Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Books & tools for transformer design and building
Dear RecAudioTubers,
I`d like to know opinions and hints on the following: - Books / Manuales on audio and power transformer design and building, for tube audio amps. I am interested in a hands-on approach. - Tools to set up a minimun shop to build small batches o hand made transformers. I have previous experience in winding some transformers, but with no formal training and with so-so results. Any help would be appreciated. By the way, I already know the site of Patrick Turner, that has information on general transformer design and building, mainly with reference to Radiotron Designer's Handbook. What I am looking for is an up to date reference that can make use of the advances in magnetics, desirable including toroid cores. My final idea is to evaluate making a audio amplifier construction shop, in the way of a "boutique" dedicated to hi-end made equipment. Patrick and other don`t worry, I live in Chile so my market is far away from yours. Best regards and keep ideas coming. |
#2
|
|||
|
|||
Southern Winds wrote: Dear RecAudioTubers, I`d like to know opinions and hints on the following: - Books / Manuales on audio and power transformer design and building, for tube audio amps. I am interested in a hands-on approach. - Tools to set up a minimun shop to build small batches o hand made transformers. I have previous experience in winding some transformers, but with no formal training and with so-so results. Any help would be appreciated. By the way, I already know the site of Patrick Turner, that has information on general transformer design and building, mainly with reference to Radiotron Designer's Handbook. What I am looking for is an up to date reference that can make use of the advances in magnetics, desirable including toroid cores. My final idea is to evaluate making a audio amplifier construction shop, in the way of a "boutique" dedicated to hi-end made equipment. Patrick and other don`t worry, I live in Chile so my market is far away from yours. I wouldn't have placed the information at my site if I worried ppl might steal it. but if you can wind a smaple of OPT No1 shown at my site exactly as I have specified, then you'd be doing well. Then you will find out how difficult it is, and how competitive the market is and you will have second thougts about botique hi-end audio. Men like myself couldn't ever marry because we simply never generate enough income, and the ladies think we are losers. But its seems that nobody much understands what I have placed there so i don't worry. I have had two chinese winders who spammed me about trannies, then I referred to my site, asked them to wind me a sample and sent it to me for appraisal, then heard nothing, so they are mostly quite incompetent. Nothing I have seen in chinese amplifiers inspires me to give up what i am doing myself. You won't find many folks posting details of the " use of the advances in magnetics, desirable including toroid cores." Plitron, Lundahl, and many many others simply will not tell you how many turns of what size wire on what type of core and how many interleavings in a manner which could be ever easily understood. There are few amateurs willing to spend the time to perfect their toroidal techniques and issue a list of formulas and methods that work. So really its all down to you to wind a few trannies and find out yourself like i did, so lock yourself up in your work shop and don't come out until you have something to show for your time. This means you have to be particularly able to measure and test your own work without prejudice, after carefully layer winding samples on a suitable machine. A toroidal winder is not something easily made yourself, and is going to cost a couple of grand usd even for something from china. E&I trannies do allow far more easy flexibility for the small volume maker, so take my advice, stay away from toroidal transformers. For OPTs, the iron u is very high, and thus easily saturated by out of balance DC bias currents in PP amps. For SE amps, toroidal OPTs are pointless. Amorphous cores offer splendid HF, and there are glowing reviews at the Lundahl site, but they are trying to interest buyers, and the technical details and distortion figures are all missing, so not a single claim is proven, so i assume amorphous cores offer nothing but a an emptier pocket. Best regards and keep ideas coming. Keep up the good work. Patrick Turner. |
#3
|
|||
|
|||
I'm still convinced there is a market for McIntosh unity coupled
transformers, which I think Patrick (and other UK/A/NZ/BritCom winders) will disdain for purely nationalistic if no other reasons, and which just are not that tough to wind. I won't debate the technical pros and cons here, just the market appeal. |
#4
|
|||
|
|||
"Southern Winds" wrote
I`d like to know opinions and hints on the following: - Books / Manuales on audio and power transformer design and building, for tube audio amps.... Menno van der Veen, who is consultant to Plitron and who designed their range of torroids, has published a few papers, some of which are available via Plitron, who has a site somewhere. His book "Modern High-End Valve Amplifiers" published by Elektor has a few chapters on the electrical parameters of transformers and how they affect amplifier performance. It makes a good summary if you are into transfer functions. But how electrical parameters are related to winding details is something you will find very difficult to find. The basics in RDH and other works of that era haven't changed, however. You can put in the characteristics of newer materials and do the same old calculations. Beyond that, precise winding details, especially for toroids, which force a different winding geometry, are what you will have to experiment with, or copy. If you find a good way of doing it you will keep it a secret, like everyone else. ...My final idea is to evaluate making a audio amplifier construction shop, in the way of a "boutique" dedicated to hi-end made equipment. You face a lot of competition from people who already know what they are doing. What advantage do you think there would be in making your own transformers? cheers, Ian |
#6
|
|||
|
|||
The market for replacements is small but for DIYers intent on cloning
and possibly from MI OEMs is at least reasonable. Copy Mac or copy the overrated RDH 4-what's the difference? One is reminded of the basically nasty Tim de Paravicini, who refers to everyone else as "dangerous amateurs" when in fact his work is largely derivative. Personally I think the Marantz and amplifiers built in that 'line of sponsorship' (VTL et seq) sound better for domestic use with efficient speakers, but they won't do what a Mac will in terms of sustained electrical abuse and high power output. It should also be said that the stock Mac uses a mickeymouse power supply and cheap components, so that a determined DIYer could build a better Mac than Mac. |
#7
|
|||
|
|||
On Mon, 18 Apr 2005 03:04:45 GMT, Patrick Turner
wrote: McIntosh are hard to wind; first you need triple insulated hi-temp wire since the cathode and anode windings are wound bifilar, so there is a constant 450v potential between the two windings at all times. Bifilar are a pita to wind. Get a failed one and cut it open to find out how they really did it. Indeed. The McIntosh transformers were bifilar, but otherwise simply (not!-sectioned) wound. This was their great beauty in 1949 -era design thinking. A commercial design with low leakage inductance allowed lower (!) idle currents. Flip side of course is that they're still around to be discussed in 2005. But McIntosh are class AB2 beam tetrode amps which have a lot of NFB applied in the output stage. Many would argue that that's not a negative, but your argument is surely that the implied high drive voltage is challenging. Mc used an elaborate combination of tricks, including positive current feedback, etc. to make the whole thing work from a single power HV DC supply. Not! modern thinking. 'Course, we're not trying to sell amplifiers in 195-something. Very little class A power is produced by a pair of 6550 in a model 275. The open loop distortion without NFB is quite high. Not terrible, but not what anyone would want to do nowadays. Different goals, different choices. Then another lot of global NFB so a total of about 40 dB is applied. Not, of course, long-loop feedback, which was only about ballpark for any other commercially produced amplifier that was stable into *any* load. Your meaning was clear, but generally, perhaps discussions using the term "feedback" could usefully include a description of the exact meaning. It seems to have ballooned somewhat hereabouts. I only mention this because I believe that issues of intrinsic linearity may be important. And that *very* local feedback falls into a gray area (of intrinsic-osity-ness-ish). Good fortune, Chris Hornbeck 6x9=42 April 29 |
#8
|
|||
|
|||
The Macs were cheap to build, and they sounded okay, and the tubes ran
forever, and most importantly people who like them really like them, consider in fact anything else to be lesser. I don't personally think they are the best sounding amps ever built, but believe it or not some do. RDH was published and printed Down Under, and that's a big part of Mr. Turner's interest in it. And he grew up around Brit amplifiers and transformers like Partridges. He does what he does essentially to amuse himself, probably considers real success gauche and crude-tall poppies and all that. |
#9
|
|||
|
|||
Chris Hornbeck wrote: On Mon, 18 Apr 2005 03:04:45 GMT, Patrick Turner wrote: McIntosh are hard to wind; first you need triple insulated hi-temp wire since the cathode and anode windings are wound bifilar, so there is a constant 450v potential between the two windings at all times. Bifilar are a pita to wind. Get a failed one and cut it open to find out how they really did it. Indeed. The McIntosh transformers were bifilar, but otherwise simply (not!-sectioned) wound. This was their great beauty in 1949 -era design thinking. A commercial design with low leakage inductance allowed lower (!) idle currents. Flip side of course is that they're still around to be discussed in 2005. The McI amps had plenty of interleaving between the bifilar a&k windings and the speaker secondaries. afaik, there were 4 P sections each with two bifilar wound a&k windings, and 5 speaker sections. They also used C-cores and were potted. Failure rates were low. But McIntosh are class AB2 beam tetrode amps which have a lot of NFB applied in the output stage. Many would argue that that's not a negative, but your argument is surely that the implied high drive voltage is challenging. One needs 120 vrms at each output tube grid. The distortion of the voltage amp is as large as that of many class A output stages without any NFB. So the 20 db of global NFB gives a similar result to many other amps without NFB in the output stage. Mc used an elaborate combination of tricks, including positive current feedback, etc. to make the whole thing work from a single power HV DC supply. Not! modern thinking. 'Course, we're not trying to sell amplifiers in 195-something. I don't recall seeing positive current FB in their circuits; wasn't it Bogen that used PCNFB? Its an easy way to reduce Ro to zero ohms, or even to a negative figure, ie, voltage output rises with a reduction of load value, but the downside is more thd and instability, so its never ever used these days. Very little class A power is produced by a pair of 6550 in a model 275. The open loop distortion without NFB is quite high. Not terrible, but not what anyone would want to do nowadays. Different goals, different choices. McI did what most SS desingerns now do routinely. Then another lot of global NFB so a total of about 40 dB is applied. Not, of course, long-loop feedback, which was only about ballpark for any other commercially produced amplifier that was stable into *any* load. McI are stable. The low phase shift rate of turnover the output stage with its shirtload of NFB with high Lp and low LL was as good as a well done Williamson, so an additional 20 db of global was possible. Your meaning was clear, but generally, perhaps discussions using the term "feedback" could usefully include a description of the exact meaning. It seems to have ballooned somewhat hereabouts. I only meant series voltage negative feedback. That is the only type used in a 275. There are two loops, a balanced one in the OPT, and the global one. I only mention this because I believe that issues of intrinsic linearity may be important. And that *very* local feedback falls into a gray area (of intrinsic-osity-ness-ish). Triode operation has as much intrinsiticationality as one could want without external loops of NFB. Mc I apply 50% of the tube output voltage in series with the grid input voltage. So the parameters of the tubes are effectively changed by the transformer's CFB application so that one would think that u of the tubes was about 1.0, and Ra was inded a lot lower than a triode's Ra. We can work it out. Ra for 6550 = 15k, Gm = 0.01 A/V, u = 150. These are approximtae figures since all but u change a lot with Ia. With series VNFB applied, Ra' = Ra / ( 1 + [ u x B ] ), where u = amplification factor and B = the fraction of output fed back. So Ra of the McI connection = 15,000 / ( 151 x 0.5 ) = 200 ohms Their OPTs are typically 4k a-a to 8 ohms, or 500:1 ratio so with CFB only the Ro at the speaker terminals is 0.4 ohms, and the global NFB reduces that another 10 times. EAR509 has 3 NFB loops, and doesn't include the OPT sec in the global so Ro is a high 0.5 ohms despite so much NFB being used And then out Tim uses PL509 in mainly class B and tries to wring 120 watts from such crummy tubes which have high open loop thd, thus they need every bit of NFB they an muster, like a transistor amp, so the EAR 509 never sounds as well or measures as well as a McI or anything you or I could make when we try. I have thought that taking the screens of the McI directly to a fixed B+ voltage would effectively make their output tubes work with the equivalent of 50% UL taps. The power output would be 15% lower but I think they'd be a better amp. Patrick Turner. Good fortune, Chris Hornbeck 6x9=42 April 29 |
#11
|
|||
|
|||
On Mon, 18 Apr 2005 05:13:49 GMT, Patrick Turner
wrote: I don't recall seeing positive current FB in their circuits; wasn't it Bogen that used PCNFB? Sorry for not being clearer, I meant in the drivers. McI did what most SS desingerns now do routinely. Ouch. Harsh, but true. So Ra of the McI connection = 15,000 / ( 151 x 0.5 ) = 200 ohms Their OPTs are typically 4k a-a to 8 ohms, or 500:1 ratio so with CFB only the Ro at the speaker terminals is 0.4 ohms, and the global NFB reduces that another 10 times. It's very late here in the Spring World, but doesn't that count the feedback twice? Sorry if I'm too fuzzy. I have thought that taking the screens of the McI directly to a fixed B+ voltage would effectively make their output tubes work with the equivalent of 50% UL taps. The power output would be 15% lower but I think they'd be a better amp. And would allow a higher plate voltage for any particular screen voltage, and an opportunity to regulate the screen voltage. And, a better amp; yeah, you betcha. Chris Hornbeck 6x9=42 April 29 |
#12
|
|||
|
|||
On Mon, 18 Apr 2005 05:13:49 GMT, Patrick Turner
wrote: The McI amps had plenty of interleaving between the bifilar a&k windings and the speaker secondaries. afaik, there were 4 P sections each with two bifilar wound a&k windings, and 5 speaker sections. They also used C-cores and were potted. Sorry, missed this earlier. Let me research Frank Mc.'s papers before I respond. My memory is very different, but often wrong. Potted, certainly, but not-interleaving was specifically a design goal (as my notoriously bad memory has it). Back with more later. Chris Hornbeck 6x9=42 April 29 |
#13
|
|||
|
|||
"Southern Winds" wrote in message
. 18... Dear RecAudioTubers, I`d like to know opinions and hints on the following: - Books / Manuales on audio and power transformer design and building, for tube audio amps. I am interested in a hands-on approach. - Tools to set up a minimun shop to build small batches o hand made transformers. Hi A great cheap way of learning the basics is with Babani's "Build your own coils and transformers". This little book has been in print for decades. It doesn't cost much either (about $8 USD I think). I got mine from a local electronic hobby shop. It covers pretty much everything you need to know about winding your own transformers, including power, choke and audio output types. Cheers. Doug |
#14
|
|||
|
|||
On Mon, 18 Apr 2005 09:04:45 GMT, "Doug Flynn"
wrote: A great cheap way of learning the basics is with Babani's "Build your own coils and transformers". This little book has been in print for decades. It doesn't cost much either (about $8 USD I think). I got mine from a local electronic hobby shop. It covers pretty much everything you need to know about winding your own transformers, including power, choke and audio output types. Hi Doug I've got that book, it's very empirical if you know what I mean. doesn't give a real insight into what is going on. I do have a couple of books that are self published by Robert Wolpert. One for output trannies, the other for the power kind. They come complete with worked examples, they explain the process and the theory really well. hope this helps. Bill. ps. Robert wolpert used to make these transformers for a living. These books [actually they are a couple of folders that are spiral bound, but lets not quibble] are built on a lifetimes experiences of making transformers. |
#15
|
|||
|
|||
Chris Hornbeck wrote: On Mon, 18 Apr 2005 05:13:49 GMT, Patrick Turner wrote: I don't recall seeing positive current FB in their circuits; wasn't it Bogen that used PCNFB? Sorry for not being clearer, I meant in the drivers. What you call positive current FB in the driver tubes driving the outputs is what McIntosh and other engineers call bootstrapping. The 12BH7 plate loads are taken to some point of similarly phased signal on the OPT, to trick the 12BH7 into thinking they have a much higher RL than they actually have. Its not positive current FB. It is, however, positive voltage FB, but fairly benign in this case. Any distortion at the supply end of the bootstrapped RL is divided down by the RL and the 12BH7 Ra, so some thd is fed positively into the amp, and slightly negates the effect of the NFB in the output stage, but it does allow greater linearity of the 12BH7, so the net effect of the bootstrapping makes the amp measure better. And it makes it possible to produce the drive voltages because without the bootstrapping its be impossible to get the voltage swings with a purely resistive load. I would never use bootstrapping from an output stage to a driver stage because its a 3 steps forward one step backward approach, and its better to use passive high impedance loading like a choke or CCS to get a large low thd voltage swing from a driver tube, and thus totally unaffected by the thd of the output stage, or load change effects. But the aproach used by McIntosh allowed a total amount of NFB higher than almost anyone else did with tubes and still gave good stability. McI did what most SS designers now do routinely. Ouch. Harsh, but true. Tim De P designed EAR509 tongue in cheek to show the world SS principles could be applied to tube amps. McIntosh had done it all 30 years before. Start with a rather non linear output stage, and apply local FB. Then use 3 stages of gain and apply other loops of FB to make a total of 44 dB. So Ra of the McI connection = 15,000 / ( 151 x 0.5 ) = 200 ohms Their OPTs are typically 4k a-a to 8 ohms, or 500:1 ratio so with CFB only the Ro at the speaker terminals is 0.4 ohms, and the global NFB reduces that another 10 times. It's very late here in the Spring World, but doesn't that count the feedback twice? Sorry if I'm too fuzzy. Indeed it does count something, but not twice, not really, and one has to be able to think of at least 4 things at once when considering anything in electronics..... The McIntosh have a balanced loop of NFB in the output stage which acts independantly of the global NFB. Just forget the GNFB for a moment. This output stage NFB is a "local" loop in the OPT, where there are two separate centre tapped windings of equal turns each. One winding has its CT grounded, and the ends taken to output tube cathodes, the other winding CT is taken to the B+, and each end to the tube anodes, and the screens of the output tubes taken to the oposite end of the anode winding to give the screen the same signal phase as the cathode signal. The cathode signal at each cathode is exactly 1/2 the total signal across the tube. Consider the voltages present when 75 watts is being produced into 4k "a-a" as we would say with any conventioal amp. 75 watts into 4k = 548 vrms " anode to anode ". In the McI we have 274 vrms acting across each tube and there is +137 vrms at the anode, and -137vrms at the cathode of one output tube, and -137vrms at a and +137vrms at the k of the other. Since the McI is largely a class B amp, the load seen for most of the cycle is 1/4 of RLa-a, or 1/4 of 4k which is 1k. Beam tetrode gain = u x RL / ( RL + Ra ). so 6550 gain = 150 x 1k / ( 15k + 1k ) = 9.375. Now to get the anode and cathode voltage we see we still need a voltage between grid and k to cause the change of a to k voltage which is across the load. So to make 274 vrms, we need 274 / 9.375 vrms = 29vrms. The beginning part of the wave cycle is in class A so the gain is a little higher so in fact Vg-k is about 27 vrms. Now if the cathode swings +137v, the grid must swing +137 + 27vrms to get that total swing a-k of 274v, so input voltage at each grid is thus 164 vrms. So 164 v of input produces a total 274 v of tube output. Gain is actually larger than 1, and is 1.67. The reduction from gain = 10 is 10/1.67 = 5.98 = about 16 dB. Its a heck of a lot higher when no RL is present because the gain without a load is near u, about 150 times, not 10 like it is with the rated load, so without a load the local NFB in the output stage is about 37 dB. The output stage of the McI allows the driver amp to produce a nearly constant amplitude driver signal to the output stage where the bulk of the error correction takes place. The earlier post calculated the Ro of the amp with no global NFB as about 0.4 ohms, and this is a very roughly calculated figure. Maybe in fact it is slightly more than 0.4 ohms, more like an ohm. Its a decent result since without any local CFB from the OPT the 6550 beam tetrode gives Ro = 15,000 / 125 = 120 ohms. When in class B only one tube is connected to the 8 ohm load from half the OPT since one tube is cut off whilethe other one soldiers on alone being turned on to produce the power. While in class A the two tubes conduct and the Ro is thus lower, but not much because the Ra is higher when Ia is lower. What we get with the McIntosh is an amp where the output stage acts as if it is a triode like one where the u = about 2.0, and the Ra is about 200 ohms. Gm remains the same 0.01A/V. Then there is the 20 dB of global NFB from the special winding on the OPT devoted solely to give a low voltage FB voltage to feed back to the cathode of V1. This is a good technique of the McI because it allows the FB to be set up independantly of any load considerations; the amount of FB is constant for the range of loads able to be matched from the available secondary windings. So we have a lot of gain between the amp input and output stage grids where a total of 328 vrms is needed g-g to work the output stage. My Tremain Audio Encyclopedia shows a 12AX7 SET input tube followed by a 12AU7 LTP followed by a 12BH7 balanced amp giving a gain of about 60 x 14 x 10 = 8,400. So only 0.04vrms is needed for full power without global NFB. Unfortunately, Tremain has no working signal voltages shown along the schematic, leaving mortals like us to argue about it 40 years later, and since I dunno what voltage is produced by the NFB winding, I can only assume the amp is set for 1 v to make the rated output, so about 26 db of global NFB seems to be used. So 26dB of global plus 16dB of local dB in the output totals 42 dB if I have counted correctly. But its not like 42 db has been applied globally. McI knew that much global FB around all those stages including an OPT no matter how good would be impossible without instability. But the two loops of NFB allow the stability to be OK. Its an early form of "nested feedback", a technique much later re-invented by one Professor Ed Cherry, who used the technique to get astonishingly low thd in SS amps in the 1980s. The McI needs all the NFB is can muster because it has to make this enormously high drive voltage, about as much as a 10 watt power amp does with a pair of 6BQ5 with an 11 k a-a load. While everyone may think its a high drive voltage, remember that triode tubes are intrinsically linear, and they are **voltage** operated devices, with ****y little current changes, and they are very good at all this. McI boasts about 0.2% of thd at 70 watts and so without the 26 dB of global NFB the thd would be about 4%, with about 2% of that produced by the output stage, so in producing the 328 vrms g-g drive voltage only 2% thd must be being made. by the driver stage. I have thought that taking the screens of the McI directly to a fixed B+ voltage would effectively make their output tubes work with the equivalent of 50% UL taps. The power output would be 15% lower but I think they'd be a better amp. And would allow a higher plate voltage for any particular screen voltage, and an opportunity to regulate the screen voltage. And, a better amp; yeah, you betcha. Let us not count our chickens before the eggs have been laid. I haven't done such a thing to a McI. The output tube gain would drop if the screens were not used as they are, so 4 dB higher drive voltage is needed. But 60 watts from a pair of 6550 would be achievable. One should remember that the McI would have a linear reduction of distortion with output voltage. Most folks would use about an average of 1 watt most days, so the voltage is 0.11 times what it is at 70 watts, so thd should measure 0.02% at a watt as the amp is. This is the same as I get with all my amps including the SE types with much lower total amount of NFB, and at least one stage less of gain. McIntosh appealed to engineer types, who like the lower Ro they had compared to the rest, and this was simply due to a lotta FB. Having Ro = 0.1 ohms or whatever does not give any better sound than if Ro was say 0.8 ohms, where the DF would be 10, imho. Patrick Turner. Chris Hornbeck 6x9=42 April 29 |
#16
|
|||
|
|||
Chris Hornbeck wrote: On Mon, 18 Apr 2005 05:13:49 GMT, Patrick Turner wrote: The McI amps had plenty of interleaving between the bifilar a&k windings and the speaker secondaries. afaik, there were 4 P sections each with two bifilar wound a&k windings, and 5 speaker sections. They also used C-cores and were potted. Sorry, missed this earlier. Let me research Frank Mc.'s papers before I respond. My memory is very different, but often wrong. Potted, certainly, but not-interleaving was specifically a design goal (as my notoriously bad memory has it). Back with more later. McIntosh used the bifilar primary windings to couple the halves devoted to each tube closely together to give freedom from switching distortion in class AB amps with very low bias current. I have built many amps, and tested many built by others and none have bifilar primary windings not once have I ever seen the oscilligrams that I have seen reported by those who have said you get switching distortions without close primary half couplings. But all the amp Iv'e measured had at least some idle current, and even when over biasing the amp for class C there was not the severe switching spikes and back emfs caused by inductances suddenly being switched right off. The OPTs I make myself seem to be able to be used quite OK with class AB without switching distortion spikes. Sure there is some extra distortions which start at the crossover point between A and AB, and thse manifest themself usually with the appearance of 5H as well as more 3H. Its actually very difficult to couple half primaries together closely in a conventional OPT, since it would involve having adjacent anode layers from each tube close to each other and between sections of secondary, and the capacitance betwen close layers of primaries from each of two tubes with opposite phased signals would be immense. McIntosh found a way around what he considered a problem. There was no interleaving of McI 1/2 primaries; they were coupled far closer than interleaving by means of bifilar winding. The Circlotron also achieves a very similar outcome by floating two B+ supplies and having one OPT primary winding instead of having McI's one fixed B+ and two primary windings with a CT each. But all these primary winding techniques still need an interleaved secondary with the primary. Without the interleaving of the secondary amoungst the primaries no matter what way they are arranged, there isn't close enough coupling from P to S and the phase shift is high and the amount of global NFB applyable is low. McI used a separate NFB winding to also over come this limitation afaik. Patrick Turner. Chris Hornbeck 6x9=42 April 29 |
#17
|
|||
|
|||
Doug Flynn wrote: "Southern Winds" wrote in message . 18... Dear RecAudioTubers, I`d like to know opinions and hints on the following: - Books / Manuales on audio and power transformer design and building, for tube audio amps. I am interested in a hands-on approach. - Tools to set up a minimun shop to build small batches o hand made transformers. Hi A great cheap way of learning the basics is with Babani's "Build your own coils and transformers". This little book has been in print for decades. It doesn't cost much either (about $8 USD I think). I got mine from a local electronic hobby shop. It covers pretty much everything you need to know about winding your own transformers, including power, choke and audio output types. Cheers. Doug I have that book Doug, and its a good reference, but for the best design of chokes and OPTs you can't better the design theory in RDH4 for OPTs and the choke design method by Hanna in the same book. Patrick Turner. |
#18
|
|||
|
|||
Patrick Turner wrote:
Chris Hornbeck wrote: On Mon, 18 Apr 2005 05:13:49 GMT, Patrick Turner wrote: I don't recall seeing positive current FB in their circuits; wasn't it Bogen that used PCNFB? Sorry for not being clearer, I meant in the drivers. What you call positive current FB in the driver tubes driving the outputs is what McIntosh and other engineers call bootstrapping. The 12BH7 plate loads are taken to some point of similarly phased signal on the OPT, to trick the 12BH7 into thinking they have a much higher RL than they actually have. Its not positive current FB. It is, however, positive voltage FB, but fairly benign in this case. Any distortion at the supply end of the bootstrapped RL is divided down by the RL and the 12BH7 Ra, so some thd is fed positively into the amp, and slightly negates the effect of the NFB in the output stage, but it does allow greater linearity of the 12BH7, so the net effect of the bootstrapping makes the amp measure better. And it makes it possible to produce the drive voltages because without the bootstrapping its be impossible to get the voltage swings with a purely resistive load. I would never use bootstrapping from an output stage to a driver stage because its a 3 steps forward one step backward approach, and its better to use passive high impedance loading like a choke or CCS to get a large low thd voltage swing from a driver tube, and thus totally unaffected by the thd of the output stage, or load change effects. But the aproach used by McIntosh allowed a total amount of NFB higher than almost anyone else did with tubes and still gave good stability. McI did what most SS designers now do routinely. Ouch. Harsh, but true. Tim De P designed EAR509 tongue in cheek to show the world SS principles could be applied to tube amps. McIntosh had done it all 30 years before. Start with a rather non linear output stage, and apply local FB. Then use 3 stages of gain and apply other loops of FB to make a total of 44 dB. So Ra of the McI connection = 15,000 / ( 151 x 0.5 ) = 200 ohms Their OPTs are typically 4k a-a to 8 ohms, or 500:1 ratio so with CFB only the Ro at the speaker terminals is 0.4 ohms, and the global NFB reduces that another 10 times. It's very late here in the Spring World, but doesn't that count the feedback twice? Sorry if I'm too fuzzy. Indeed it does count something, but not twice, not really, and one has to be able to think of at least 4 things at once when considering anything in electronics..... The McIntosh have a balanced loop of NFB in the output stage which acts independantly of the global NFB. Just forget the GNFB for a moment. This output stage NFB is a "local" loop in the OPT, where there are two separate centre tapped windings of equal turns each. One winding has its CT grounded, and the ends taken to output tube cathodes, the other winding CT is taken to the B+, and each end to the tube anodes, and the screens of the output tubes taken to the oposite end of the anode winding to give the screen the same signal phase as the cathode signal. The cathode signal at each cathode is exactly 1/2 the total signal across the tube. Consider the voltages present when 75 watts is being produced into 4k "a-a" as we would say with any conventioal amp. 75 watts into 4k = 548 vrms " anode to anode ". In the McI we have 274 vrms acting across each tube and there is +137 vrms at the anode, and -137vrms at the cathode of one output tube, and -137vrms at a and +137vrms at the k of the other. Since the McI is largely a class B amp, the load seen for most of the cycle is 1/4 of RLa-a, or 1/4 of 4k which is 1k. Beam tetrode gain = u x RL / ( RL + Ra ). so 6550 gain = 150 x 1k / ( 15k + 1k ) = 9.375. Now to get the anode and cathode voltage we see we still need a voltage between grid and k to cause the change of a to k voltage which is across the load. So to make 274 vrms, we need 274 / 9.375 vrms = 29vrms. The beginning part of the wave cycle is in class A so the gain is a little higher so in fact Vg-k is about 27 vrms. Now if the cathode swings +137v, the grid must swing +137 + 27vrms to get that total swing a-k of 274v, so input voltage at each grid is thus 164 vrms. So 164 v of input produces a total 274 v of tube output. Gain is actually larger than 1, and is 1.67. The reduction from gain = 10 is 10/1.67 = 5.98 = about 16 dB. Its a heck of a lot higher when no RL is present because the gain without a load is near u, about 150 times, not 10 like it is with the rated load, so without a load the local NFB in the output stage is about 37 dB. The output stage of the McI allows the driver amp to produce a nearly constant amplitude driver signal to the output stage where the bulk of the error correction takes place. The earlier post calculated the Ro of the amp with no global NFB as about 0.4 ohms, and this is a very roughly calculated figure. Maybe in fact it is slightly more than 0.4 ohms, more like an ohm. Its a decent result since without any local CFB from the OPT the 6550 beam tetrode gives Ro = 15,000 / 125 = 120 ohms. When in class B only one tube is connected to the 8 ohm load from half the OPT since one tube is cut off whilethe other one soldiers on alone being turned on to produce the power. While in class A the two tubes conduct and the Ro is thus lower, but not much because the Ra is higher when Ia is lower. What we get with the McIntosh is an amp where the output stage acts as if it is a triode like one where the u = about 2.0, and the Ra is about 200 ohms. Gm remains the same 0.01A/V. Then there is the 20 dB of global NFB from the special winding on the OPT devoted solely to give a low voltage FB voltage to feed back to the cathode of V1. This is a good technique of the McI because it allows the FB to be set up independantly of any load considerations; the amount of FB is constant for the range of loads able to be matched from the available secondary windings. So we have a lot of gain between the amp input and output stage grids where a total of 328 vrms is needed g-g to work the output stage. My Tremain Audio Encyclopedia shows a 12AX7 SET input tube followed by a 12AU7 LTP followed by a 12BH7 balanced amp giving a gain of about 60 x 14 x 10 = 8,400. So only 0.04vrms is needed for full power without global NFB. Unfortunately, Tremain has no working signal voltages shown along the schematic, leaving mortals like us to argue about it 40 years later, and since I dunno what voltage is produced by the NFB winding, I can only assume the amp is set for 1 v to make the rated output, so about 26 db of global NFB seems to be used. So 26dB of global plus 16dB of local dB in the output totals 42 dB if I have counted correctly. But its not like 42 db has been applied globally. McI knew that much global FB around all those stages including an OPT no matter how good would be impossible without instability. But the two loops of NFB allow the stability to be OK. Its an early form of "nested feedback", a technique much later re-invented by one Professor Ed Cherry, who used the technique to get astonishingly low thd in SS amps in the 1980s. The McI needs all the NFB is can muster because it has to make this enormously high drive voltage, about as much as a 10 watt power amp does with a pair of 6BQ5 with an 11 k a-a load. While everyone may think its a high drive voltage, remember that triode tubes are intrinsically linear, and they are **voltage** operated devices, with ****y little current changes, and they are very good at all this. McI boasts about 0.2% of thd at 70 watts and so without the 26 dB of global NFB the thd would be about 4%, with about 2% of that produced by the output stage, so in producing the 328 vrms g-g drive voltage only 2% thd must be being made. by the driver stage. I have thought that taking the screens of the McI directly to a fixed B+ voltage would effectively make their output tubes work with the equivalent of 50% UL taps. The power output would be 15% lower but I think they'd be a better amp. And would allow a higher plate voltage for any particular screen voltage, and an opportunity to regulate the screen voltage. And, a better amp; yeah, you betcha. I did exactly that in an amp published in AudioXpress magazine August 2004. Using a PP pair of 6LU8's with 400v plate & 320v screen it manages 37 watts at clipping. The total cost is about 200.00 USD. More recently an experimental amp of this kind managed a DF of 5 with no loop NFB at all. The circuit is a winner for a number of reasons. John Stewart Let us not count our chickens before the eggs have been laid. I haven't done such a thing to a McI. The output tube gain would drop if the screens were not used as they are, so 4 dB higher drive voltage is needed. But 60 watts from a pair of 6550 would be achievable. One should remember that the McI would have a linear reduction of distortion with output voltage. Most folks would use about an average of 1 watt most days, so the voltage is 0.11 times what it is at 70 watts, so thd should measure 0.02% at a watt as the amp is. This is the same as I get with all my amps including the SE types with much lower total amount of NFB, and at least one stage less of gain. McIntosh appealed to engineer types, who like the lower Ro they had compared to the rest, and this was simply due to a lotta FB. Having Ro = 0.1 ohms or whatever does not give any better sound than if Ro was say 0.8 ohms, where the DF would be 10, imho. Patrick Turner. Chris Hornbeck 6x9=42 April 29 |
#19
|
|||
|
|||
|
#20
|
|||
|
|||
Yves wrote: "Patrick Turner" a écrit dans le message de news: ... [ . . . ] The Circlotron also achieves a very similar outcome by floating two B+ supplies and having one OPT primary winding instead of having McI's one fixed B+ and two primary windings with a CT each. I find this solution much more "elegant" ! The overhead in PSU looks easier to manage than the one at the OPT I suppose that your magistral description about McI few posts ahead applies to Circlotron as well ? You need a highish voltage drive for the output grids of a Circlotron. if the PS are each +470v, the tube load seen by each of two 6550 tubes is the same as in a McI, the voltages across the tubes will be the same. Probably it is easier to have two HT windings on the PS tranny and two extra elcaps and an extra choke than go to all the trouble of the McI OPT. Patrick Turner. Yves. But all these primary winding techniques still need an interleaved secondary with the primary. Without the interleaving of the secondary amoungst the primaries no matter what way they are arranged, there isn't close enough coupling from P to S and the phase shift is high and the amount of global NFB applyable is low. McI used a separate NFB winding to also over come this limitation afaik. Patrick Turner. |
#21
|
|||
|
|||
On Mon, 18 Apr 2005 12:17:34 GMT, Patrick Turner
wrote: What you call positive current FB in the driver tubes driving the outputs is what McIntosh and other engineers call bootstrapping. Its not positive current FB. It is, however, positive voltage FB, but fairly benign in this case. You're right; it's voltage-sensitive feedback, but applied in parallel to signal, rather than series, so should still be called voltage feedback. Bootstrapping is as good a name as any; I shoulda just used it. The earlier post calculated the Ro of the amp with no global NFB as about 0.4 ohms, and this is a very roughly calculated figure. Maybe in fact it is slightly more than 0.4 ohms, more like an ohm. Its a decent result since without any local CFB from the OPT the 6550 beam tetrode gives Ro = 15,000 / 125 = 120 ohms. When in class B only one tube is connected to the 8 ohm load from half the OPT since one tube is cut off whilethe other one soldiers on alone being turned on to produce the power. While in class A the two tubes conduct and the Ro is thus lower, but not much because the Ra is higher when Ia is lower. What we get with the McIntosh is an amp where the output stage acts as if it is a triode like one where the u = about 2.0, and the Ra is about 200 ohms. Gm remains the same 0.01A/V. Yeah, you're counting it twice. If you'd recalculated the Gm, I'd start to agree. My Tremain Audio Encyclopedia shows a 12AX7 SET input tube followed by a 12AU7 LTP followed by a 12BH7 balanced amp Later designs had direct-coupled cathode follower drivers for the output valves. Telefunken 12AX7's run at over 400 volts last for more than 40 years (wow). Let us not count our chickens before the eggs have been laid. I haven't done such a thing to a McI. The output tube gain would drop if the screens were not used as they are, so 4 dB higher drive voltage is needed. 4 dB? Sounds high... but not a deal-breaker. In America, Uncie Ned is threatening to release a smaller cored version of the Dyna 441 transformer with 43% tertiary winding, giving a total primary Z about 6600 ohms, for about $110US. These designs may get interesting. Thanks, Chris Hornbeck 6x9=42 April 29 |
#22
|
|||
|
|||
On Tue, 19 Apr 2005 00:40:35 GMT, Chris Hornbeck
wrote: tertiary winding, giving a total primary Z about 6600 ohms, Sorry, 8800 ohms. Jeez, can't get anything right today. Chris Hornbeck 6x9=42 April 29 |
#23
|
|||
|
|||
Chris Hornbeck wrote: On Mon, 18 Apr 2005 12:17:34 GMT, Patrick Turner wrote: What you call positive current FB in the driver tubes driving the outputs is what McIntosh and other engineers call bootstrapping. Its not positive current FB. It is, however, positive voltage FB, but fairly benign in this case. You're right; it's voltage-sensitive feedback, but applied in parallel to signal, rather than series, so should still be called voltage feedback. Bootstrapping is as good a name as any; I shoulda just used it. Bootstrapping is an awkward term adopted by engineers to describe el cheapo efforts to increase amp gain. The u follower topology with two triodes was once always called a bootstrapped follower, and the word bootstrapping means generally to try to pick oneself up by tugging at one's bootstraps, which is as useless a way as any to try to cause levitation. In a bootstrapped preamp stage with two triodes the larger the value of the resistor between top triode's cathode and bottom triode's anode, and the higher the load value connected to the top tube cathode, the more "bootstrapping" you get, and the voltage change across the K-A resistor gets lower, and the bottom tube sees a load tending toward a CCS. But the load can't ever be more than {Ra + [ ( u + 1 ) x Rk ]} where the two tubes are matched triodes of the same u. So if you have 6SN7, and you have say 15k between K and A, the max load seen by the bottom tube = approx 325k But its high enough to make the gain of the bottom tube = 19.4, very close to 20, which is the u. So if you swing 32vrms at the bottom tube anode you get only 0.1 mA of Ia change, and if you plot this much change across the plate curves on a 6SN7, you will see just how linear a triode can be. The earlier post calculated the Ro of the amp with no global NFB as about 0.4 ohms, and this is a very roughly calculated figure. Maybe in fact it is slightly more than 0.4 ohms, more like an ohm. Its a decent result since without any local CFB from the OPT the 6550 beam tetrode gives Ro = 15,000 / 125 = 120 ohms. When in class B only one tube is connected to the 8 ohm load from half the OPT since one tube is cut off whilethe other one soldiers on alone being turned on to produce the power. While in class A the two tubes conduct and the Ro is thus lower, but not much because the Ra is higher when Ia is lower. What we get with the McIntosh is an amp where the output stage acts as if it is a triode like one where the u = about 2.0, and the Ra is about 200 ohms. Gm remains the same 0.01A/V. Yeah, you're counting it twice. If you'd recalculated the Gm, I'd start to agree. But if we apply the revised u and Ra for the McI set up, we could say that for a 1k load, gain with the local CFB, A' = 2 x 1,000 / ( 1,000 + 200 ) = 1.67. gm = u / Ra, so here the revised gm = 2 / 200 = 0.01 A/V, so gm doesn't change. gm = u / Ra u = Ra x gm. so gain can be expressed as = Ra x gm x RL / ( RL + Ra ) when Ra is low compared to RL, we cannot just say gain = gm x RL, which gives the approximate gain for a beam tetrode or pentode because the Ra is so high. My Tremain Audio Encyclopedia shows a 12AX7 SET input tube followed by a 12AU7 LTP followed by a 12BH7 balanced amp Later designs had direct-coupled cathode follower drivers for the output valves. Telefunken 12AX7's run at over 400 volts last for more than 40 years (wow). I have not referred to the CF buffers after the 12BH7 because the CF gain is just under 1.0, and has an insignificant value for gain calculations. One wonders why McI chose 12AX7 rather than say 12AU7, which would allow higher grid currents to the 6550, but no need to have such current ability. Just as long as the X7 just does the job, no need to have a driver capable of sending too much current in a fault situation. But yeah, they last OK. Let us not count our chickens before the eggs have been laid. I haven't done such a thing to a McI. The output tube gain would drop if the screens were not used as they are, so 4 dB higher drive voltage is needed. 4 dB? Sounds high... but not a deal-breaker. Maybe not. If ther screens are taken to a B+ fixed voltage, the gain is the same at UL with 50% taps, and probably 6 instead of about the 10 I calculated. The output voltage would be limited down a bit compared to beam tetrode, so I guess the drive g-g with the CFB would only increase marginally. In America, Uncie Ned is threatening to release a smaller cored version of the Dyna 441 transformer with 43% tertiary winding, giving a total primary Z about 6600 ohms, for about $110US. These designs may get interesting. It wouldn't be a McI based design then. I have found little benefit with running UL amps with screens at lower supply voltage than the anode supply. It might be a way of increasing the class B part of the AB power a lot, by swinging a larger voltage across larger load. Presumably one could have 800v for the anode supply and 400v for the screen supply. Say you had 650 peak v swing at each anode for a load of 7k a-a, then a-a swing is 919 vrms a-a, so you get about 120 watts from a pair of 6550. If the screen voltage swing was 43%, then expect 279 pk v swing at each screen, and maybe it works OK. Almost no pure class A power though. TT21 would be a better tube for this sort of thing. I prefer to use the tubes for far less maximum AB power. Since I don't like running 6550 at more than 30 pD, and since efficiency in class A UL or CFB is about 40%, 12 watts from a tube of pure class A is all that can be had so it takes a quad of 6550 to make around 50 watts of pure class A There is no need to use more than +500v for Ea, since the class A efficiency does not get much better with Ea above 500v. In my 8585 there is 12.5% of the total primary voltage fed back as CFB. Screens are taken to +330v fixed, and Ea = 470v. THD is lower than UL in all modes, class A or UL and spectra is good. Effective Ra is lower than UL, and gain with the CFB is about the same as triode. The 300 watt amps I made use 20% of the primary voltage as CFB with screens all taken to a fixed supply, and drive voltage is up to 150 vrms g-g, and very easy done at very low thd with EL84 in triode with a choke plus R load to each side in an LTP. Since the OPTs have so much interleaving the bandwidth is 270 kHz, and this is quite sufficient to allow easy tailoring of the rest of the amp to use an additional 8 db of global NFB with excellent stability even into any value of capacitor load. Patrick Turner. Thanks, Chris Hornbeck 6x9=42 April 29 |
#24
|
|||
|
|||
I maintain there is at least another 20 to 30 years' in the state of
the art from RDH4. It essentially documents best practices circa WWII. Peerless, UTC Linear Standard, Triad, and others were all well more advanced than RDH4 describes. |
#25
|
|||
|
|||
On Tue, 19 Apr 2005 02:53:24 GMT, Patrick Turner
wrote: So if you swing 32vrms at the bottom tube anode you get only 0.1 mA of Ia change, and if you plot this much change across the plate curves on a 6SN7, you will see just how linear a triode can be. Damnstraight. Despite some local press, hot-cathode devices have spectacularly linear transfer curves. And, devices with truely linear transfer curves retain linearity into reactive loads, something not discussed much these days. If you'd recalculated the Gm, I'd start to agree. But if we apply the revised u and Ra for the McI set up, we could say that for a 1k load, Wait; lets start here. 1K ohm? 4 dB? Sounds high... but not a deal-breaker. Maybe not. If ther screens are taken to a B+ fixed voltage, the gain is the same at UL with 50% taps, and probably 6 instead of about the 10 I calculated. The output voltage would be limited down a bit compared to beam tetrode, so I guess the drive g-g with the CFB would only increase marginally. Perzactly. Maybe a dB or so. It wouldn't be a McI based design then. Yeah, more like a QUAD, maybe. I have found little benefit with running UL amps with screens at lower supply voltage than the anode supply. The opportunity to regulate G2 voltage might prove interesting; remains to be seen. I'm a triode hardcore asshole, personally, so don't have any practical experience, recently enough to matter. It might be a way of increasing the class B part of the AB power a lot, by swinging a larger voltage across larger load. Presumably one could have 800v for the anode supply and 400v for the screen supply. Say you had 650 peak v swing at each anode for a load of 7k a-a, then a-a swing is 919 vrms a-a, so you get about 120 watts from a pair of 6550. If the screen voltage swing was 43%, then expect 279 pk v swing at each screen, and maybe it works OK. Almost no pure class A power though. TT21 would be a better tube for this sort of thing. In America, the 1625 is the last(?) overlooked made-before- ye-were-born 6L6 family bottle. Like all these old RF beam tubes, G2 voltage is a really tricky issue. Never seen a TT21; same issues? The 300 watt amps I made use 20% of the primary voltage as CFB with screens all taken to a fixed supply, and drive voltage is up to 150 vrms g-g, and very easy done at very low thd with EL84 in triode with a choke plus R load to each side in an LTP. Sounds just right. Chris Hornbeck 6x9=42 April 29 |
#26
|
|||
|
|||
|
#27
|
|||
|
|||
Chris Hornbeck wrote: On Tue, 19 Apr 2005 02:53:24 GMT, Patrick Turner wrote: So if you swing 32vrms at the bottom tube anode you get only 0.1 mA of Ia change, and if you plot this much change across the plate curves on a 6SN7, you will see just how linear a triode can be. Damnstraight. Despite some local press, hot-cathode devices have spectacularly linear transfer curves. And, devices with truely linear transfer curves retain linearity into reactive loads, something not discussed much these days. If you'd recalculated the Gm, I'd start to agree. But if we apply the revised u and Ra for the McI set up, we could say that for a 1k load, Wait; lets start here. 1K ohm? 1k is the load seen by each tube in a McI amp while it works in classB. 4 dB? Sounds high... but not a deal-breaker. Maybe not. If ther screens are taken to a B+ fixed voltage, the gain is the same at UL with 50% taps, and probably 6 instead of about the 10 I calculated. The output voltage would be limited down a bit compared to beam tetrode, so I guess the drive g-g with the CFB would only increase marginally. Perzactly. Maybe a dB or so. It wouldn't be a McI based design then. Yeah, more like a QUAD, maybe. I have found little benefit with running UL amps with screens at lower supply voltage than the anode supply. The opportunity to regulate G2 voltage might prove interesting; remains to be seen. I'm a triode hardcore asshole, personally, so don't have any practical experience, recently enough to matter. Nothing wrong with triodes. If the tubes in the McI are trioded, the amp will work in triode AB2, but maybe a lot more grid current is needed, so 12AU7 instead of 12AX7 as the buffer would be better. 6L6 in triode AB2 can make nearly double the po compared to AB1. It might be a way of increasing the class B part of the AB power a lot, by swinging a larger voltage across larger load. Presumably one could have 800v for the anode supply and 400v for the screen supply. Say you had 650 peak v swing at each anode for a load of 7k a-a, then a-a swing is 919 vrms a-a, so you get about 120 watts from a pair of 6550. If the screen voltage swing was 43%, then expect 279 pk v swing at each screen, and maybe it works OK. Almost no pure class A power though. TT21 would be a better tube for this sort of thing. In America, the 1625 is the last(?) overlooked made-before- ye-were-born 6L6 family bottle. Like all these old RF beam tubes, G2 voltage is a really tricky issue. Never seen a TT21; same issues? TT21 is a KT88 with an anode top cap. No arcing between pin 3 and the heaters. Pin 4 is still the screen, and in RFamps is often at a fixed voltage of 300v only, with say 800v quite OK on the anode. Its a good tube the TT21. The 300 watt amps I made use 20% of the primary voltage as CFB with screens all taken to a fixed supply, and drive voltage is up to 150 vrms g-g, and very easy done at very low thd with EL84 in triode with a choke plus R load to each side in an LTP. Sounds just right. EL84 in triode is about equal to 5 halves of a 6SN7 in parallel, u = 20, Ra = 2.2k at 14 mA of Ia, and even lower at Ia = 25 mA. EL86 has u = 11, Ra = 1.4k at 14 mA. Its also a Prince of Linearity. Patrick Turner. Chris Hornbeck 6x9=42 April 29 |
#28
|
|||
|
|||
Patrick
I'm a triode hardcore asshole, personally, so don't have any practical experience, recently enough to matter. Nothing wrong with triodes. If the tubes in the McI are trioded, the amp will work in triode AB2, but maybe a lot more grid current is needed, so 12AU7 instead of 12AX7 as the buffer would be better. 6L6 in triode AB2 can make nearly double the po compared to AB1. I've done many measurements on this circuit while the OP was connected pentode, 50 % UL & triode. The bottom line is that triode operation offers no advantage at all. That is because most of the output stage gain is used in local NFB. The spread of output stage gain from triode to pentode operation is less than 10% & so is the resultant DF. DF's of 5 or so are no problem at all without global NFB, even while running the OP as pentodes. UL offers the possibility to run the plates at higher voltage than the screens. As a result, more power is possible. The circuits I've built sofar exhibit IMD's in the order of -50 db at all levels up to within one db of clipping. Try that with your SET amp! Cheers, John Stewart |
#29
|
|||
|
|||
On Tue, 19 Apr 2005 16:42:49 GMT, Patrick Turner
wrote: 1k is the load seen by each tube in a McI amp while it works in classB. Well, it's the load on each winding of the OPT. Not at all following you here. If the tubes in the McI are trioded, the amp will work in triode AB2, but maybe a lot more grid current is needed, so 12AU7 instead of 12AX7 as the buffer would be better. 6L6 in triode AB2 can make nearly double the po compared to AB1. And, given the same OPT, and unlimited drive, as much output as the pentode originals. If you'd ever had a lifetime supply of 211's you might, like me, try to find excuses for AB2. Arf. TT21 is a KT88 with an anode top cap. No arcing between pin 3 and the heaters. Pin 4 is still the screen, and in RFamps is often at a fixed voltage of 300v only, with say 800v quite OK on the anode. Its a good tube the TT21. Yeah, I *have* heard of those. Some years back they appeared (briefly) on the American surplus market. All gone now, of course. Were they actually made at the Genelex plant? Hard to say, of course. EL86 has u = 11, Ra = 1.4k at 14 mA. Its also a Prince of Linearity. I'll have to check it out. Thanks for the heads-up. Chris Hornbeck "Hum is more than just not knowing the words." -ha |
#30
|
|||
|
|||
John Stewart wrote: Patrick I'm a triode hardcore asshole, personally, so don't have any practical experience, recently enough to matter. Nothing wrong with triodes. If the tubes in the McI are trioded, the amp will work in triode AB2, but maybe a lot more grid current is needed, so 12AU7 instead of 12AX7 as the buffer would be better. 6L6 in triode AB2 can make nearly double the po compared to AB1. I've done many measurements on this circuit while the OP was connected pentode, 50 % UL & triode. The bottom line is that triode operation offers no advantage at all. That is because most of the output stage gain is used in local NFB. Well, the local NFB in this case being within the tube, electrostatically. I am tolerant of triode connections, but I rarely build a triode amp when the UL or CFB connection along with global NFB seem to produce the goodies just the same. The spread of output stage gain from triode to pentode operation is less than 10% & so is the resultant DF. DF's of 5 or so are no problem at all without global NFB, even while running the OP as pentodes. DF = 5 means the Ro would be 1.6 ohms and RL = 8 ohms. This is possible with a pair of triodes without any loop FB whatever, just have the triodes in class A and with a high OPT ratio. Or have CFB in the OPT, with screens to a B+, so that its like the Quad II which boasted its output stage was like a pair of triodes bu with beam tube power. the 10% of CFN applied in the Quad II amp barely lets the output stage to behave as a triode one and the DF isn't5 with no other FB. So one has to apply maybe 25% CFB with most beam tets or pents to get at least 1.6 ohms Ro at the output. I calculated the Ra effective of the McI was 200ohms, so 400 a-a so if the OPT was nominally 4k : 8, the Ro would be 400 / 500 = 0.8, and McI use a lot of CFB; some 50% of the primary voltage is fed back to the cathode. and you need 300vrms drive g-g. But 150 g-g is easier to do at low thd, say only 0.1 % thd, so it is possible to do 25% CFB and attain your DF of 5. UL offers the possibility to run the plates at higher voltage than the screens. As a result, more power is possible. Only if you have a trannie with the required screen winding at a lower DC potential or you have zener diodes and caps off the screen taps. Whenever I have tried it it didn't give very good results. Really high B+ for anode supply only leads to less class A and a nearly class B amp. The circuits I've built sofar exhibit IMD's in the order of -50 db at all levels up to within one db of clipping. Try that with your SET amp! I don't have a fancy hi-fi SET to try it with except in a tube radio where a lone EL34 does a great job making music, and since 0.25 watts is all I use I don't care what the imd might measure at 5 watts. That EL34 has global NFB though. Cheers, John Stewart |
#31
|
|||
|
|||
Chris Hornbeck wrote: On Tue, 19 Apr 2005 16:42:49 GMT, Patrick Turner wrote: 1k is the load seen by each tube in a McI amp while it works in classB. Well, it's the load on each winding of the OPT. Not at all following you here. The McIntosh confuses many people. But it is a nearly class B amp. Think for a minute about an amp with a normal single CT winding only in the anode circuit. When biased for near class B operation, as soon as one tube turns on, the other turns off. most of the wave cycle power away from the zero crossing is produced by one tube or the other, but certainly not by the two tubes together. So if you have a 4K : 8 ohm normal OPT, when its in class B with only one tube conducting, only half the primary has an AC flow so the turn ratio is 1/2 the P to the whole of the S, and since the impedance ratio of the OPT is the square of the turn ratio the load seen by the single tube is 1/4 of the primary load normally seen by two tubes when both are conducting, which only happens at low power levels when the amp is doing its small amount of class A power. McI have the primary arranged so each tube's half of the OPT is further split into two halves, one in the anode circuit and an equal one in the catode circuit. The catode circuit provides the local NFB. But if you were a tube you wouldn't feel any different working in a McI than you would in a "normal" amp with normal windings. So the McI load is 1k per tube most of the time. If the McI is biased up to work mostly in class A, then each tubes sees 2k as its load. Its a very confusing issue, but if you measure a few amps you will see what the conventional anode to anode load description means and what it implies in class A, AB, or B operation. If the tubes in the McI are trioded, the amp will work in triode AB2, but maybe a lot more grid current is needed, so 12AU7 instead of 12AX7 as the buffer would be better. 6L6 in triode AB2 can make nearly double the po compared to AB1. And, given the same OPT, and unlimited drive, as much output as the pentode originals. If you'd ever had a lifetime supply of 211's you might, like me, try to find excuses for AB2. Arf. I prefer class A, AB1 myself.... TT21 is a KT88 with an anode top cap. No arcing between pin 3 and the heaters. Pin 4 is still the screen, and in RFamps is often at a fixed voltage of 300v only, with say 800v quite OK on the anode. Its a good tube the TT21. Yeah, I *have* heard of those. Some years back they appeared (briefly) on the American surplus market. All gone now, of course. Were they actually made at the Genelex plant? Hard to say, of course. EL86 has u = 11, Ra = 1.4k at 14 mA. Its also a Prince of Linearity. I'll have to check it out. Thanks for the heads-up. The humble 807 is usable safely with higher voltages because it has a most sensibly laid out pin arrangement and an anode top cap. With Ea = 600v, Eg2 at 300v, and in beam mode, 80 watts AB2 is quite possible, but thd is around 13% without NFB. Patrick Turner. Chris Hornbeck "Hum is more than just not knowing the words." -ha |
#32
|
|||
|
|||
On Wed, 20 Apr 2005 01:15:22 GMT, Patrick Turner
wrote: The McIntosh confuses many people. snippage for sanity So the McI load is 1k per tube most of the time. If the McI is biased up to work mostly in class A, then each tubes sees 2k as its load. Its a very confusing issue, but if you measure a few amps you will see what the conventional anode to anode load description means and what it implies in class A, AB, or B operation. If the tubes in the McI are trioded, the amp will work in triode AB2, but maybe a lot more grid current is needed, so 12AU7 instead of 12AX7 as the buffer would be better. 6L6 in triode AB2 can make nearly double the po compared to AB1. And, given the same OPT, and unlimited drive, as much output as the pentode originals. If you'd ever had a lifetime supply of 211's you might, like me, try to find excuses for AB2. Arf. I prefer class A, AB1 myself.... TT21 is a KT88 with an anode top cap. No arcing between pin 3 and the heaters. Pin 4 is still the screen, and in RFamps is often at a fixed voltage of 300v only, with say 800v quite OK on the anode. Its a good tube the TT21. Yeah, I *have* heard of those. Some years back they appeared (briefly) on the American surplus market. All gone now, of course. Were they actually made at the Genelex plant? Hard to say, of course. EL86 has u = 11, Ra = 1.4k at 14 mA. Its also a Prince of Linearity. I'll have to check it out. Thanks for the heads-up. The humble 807 is usable safely with higher voltages because it has a most sensibly laid out pin arrangement and an anode top cap. With Ea = 600v, Eg2 at 300v, and in beam mode, 80 watts AB2 is quite possible, but thd is around 13% without NFB. Patrick Turner. Chris Hornbeck "Hum is more than just not knowing the words." -ha Chris Hornbeck "Hum is more than just not knowing the words." -ha |
#33
|
|||
|
|||
On Wed, 20 Apr 2005 01:15:22 GMT, Patrick Turner
wrote: The McIntosh confuses many people. snipped for sanity So the McI load is 1k per tube most of the time. Loading is just a matter of turns ratios and down-stream externals. But I do know where you're coming from. The actual impedance at each winding in the old classic McIntosh's is 250 ohms. We've both conflated memory and practice. Happens (more than I'd like...) I prefer class A, AB1 myself.... Me, too, of course. And I meant to say A2, but machs nicht. Thanks, Chris Hornbeck "Hum is more than just not knowing the words." -ha |
#34
|
|||
|
|||
Patrick Turner wrote:
John Stewart wrote: Patrick I'm a triode hardcore asshole, personally, so don't have any practical experience, recently enough to matter. Nothing wrong with triodes. If the tubes in the McI are trioded, the amp will work in triode AB2, but maybe a lot more grid current is needed, so 12AU7 instead of 12AX7 as the buffer would be better. 6L6 in triode AB2 can make nearly double the po compared to AB1. I've done many measurements on this circuit while the OP was connected pentode, 50 % UL & triode. The bottom line is that triode operation offers no advantage at all. That is because most of the output stage gain is used in local NFB. Well, the local NFB in this case being within the tube, electrostatically. I didn't think you would ever believe that paper on internal NFB in triodes I brought to the attention of the NG more than a year ago. What changed your mind? I am tolerant of triode connections, but I rarely build a triode amp when the UL or CFB connection along with global NFB seem to produce the goodies just the same. The spread of output stage gain from triode to pentode operation is less than 10% & so is the resultant DF. DF's of 5 or so are no problem at all without global NFB, even while running the OP as pentodes. DF = 5 means the Ro would be 1.6 ohms and RL = 8 ohms. I never would have thought of that! (Smiles) Patrick likes to state the obvious. This is possible with a pair of triodes without any loop FB whatever, just have the triodes in class A and with a high OPT ratio. True enough, but the available audio power drops. For example, max power loading for a 2A3 occurs at about 3Rp, so if the OPT is perfect the DF will be 3. If the loading is 5Rp, than DF is 5 but power available drops by 20%. That doesn't happen in the McIntosh circuit. Or have CFB in the OPT, with screens to a B+, so that its like the Quad II which boasted its output stage was like a pair of triodes bu with beam tube power. the 10% of CFN applied in the Quad II amp barely lets the output stage to behave as a triode one and the DF isn't5 with no other FB. That's all very nice Patrick & a great credit to the original designers, not you. But how is an ordinary experimenter to avail himself of the specially wound OPT? You keep ranting on in respect to how good this circuit is & how often you use it but that is of little help to most of those here in the NG. So one has to apply maybe 25% CFB with most beam tets or pents to get at least 1.6 ohms Ro at the output. Good for you. I guess we will have to listen again to the long analysis of 'Why'. I calculated the Ra effective of the McI was 200ohms, so 400 a-a so if the OPT was nominally 4k : 8, the Ro would be 400 / 500 = 0.8, and McI use a lot of CFB; some 50% of the primary voltage is fed back to the cathode. and you need 300vrms drive g-g. But 150 g-g is easier to do at low thd, say only 0.1 % thd, so it is possible to do 25% CFB and attain your DF of 5. Yadda, Yadda! UL offers the possibility to run the plates at higher voltage than the screens. As a result, more power is possible. Only if you have a trannie with the required screen winding at a lower DC potential or you have zener diodes and caps off the screen taps. Whenever I have tried it it didn't give very good results. Or you can do it like Crowhurst did. I followed his lead & a very good amp resulted at low cost. Really high B+ for anode supply only leads to less class A and a nearly class B amp. Really. What do you want to bore us with next? The circuits I've built sofar exhibit IMD's in the order of -50 db at all levels up to within one db of clipping. Try that with your SET amp! I don't have a fancy hi-fi SET to try it with except in a tube radio where a lone EL34 does a great job making music, and since 0.25 watts is all I use I don't care what the imd might measure at 5 watts. That EL34 has global NFB though. No new ideas yet? Your circuits really are stuck in the 50's! John Stewart Cheers, John Stewart |
#35
|
|||
|
|||
Chris Hornbeck wrote: On Wed, 20 Apr 2005 01:15:22 GMT, Patrick Turner wrote: The McIntosh confuses many people. snipped for sanity So the McI load is 1k per tube most of the time. Loading is just a matter of turns ratios and down-stream externals. But I do know where you're coming from. The actual impedance at each winding in the old classic McIntosh's is 250 ohms. We've both conflated memory and practice. Happens (more than I'd like...) Ppl consider the McI OPT as a one winding tranny with two bifilar coils to make the winding, so its 1k end to end of the tranny. since the windings are centretapped, they think each half winding is 250 ohms. But the individual class B tube load is 1k, and the RL a-a in the classical manner of understanding is 4 k. I prefer class A, AB1 myself.... Me, too, of course. And I meant to say A2, but machs nicht. ; Patrick Turner. Thanks, Chris Hornbeck "Hum is more than just not knowing the words." -ha |
#36
|
|||
|
|||
John Stewart wrote: Patrick Turner wrote: John Stewart wrote: Patrick I'm a triode hardcore asshole, personally, so don't have any practical experience, recently enough to matter. Nothing wrong with triodes. If the tubes in the McI are trioded, the amp will work in triode AB2, but maybe a lot more grid current is needed, so 12AU7 instead of 12AX7 as the buffer would be better. 6L6 in triode AB2 can make nearly double the po compared to AB1. I've done many measurements on this circuit while the OP was connected pentode, 50 % UL & triode. The bottom line is that triode operation offers no advantage at all. That is because most of the output stage gain is used in local NFB. Well, the local NFB in this case being within the tube, electrostatically. I didn't think you would ever believe that paper on internal NFB in triodes I brought to the attention of the NG more than a year ago. What changed your mind? I am tolerant of triode connections, but I rarely build a triode amp when the UL or CFB connection along with global NFB seem to produce the goodies just the same. The spread of output stage gain from triode to pentode operation is less than 10% & so is the resultant DF. DF's of 5 or so are no problem at all without global NFB, even while running the OP as pentodes. DF = 5 means the Ro would be 1.6 ohms and RL = 8 ohms. I never would have thought of that! (Smiles) Patrick likes to state the obvious. This is possible with a pair of triodes without any loop FB whatever, just have the triodes in class A and with a high OPT ratio. True enough, but the available audio power drops. Sigh, yes, it does. But if yo are listening to a fine stradivarus cello, you don't need too much power. Add a few more triodes and the whole orchestra sounds great. and maybe a few more and the works of Mahler with 1,000 singers becomes real Just use enough. Triodaholics know all about this; some cheat, and use horns!! :-] For example, max power loading for a 2A3occurs at about 3Rp, so if the OPT is perfect the DF will be 3. If the loading is 5Rp, than DF is 5 but power available drops by 20%. That doesn't happen in the McIntosh circuit. Well actually anyone using a 2A3 won't care too much power he misses out on if he goes for the best watt of power for his 100 dB/W horn speakers. Power is irrelevant to such ppl. I have a 12" speaker and a dome tweeter in my kitchen radio and sensitivity is probably 93 dB/W easy, and I'd be happy with 2 not the 5 I have now from a trioded EL34. I sure don't need a McIntosh. But I do have 50 watt amps for the lounge speakers which are 89 dB/W and which I sometimes like a little loud after a drink. If you plot the power curve at clipping for a McIntosh you wil get a typical peak at maybe 80 watts with a rapid fall each side of the peak because the amp is voltage limited as RL rises, and current limited as RL falls, and so if you desperately needed 80 watts of the maximum possible Mc I po, ( I am not sure what the actual max po is ), then raising the RL value would be just as much a dissapointment with a McI as it is with a 2A3, if going from 3.5 watts to 2 watts was unacceptable. Or have CFB in the OPT, with screens to a B+, so that its like the Quad II which boasted its output stage was like a pair of triodes bu with beam tube power. the 10% of CFN applied in the Quad II amp barely lets the output stage to behave as a triode one and the DF isn't5 with no other FB. That's all very nice Patrick & a great credit to the original designers, not you. But how is an ordinary experimenter to avail himself of the specially wound OPT? Winding it himself like i do, or engaging a specialist winder. Sowter are good at this, Ian Iveson bought a beauty to use with his amp which has 4 x 6CL6? outputs. You keep ranting on in respect to how good this circuit is & how often you use it but that is of little help to most of those here in the NG. Then use the next best, plain old UL. Hammond make a range of affordable types...... There is a mountain of information already existing on the more ordinary tube amp. I like exploring the slightly different variety..... So one has to apply maybe 25% CFB with most beam tets or pents to get at least 1.6 ohms Ro at the output. Good for you. I guess we will have to listen again to the long analysis of 'Why'. Not this time. From what I have said elsewhere, ppl should have the general ability to grasp the idea of dynamic plate resistance, Ra, and how an applied amount of NFB effectively reduces the Ra, and by now they should realise that if you have an OPT to a load, then the source resistance in series with the primary becomes transformed into a lower source resistance at the secondary. I calculated the Ra effective of the McI was 200ohms, so 400 a-a so if the OPT was nominally 4k : 8, the Ro would be 400 / 500 = 0.8, and McI use a lot of CFB; some 50% of the primary voltage is fed back to the cathode. and you need 300vrms drive g-g. But 150 g-g is easier to do at low thd, say only 0.1 % thd, so it is possible to do 25% CFB and attain your DF of 5. Yadda, Yadda! UL offers the possibility to run the plates at higher voltage than the screens. As a result, more power is possible. Only if you have a trannie with the required screen winding at a lower DC potential or you have zener diodes and caps off the screen taps. Whenever I have tried it it didn't give very good results. Or you can do it like Crowhurst did. I followed his lead & a very good amp resulted at low cost. I have no argument there..... Really high B+ for anode supply only leads to less class A and a nearly class B amp. Really. What do you want to bore us with next? Facts are facts. Ppl rush at building a tube amp and think maximum power ability and overload behaviour, but in practice they'y never ever use max po or ever overload the amp. So the first 10 watts are the most important. Some folks like to build a boat of great visual beauty with curves everywhere and timber work just to sail around the bay. Others will take the cheap option and weld some horror together in a week and slap a coat of paint on.... Each unto their own. The circuits I've built sofar exhibit IMD's in the order of -50 db at all levels up to within one db of clipping. Try that with your SET amp! I don't have a fancy hi-fi SET to try it with except in a tube radio where a lone EL34 does a great job making music, and since 0.25 watts is all I use I don't care what the imd might measure at 5 watts. That EL34 has global NFB though. No new ideas yet? Your circuits really are stuck in the 50's! John Stewart Isn't that when Crowshurst wrote most of his stuff? Some of my ideas follow the Quad II example. I implement the idea rather a lot better than Quad ever did. If you stroll around my ancient website, there are a few circuits that have definately post 1990 origins, but who cares, nearly every SS amp has been done before and there is really no inovations to be made except perhaps to make a digital amp which converts the digital signal right at its output. I don't give a damn about originality. To be completely original in tube topologies is quite impossible. Everyone is in the same boat here so nobody else can be more original than I can, but we would welcome it if they manage it. One could try to build a steam powered motor car with a rotary steam engine. It sounds like it hasn't been done yet, but it should work if the Wankel petrol engine is anything to go by. No reciprocating parts, must be lighter, more efficient. But take a look at Douglas Self's site, he has oodles of material on rotary steam engines, and how all attempts failed to make one better than a piston engine. Steam turbines were tried on locomotives, and that led knowhere except to cranky loco mechanics..... Not too bad on ships I'm told, oh, and in power stations. We could try to build an original titanium dining table. I'd still prefer a wooden one. Christ was a carpenteer, and today's carpenteers follow Him even if they are Moslem. There's something wondrous about building a beuatiful OPT, or chassis, or tubing up.... The cheapistivists are welcome to save their pennies, but some folks like to spend theirs. There is no shame in using tubes, and none in making retro equipment. And no shame in using some modern techniques such as transistor CCS, whoppingly large SMPS electrolytic caps, etc..... The sound justifies the means. The means can have beauty. Patrick Turner. Cheers, John Stewart |
#37
|
|||
|
|||
On Wed, 20 Apr 2005 14:24:58 GMT, Patrick Turner
wrote: Ppl consider the McI OPT as a one winding tranny with two bifilar coils to make the winding, so its 1k end to end of the tranny. since the windings are centretapped, they think each half winding is 250 ohms. But the individual class B tube load is 1k, and the RL a-a in the classical manner of understanding is 4 k. Yes, we're just babbling on about the classic push-pull transformer naming conventions. We probably ought to just specify the turns ratio and let interested readers do their own arithmetic. Thanks, Chris Hornbeck |
#38
|
|||
|
|||
Chris Hornbeck wrote: On Wed, 20 Apr 2005 14:24:58 GMT, Patrick Turner wrote: Ppl consider the McI OPT as a one winding tranny with two bifilar coils to make the winding, so its 1k end to end of the tranny. since the windings are centretapped, they think each half winding is 250 ohms. But the individual class B tube load is 1k, and the RL a-a in the classical manner of understanding is 4 k. Yes, we're just babbling on about the classic push-pull transformer naming conventions. We probably ought to just specify the turns ratio and let interested readers do their own arithmetic. You will find the McIntosh OPT has about the same number of primary turns as any other well made "normal" OPT which has a 4k to 4-8-16 set up. Half the McI P turns are between cathodes and a grounded CT, and half between anodes and CT at the B+. All these turns are bifilar wound, which is about as close wound as one can get. Afaik, the primary is arranged into 4 sections, and if we consider the total number of P turns in a and k windings as N turns, then each P section contains N/8 turns for anode, and N/8 turns for cathode, and probably in 4 layers of the bifilar wound wire. So I suspect sec is distributed in 5 sections between the P, so a pattern of S-P-S-P-S-P-S-P-S is the build up of the interleavings. The 4P-5S interleaving will give very good HF response and allow a lot of NFB, although the 275 McI schematic I have shows a separate widing for the global FB. Such a winding need not be thick wire, and can be like the thin primary wire, and wound over the thick speaker secs and laying between each speaker sec thick wire turn so it is indeed very closely coupled to the secondary signal so that the OPT winding resistance is included in the NFB loop, and this resistance is effectively reduced by the NFB as well as the Ra. McIntosh have used thickly insulated wire for the a and k windings. I am not sure if such wire is readily available for a diyer, but my supplier lists 3 grades of polyesterimide 0.4 mm dia wire which I would use for a primary coil each with 0.025mm, 0.038mm, and 0.058mm of insulation thickness. The grade 3 is hard to get, and grade 2 is all they have. But maybe McI had even thicker insulation on their wires. If one were to build a Circlotron using the same output tubes as the Mc to get the same po, one would indeed need only 1/2 the total P turns used in a McI or conventional OPT. The wire would have twice the sectional area, and thus the dia would be 0.56 mm instead of 0.4 mm, and it'd be a lot easier to wind. But the same high drive voltage g-g is needed and there are 2 floating PS which isn't too hard for the diyer to achieve, except that there are not many power trannies around with suitably rated twin HT windings. So a pair of smaller power trannies can be used for one channel. This is OK because if a diyer asks no nore than 50 watts from his pair of 6550 then each PT only has to ever provide around 45 watts of plate power, which is half the 90 watts of plate input power when the amp is in class AB making 50 watts. I don't know how important screening the HT winding from stray capacitance coupling to other windings is in a circlotron. But I do see that they could possibly suffer from switching noise because the whole B+ floats on the tube anode and cathode circuits separated effectively from 0V only by the anode-cathode load to the grounded CT. So I'd assume a circlotron would be best designed when the PT had a pair of fully screened off HT windings, both screened from the mains winding, and from each other. Diying a McI or a Circlotron takes some doing, and most think the results aren't worth the trouble. In my 300 watters, there is a 5P x 6S interleaving pattern, and devoting one section of P for a CFB winding is extremely easy needing only 10 minutes more winding work. And the result is that the Ro of the amp without loop FB is less than triode while retaining ease of drive to the output tubes and only 150 vrms g-g is needed. So for a total of 250 vrms of output voltage on one tube a to k, there is 75 vrms applied to the g1, so A' = 3.33, or just less than triode, because 6550 has u = about 5 in triode. I once saw a Lincoln-Walsh schematic which had CFB and an IST. Anyone have that schematic? Patrick Turner. Thanks, Chris Hornbeck |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
The Voices of Pro Audio Engineers | High End Audio | |||
KISS 113 by Andre Jute | Vacuum Tubes | |||
"Competent design" | High End Audio | |||
are there any books to read to get good at analog electronics? | Pro Audio | |||
Adapting an older design? (was Tube AM Tuner...) | Vacuum Tubes |