i have been toying with this idea for a long time....but knowing that it takes more work, i balked....😀
gorgon, what you say is true for power transform. Information transform is a bit more subtle and it is this that we are interested in for audio transformers. Power is important, for power usage, but you must think about the other half of the electrical field, the electrostatic moment. If you must dilute the electrostatic coupling, you will loose information
Again, all three forms of core can provide the very best performance possible, but they will have strengths and weaknesses. Personally, I would prefer a tape wound J core with the core divided into thirds and manipulated. E/I core provides a serviceable alternative that allows me to build audio transformers to suit my interests.
Many people are appreciative of the information transform that can be gotten out of commercial core. An at least equal number prefer a leaner less colorful response, without deeply revealed gradient structures, to transients and tones. Once you have solved the balance of information to frequency response for a particular core form, you will find that other formats are essentially equal, but satisfy different musical information tastes.
Bud
Again, all three forms of core can provide the very best performance possible, but they will have strengths and weaknesses. Personally, I would prefer a tape wound J core with the core divided into thirds and manipulated. E/I core provides a serviceable alternative that allows me to build audio transformers to suit my interests.
Many people are appreciative of the information transform that can be gotten out of commercial core. An at least equal number prefer a leaner less colorful response, without deeply revealed gradient structures, to transients and tones. Once you have solved the balance of information to frequency response for a particular core form, you will find that other formats are essentially equal, but satisfy different musical information tastes.
Bud
Pardon me, I have to stand corrected: 54 mm of course equals to 2 1/8 inches...
Best regards!
Give me a break!
They offer silicon steel c-cores in a good quality M4; with the available sizes you could make whatever core set up (might require some creativity though...).
Alphacore even offers amorphous c-cores, so I really don't understand the "somewhat limited offerings"....
I source cores in somewhat larger quantities nowadays which means that I have to pay attention to price, and especially price of transport, as cores are heavy.
That is a different situation.
With the stuff Alphacore has on offer you can try your winding efforts and development of your skills for years to come IMO, at least that was my own experience when I had to source single (and therefore more expensive) pieces here in Europe years ago.
Eddy currents are not a result of magnetostriction but simply induced current.
How else could we reduce eddy current with the help of laminating
the core and/or by increasing the resistivity of the core?
Would you explain how to TRANSFORM a electrical field in a transformer?
If you are talking about the UNWANTED dielectric properties of audiotransformers winding arrangements that certainly can also result in capacitive coupling you sure have a mystical way to do that.
By choosing a lamination scheme that is not deliberately used to make a power transformer as inert to frequencies other than 50/60 Hz as possible.
Do you think that dielectrics should be eliminated from audio transformers? Would this somehow make them sound better?
I am just working with the possibility that capacitive coupling can be manipulated, that you can use simple physics to think about E Field coupling, how to improve what you want of it and limit the effect of what you do not want. I will make another mystical statement for you to be upset about. Capacitance is the most important factor in the reproduction of tone and transient gradients. These gradients are the heart of portraying reproductions of artistic expression and without them reproduced music is sterile.
A push-pull OPT has to take into account dc imbalance, so there is only 1 lamination scheme possible. You have to introduce a CONTROLLED airgap. Anyway, if you choose to compromise and "bridge" the unavoidable gap of the EI intersection (i know, most do so to achieve a higher inductance) you will introduce flux-crowding in the lamination that bridges the airgap. Do to fluxcrowding , and therefore higher flux in those areas, they will also be the first areas to saturate (especially with even a small amount of dc imbalance and/or remanence and under "first pulse" conditions where flux can more than double). Naturally eddy current losses and hysteresislosses are also significant above average also at the lower flux-levels.
Yes I do. The ideal would be to have a transformer with no capacitances at all (this would allow for really thigth coupling and a first order roll off at the high frequency end without even the possibility of peaking). Offcourse this is not possible, we have no way to get rid of the permittivity of free space. All we can do is to try to get as close as possible. I strive for the lowest possible Dk and lossfactor. I also strive for the maximum of frequency and field-strength independence of this factors. Also, to minimize unlinearity, memory-effects should be minimized in the core as well as in the isolation.
yes, capacitance is important, MINIMUM CAPACITANCE that is.
It would not make sense to have a step-down transformer and try to couple it also capacitively. You can only DIVIDE the voltage capacitively, you CANNOT multiply the current! You can only make a capacitive divider, not a tranformer. Capacitive coupling makes only sense in resonated narrow band transformers.
As to all the artistic magic you like to weave into a purely technical subject,
i suggest you digg more into hysteresis related stuff (both core AND dielectricum). Also the totally different mechanism of magnetostriction (and its effects) at low versus high flux levels is worth to consider.
Last edited:
Don't agree; in my 1:1 very wide band interstage transformers I use capacitive coupling to my advantage.
Gorgon, you claim to have some knowledge on magnetics.
Could you tell us a bit more about your experience winding transformers?
In other words: to what extent has your theories been confirmed by practical experience?
Do you claim to have the same experience as guys like DaveS, BudP, Lundahl, me maybe, and some others?
Bud, as to sound "better".
In a audiotransformer, i strive for measures better under the conditions it is used, and "does not have any sound of its own".
In a audiotransformer, i strive for measures better under the conditions it is used, and "does not have any sound of its own".
I have not "wound" transformers myself, other than transformers for measuring equippement with a BW 500Hz-500kHz 10W flat over the band, and some very high powered water cooled narrow band stuff from 50Hz up to 2Mhz
Piet, i did not make any claims, i stated facts, until now only simple and clear cut stuff.
I am sure all of those you mentioned have plenty more of experience and knowledge on how to efficiently produce a COMMERCIAL competent audio transformer than i have. That is not the point here, almost any commercial product can be improved IMMENSLY if commercial interests and limitations of mass-production are putted aside. My knowledge in transformers and level of expirience is different to yours, or BudP, or who ever. Naturally so, since i spent over 50years of my life on either improving commercially produced products (ranging from measuring equippement to high power induktiv and dielectric heating ), or troubleshooting ill designs, or designed and building the first ever working prototype. There where not many days in my life where i would have been involved in routinework or where limited by commercial issues. There was a time in my life when tubes where out of fashion. Trust me, back then it was almost impossible to convince anybody that a triodeamp with a proper transformer sounds better than the 0.01% 100 watt semiconductor stuff of those days. Producers of OPTs for audio where almost totally out of busnes and only much later have been re-invented. The stuff i designed and/or builded where all of the type money could not buy, prototypes or unique pieces of equippement wich always necessated a "out of the box" way of thincking and lots of work.
After i left wissentschaftliche werkstätten in german, i worked at the philips industrial service department in finland. When i left there in 1972 to start my own busnes, philips offered me to take over theyr industrial heating busnes here. At the time i had some knowledge of this special branch, but not nearly enough, especially regarding properties under high frequency heating conditions. Since there where also only very limited info in literature, i had to digg into it myself. After spending many month in my workshop, building measuring equippement, taking measurements a.s.o. i gained knowledge like at what frequency a certain type of material has what losses, what dielectric constant, a.s.o.
I also quickly found out that isolation materials of a typical trade name but from a different manufacturer may differ quite a bit. Stuff you have to work out yourself. I did not stop there, over the years i also went into studying single and multilayer coil geometries and theyr properties, self resonance related stuff, increase of Rac with frequency, increase of Rac at the edges of a coil or do to the influence of fringe-flux of a gap a.s.o.
What i did not investigate in deep, and what would interst me now would be the dielectric properties of different isolation-materials of the avaiable magnet wires and theyr variation, be it do to differnt suppliers and/or at different operation conditions.
Same goes for the various impregnation materials and isolation materials avaiable , alltough my approach is to reduce theyr use to the absolute minimum and replace them with air. Even i propably know all that can be found published about this materials i would like to hear if any of you studied those in more deep and some measuring results. From what i know about the properties of various plastic,wood, fillers and glues i find it very likely that equally VERY FREQUENCY DEPENDEND properties can exist in especially paperbased isolation material.
Any input on data that has been obtained other than from the manufactorers datasheets would be very welcome.
I am sure all of those you mentioned have plenty more of experience and knowledge on how to efficiently produce a COMMERCIAL competent audio transformer than i have. That is not the point here, almost any commercial product can be improved IMMENSLY if commercial interests and limitations of mass-production are putted aside. My knowledge in transformers and level of expirience is different to yours, or BudP, or who ever. Naturally so, since i spent over 50years of my life on either improving commercially produced products (ranging from measuring equippement to high power induktiv and dielectric heating ), or troubleshooting ill designs, or designed and building the first ever working prototype. There where not many days in my life where i would have been involved in routinework or where limited by commercial issues. There was a time in my life when tubes where out of fashion. Trust me, back then it was almost impossible to convince anybody that a triodeamp with a proper transformer sounds better than the 0.01% 100 watt semiconductor stuff of those days. Producers of OPTs for audio where almost totally out of busnes and only much later have been re-invented. The stuff i designed and/or builded where all of the type money could not buy, prototypes or unique pieces of equippement wich always necessated a "out of the box" way of thincking and lots of work.
After i left wissentschaftliche werkstätten in german, i worked at the philips industrial service department in finland. When i left there in 1972 to start my own busnes, philips offered me to take over theyr industrial heating busnes here. At the time i had some knowledge of this special branch, but not nearly enough, especially regarding properties under high frequency heating conditions. Since there where also only very limited info in literature, i had to digg into it myself. After spending many month in my workshop, building measuring equippement, taking measurements a.s.o. i gained knowledge like at what frequency a certain type of material has what losses, what dielectric constant, a.s.o.
I also quickly found out that isolation materials of a typical trade name but from a different manufacturer may differ quite a bit. Stuff you have to work out yourself. I did not stop there, over the years i also went into studying single and multilayer coil geometries and theyr properties, self resonance related stuff, increase of Rac with frequency, increase of Rac at the edges of a coil or do to the influence of fringe-flux of a gap a.s.o.
What i did not investigate in deep, and what would interst me now would be the dielectric properties of different isolation-materials of the avaiable magnet wires and theyr variation, be it do to differnt suppliers and/or at different operation conditions.
Same goes for the various impregnation materials and isolation materials avaiable , alltough my approach is to reduce theyr use to the absolute minimum and replace them with air. Even i propably know all that can be found published about this materials i would like to hear if any of you studied those in more deep and some measuring results. From what i know about the properties of various plastic,wood, fillers and glues i find it very likely that equally VERY FREQUENCY DEPENDEND properties can exist in especially paperbased isolation material.
Any input on data that has been obtained other than from the manufactorers datasheets would be very welcome.
Last edited:
gorgon,
I too have spent the last 40 years in industrial magnetics, which means power magnetics and have 3 decades of experience in switch mode wide band design. The knowledge developed in those fields is not applicable here, in audio, at anything greater than line level voltages.
To effectively emphasize direct capacitive coupling over indirect coupling (called distributed capacitance) you must utilize dielectric materials that cause the "work" done by the indirect coupling to be at least twice that of the dielectric material that separates the two antenna plates of primary and secondary.
You should not use film plastic interlayer insulation due to the E Field acceptance and release characteristics of the film and to the leakage inductance caused by the layers of film. Use of electrical grade Kraft paper eliminates the film draw backs, but does not get rid of the leakage inductance. Use of coil wire to coil wire, without interlayer inserts, requires both a high strength (class 155) coating and a 100% solids vacuum impregnated material that remains in the coil during the cure cycle. If you do not use film plastics between primary and secondary antenna plates, your coil rise time will easily keep pace with that of the very best core materials.
Do not hesitate to use a full vacuum impregnate material that is 100% solids and has a catalyzed dielectric constant of 4 to 6. Use nylon with glass fill winding forms and wrap the finished coil with an equivalent dielectric "work" load of polyesterterepthalate. For most end wraps this equates to 6 to 10 mills total of mylar, in two complete turns. Doing this, along with the other activities, will allow you to have audio transformers with 250pf or less of indirect coupling, with direct coupling in the tens of nanofarads.
This will not allow 100 kHz bandwidth. However, the finished product will have phase stability out to 45 kHz. This is exactly what you want. Your OPT's will not be made for test equipment and it is very easy to overlook this and assume that faster is better, when it is not.
When combined with a core construction that passively demagnetizes from hysteresis saturation, does not have a square wave leading edge overshoot and thus has no crossover distortion when used in push pull, you will have OPT's that are clear and so full of detail that they cause directly compared amorphous core OPT's to sound how animated movie versions of live events look.
The reason for using commercial core material is to avoid the serious problem encountered with high performance power core, like amorphous or the nickle options. With these materials, all capacitance is bad, with distributed being the worst of the numerous evils. You end up having to throw away the E Field and so you also throw away the information contained in these fields, that occur every time there is a vector change in the B Field, due to the signal that is driving both fields.
It is a first order mistake to assume that what you learned in high frequency magnetics can be applied to audio magnetics. Take it to heart that the "old school" of audio designers used "audio grade" core in their tube amplifiers. Audio grade is 26 gauge M19 material. I use M6 and M3, but must pay careful attention to the coil materials and construction to avoid 60 kHz spikes of the sort you get when using amorphous core with any capacitance between primary and secondary.
I would also council against "air" dielectric construction. Give heed to corona here. Voltages above 30 vac begin to degrade the dielectric coatings on wires. Everywhere there is a dielectric discontinuity, like applying a piece of tape to hold a coil wire in place, the activity of corona is magnified threefold. The lifetime of an OPT with 450 vdc and up to 152 vac rms in the primary will be less than 10 k hours.
As for audible differences in wire coatings, stick with poly nylon, the higher temperature coatings have dielectric constants above 4 and in an OPT with poor plate to plate coupling, this becomes one of the controlling factors in rise time of the coil.
All of the above comes from a clear knowledge of the basic first order approximations that are the transformer derivatives of Maxwell. They are not enough and audio transformers are not a purely technical category. You will find yourself forming "attitudes" about materials used in core and coil construction that are not supportable even with complex signal portrayal, what you need to find will be lost in the "grass". If you are going to pursue audio output design, you are going to have to look beyond what has worked for you in power solutions and when you get around to interstage transformers, may the good lord grant you grace!!!
Bud
I too have spent the last 40 years in industrial magnetics, which means power magnetics and have 3 decades of experience in switch mode wide band design. The knowledge developed in those fields is not applicable here, in audio, at anything greater than line level voltages.
To effectively emphasize direct capacitive coupling over indirect coupling (called distributed capacitance) you must utilize dielectric materials that cause the "work" done by the indirect coupling to be at least twice that of the dielectric material that separates the two antenna plates of primary and secondary.
You should not use film plastic interlayer insulation due to the E Field acceptance and release characteristics of the film and to the leakage inductance caused by the layers of film. Use of electrical grade Kraft paper eliminates the film draw backs, but does not get rid of the leakage inductance. Use of coil wire to coil wire, without interlayer inserts, requires both a high strength (class 155) coating and a 100% solids vacuum impregnated material that remains in the coil during the cure cycle. If you do not use film plastics between primary and secondary antenna plates, your coil rise time will easily keep pace with that of the very best core materials.
Do not hesitate to use a full vacuum impregnate material that is 100% solids and has a catalyzed dielectric constant of 4 to 6. Use nylon with glass fill winding forms and wrap the finished coil with an equivalent dielectric "work" load of polyesterterepthalate. For most end wraps this equates to 6 to 10 mills total of mylar, in two complete turns. Doing this, along with the other activities, will allow you to have audio transformers with 250pf or less of indirect coupling, with direct coupling in the tens of nanofarads.
This will not allow 100 kHz bandwidth. However, the finished product will have phase stability out to 45 kHz. This is exactly what you want. Your OPT's will not be made for test equipment and it is very easy to overlook this and assume that faster is better, when it is not.
When combined with a core construction that passively demagnetizes from hysteresis saturation, does not have a square wave leading edge overshoot and thus has no crossover distortion when used in push pull, you will have OPT's that are clear and so full of detail that they cause directly compared amorphous core OPT's to sound how animated movie versions of live events look.
The reason for using commercial core material is to avoid the serious problem encountered with high performance power core, like amorphous or the nickle options. With these materials, all capacitance is bad, with distributed being the worst of the numerous evils. You end up having to throw away the E Field and so you also throw away the information contained in these fields, that occur every time there is a vector change in the B Field, due to the signal that is driving both fields.
It is a first order mistake to assume that what you learned in high frequency magnetics can be applied to audio magnetics. Take it to heart that the "old school" of audio designers used "audio grade" core in their tube amplifiers. Audio grade is 26 gauge M19 material. I use M6 and M3, but must pay careful attention to the coil materials and construction to avoid 60 kHz spikes of the sort you get when using amorphous core with any capacitance between primary and secondary.
I would also council against "air" dielectric construction. Give heed to corona here. Voltages above 30 vac begin to degrade the dielectric coatings on wires. Everywhere there is a dielectric discontinuity, like applying a piece of tape to hold a coil wire in place, the activity of corona is magnified threefold. The lifetime of an OPT with 450 vdc and up to 152 vac rms in the primary will be less than 10 k hours.
As for audible differences in wire coatings, stick with poly nylon, the higher temperature coatings have dielectric constants above 4 and in an OPT with poor plate to plate coupling, this becomes one of the controlling factors in rise time of the coil.
All of the above comes from a clear knowledge of the basic first order approximations that are the transformer derivatives of Maxwell. They are not enough and audio transformers are not a purely technical category. You will find yourself forming "attitudes" about materials used in core and coil construction that are not supportable even with complex signal portrayal, what you need to find will be lost in the "grass". If you are going to pursue audio output design, you are going to have to look beyond what has worked for you in power solutions and when you get around to interstage transformers, may the good lord grant you grace!!!
Bud
Last edited:
there appear to be two main factions in this Thread.
Those that want to learn and share and disseminate information.
Those that want to keep the snake oil in the bottle and tell nothing of their findings.
Those that want to learn and share and disseminate information.
Those that want to keep the snake oil in the bottle and tell nothing of their findings.
How amusing Andrew. Do you have a job? Is your job in any way threatened by intellectual property pirates?
I haven't any problem telling all I know, to someone whom I trust and have reason to think will use the information I provide, in a market place I cannot enter for one reason or another, usually having to do with govt. protection of local manufacturers. Check out Alien Amplifiers in Brazil, for someone who has taken my real world, easily produced transformer designs and made himself a company from them. All he had to do was ask politely and privately. No fees, no infringement suits, but his market is not for me.
Then, check out O-Netics and perhaps discover what my customers have to say. If I provide any more explicit information than I have I will endanger my ability to live indoors and feed my wife of 45 years. I will also be unable play around in my own pea patch and develop such novel concepts as EnABL and Ground Side Electron support.
The Ground Side Electron thread, where I divulged what I know, was pirated by an unscrupulous person here in the US and foolishly provided at a small fraction of the price required to market Ground Control through commercial online sales sites. Because of this person and others just like him I have no reason to provide you and the rest of the world with details beyond what I have done.
Transformers are not rocket science. They are not a "black art". They just don't have any moving parts and do their work without any obvious physical forces. But they are mud simple to make. That audio transformers that are advanced in real world performance that exceeds what was produced in the 40's and 50's. require a more subtle approach doesn't mean that approach is beyond your ability to comprehend or that there is snake oil involved. If you actually educate yourself in the few simple formulas required to design transformers, build a couple to get an idea of the physical aspects involved and then sit and think about what I have written, you will discover what I implement in the real world. If you then try out your discoveries you will find that none of this is the dreaded snake oil you seem fond of.
Bud
I haven't any problem telling all I know, to someone whom I trust and have reason to think will use the information I provide, in a market place I cannot enter for one reason or another, usually having to do with govt. protection of local manufacturers. Check out Alien Amplifiers in Brazil, for someone who has taken my real world, easily produced transformer designs and made himself a company from them. All he had to do was ask politely and privately. No fees, no infringement suits, but his market is not for me.
Then, check out O-Netics and perhaps discover what my customers have to say. If I provide any more explicit information than I have I will endanger my ability to live indoors and feed my wife of 45 years. I will also be unable play around in my own pea patch and develop such novel concepts as EnABL and Ground Side Electron support.
The Ground Side Electron thread, where I divulged what I know, was pirated by an unscrupulous person here in the US and foolishly provided at a small fraction of the price required to market Ground Control through commercial online sales sites. Because of this person and others just like him I have no reason to provide you and the rest of the world with details beyond what I have done.
Transformers are not rocket science. They are not a "black art". They just don't have any moving parts and do their work without any obvious physical forces. But they are mud simple to make. That audio transformers that are advanced in real world performance that exceeds what was produced in the 40's and 50's. require a more subtle approach doesn't mean that approach is beyond your ability to comprehend or that there is snake oil involved. If you actually educate yourself in the few simple formulas required to design transformers, build a couple to get an idea of the physical aspects involved and then sit and think about what I have written, you will discover what I implement in the real world. If you then try out your discoveries you will find that none of this is the dreaded snake oil you seem fond of.
Bud
Last edited:
Hey BudP,
like in another thread, i would like to say you have the patience of a saint......
although i do not fully comprehend all of your post, i am making a compilation of them in a word document, so that i can refer to them from time to time......
i am now making my own opt's not so much to gain from them commercially but to support this hobby......
please keep sharing.......thanks....
like in another thread, i would like to say you have the patience of a saint......
although i do not fully comprehend all of your post, i am making a compilation of them in a word document, so that i can refer to them from time to time......
i am now making my own opt's not so much to gain from them commercially but to support this hobby......
please keep sharing.......thanks....
I just noticed Tony's invitation message from a couple of weeks ago. I don't have a lot of experience on this stuff, just a few small projects from time to time. I think I agree with most of what gorgon53 has said, seems this is the conventional knowledge from texts and app. notes.
On the splitting of C cores, I can see where that would help equalize the magnetic path lengths. The toroid doesn't have that option, but it's performance on leakage L (and low distributed C if progressive wound or other low C variants) is still going to be tops. Using a long single C core set with two thin windings (each side) would give pretty good low leakage results. (long thin windings give low leakage as long as the core is not getting near saturation, this forces the air leakage pathes to be as long as the iron path, so the common high Mu iron path gets preference, the toroid is the best case if it's not overwound)
I've never seen any toroids made from multiple lamination strips, always a long single strip. I suppose that might be a technique for very weakly gapping it. I would think that most partial gapped toroids just use some interleave film or a coating on one side of the lam. strip. Actual cut toroids would need some darn strong bonding material to keep it from springing apart or resonating.
I'm not really following BudP on the "antenna event" thing. I know that the leakage L and distributed C need to have their Z matched to the effective load Z at their resonance freq. for critical damping. So optimising just one or the other will cause ringing.
For a bifilar wound setup, there could be traveling wave phenomina, but audio is too low a freq. to be of concern typically. Distributed capacitance loading is the big issue, so either matched signals needed as for bifilar or electrical isolation for the usual case.
It's true that the wave velocity of a transmission line is slowed by c/Sqrt(Mu,Eo) [relative values], (which could make audio frequency wavelengths a concern, 100X slowing with 3000 Mu and 3 Epsilon) but the magnetic core coupling in the xfmr is simultaneous to all turns (so no wave effects). The leakage L is not, but it's at air permeability.
As far as materials go, one can optimise some properties but usually with the trade-off of others or high price. Plain old M6 should work fine if the driving impedance is kept low enough to overcome inductive current, losses and variable Mu. Keeping the grain orientation in the right direction is a big big plus (toroids, cut cores, some exotic long E double I lams) Adding some permalloy lams in with M6 sounds like an interesting idea, don't know if that puts a noticeable kink in the response when it saturates though. Would also make DC balance more critical.
Ferrite material requires around 4X the material for the same power, making winding optimisation a problem. The HF carrier schemes (Berning or others like mine) can overcome the ferrite material bulk limitation and remove hysteresis even, but suffers from leakage L problems at the high carrier frequency. Then it needs filtering, making for feedback or damping problems. There is also a switched capacitor OT scheme, but that's getting off subject.
As far as building the amp to optimise the OT, the Circlotron is the best due to half the turns and the lack of center tapped side to side (P-P) class B coupling problems. And then using feedbacks from the plates to optimise the electronics linearity without OT bandwidth concerns. Then just enough global Fdbk to fix the xfmr deficiencies, freq. response, damping factor...
For SE xfmrs, there are adequate circuit design means now to get the same results using a P-P OT (anti-triode schemes) which give twice the power with no air gapping or inductance problems or weight issues, or bandwidth issues, or cost issues or power supply issues.... SE OTs should be relegated to the dust bin of history in my opinion.
There are some exotic xfmr techniques which derive from broadband xfmrs for RF. Not seen in audio, but adaptable (not cheap though). Everyone can make a high performance 1:1 bifilar transformer. You can just take N 1:1 bifilar xfmrs and connect the secondaries in parallel using common mode chokes. The primary sides get connected in series (good capacitive isolation from the core required). The common mode chokes have to increase in current handling capability as they get nearer the speaker terminals, and each has to handle the AC voltage across a single winding level.
This scheme obviously uses a lot more copper than a normal OT, and a considerably bigger core to fit it all on (the common mode inductors and all the 1:1's can be put on the same core as long as they are phased correctly and adequately isolated capacitance wise.)
-------------------------------------------------------------------------------
Another RF approach is N transmission lines with magnetic cores strung over the outside along the length (forming common mode chokes effectively, these can also just be wound on a bigger core).
On one end the transmission lines they get series connected for the high Z primary, and on the other end the lines get parallel connected for the speaker termination. Autotransformer mode obviously.
It turns out there is a lot of redundancy in this transmission line setup that can be simplified. Less common mode choking is required for the lines nearest the speaker ground terminal. So a triangular section of cores (and line length) can be removed from the array. Then close observation of the voltages on the lines shows that the longer lines only need to work from where the last line left off. So then cutting the line sections to equal lenths and rewiring in a ladder like or step like manner.
What remains can be shown to be equivalent to N bifilar xfmrs, with the two bifilar windings connected in series (ie, center tapped bifilar). Each bifilar xfmr has one of it's windings connected to the adjacent xfmr top side and the other winding to the bottom side xfmr, forming a step like ladder thru all the xfmrs. Then one can consider the connected windings between the adjacent xfmrs to be the same winding if the xfmrs are all combined on the same large core.
Final result of these manipulations is a pyramid (cross section) of adjacent bifilar windings (but no proximity between non adjacent layers in the pyramid). This can be physically realized by a pyramid (cross section) of small wires. Starting with one wire at the peak, then two side by side wires in the next layer down, then 3 in the next layer, then 4 ... till N wires across in the bottom layer of the pyramid. All wires in a layer are paralleled. The ends of each layer get the ladder connections from earlier (now looks like just an auto-xfmr).
The bottom layer is the final high current secondary winding for the speaker terminals. The triangular bundle of wires then gets wound on the core so that it has enough turns for the secondary. Good insulation is required between adjacent bundles on the core or between multiple bundle layers on the core. Obviously one will want to make use of the triangular gaps left between the bundle turns on the core by winding a second bundle layer (with inverted triangle cross section) on top in the gaps (but insulated between the bundles).
This arrangement gives bifilar like coupling across N:1 turns ratio with transmission line impedance along the way varying in step fashion from high Z to low Z (optimum for RF to avoid reflections). Distributed capacitance is incorporated into the transmission line Z. Leakage inductance is extremely low due to the bifilar coupling everywhere. No more copper is used than an ordinary OT, but good luck trying to wind this monster!
On the splitting of C cores, I can see where that would help equalize the magnetic path lengths. The toroid doesn't have that option, but it's performance on leakage L (and low distributed C if progressive wound or other low C variants) is still going to be tops. Using a long single C core set with two thin windings (each side) would give pretty good low leakage results. (long thin windings give low leakage as long as the core is not getting near saturation, this forces the air leakage pathes to be as long as the iron path, so the common high Mu iron path gets preference, the toroid is the best case if it's not overwound)
I've never seen any toroids made from multiple lamination strips, always a long single strip. I suppose that might be a technique for very weakly gapping it. I would think that most partial gapped toroids just use some interleave film or a coating on one side of the lam. strip. Actual cut toroids would need some darn strong bonding material to keep it from springing apart or resonating.
I'm not really following BudP on the "antenna event" thing. I know that the leakage L and distributed C need to have their Z matched to the effective load Z at their resonance freq. for critical damping. So optimising just one or the other will cause ringing.
For a bifilar wound setup, there could be traveling wave phenomina, but audio is too low a freq. to be of concern typically. Distributed capacitance loading is the big issue, so either matched signals needed as for bifilar or electrical isolation for the usual case.
It's true that the wave velocity of a transmission line is slowed by c/Sqrt(Mu,Eo) [relative values], (which could make audio frequency wavelengths a concern, 100X slowing with 3000 Mu and 3 Epsilon) but the magnetic core coupling in the xfmr is simultaneous to all turns (so no wave effects). The leakage L is not, but it's at air permeability.
As far as materials go, one can optimise some properties but usually with the trade-off of others or high price. Plain old M6 should work fine if the driving impedance is kept low enough to overcome inductive current, losses and variable Mu. Keeping the grain orientation in the right direction is a big big plus (toroids, cut cores, some exotic long E double I lams) Adding some permalloy lams in with M6 sounds like an interesting idea, don't know if that puts a noticeable kink in the response when it saturates though. Would also make DC balance more critical.
Ferrite material requires around 4X the material for the same power, making winding optimisation a problem. The HF carrier schemes (Berning or others like mine) can overcome the ferrite material bulk limitation and remove hysteresis even, but suffers from leakage L problems at the high carrier frequency. Then it needs filtering, making for feedback or damping problems. There is also a switched capacitor OT scheme, but that's getting off subject.
As far as building the amp to optimise the OT, the Circlotron is the best due to half the turns and the lack of center tapped side to side (P-P) class B coupling problems. And then using feedbacks from the plates to optimise the electronics linearity without OT bandwidth concerns. Then just enough global Fdbk to fix the xfmr deficiencies, freq. response, damping factor...
For SE xfmrs, there are adequate circuit design means now to get the same results using a P-P OT (anti-triode schemes) which give twice the power with no air gapping or inductance problems or weight issues, or bandwidth issues, or cost issues or power supply issues.... SE OTs should be relegated to the dust bin of history in my opinion.
There are some exotic xfmr techniques which derive from broadband xfmrs for RF. Not seen in audio, but adaptable (not cheap though). Everyone can make a high performance 1:1 bifilar transformer. You can just take N 1:1 bifilar xfmrs and connect the secondaries in parallel using common mode chokes. The primary sides get connected in series (good capacitive isolation from the core required). The common mode chokes have to increase in current handling capability as they get nearer the speaker terminals, and each has to handle the AC voltage across a single winding level.
This scheme obviously uses a lot more copper than a normal OT, and a considerably bigger core to fit it all on (the common mode inductors and all the 1:1's can be put on the same core as long as they are phased correctly and adequately isolated capacitance wise.)
-------------------------------------------------------------------------------
Another RF approach is N transmission lines with magnetic cores strung over the outside along the length (forming common mode chokes effectively, these can also just be wound on a bigger core).
On one end the transmission lines they get series connected for the high Z primary, and on the other end the lines get parallel connected for the speaker termination. Autotransformer mode obviously.
It turns out there is a lot of redundancy in this transmission line setup that can be simplified. Less common mode choking is required for the lines nearest the speaker ground terminal. So a triangular section of cores (and line length) can be removed from the array. Then close observation of the voltages on the lines shows that the longer lines only need to work from where the last line left off. So then cutting the line sections to equal lenths and rewiring in a ladder like or step like manner.
What remains can be shown to be equivalent to N bifilar xfmrs, with the two bifilar windings connected in series (ie, center tapped bifilar). Each bifilar xfmr has one of it's windings connected to the adjacent xfmr top side and the other winding to the bottom side xfmr, forming a step like ladder thru all the xfmrs. Then one can consider the connected windings between the adjacent xfmrs to be the same winding if the xfmrs are all combined on the same large core.
Final result of these manipulations is a pyramid (cross section) of adjacent bifilar windings (but no proximity between non adjacent layers in the pyramid). This can be physically realized by a pyramid (cross section) of small wires. Starting with one wire at the peak, then two side by side wires in the next layer down, then 3 in the next layer, then 4 ... till N wires across in the bottom layer of the pyramid. All wires in a layer are paralleled. The ends of each layer get the ladder connections from earlier (now looks like just an auto-xfmr).
The bottom layer is the final high current secondary winding for the speaker terminals. The triangular bundle of wires then gets wound on the core so that it has enough turns for the secondary. Good insulation is required between adjacent bundles on the core or between multiple bundle layers on the core. Obviously one will want to make use of the triangular gaps left between the bundle turns on the core by winding a second bundle layer (with inverted triangle cross section) on top in the gaps (but insulated between the bundles).
This arrangement gives bifilar like coupling across N:1 turns ratio with transmission line impedance along the way varying in step fashion from high Z to low Z (optimum for RF to avoid reflections). Distributed capacitance is incorporated into the transmission line Z. Leakage inductance is extremely low due to the bifilar coupling everywhere. No more copper is used than an ordinary OT, but good luck trying to wind this monster!
Last edited:
Antenna event?
When a current passes through, a wire a magnetic field is emitted. When the current vector is altered the magnetic field ceases and an electrostatic field is emitted. If another plate, at a different potential, is adjacent to the emitting wire it receives both of these fields, one directly and one through a dielectric conductor. Being in a ferrous bounding perimeter collapses and concentrates the magnetic field 1000 times permeability but the same antenna event occurs. Even autoformers work in this fashion. Due to their diverse and conflicting fields autoformers are the least useful audio transformer devices. They are fine for power transformers.
Does this help?
Bud
When a current passes through, a wire a magnetic field is emitted. When the current vector is altered the magnetic field ceases and an electrostatic field is emitted. If another plate, at a different potential, is adjacent to the emitting wire it receives both of these fields, one directly and one through a dielectric conductor. Being in a ferrous bounding perimeter collapses and concentrates the magnetic field 1000 times permeability but the same antenna event occurs. Even autoformers work in this fashion. Due to their diverse and conflicting fields autoformers are the least useful audio transformer devices. They are fine for power transformers.
Does this help?
Bud
- Status
- Not open for further replies.