in the output transformer is that it can cause the transformer to saturate under some conditions, producing distortion and restricting the power capability at low frequencies. An alternate approach to the problem of current imbalance in the output transformer is to use a simple circuit and deal with the current imbalance by over specifying the output transformer so that saturation won't occur with a reasonably balanced pair of output tubes. While this approach may cost a bit more, it is only a hobby, and is harmonious with Al's maxim that "a penny saved is a world lost". In Tremains's Audio Encyclopedia there is some info on the effects of dc offset in a PP OPT. None of the info praises imbalance, and imbalance if over 10% can have terrible effects on sound quality even at low levels simply because the THD / IMD rises dramatically. In many Quad-II amps I have seen Ia in OP tubes with one at 90mA, and the other at 40mA, with the hot one tending to glow a little red in its anode. Sound is ****. Replacing the silly 180 common Rk and small value electro bypass cap with 470 ohms each side and each bypassed with 470uF at least will usually reduce the OPT tube imbalance to les than 10% even if there is a huge imbalance with a common Rk. Owners are notoriously stupid about servicing their amps and wait until smoke and fires break out inside the amp before they replace anything. They hate having to take the amps for an annual check up more than they hate their dentist. So positive grid current develops in one tube and its sends Ia high, thus generating a more +ve voltage across the common Rk, which turns off the current in the other tube. What was Peter Walker doing when he invented this doozy? In bed with the bimbo accountant at Quad? Anyway, the easiest solution is dual Rk&Ck per output tube where there is nearly half the total PO as classA PO. Fixed bias is an oxymoron for owners because if its fixed, then why da ya hafta adjust it? They learn bias is important after a smoke filled room even occurs. But before the smoke, the OPT and sound labours away with an imbalance present, and THD are 6 times higher than normal, and this doesn't get noticed by golden eared audiophiles either. The NFB keeps the THD&IMD at levels at least below what it is without any GNFB, so the malfunctioning imbalanced circuit gets away with murder. There ain't nothing special about a Garter circuit. Its just one of a few options, perhaps marginally better technically than some others but in many ways much worse than other biasing methods I know. I just don't know how the hell all these claims can be made about much improved bass occurs with a garter biasing network. You could still have two OP tubes with a bit of age and which have drifted well away from their new condition matchedness. The Garter won't bring them as close togther for Ia as would an adjustable balance network I like to use with a couple of LED to indicate balance. Such LED will show a 3mA Ia imbalance easily. And you'll KNOW always if the amp is balanced each time you use it. And bass will be fabulous, and the rest of the band won't include ugly sounding IMD products because of imbalances. Patrick Turner. -- Regards, John Byrns Surf my web pages at, http://fmamradios.com/ |
#13
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-P amp
In article ,
flipper wrote: On Fri, 10 Oct 2008 19:43:02 -0500, John Byrns wrote: In article , flipper wrote: On Thu, 09 Oct 2008 08:17:24 -0500, John Byrns wrote: In article , tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Doesn't meet the KISS principle, why use "eight resistors and four capacitors" when only one resistor and one capacitor will do the job? The first task is, or should be, to define 'the job' and if current balance is a part of it then the "one resistor and one capacitor" doesn't do it. Al's definition of the job seems to be first preventing "current imbalance in the output transformer", because it "gobbles bandwidth and sensativity", and secondly insuring that "it sounds wonderful". Presumably Al's point is that if "current imbalance in the output transformer gobbles bandwidth" then it won't sound so wonderful. While that seems reasonable, I'm not sure how gobbling "sensativity", what ever that is, will keep an amplifier from sounding wonderful? My requirements for the job are a little different than Al's. I start with the same requirement that the amplifier sound wonderful, however I don't feel that current balance in the output transformer is a necessary condition for an amplifier to sound wonderful, so I don't make current balance a requirement of the job. Secondly for numerous reasons I prefer simplicity, so I don't like solutions that take extra parts to implement, KISS. Presumably the problem with current imbalance in the output transformer is that it can cause the transformer to saturate under some conditions, producing distortion and restricting the power capability at low frequencies. An alternate approach to the problem of current imbalance in the output transformer is to use a simple circuit and deal with the current imbalance by over specifying the output transformer so that saturation won't occur with a reasonably balanced pair of output tubes. While this approach may cost a bit more, it is only a hobby, and is harmonious with Al's maxim that "a penny saved is a world lost". I don't know why you think using a hundred words for 10 'improves' anything. You don't care about balance, fine. He does so it is simply not true that "one resistor and one capacitor will do the job" he set out to do. I also don't see why you feel the need to do nothing but denigrate what he clearly finds to be an improvement to his amp. Exactly how did I "denigrate what he clearly finds to be an improvement to his amp"? What I said was that I preferred a simpler solution to the problem, I didn't say his approach wasn't an improvement, what I said was it was too complex for my taste. Read the thread again, if there was any denigration going on it was Al's response to my first post where I indicated my preference for a KISS solution. But speaking of denigrating people, what is it with you today? First you attack poor old Patrick, and now you are after me, clearly you are having some sort of mood problem today. But then I suppose I shouldn't expect anything less from someone who hides behind a fishy alias, and is afraid to allow his posts to be archived. -- Regards, John Byrns Surf my web pages at, http://fmamradios.com/ |
#14
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
flipper wrote: On Thu, 09 Oct 2008 22:53:31 GMT, Patrick Turner wrote: flipper wrote: On Thu, 9 Oct 2008 03:44:21 -0700 (PDT), tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Hehe. I can appreciate that. One advantage the current mirror has is it doesn't alter the rest of the amp. By that I mean, it 'fits' inside the existing bias voltage so the rest of the amp is unaware there's been a 'change'. Of course, it's also more accurate. I can also see how some might want a 'no sand', 'purist', amp too, though. The current mirror is unecessary. You love to 'argue' with things never said. It maybe easier A pair of current sinks is neither fewer components nor 'easier'. and just as good And it isn't "just as good" either. to have two separate ss current sinks, one for each cathode and each can be arranged to get less than 1mA of current. But the trouble is when the amp works in class AB. Then you get an even larger rise in Ek than with plain resistors used for cathode current, and the Ek rise occurs on each cathode, and the amp effectively becomes biased for class B, or even class C, and with a dramatic increase in odd order distortion, ie, crossover distortion. Between class A and B the Ek of each OP tube lurches around. Another of your favorite 'arguments'. Switch off to a different circuit and then discuss *it's* supposed 'failings'. Whether or not one uses the Garter biasing circuit should be discussed with pros, cons, and all relevant issues raised. Having rising Ek on both OP tubes when the PP amp works in class AB beyound the pure class A portion is a rotten thing for the music. I have listed the reasons here more than once. Trouble is in many PP amps, there is very little class A content before the amp begins to work in class AB1, so bias lurches around and it don't help much if the current is still balanced; wrongly biased OPT tubes are plain wrong. r As for class AB action, the current mirror behaves like the typical cathode bias, except it's *balanced*. So what? The Ek of both OP cathodes rise and the amp becomes wrongly biased. I have spelt out a way of dynamic bias stabilization to overcome this Ek variation when cathode biasing is used see my last post. That isn't the purpose of either his circuit or mine so, again, you're braying in the dark. Whatever the purpose of his or yours is something I do understand. I am not barking inthe dark to point out the problems with the Garter or the other schemes such as those presented by Broskie. There are what I think are better ways to bias OP tubes. As long as I live, I'll keep saying so and draw people to my website so they may find all the details fully explained and try it out in their own amps like I have. I won't accept any BS about my methods sounding worse because after hearing so many amps of other peoples and my own, I know my methods will sound well if anyone uses them. I don't care if non technically minded people find fault in whatever they might try from my website. Usually thay have made a mistake in implementation, and as you should know, the sound at speakers is determined by a large number of factors, and not by just the darn biasing method. But biasing is a nightmare for many owners of tube gear to have to ever fully undertstand. The only other thing that *IS SIMPLE* that works against Ek rise in class AB is fixed bias, and then the use of an active balancing circuit is OK, either with an LTP made using a pair of MJE350 to balance the grid bias voltage, or else have a 12AX7 set up as an LTP with a -300V supply to so the same thing. I've worked with that too but fixed bias doesn't auto bias, like cathode bias does. But you *can have* fixed bias and self balancing at the same time. The LTP control of bias for each tube merely keeps the balance after you have set the bias for **both** OP tubes. because its fixed bias, there is no rise in Ek and mis-biasing, and the LTP tries to maintain Ia balance even though Ia varies a lot, even in a nearly class B amp. But differential control of biasing can be tricky at LF because you'll have a phase shift oscillator really easily, especially with GNFB present. If you built a breadboard of such an amp you'd find out about. If you don't, you'll stay ignorant. Come up with a Lincoln Walsh style auto fixed bias combined with DC balance, figure out how to make it all stable, and then you'll really have something. Yeah, I had a look at that and now forget exactly how LW did it. I must have thought what I know already probably is better. But I have never thought any amp needs active bias control beyond simple cathode biase with R&C, or some method to allow the good Idc regulation of cathode bias but the benefits of fixed bias which both are present in my active bias stabilizing method where excess Iac is acively bypassed at the cathode instead of being allowed to charge up a bias Ck. Where you have 4 or more OP tubes, then balancing is BS because how does one make them all interactive and aquire equal Idc, and maintain equal Ek for correct bias? A Garter circuit in my 300W amps with 12 x 6550 would be quite useless unless you had the amp set up as 6 balanced pairs of tubes. Each pair might have very different total bias to the next pair along... BUT these arrangements are prone to make anyone's amp into a LF phase shift oscillator, especially with GNFB. The current mirror is no more prone to LF oscillation than cathode bias is. It should not be. But at LF, instead of having a finite R present, say 470 ohms, you have the high reactance of the C, and the added phase shift. If the C = 1,000 uF, at 10Hz ZC = 160 ohms and at 1Hz = 1k6, and at 0.1Hz its 16k, anyway, Not sure I like the CC current sink. Often you'll find an amp that had a common Rk and is changed to dual Rk&Ck becomes less stable at LF. Usually you have to apply a LF gain shelving network to overcome the dumbass shortcoming of design in the original amp which barely had enough LF stability margin anyway with the amount of GNFB applied. One has to be prepared to over-ride the deficiencies in old designs as one encounters them. I found that simply having fixed bias and with a pair of LED to indicate the bias balance worked just fine. A small pot on top of the chassis and workable with a fingernail for bias balance adjustment I've told you before I like the circuit but it isn't ZERO user adjustment. It's just a poor man's meter. Often rich men buy my amps. They always lie about it, and tell me how poor they are. Common now, anyone can read two LEDs for bias balance in a channel and bias is easily kept within a few mA of balance and its all much better than having a proper meter fitted or needing to buy a meter then fiddle around with screw drivers up dark holes into tiny slots. I learnt very early just how stupid and non-technical 95% of music lovers are. If I see an audiophile with a screw driver in hand or a soldering iron, I snatch them away, and say "NO!", because I know that soon afterwards there will be smoke, confusion, helplessness, and a big bill to pay. The LED method of balancing bias is the best simple thing for fixed bias amps because you ** see the LED ** all the time while using the amp wheras if you need to meter the bias you can't be sure its adjusted right. In ARC and Manley labs amps where I have chucked out the horrible circuits and replaced them with my own, I have used LED bias indicators. No meters necessary, HOORAY!!! In such cases with ARC VT100 with 4 x 6550 OP tubes, each tube has two led, and a fingernail adjust screw. All you do is adjust each screw which is flush with the case covers on the side of the amp so that both green an red LED are not alight for bias to be right. If there's slight green showing, bias Ia is OK but a little low, and if red bias is a little high, and perhaps needs a turn down. Fixed bias means you can have very variable bias if the mains voltage changes and if that change id say 5%, say from 240Vac to 252Vac as I often se here, the B+ can rise +5% from say +480Vdc to +504Vdc, and in UL or triode amps the Ea = Eg2, and Ra of a 6550 is maybe 1,200 ohms at idle so 24Vdc extra Ea volts causes 20mA of extra Ia. So if you had tubes biased for 30W with +480V, Ia might be 63mA, then with 504Vdc, Ia = 83mA, and Pda becomes 42 watts, and damned dangerously close to tube meltdown. So I always bias tubes fairly low rather than high because I know smoke happens too often the other way. Why to ARC and Manley Labs and VTL and Quad, Leak, and all the many others end up on my bench? Its because of smoke, clouds of it, and angry customers because they trusted some jerk. I de-jerkize the designs, and then ahhhhhhh, moooozzzic, and never any smoke. gives all the active Ia dc balancing anyone would want for hi-fi becaue if theamp make a max of say 40W class AB, and 20W in class A before moving to AB, then that's enough class A power for 99% of listeners and the amp just does not de-balance Oh, yeah, sure. Tubes never drift and a retube will be 'just like' the ones you replaced. I didn't say tubes don't drift. They do. And a lot in the life some owners expect out of them. Designers of amps *other than myself* seem to put a great big effort into minimizing tube life and upsetting customer music. Make any amp using schematics from my website and you'll get a longer life from tubes and know what they are up to all the time you have the amp turned on. Peace of mind. because there are no RC networks on each cathode. Bias unbalance could come from serious clipping when the coupling caps charge up due to grid current, but by that point most listeners are being deafened, or there is a serious amp problem. The unblance will register in the LEDs used to monitor the balance. So a shorted speaker lead is easily spotted, or bias failure. See the ideas and design detials at http://www.turneraudio.com.au/5050Integrated.html My ideas Now comes the "I am a genius" Turner advertisement. Got a better alternative? I'm just an average bloke. Definately not a ****ing genius. on this have not been improved by anyone, or used commercially by anyone because most designers of commercial amps are dumbasses, stubborn, and in love with absurd complexity which are monuments to their ego but not to fine music, and they don't care if their products blow up and cause much regret amoung the buying public, and bring a bad name to the use of tubes. But with my designs you always get superlative sound and reliablity and ease of bias management and you keep an eye on the bias whenever the amps are on. And if you go out to buy some beer and leave the amps turned on, that will always be the time when a tube decides to fail to a short, ot just conduct far too much Idc, then my amps all just turn themselves off politely instead of burning a house down. Don't ask me if the margin of LF stability is improved with the Blumlein circuit, but probably not. Since it wasn't intended to address strawman 'LF stability' one wonders why you think it should 'improve' it. You guys with simulators might be able to answer that; whether you'd be right needs a check with the breadboard - of course!. What the HELL are you braying about now? He BUILT the damn thing. I didn't see any reported measurements of stability margins. Al just does things he likes; maybe he un-does what he doesn't. But sure, he built it, but that's different from a breadboard design and analysing the hell out of it. The Garter isn't something likely to cause smoke; its dead simple, and little analysis is needed by most diyers as long as it is stable in their amps and no smoke if produced. When I build something, I ***always analyse and test it all***, and look for possible dysfunctions anywhere and everywhere. And I look for improvements in technical function and if that's possible without huge added complexity of FB paths then I will explore the possibility fully with a prototype on a board or a chassis. I prefer to use real chassis for "breadboarding". Sometimes the test amp prototype is the very thing you end up selling to a customer, so you have to be very careful how you set up and test each stage to really understand what you are doing and why, and that's how I developed the 845 amps. I did have all sorts of ideas about current sinks for the cathodes in the 845 amps I finished last winter, but I thought a well rated and heatsinked 2k2 for each Rk amps would be quite excellent, and it turned out that way. There is active protection in the amps if the bias rises from 72mA per 845 to 100mA. The 845 tube won't meltdown internally if its bias current drifts upwards, but it sure will cause the amp to politely turn off. Even in cathode biased amps, a crook tube will still just "take off" with excessive bias current usually at the end of its life. The high Ek does not hold down the Ia. With CC sink, the Ek would just zoom right up unimpeded by any Rk, and you'd still need the active protection that nobody else bothers to fit into their amps. In the case of my 845 amps with 2 x 845 in each channel in parallel, the bias balance in each pair of tubes was good, but its not very important because the amps are SE, and don't need Ia balance to avoid Idc offsets. I was able to use a chinese 845 and a KR845 in the amps and the PO, distortion, bias setting and music quality remained the same. Cathode biasing including the Garter method offers some protection against rise in mains Vac and hence rise in Ea. The effective Ra at dc in triode or UL amps becomes Ra + [(µ+1 x Rk]. So with an 845, Ra = 2k2 approx. Its quite low, and a small Ea change could cause Pda to rise alarmingly. But Ra with Rk = 2k2 = 13k2, and the Pda just won't increase dangerously with a mains rise. In class A SE amps, cathode bias is terrific. Same goes for class A PP amps. But none are ever actually made. They are always class AB, and in these latter days of design by accountants and marketing experts, usually the PP amp is a glorified high power class B POS PA amp with medium NFB to win the war of watts. Its all that is won. Music is compromised, and the occasional house gets burnt down. Patrick Turner. Perhaps, in the fullness of time. Perhaps. Happy Ears! Al On Oct 9, 1:29?am, flipper wrote: Glad to hear it was a success and I, too, think current balancing makes a difference. Why not also try the current mirror approach? |
#15
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
flipper wrote: On Fri, 10 Oct 2008 19:43:02 -0500, John Byrns wrote: In article , flipper wrote: On Thu, 09 Oct 2008 08:17:24 -0500, John Byrns wrote: In article , tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Doesn't meet the KISS principle, why use "eight resistors and four capacitors" when only one resistor and one capacitor will do the job? The first task is, or should be, to define 'the job' and if current balance is a part of it then the "one resistor and one capacitor" doesn't do it. Al's definition of the job seems to be first preventing "current imbalance in the output transformer", because it "gobbles bandwidth and sensativity", and secondly insuring that "it sounds wonderful". Presumably Al's point is that if "current imbalance in the output transformer gobbles bandwidth" then it won't sound so wonderful. While that seems reasonable, I'm not sure how gobbling "sensativity", what ever that is, will keep an amplifier from sounding wonderful? My requirements for the job are a little different than Al's. I start with the same requirement that the amplifier sound wonderful, however I don't feel that current balance in the output transformer is a necessary condition for an amplifier to sound wonderful, so I don't make current balance a requirement of the job. Secondly for numerous reasons I prefer simplicity, so I don't like solutions that take extra parts to implement, KISS. Presumably the problem with current imbalance in the output transformer is that it can cause the transformer to saturate under some conditions, producing distortion and restricting the power capability at low frequencies. An alternate approach to the problem of current imbalance in the output transformer is to use a simple circuit and deal with the current imbalance by over specifying the output transformer so that saturation won't occur with a reasonably balanced pair of output tubes. While this approach may cost a bit more, it is only a hobby, and is harmonious with Al's maxim that "a penny saved is a world lost". I don't know why you think using a hundred words for 10 'improves' anything. You don't care about balance, fine. He does so it is simply not true that "one resistor and one capacitor will do the job" he set out to do. I also don't see why you feel the need to do nothing but denigrate what he clearly finds to be an improvement to his amp. We all know the Garter improves some amps. But how many commercially successful brands employ Garter? Many factors affect the sound. Patrick Turner. |
#16
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
John Byrns wrote: In article , flipper wrote: On Fri, 10 Oct 2008 19:43:02 -0500, John Byrns wrote: In article , flipper wrote: On Thu, 09 Oct 2008 08:17:24 -0500, John Byrns wrote: In article , tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Doesn't meet the KISS principle, why use "eight resistors and four capacitors" when only one resistor and one capacitor will do the job? The first task is, or should be, to define 'the job' and if current balance is a part of it then the "one resistor and one capacitor" doesn't do it. Al's definition of the job seems to be first preventing "current imbalance in the output transformer", because it "gobbles bandwidth and sensativity", and secondly insuring that "it sounds wonderful". Presumably Al's point is that if "current imbalance in the output transformer gobbles bandwidth" then it won't sound so wonderful. While that seems reasonable, I'm not sure how gobbling "sensativity", what ever that is, will keep an amplifier from sounding wonderful? My requirements for the job are a little different than Al's. I start with the same requirement that the amplifier sound wonderful, however I don't feel that current balance in the output transformer is a necessary condition for an amplifier to sound wonderful, so I don't make current balance a requirement of the job. Secondly for numerous reasons I prefer simplicity, so I don't like solutions that take extra parts to implement, KISS. Presumably the problem with current imbalance in the output transformer is that it can cause the transformer to saturate under some conditions, producing distortion and restricting the power capability at low frequencies. An alternate approach to the problem of current imbalance in the output transformer is to use a simple circuit and deal with the current imbalance by over specifying the output transformer so that saturation won't occur with a reasonably balanced pair of output tubes. While this approach may cost a bit more, it is only a hobby, and is harmonious with Al's maxim that "a penny saved is a world lost". I don't know why you think using a hundred words for 10 'improves' anything. You don't care about balance, fine. He does so it is simply not true that "one resistor and one capacitor will do the job" he set out to do. I also don't see why you feel the need to do nothing but denigrate what he clearly finds to be an improvement to his amp. Exactly how did I "denigrate what he clearly finds to be an improvement to his amp"? What I said was that I preferred a simpler solution to the problem, I didn't say his approach wasn't an improvement, what I said was it was too complex for my taste. Read the thread again, if there was any denigration going on it was Al's response to my first post where I indicated my preference for a KISS solution. But speaking of denigrating people, what is it with you today? First you attack poor old Patrick, and now you are after me, clearly you are having some sort of mood problem today. But then I suppose I shouldn't expect anything less from someone who hides behind a fishy alias, and is afraid to allow his posts to be archived. -- Regards, John Byrns I would not worry about our fiesty fish amoung us. He ain't a shark. He asks some pithy questions and makes some contentious comments, but I believe he's actually fairly good natured most of the time. We all learn from good natured argy bargy. But you have me right, I'm poor, and I'm old, well, damn well not 30 anymore, but I do get over 500 hits a day at my website and somebody finds it interesting. Maybe its not the best tube audio site but together with others we make up a wealth of info on how-to-do issues on the Internet, and its all completely free of charge. Is the Garter better than two plain old R&C bias parts per tube in a substantially class A PP amp? I don't see it being much better. But if it is, then the betterment is marginal, and will only cost 1 extra resistor per tube, for the two series R for each Rk. The only other bother is that the total value of the two bias Rk must be twice what you'd normally have with just two sets of separate R&C. The grid voltage is determined by the voltage half way down the total Rk. So if you have a circuit for class AB1 KT66 with Ea = 365V, and 30V of grid bias voltage across 470ohms, then Rk total will have to be 940 ohms, and the Ek will be +60Vdc, Eg1 = +30Vdc. The doubling of the total value of the Rk definitely will improve natural idle current NFB regulation of Ia by Rk. But you'll dissipate an extra 1.9W per cathode resistance but that's not too much a price to pay. I didn't consider this aspect before. But Ek rise due to class AB charge up currents in OP tubes will be worse, but then if you also used my bias stabilizing circuit with a power transistor and about 3 x R for each OP tube then you'd control the Ek rise with class AB abd keep the circuit balanced at the same time. Maybe this is the best, I don't know the full answer, because so many ways can be thought of for achieving an outcome. Patrick Turner. Surf my web pages at, http://fmamradios.com/ |
#17
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
flipper wrote: On Sat, 11 Oct 2008 05:23:27 GMT, Patrick Turner wrote: flipper wrote: On Fri, 10 Oct 2008 19:43:02 -0500, John Byrns wrote: In article , flipper wrote: On Thu, 09 Oct 2008 08:17:24 -0500, John Byrns wrote: In article , tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Doesn't meet the KISS principle, why use "eight resistors and four capacitors" when only one resistor and one capacitor will do the job? The first task is, or should be, to define 'the job' and if current balance is a part of it then the "one resistor and one capacitor" doesn't do it. Al's definition of the job seems to be first preventing "current imbalance in the output transformer", because it "gobbles bandwidth and sensativity", and secondly insuring that "it sounds wonderful". Presumably Al's point is that if "current imbalance in the output transformer gobbles bandwidth" then it won't sound so wonderful. While that seems reasonable, I'm not sure how gobbling "sensativity", what ever that is, will keep an amplifier from sounding wonderful? My requirements for the job are a little different than Al's. I start with the same requirement that the amplifier sound wonderful, however I don't feel that current balance in the output transformer is a necessary condition for an amplifier to sound wonderful, so I don't make current balance a requirement of the job. Secondly for numerous reasons I prefer simplicity, so I don't like solutions that take extra parts to implement, KISS. Presumably the problem with current imbalance in the output transformer is that it can cause the transformer to saturate under some conditions, producing distortion and restricting the power capability at low frequencies. An alternate approach to the problem of current imbalance in the output transformer is to use a simple circuit and deal with the current imbalance by over specifying the output transformer so that saturation won't occur with a reasonably balanced pair of output tubes. While this approach may cost a bit more, it is only a hobby, and is harmonious with Al's maxim that "a penny saved is a world lost". I don't know why you think using a hundred words for 10 'improves' anything. You don't care about balance, fine. He does so it is simply not true that "one resistor and one capacitor will do the job" he set out to do. I also don't see why you feel the need to do nothing but denigrate what he clearly finds to be an improvement to his amp. We all know the Garter improves some amps. I would have thought so but that isn't what Byrns said. But how many commercially successful brands employ Garter? Very few I'll wager, because before SS it cost money and after SS there are more effective methods. Garter only needs 2 extra R per channel with 2 output tubes. The cost of 2R and a higher voltage rated electro cap was enough to make a bean counter froth and bubble at the mouth. See my other post in reply to JB about this issue. After SS, the most commercially effective method to increase profits was to dump tubes altogether. Those makers who stayed with tubes often gradually increased complexity and huge amounts of NFB and all sorts of trick, bells, and whistles and features. No active protection, troublesome bias setting, class A amount is tiny, and the stage is set for smoke. But, then, Byrns didn't suggest a 'better' balance method, he said a single resistor and cap was just as good when it clearly isn't. The single R&C is only slightly marginally worse than Garter. Remember, garter needs RK to be twice the normal Rk value where there are two series R per Rk for one tube. My other post explains it. So there is some better regulation of the idle current going on in Garter but I find normal single R&C just fine, and tubes have usually near enough Ia most of their life. Fixed bias with balance indication LED and an adjust pot is best for class AB with two tubes, because the tube conditions are nicely monitored all the time. My 8585 has 4 single adjust pots on the front panel of the amp per channel, but also LED which turn on to tell you if the balance is wrong by more than a few mA. Yes, a volt meter must be connected, but that's easy, its by probing non lethal contacts at the front of the amp while twiddling pot shafts adjacent to the pots, nice tough 10k x 3W wire wounds with 1/4" shaft with a slot cut in it, no ****y little cermet ****e. When the amp is adjusted right, the LED remain extinguished. The amp as a whole is remarkably bias stable. Mind you, I don't have either Ea or Ia too high like many makers. I only ask 85 watts from 4 x 6550 instead of the 150W that other makers would try to get. Many factors affect the sound. True but those 'other factors' were not the subject at hand. Garter effect on biasing was the subject at hand. But I can't think of Garter only without thinking of other biasing schemes. Garter is best in pure class A amps. Perhaps it can be slightly improved upon by including a transistor or two to make the circuit balance better than it does. As it is naturally, Ek of one tube rises, so the other grid cops 1/2 that rise, and 1/2 that rise is thenj applied to the first tube which is dc positive FB, not so good, and a better method would be welcome, no? Patrick Turner. Patrick Turner. |
#18
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
On Oct 11, 8:34�am, Patrick Turner wrote:
I would not worry about our fiesty fish amoung us. He ain't a shark. He asks some pithy questions and makes some contentious comments, but I believe he's actually fairly good natured most of the time. We all learn from good natured argy bargy. But you have me right, I'm poor, and I'm old, well, damn well not 30 anymore, but I do get over 500 hits a day at my website and somebody finds it interesting. Maybe its not the best tube audio site but together with others we make up a wealth of info on how-to-do issues on the Internet, and its all completely free of charge. Is the Garter better than two plain old R&C bias parts per tube in a substantially class A PP amp? I don't see it being much better. But if it is, then the betterment is marginal, and will only cost 1 extra resistor per tube, for the two series R for each Rk. The only other bother is that the total value of the two bias Rk must be twice what you'd normally have with just two sets of separate R&C. The grid voltage is determined by the voltage half way down the total Rk. So if you have a circuit for class AB1 KT66 with Ea = 365V, and 30V of grid bias voltage across 470ohms, then Rk total will have to be 940 ohms, and the Ek will be +60Vdc, Eg1 = +30Vdc. The doubling of the total value of the Rk definitely will improve natural idle current NFB regulation of Ia by Rk. But you'll dissipate an extra 1.9W per cathode resistance but that's not too much a price to pay. I didn't consider this aspect before. But Ek rise due to class AB charge up currents in OP tubes will be worse, but then if you also used my bias stabilizing circuit with a power transistor and about 3 x R for each OP tube then you'd control the Ek rise with class AB abd keep the circuit balanced at the same time. Maybe this is the best, I don't know the full answer, because so many ways can be thought of for achieving an outcome. Patrick Turner. Hi RATs! Yup. Actually, Mr. Blumlein spoke only of balancing the current at zero signal, OWTTE. Seems DC current offset magnetizes the iron, one way or the other doesn't matter, it just changes the path seen by the occasional signal I do not know where I found his writing, but, it was on the web, I have no access to real books in my decrepitude My latest amp *******ized is getting better and better, he said, wishing he was not being silly As Mr. B was only concerned with the output transformer at rest, a commercial amp with multiple output tubes uses the circuit and enjoys the "no user effort required". Patrick is not the first source I have heard say 10% is the only required matching of output tubes. My first, lo these many years ago, heard my first implementation and was perfetly willing to admit it sounded wonderful, but, not to change his idea. At Univac, in the 60's, more than one hardware guy said their worst nightmare was a software guy with a screwdriver This was back when changing a discrete transistor required changing the length of the wires leading to it, to tune the circuit to the larger context. They had a straightforward procedure for this process. It was just all in a days work. No wonder mainframes cost more than Hollywood movies And the old vacuum tube computers did not require SECDED logic, they were too slow and the signals were too huge to be bothered by sunspots Oh, SECDED is Sigle Error Correct, Double Error Detect, in the original Cray, now in the Smithsonian ______ I would like to publically thank Phillip Bach for teaching me to tune a piano. He was once the president of the Piano Tuner's Guild and pretty dang clever. I am not a piano tuner. I can do it, but, I take so long it drifts off while I slave There were pianos even I was able to help But, I hear piano music like never before. Not better, just different ... ______ The guitar recording now playing features a talented guitarist playing a fine instrument. I was never that good, and no guitar I ever touched had such a complex voice. There used to be a violin repair shop in Minneapolis, Drew's. Might still be, I dunno, but, he repaired a guitar for me and it was amazing what he accomplished with a simple modification. ______ Such experiences taught me to seek better sound, everywhere. It is fun, even when I have the ocassional dry decade, or two Happy Ears! Al |
#19
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
On Oct 10, 7:43�pm, John Byrns wrote:
Al's definition of the job seems to be first preventing "current imbalance in the output transformer", because it "gobbles bandwidth and sensativity", and secondly insuring that "it sounds wonderful". �Presumably Al's point is that if "current imbalance in the output transformer gobbles bandwidth" then it won't sound so wonderful. �While that seems reasonable, I'm not sure how gobbling "sensativity", what ever that is, will keep an amplifier from sounding wonderful? Hi RATs! OK, "sensitivity" - my NG editor doesn't do spelling I intentionally mispelled "college" in fifth grade because some egomaniac whispered the correct spelling to me from behind during the actual Spelling Bee. I certainly showed that clown ... I shall never cheat! ... I no longer spell things wrong on purpose, just wrong. Sigh. Mark Twain noticed you really have to admire a man who can spell one word many ways ... As Blumlein stated, somewhere (?), any stable imbalance in the Zero Signal state of an output transformer magnetizes the core of that output transformer. This quickly disables low frequency throughput, and also requires more input signal for a comparable output signal. Putting in the garters allows the driving stage not to be forced to shout. Any shouting in the signal path gives me nightmares. I listen while asleep more often than I would like. It is the rest I don't get while sleeping (with or without music ) which is the brute strength of this little malady of mine. And, as I approach eleven years, brute seems quite a good choice of word Happy Ears! Al |
#20
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-P amp
In article ,
flipper wrote: On Sat, 11 Oct 2008 05:20:44 GMT, Patrick Turner wrote: Often you'll find an amp that had a common Rk and is changed to dual Rk&Ck becomes less stable at LF. IMO it's because you have two different time constants even though the resistors and capacitors are supposedly the same. In reality they aren't and, of course, neither are the tubes. Your opinion is wrong, it has nothing to do with having two different time constants. When both cathodes of a push pull amplifier are tied together with a single resistor and capacitor used to provide cathode bias, the time constant of the cathode circuit does not affect the fundamental signal frequency and stability. When the two cathodes are separated and each is provided with its own resistor and capacitor, the time constant, even if both are identical, enters into the stability equation because when there are separate capacitors they affect the fundamental signal frequency. -- Regards, John Byrns Surf my web pages at, http://fmamradios.com/ |
#21
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
tubegarden wrote: On Oct 11, 8:34�am, Patrick Turner wrote: I would not worry about our fiesty fish amoung us. He ain't a shark. He asks some pithy questions and makes some contentious comments, but I believe he's actually fairly good natured most of the time. We all learn from good natured argy bargy. But you have me right, I'm poor, and I'm old, well, damn well not 30 anymore, but I do get over 500 hits a day at my website and somebody finds it interesting. Maybe its not the best tube audio site but together with others we make up a wealth of info on how-to-do issues on the Internet, and its all completely free of charge. Is the Garter better than two plain old R&C bias parts per tube in a substantially class A PP amp? I don't see it being much better. But if it is, then the betterment is marginal, and will only cost 1 extra resistor per tube, for the two series R for each Rk. The only other bother is that the total value of the two bias Rk must be twice what you'd normally have with just two sets of separate R&C. The grid voltage is determined by the voltage half way down the total Rk. So if you have a circuit for class AB1 KT66 with Ea = 365V, and 30V of grid bias voltage across 470ohms, then Rk total will have to be 940 ohms, and the Ek will be +60Vdc, Eg1 = +30Vdc. The doubling of the total value of the Rk definitely will improve natural idle current NFB regulation of Ia by Rk. But you'll dissipate an extra 1.9W per cathode resistance but that's not too much a price to pay. I didn't consider this aspect before. But Ek rise due to class AB charge up currents in OP tubes will be worse, but then if you also used my bias stabilizing circuit with a power transistor and about 3 x R for each OP tube then you'd control the Ek rise with class AB abd keep the circuit balanced at the same time. Maybe this is the best, I don't know the full answer, because so many ways can be thought of for achieving an outcome. Patrick Turner. Hi RATs! Yup. Actually, Mr. Blumlein spoke only of balancing the current at zero signal, OWTTE. Seems DC current offset magnetizes the iron, one way or the other doesn't matter, it just changes the path seen by the occasional signal I do not know where I found his writing, but, it was on the web, I have no access to real books in my decrepitude My latest amp *******ized is getting better and better, he said, wishing he was not being silly As Mr. B was only concerned with the output transformer at rest, a commercial amp with multiple output tubes uses the circuit and enjoys the "no user effort required". Patrick is not the first source I have heard say 10% is the only required matching of output tubes. My first, lo these many years ago, heard my first implementation and was perfetly willing to admit it sounded wonderful, but, not to change his idea. To speak more clearly what a 10% tube mismatch means is this - tubes can have a 10% difference in signal gm or bias voltage required to hold down a given Ia. In the amp circuit, such mismatched tubes *need* to be biased for equal Ia in each to avoid dc magnetization of the core. And such mismatched tubes are preferable in a largely class A amp, where as far as the tubes are concerned, they each see their work being that of an SE tube working in class A but with slightly different loads on each. If you measured 2 tubes in any class A or AB PP amp, you probably will find very close anode signal voltages at low levels, say 20Vrms at each anode because of the magnetic "locking" of each half of the primary but if you have a 10 ohm R between anode and promary then you'd maybe measure different currents in each 10 ohms, and load seen by each tube = voltage at load or OPT primary connection / current between anode and load. The mismatch will produce some 2H which is normally fully cancelled if the OP tubes are matched, but usually this 2H is less than the major distortion product of 3H. At Univac, in the 60's, more than one hardware guy said their worst nightmare was a software guy with a screwdriver This was back when changing a discrete transistor required changing the length of the wires leading to it, to tune the circuit to the larger context. They had a straightforward procedure for this process. It was just all in a days work. No wonder mainframes cost more than Hollywood movies And the old vacuum tube computers did not require SECDED logic, they were too slow and the signals were too huge to be bothered by sunspots Oh, SECDED is Sigle Error Correct, Double Error Detect, in the original Cray, now in the Smithsonian ______ I would like to publically thank Phillip Bach for teaching me to tune a piano. He was once the president of the Piano Tuner's Guild and pretty dang clever. I am not a piano tuner. I can do it, but, I take so long it drifts off while I slave There were pianos even I was able to help But, I hear piano music like never before. Not better, just different ... ______ The guitar recording now playing features a talented guitarist playing a fine instrument. I was never that good, and no guitar I ever touched had such a complex voice. There used to be a violin repair shop in Minneapolis, Drew's. Might still be, I dunno, but, he repaired a guitar for me and it was amazing what he accomplished with a simple modification. ______ Such experiences taught me to seek better sound, everywhere. It is fun, even when I have the ocassional dry decade, or two Well, here's to wet decades. Just what that might mean is open to interpretation.... Patrick Turner. d Happy Ears! Al |
#22
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
flipper wrote: On Sat, 11 Oct 2008 05:20:44 GMT, Patrick Turner wrote: flipper wrote: On Thu, 09 Oct 2008 22:53:31 GMT, Patrick Turner wrote: flipper wrote: On Thu, 9 Oct 2008 03:44:21 -0700 (PDT), tubegarden wrote: Hi RATs! Took me three and a half years to try eight resistors and four capacitors Hehe. I can appreciate that. One advantage the current mirror has is it doesn't alter the rest of the amp. By that I mean, it 'fits' inside the existing bias voltage so the rest of the amp is unaware there's been a 'change'. Of course, it's also more accurate. I can also see how some might want a 'no sand', 'purist', amp too, though. The current mirror is unecessary. You love to 'argue' with things never said. It maybe easier A pair of current sinks is neither fewer components nor 'easier'. and just as good And it isn't "just as good" either. to have two separate ss current sinks, one for each cathode and each can be arranged to get less than 1mA of current. But the trouble is when the amp works in class AB. Then you get an even larger rise in Ek than with plain resistors used for cathode current, and the Ek rise occurs on each cathode, and the amp effectively becomes biased for class B, or even class C, and with a dramatic increase in odd order distortion, ie, crossover distortion. Between class A and B the Ek of each OP tube lurches around. Another of your favorite 'arguments'. Switch off to a different circuit and then discuss *it's* supposed 'failings'. Whether or not one uses the Garter biasing circuit should be discussed with pros, cons, and all relevant issues raised. The key word is "relevant" and the topic was OPT balance, not Ek. Ek for each OP cathode is relevant to balance because where you have two Rk, balance can only be when Ek are equal in each OP tube. With CCS for each cathode, Ik balance exists even if Ek vary. Having rising Ek on both OP tubes when the PP amp works in class AB beyound the pure class A portion is a rotten thing for the music. Not when there's sufficient Class A. Agreed, but the trend in the war of watts to get better sales figures in many amps is to have the amp produce maybe only 2 watts in class A and then there is a total of say 65 watts. People pay money to me to reduce this idiotic design detail so that maybe 15 watts is class A and maximum PO is 40W. I have listed the reasons here more than once. Trouble is in many PP amps, there is very little class A content before the amp begins to work in class AB1, so bias lurches around and it don't help much if the current is still balanced; wrongly biased OPT tubes are plain wrong. You have no evidence whatsoever that the amp in question has " little class A content." Agreed, but for anyone considering trying to use the Garter, then doncha thik its relevant to talk about some modern amp in which it may be used rather than something made in 1935, and likely to be pure class A triodes? And with crummy iron in most cases. As for class AB action, the current mirror behaves like the typical cathode bias, except it's *balanced*. So what? The "so what" is because the current mirror, and garter circuit that started this thread, are intended to address current balance, not Ek. Indeed. I like to address both issues, balance and Ek stability. Does it matter that I stretch the thread a bit? The Ek of both OP cathodes rise and the amp becomes wrongly biased. Usually not. True if levels are low, but if levels ever are high, Ek will move, and bias is dynamically wandering around. Its not a desirable amp feature. I have spelt out a way of dynamic bias stabilization to overcome this Ek variation when cathode biasing is used see my last post. That isn't the purpose of either his circuit or mine so, again, you're braying in the dark. Whatever the purpose of his or yours is something I do understand. If you did you wouldn't need to say "whatever the purpose." I am not barking inthe dark to point out the problems with the Garter or the other schemes such as those presented by Broskie. Then you need to reread Broskie because what you lump all together as "other schemes" have different purposes. Some are to address balance and others are intended to keep Ek constant. There are what I think are better ways to bias OP tubes. As long as I live, I'll keep saying so and draw people to my website so they may find all the details fully explained and try it out in their own amps like I have. I won't accept any BS about my methods sounding worse because after hearing so many amps of other peoples and my own, I know my methods will sound well if anyone uses them. I don't care if non technically minded people find fault in whatever they might try from my website. Usually thay have made a mistake in implementation, and as you should know, the sound at speakers is determined by a large number of factors, and not by just the darn biasing method. But biasing is a nightmare for many owners of tube gear to have to ever fully undertstand. As I previously mentioned, you love to 'argue' with things never said. The only other thing that *IS SIMPLE* that works against Ek rise in class AB is fixed bias, and then the use of an active balancing circuit is OK, either with an LTP made using a pair of MJE350 to balance the grid bias voltage, or else have a 12AX7 set up as an LTP with a -300V supply to so the same thing. I've worked with that too but fixed bias doesn't auto bias, like cathode bias does. But you *can have* fixed bias and self balancing at the same time. I didn't say "self balancing." I said auto bias. The LTP control of bias for each tube merely keeps the balance after you have set the bias for **both** OP tubes. because its fixed bias, there is no rise in Ek and mis-biasing, and the LTP tries to maintain Ia balance even though Ia varies a lot, even in a nearly class B amp. But differential control of biasing can be tricky at LF because you'll have a phase shift oscillator really easily, especially with GNFB present. If you built a breadboard of such an amp you'd find out about. If you don't, you'll stay ignorant. Right back attcha. Come up with a Lincoln Walsh style auto fixed bias combined with DC balance, figure out how to make it all stable, and then you'll really have something. Yeah, I had a look at that and now forget exactly how LW did it. I must have thought what I know already probably is better. I wouldn't be a bit surprised. But I have never thought any amp needs active bias control beyond simple cathode biase with R&C, or some method to allow the good Idc regulation of cathode bias but the benefits of fixed bias which both are present in my active bias stabilizing method where excess Iac is acively bypassed at the cathode instead of being allowed to charge up a bias Ck. Where you have 4 or more OP tubes, then balancing is BS because how does one make them all interactive and aquire equal Idc, and maintain equal Ek for correct bias? A Garter circuit in my 300W amps with 12 x 6550 would be quite useless unless you had the amp set up as 6 balanced pairs of tubes. Each pair might have very different total bias to the next pair along... Nobody said you 'should use' a garter circuit. BUT these arrangements are prone to make anyone's amp into a LF phase shift oscillator, especially with GNFB. The current mirror is no more prone to LF oscillation than cathode bias is. It should not be. But at LF, instead of having a finite R present, say 470 ohms, you have the high reactance of the C, and the added phase shift. If the C = 1,000 uF, at 10Hz ZC = 160 ohms and at 1Hz = 1k6, and at 0.1Hz its 16k, anyway, Not sure I like the CC current sink. Your capacitor discussion applies to cathode bias in general so I have no idea why you imply it's only with a CC sink. Often you'll find an amp that had a common Rk and is changed to dual Rk&Ck becomes less stable at LF. IMO it's because you have two different time constants even though the resistors and capacitors are supposedly the same. In reality they aren't and, of course, neither are the tubes. Usually you have to apply a LF gain shelving network to overcome the dumbass shortcoming of design in the original amp which barely had enough LF stability margin anyway with the amount of GNFB applied. One has to be prepared to over-ride the deficiencies in old designs as one encounters them. Then it isn't a 'defect' of the balancing scheme but of the amp. Depends. Where a common Rk was used and a single Ck, an amp might barely have enough LF stability. With two R&C networks, the must be more phase shift for the same open loop gain and if all goes unstable at LF with GNFB. The defect wasn't actually present originally, just a shortcoming, and the gain shelving network is an improvement to nearly all amps and should have been worked out and included from day 1, but usually isn't. Makers go by the idea of near enough is good enough, and supply the public with amps that oscillate at LF quite well with no speaker connected. Makers don't take into account some owners will leave their amps turned on without a load sometimes or use a 16 ohm speaker on the 4 ohm tap thus raising OLG and thus allowing oscillation. Unconditional stability is to me the only acceptable technical stability feature. I found that simply having fixed bias and with a pair of LED to indicate the bias balance worked just fine. A small pot on top of the chassis and workable with a fingernail for bias balance adjustment I've told you before I like the circuit but it isn't ZERO user adjustment. It's just a poor man's meter. Often rich men buy my amps. They always lie about it, and tell me how poor they are. Of course. Common now, anyone can read two LEDs for bias balance in a channel and bias is easily kept within a few mA of balance and its all much better than having a proper meter fitted or needing to buy a meter then fiddle around with screw drivers up dark holes into tiny slots. I learnt very early just how stupid and non-technical 95% of music lovers are. If I see an audiophile with a screw driver in hand or a soldering iron, I snatch them away, and say "NO!", because I know that soon afterwards there will be smoke, confusion, helplessness, and a big bill to pay. The LED method of balancing bias is the best simple thing for fixed bias amps because you ** see the LED ** all the time while using the amp wheras if you need to meter the bias you can't be sure its adjusted right. In ARC and Manley labs amps where I have chucked out the horrible circuits and replaced them with my own, I have used LED bias indicators. No meters necessary, HOORAY!!! In such cases with ARC VT100 with 4 x 6550 OP tubes, each tube has two led, and a fingernail adjust screw. All you do is adjust each screw which is flush with the case covers on the side of the amp so that both green an red LED are not alight for bias to be right. If there's slight green showing, bias Ia is OK but a little low, and if red bias is a little high, and perhaps needs a turn down. Fixed bias means you can have very variable bias if the mains voltage changes and if that change id say 5%, say from 240Vac to 252Vac as I often se here, the B+ can rise +5% from say +480Vdc to +504Vdc, and in UL or triode amps the Ea = Eg2, and Ra of a 6550 is maybe 1,200 ohms at idle so 24Vdc extra Ea volts causes 20mA of extra Ia. So if you had tubes biased for 30W with +480V, Ia might be 63mA, then with 504Vdc, Ia = 83mA, and Pda becomes 42 watts, and damned dangerously close to tube meltdown. So I always bias tubes fairly low rather than high because I know smoke happens too often the other way. Why to ARC and Manley Labs and VTL and Quad, Leak, and all the many others end up on my bench? Its because of smoke, clouds of it, and angry customers because they trusted some jerk. I de-jerkize the designs, and then ahhhhhhh, moooozzzic, and never any smoke. That's a lot of words but nothing to contradict the point it isn't "ZERO user adjustment." In some amps I have cathode bias *and* bias balance adjustment, see "Reformed Quad II schematic, March 2006," about 1/2 way down the page at http://www.turneraudio.com.au/quad2powerampmods.html gives all the active Ia dc balancing anyone would want for hi-fi becaue if theamp make a max of say 40W class AB, and 20W in class A before moving to AB, then that's enough class A power for 99% of listeners and the amp just does not de-balance Oh, yeah, sure. Tubes never drift and a retube will be 'just like' the ones you replaced. I didn't say tubes don't drift. You imply such with "does not de-balance." They do. Then the amp *will* 'de-balance'. Or else you wouldn't be 'bragging' "because you ** see the LED ** all the time while using the amp." And a lot in the life some owners expect out of them. Designers of amps *other than myself* seem to put a great big effort into minimizing tube life and upsetting customer music. Of course they do. Make any amp using schematics from my website and you'll get a longer life from tubes and know what they are up to all the time you have the amp turned on. Peace of mind. because there are no RC networks on each cathode. Bias unbalance could come from serious clipping when the coupling caps charge up due to grid current, but by that point most listeners are being deafened, or there is a serious amp problem. The unblance will register in the LEDs used to monitor the balance. So a shorted speaker lead is easily spotted, or bias failure. See the ideas and design detials at http://www.turneraudio.com.au/5050Integrated.html My ideas Now comes the "I am a genius" Turner advertisement. Got a better alternative? Yeah. Simply talk about the subject. I am, and I can say to people, go to my website and compare whatever you plan to use with what I have. I'm just an average bloke. Definately not a ****ing genius. Just that everyone else is incompetent. How on earth can you make such a stupid conclusion? Everyone else is not incompetent. Many designers are incompetent, but not all. Many designers are compelled to obey the edicts of the bean counters and include only a fraction of the parts they think are really needed to make the circuit work best. Bean counters tell designers to reduce quality to that of everyone else, and to never ever give Joe Public a really great deal. All companies collude together on this quality issue; there MUST be inbuilt shortcomings and defects to make products fail some time after the warranty period expires so consumers will come back for more after a breakdown. on this have not been improved by anyone, or used commercially by anyone because most designers of commercial amps are dumbasses, stubborn, and in love with absurd complexity which are monuments to their ego but not to fine music, and they don't care if their products blow up and cause much regret amoung the buying public, and bring a bad name to the use of tubes. But with my designs you always get superlative sound and reliablity and ease of bias management and you keep an eye on the bias whenever the amps are on. And if you go out to buy some beer and leave the amps turned on, that will always be the time when a tube decides to fail to a short, ot just conduct far too much Idc, then my amps all just turn themselves off politely instead of burning a house down. Don't ask me if the margin of LF stability is improved with the Blumlein circuit, but probably not. Since it wasn't intended to address strawman 'LF stability' one wonders why you think it should 'improve' it. You guys with simulators might be able to answer that; whether you'd be right needs a check with the breadboard - of course!. What the HELL are you braying about now? He BUILT the damn thing. I didn't see any reported measurements of stability margins. Doesn't mean he didn't 'breadboard' or build the thing, which was the charge. Let's not have a court hearing as to what Al may or may not have done. Besides, why should he 'give you the answers' instead of your lazy ass getting on the bench breadboarding it for yourself so you won't "remain ignorant," as one prominent member in here likes to say. The Garter is so simple that one sees how it works without a breadboard. Speaking of which, I don't see any "reported measurements" on your bias stabilization scheme to compare against Broskie's. The bias stabilization scheme is able to be tweaked so that with sine wave testing up to clipping, Ek rise is less than 10%. I've left it open as to how ppl may wish to tweak it. In fact with low Ia bias, it can work to make the onset of AB increase Ik so that effectively the amp will move further into class A. As the signal eases, so does the bias. Ek will fall, rather than rise, and the large bypass caps smooth out the operation. The distortion profile where the circuit is set up for no reduction of EK after class AB action begins is virtually the same as with fixed bias, so there is a huge improvement THD compared to any amp which does not keep both Ek and Eg1 constant. I don't need to make a detailed public submission about how my circuits may be better than broskies. I am merely saying I wouldn't use much of what Broskie does, and I have my reasons, and they are best evaluated by members of the thinking public if they care to analyse the information at my website and broskie's, or anyone else's. Al just does things he likes; maybe he un-does what he doesn't. But sure, he built it, but that's different from a breadboard design and analysing the hell out of it. The Garter isn't something likely to cause smoke; its dead simple, and little analysis is needed by most diyers as long as it is stable in their amps and no smoke if produced. Nice self contradiction in what, for you, is a relatively short space. He didn't 'analyses the hell out of it' but "little analysis is needed." When I build something, I ***always analyse and test it all***, and look for possible dysfunctions anywhere and everywhere. Goodie for you. I believe in being scientific about developing useful fine sounding circuits. And I look for improvements in technical function and if that's possible without huge added complexity of FB paths then I will explore the possibility fully with a prototype on a board or a chassis. I prefer to use real chassis for "breadboarding". Sometimes the test amp prototype is the very thing you end up selling to a customer, Want to make any bets on which chassis I sell? I rarely make bets. When the Melbourne Cup is run soon I won't be betting. I don't have a clue about horses. so you have to be very careful how you set up and test each stage to really understand what you are doing and why, and that's how I developed the 845 amps. I did have all sorts of ideas about current sinks for the cathodes in the 845 amps I finished last winter, but I thought a well rated and heatsinked 2k2 for each Rk amps would be quite excellent, and it turned out that way. There is active protection in the amps if the bias rises from 72mA per 845 to 100mA. The 845 tube won't meltdown internally if its bias current drifts upwards, but it sure will cause the amp to politely turn off. Even in cathode biased amps, a crook tube will still just "take off" with excessive bias current usually at the end of its life. The high Ek does not hold down the Ia. With CC sink, the Ek would just zoom right up unimpeded by any Rk, and you'd still need the active protection that nobody else bothers to fit into their amps. Again, as I mentioned, you love to switch circuits and then 'argue' about the circuit only you have brought up. *I* told you when you *first* mentioned dual CCSs that they are *not*, as you suggested, 'just as good' as the current mirror yet you *still* pontificate on that very thing. Of course, you made doubly sure it was irrelevant and obtuse by switching to SE vs PP so there's no connection whatsoever to the topic at hand, friggin cathode bias PP BALANCE. Calm down, you'll blow a gasket..... In the case of my 845 amps with 2 x 845 in each channel in parallel, the bias balance in each pair of tubes was good, but its not very important because the amps are SE, and don't need Ia balance to avoid Idc offsets. I was able to use a chinese 845 and a KR845 in the amps and the PO, distortion, bias setting and music quality remained the same. Right, so it's utterly irrelevant to PP balance. And to think I did read the whole thing. The 845 amps I just built do exhibit very close biasing with plain dual R&C cathode bias networks and were they used in a PP circuit the amp woud work just fine. I doubt I'd want to use a Garter or CCS arrangement unless I also incorporated active bias stabilization because the amps would class AB if PP. One can get 100W AB1 of very nice PO from 2 x 845, with maybe 35W of Pda in each tube. Using Garter would means that twice the cathode to grid Vdc must appear across each Rk if the RK is divided into two equal R. Balance at idle would be slightly better than having two R&C networks. Cathode biasing including the Garter method offers some protection against rise in mains Vac After a whole page wandering through circuits and topologies having nothing to do with the topic at hand you finally get around to admitting what I said at the top. I don't remember that you even mentioned what effects mains voltage changes have. Maybe I was too distracted by your other nit-pickings. and hence rise in Ea. The effective Ra at dc in triode or UL amps becomes Ra + [(µ+1 x Rk]. So with an 845, Ra = 2k2 approx. Its quite low, and a small Ea change could cause Pda to rise alarmingly. But Ra with Rk = 2k2 = 13k2, and the Pda just won't increase dangerously with a mains rise. In class A SE amps, cathode bias is terrific. Same goes for class A PP amps. But none are ever actually made. They are always class AB, and in these latter days of design by accountants and marketing experts, usually the PP amp is a glorified high power class B POS PA amp with medium NFB to win the war of watts. Its all that is won. Music is compromised, and the occasional house gets burnt down. Despite your claim of being the only commercial enterprise on the planet that makes decent amps, I am not a commercial enterprise. Again your conclusions about me are plainly wrong. There are better amps around than mine. I don't see many though because maybe they are so reliable they never get to land on my bench for singing lessons and for persusasion towards polite behaviour by means of torture with a hot soldering iron, long nose pliers and side cutters. I do see a lot of crap land on my bench. Often very pretty, highly priced but still crap, like a really beautiful girl who can't cook, sing, or be useful in any way, and who complains about everything all day. Patrick Turner. Perhaps, in the fullness of time. Perhaps. Happy Ears! Al On Oct 9, 1:29?am, flipper wrote: Glad to hear it was a success and I, too, think current balancing makes a difference. Why not also try the current mirror approach? |
#23
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-P amp
In article ,
Patrick Turner wrote: I would not worry about our fiesty fish amoung us. He ain't a shark. He asks some pithy questions and makes some contentious comments, but I believe he's actually fairly good natured most of the time. We all learn from good natured argy bargy. But you have me right, I'm poor, and I'm old, well, damn well not 30 anymore, but I do get over 500 hits a day at my website and somebody finds it interesting. Maybe its not the best tube audio site but together with others we make up a wealth of info on how-to-do issues on the Internet, and its all completely free of charge. Is the Garter better than two plain old R&C bias parts per tube in a substantially class A PP amp? I don't see it being much better. But if it is, then the betterment is marginal, and will only cost 1 extra resistor per tube, for the two series R for each Rk. The only other bother is that the total value of the two bias Rk must be twice what you'd normally have with just two sets of separate R&C. The grid voltage is determined by the voltage half way down the total Rk. So if you have a circuit for class AB1 KT66 with Ea = 365V, and 30V of grid bias voltage across 470ohms, then Rk total will have to be 940 ohms, and the Ek will be +60Vdc, Eg1 = +30Vdc. The doubling of the total value of the Rk definitely will improve natural idle current NFB regulation of Ia by Rk. But you'll dissipate an extra 1.9W per cathode resistance but that's not too much a price to pay. I didn't consider this aspect before. But Ek rise due to class AB charge up currents in OP tubes will be worse, but then if you also used my bias stabilizing circuit with a power transistor and about 3 x R for each OP tube then you'd control the Ek rise with class AB abd keep the circuit balanced at the same time. Maybe this is the best, I don't know the full answer, because so many ways can be thought of for achieving an outcome. I will take this opportunity to elaborate on a couple of issues that may or may not relate directly to the above. First, several people have taken me to task for suggesting that an output transformer with unbalanced DC current in the primary can sound wonderful. In support of my contention, I would note that more than a few audiophiles use Single Ended amplifiers, most of which have unbalanced DC current in the primary of the output transformer. Having the golden ears that they do, these audiophiles wouldn't put up with these amplifiers if they didn't sound wonderful. My suggestion is that if transformers can be designed that produce wonderful sound from Single Ended amplifiers, then it should also be possible to apply the same technology to transformers for push pull amplifiers that can tolerate any desired amount DC current imbalance and still sound wonderful, without the need for "garters", "current mirrors", or other such circuits favored by home brewers. Unfortunately this requires either building your own transformers, or having custom transformers built to spec. My alternate suggestion was to note that, at least in the day, push pull output transformers were specified to tolerate a certain percentage of DC current imbalance without degrading the sound. While the allowed current imbalance in a commercial transformer may not be adequate to what we feel we need to accommodate an unbalanced pair of out put tubes, we can increase the allowable DC imbalance by using an oversized transformer, thereby insuring that we won't loose the wonderful sound. For example consider a commercial output transformer that is designed to retain it's wonderful sound with a 10% imbalance in it's primary currents. If we use this transformer in a lower powered amplifier than it was intended for, say one that operates with half the current, then we have increased the transformers tolerance for DC imbalance to 20% at the new operating point, reducing or eliminating the need for "garters" or whatnot. I have put it on my "to do" list to calculate exactly how much improvement in the balance the "garter" circuit provides. It should also be noted that the lower resistors in the "garter" circuit don't have to be equal in value to the upper resistors. Increasing the value of the lower resistors will increase the balancing effect of the "garter", although it will also increase the other, generally undesirable, side effects produced by the "garter" circuit. As to the balancing effect produced, my rating of the various cathode bias circuits is as follows, from best to worst. 1. Current mirror. 2. Garter circuit. 3. Individual cathode resistors for each tube. 4. One cathode resistor for many, typically four or more tubes. 5. One cathode resistor for two tubes. -- Regards, John Byrns Surf my web pages at, http://fmamradios.com/ |
#24
Posted to rec.audio.tubes
|
|||
|
|||
Bean Counters? (Was: Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-P amp)
In article ,
flipper wrote: On Sat, 11 Oct 2008 13:56:25 GMT, Patrick Turner wrote: Garter only needs 2 extra R per channel with 2 output tubes. Three Rs and one C vs the typical single RC. The cost of 2R and a higher voltage rated electro cap was enough to make a bean counter froth and bubble at the mouth. "Bean counters' don't give a flying fig whether the amp costs one buck or a million and they don't even care if you're in budget. That's your problem and you'll answer for it, not them. See my other post in reply to JB about this issue. I must have missed that one, but in any case for sometime I have been meaning to ask Patrick who these "bean counters" are/were that he is always railing against? It's been over 40 years since I worked in the radio design factory, so my memories are a bit dim, however I don't remember any so called "bean counters" being involved in the designs. What I do remember are guys that I will call "marketing", although they may have actually been product planners or some such. They would produce a description of the radio they wanted and pass it on to the engineers and the guys that did the cabinets. This description included things like the price point they wanted to hit, the required features and functions like the desired control knobs etc., and a few minor technical specifications like the power output for a high end radio. There were also technology features that were to be included so that they could advertise these in the front panel graphics, an example from the time was "FETs". I don't remember any "bean counters" being involved in the design process, it was up to the engineer(s) doing the design to meet the price point, using component pricing information provided by the purchasing department. Can someone please enlighten me about these "bean counters" and who they were? -- Regards, John Byrns Surf my web pages at, http://fmamradios.com/ |
#25
Posted to rec.audio.tubes
|
|||
|
|||
Alan Dower Blumlein "garter" circuits now in triode mode EL34 P-Pamp
On Oct 12, 2:15�pm, John Byrns wrote:
First, several people have taken me to task for suggesting that an output transformer with unbalanced DC current in the primary can sound wonderful.. �In support of my contention, I would note that more than a few audiophiles use Single Ended amplifiers, most of which have unbalanced DC current in the primary of the output transformer. �Having the golden ears that they do, these audiophiles wouldn't put up with these amplifiers if they didn't sound wonderful. Hi RATs! Apples and Pineapples. P-P output transformers get magnetized by unbalanced DC in the primary. SE amplifiers normally have DC in the primary, and have an air gap in the cores to avoid magnetizing. OK, except us experimenters who put a big cap in series with the primary. And a choke in the B+ path. Parafeed is a known trick, it allowed me to build SE circuits with P-P iron. I submit for consideration that if one uses a brilliantly designed P-P output transformer which allows for more DC current imbalance, Garters may improve it yet again But, we wander far afield. The garters help an existing amp sound gooder. Cheap and easy, except the issue of tiny amounts of "spare room" in some products. I added a circuit board beneath my Ella. It ain't pretty, just sounds better. My projects draw lots of laughter from real designers ... I just listen and grin Happy Ears! Al |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Alan Dower Blumlien's garters - the great balanced rest bias circuit! | Vacuum Tubes | |||
SA-DX1040 Technics Receiver switching to "Protected Mode" | Tech | |||
FA: Holland EL34 Double "D" Getter with Brown Base Tubes | Marketplace | |||
FA: Holland EL34 Double "D" Getter with Brown Base Tubes | Vacuum Tubes |