Well, thanks, now I understand a little better.
When the core magnetization is symmetric, the magnetization current must also be symmetric, then its Fourier expansion only has odd harmonics.
If we unbalanced symmetrical AC magnetization with DC magnetization, the magnetization current must also be asymmetric, then its Fourier expansion has both, odd and even harmonics.
But the relation between B and H in ferromagnetic materials is not linear.
B = H + 4π M (Gaussian units)
Then the magnetization current can't be sinusoidal, but it can be periodic, and indeed it is.
This is consistent with the above reasoning, which IS the case with ANY electrical steel.
Bingo! You're right!
Last edited:
At the risk of shooting myself in the mouth (where foot usually resides) There is an ebayer selling toroid GOSS cores of various sizes that look perfect for making toroidd outputs. He includes the TPV data and a lot of other data needed to design your own transformer. I bought a couple 500VA and he has them available from 200VA to 1500VA. I'm in no way connected with him and don't know enough about toroids to even say these would work, but one of you may be able to. If so, these may be a good deal.
Toroidal Laminated Core for 50 60Hz AC Power Transformer Wind Your Own 7 Lb | eBay
Toroidal Laminated Core for 50 60Hz AC Power Transformer Wind Your Own 7 Lb | eBay
I didn't see anyone talking about lamination thickness. Osvaldo's brilliant use of commercial toroids intended for AC operation at 60 Hz (or 50 Hz?) seems to me a bit of a waste of his valuable time, because the laminations in a core that's intended for AC 60 Hz operation, whether toroid, C, R, O or E, are going to be relatively thick - probably .30 mm - and, if I understand correctly, such thick laminations are going to cause large losses at higher frequencies - and therefor are suited for 60 Hz and not a heckuvalot higher. As a matter of fact, I talked to the engineer at Maryland Toroid once about using their toroid cores at 400 Hz (military), because I wanted to build a more efficient power supply with a smaller mains transformer (the military isn't dumb when it comes to making battlefield equipment lighter), and he cautioned me that at 400 Hz the core, with it's thick laminations intended for 60 Hz, would heat up a lot more at 400 Hz and that in the end I wouldn't really gain anything.
Upon looking into this further, I did find out that Alpha-Core, among others, does make cores with laminations .05 mm thick (thin), you just have to either special order or find a core size that happens to come with the option of the .05 mm lams.
And, I have to agree, with C-cores so easy to get and use, I just don't see any reason at all for toroids, they really ain't all they're cracked up to be, and what a PAIN to wind. I still have a big one I started winding, oh, five years ago, and I just fricking finally gave up, it was too frustrating and way too boring LOL....
Very Best, Synchro27
Upon looking into this further, I did find out that Alpha-Core, among others, does make cores with laminations .05 mm thick (thin), you just have to either special order or find a core size that happens to come with the option of the .05 mm lams.
And, I have to agree, with C-cores so easy to get and use, I just don't see any reason at all for toroids, they really ain't all they're cracked up to be, and what a PAIN to wind. I still have a big one I started winding, oh, five years ago, and I just fricking finally gave up, it was too frustrating and way too boring LOL....
Very Best, Synchro27
Lamination thickness
Lamination thickness around 0.3mm in a OPT is a good compromiss. Why? Because at the low frequency end you need all the iron you can get, you need a good fillfactor. At higher frequencies you dont need thinner lamination because flux goes down as frequency goes up. Things are totally different in a power transformer working at a single frequency and maxed out flux.
Lamination thickness around 0.3mm in a OPT is a good compromiss. Why? Because at the low frequency end you need all the iron you can get, you need a good fillfactor. At higher frequencies you dont need thinner lamination because flux goes down as frequency goes up. Things are totally different in a power transformer working at a single frequency and maxed out flux.
gorgon53, your point about fill factor is a very good one.
I think, though, I prefer to use the .05 mm in a GOSS C-core, and just over-size to get the fill factor I need for the base.
I plan to use a special winding technique with space between each wire to minimize capacitance, which will require a larger winding window anyway, so I have to oversize the core anyway.
I think, though, I prefer to use the .05 mm in a GOSS C-core, and just over-size to get the fill factor I need for the base.
I plan to use a special winding technique with space between each wire to minimize capacitance, which will require a larger winding window anyway, so I have to oversize the core anyway.
a function of the wire circular mils, you can parallel several strands to get more ampacity and get your target efficiency in the windings.....
Dear colleagues.
This is my first post on this resource.
I apologize for the error in English.
Regarding the benefits of amorphous materials of steel, I would not be optimistic unambiguous. Because of their dignity as a linear hysteresis loop in amorphous materials are greatly exaggerated. Hinge is really linear, but its contribution to the sound quality is not significant because it affects only the nonlinearity of magnetizing current. Meanwhile, the value itself of magnetizing current transformers in a well-designed iron core even at extremely low. A distortion in the tubes themselves compared to these distortions of the magnetizing current and their effect on the sound, extremely high, reaching more than 1%. Of course good to have a free option in the form of a linear magnetizing current. But if the value of the magnetizing current is negligible distortion and do so more will be void. Actually, advantage of amorphous materials are very low loss, high permeability in a wide frequency range that far exceeds the requirements even to audio transformers. In general, transformers Steel HI-B magnetic cores thickness of 0.08mm - 0.1mm (and even for EI to 0.3mm thickness) in the output transformers are successfully competing with amorphous as it is now so widely used much earlier.
The result - in audio transformer advantageous to reduce the amount of magnetizing current, reduce losses, have a high permeability over a wide frequency range. The very important to have a very high linearity of the hysteresis loop is not strange to be on the last place in this list.
This is my first post on this resource.
I apologize for the error in English.
Regarding the benefits of amorphous materials of steel, I would not be optimistic unambiguous. Because of their dignity as a linear hysteresis loop in amorphous materials are greatly exaggerated. Hinge is really linear, but its contribution to the sound quality is not significant because it affects only the nonlinearity of magnetizing current. Meanwhile, the value itself of magnetizing current transformers in a well-designed iron core even at extremely low. A distortion in the tubes themselves compared to these distortions of the magnetizing current and their effect on the sound, extremely high, reaching more than 1%. Of course good to have a free option in the form of a linear magnetizing current. But if the value of the magnetizing current is negligible distortion and do so more will be void. Actually, advantage of amorphous materials are very low loss, high permeability in a wide frequency range that far exceeds the requirements even to audio transformers. In general, transformers Steel HI-B magnetic cores thickness of 0.08mm - 0.1mm (and even for EI to 0.3mm thickness) in the output transformers are successfully competing with amorphous as it is now so widely used much earlier.
The result - in audio transformer advantageous to reduce the amount of magnetizing current, reduce losses, have a high permeability over a wide frequency range. The very important to have a very high linearity of the hysteresis loop is not strange to be on the last place in this list.
Yes, worrying about hysteresis is like worrying about distortion of the (magnetizing current) distortion (which is not even a voltage transfer distortion!). (like a few % of 1% = 0.01%) Just use low output impedance from the output tubes to minimize the effect (use some local N Fdbks).
Unless you have a top notch toroid winding machine available (capable of progressive winding), then the first order of business is to reduce the number of turns on the toroid. Use parallel tubes or high current tubes, like big TV Sweep tubes, with a lower B+. Next use the Circlotron configuration, so again 1/2 the turns needed. 3000 turn toroid windings are out of the question by hand, especially with any kind of winding optimizations. Grid 2 drive (or Twin/Crazy drive) of TV Sweep tubes is far more linear than "audio" tubes driven by grid 1.
One can get 6 inch OD, 2.7 inch ID, ferrite cores, which could be stacked onto the side(s) of a matching M6 lamination toroid to provide HF capability and high Mu at small signals, much like permalloy striping when added to E-I cores. Or could get a thin permalloy core to stack along side, but these are rare/expensive in custom large sizes.
Unless you have a top notch toroid winding machine available (capable of progressive winding), then the first order of business is to reduce the number of turns on the toroid. Use parallel tubes or high current tubes, like big TV Sweep tubes, with a lower B+. Next use the Circlotron configuration, so again 1/2 the turns needed. 3000 turn toroid windings are out of the question by hand, especially with any kind of winding optimizations. Grid 2 drive (or Twin/Crazy drive) of TV Sweep tubes is far more linear than "audio" tubes driven by grid 1.
One can get 6 inch OD, 2.7 inch ID, ferrite cores, which could be stacked onto the side(s) of a matching M6 lamination toroid to provide HF capability and high Mu at small signals, much like permalloy striping when added to E-I cores. Or could get a thin permalloy core to stack along side, but these are rare/expensive in custom large sizes.
Last edited:
Perhaps it is not clearly written.
I'll try to write short sentences.
The transfer function of the voltage transformer, no permeability.
Consequently the output transformer linearity of the hysteresis loop in practice either has no effect.
The non-linearity of the hysteresis loop can affect only the magnetizing current.
However, the magnetizing current is very small compared to the load current.
A load current of 10 000% non-linear magnetizing current.
But most importantly, according to the amplitude of the load current is also much more magnetizing current.
Therefore, the load current is much more to contribute to the total current distortion. But even this sum of both currents are not taken into account in practice. And it is accepted as self-evident. And most importantly, the lamp itself in the voltage amplifier introduces distortion much more. The distortion can affect the lamp a very small magnetizing current (compared to the load current) flowing through the lamp. So why so much talk about the influence of non-linearity in the current distortion of the magnetization loop? That is a lot of talk about the slightest impact. But most importantly, that in fact, where large distortions are silent.
In Russia there are powerful lamps 6C33C. Their internal resistance of 100 ohms is best if you turn on the load in the anode. And if you include all the loads 25Om in the cathode. I did in 1996, they are PP amplifier. The load was included in the cathodes. Then my transformer was a small C-core on the core iron tape thickness of 0.25 mm. primary inductance - 9H. For example, Japanese Transformer SE XE-20-600S for 6C33C had primary inductance - 2.2H.
In Russia, there is a choice of any modern magnetic materials, HI-B 0.08mm-0.1mm - 0.15mm. Amorphous nanocrystalline and permalloy too. But permalloy is very expensive. It has a low demand and is considered obsolete, not because of poor sound quality, but because of its high price. Mix different magnetic materials in the magnetic circuit, it makes no sense. A material which has large losses will degrade the advantages of other materials combined in this magnetic circuit. The most preferred use of nanocrystalline material. The only problem is that you need to choose the right for each particular case, as well as pipes, the best type of material.
I'll try to write short sentences.
The transfer function of the voltage transformer, no permeability.
Consequently the output transformer linearity of the hysteresis loop in practice either has no effect.
The non-linearity of the hysteresis loop can affect only the magnetizing current.
However, the magnetizing current is very small compared to the load current.
A load current of 10 000% non-linear magnetizing current.
But most importantly, according to the amplitude of the load current is also much more magnetizing current.
Therefore, the load current is much more to contribute to the total current distortion. But even this sum of both currents are not taken into account in practice. And it is accepted as self-evident. And most importantly, the lamp itself in the voltage amplifier introduces distortion much more. The distortion can affect the lamp a very small magnetizing current (compared to the load current) flowing through the lamp. So why so much talk about the influence of non-linearity in the current distortion of the magnetization loop? That is a lot of talk about the slightest impact. But most importantly, that in fact, where large distortions are silent.
In Russia there are powerful lamps 6C33C. Their internal resistance of 100 ohms is best if you turn on the load in the anode. And if you include all the loads 25Om in the cathode. I did in 1996, they are PP amplifier. The load was included in the cathodes. Then my transformer was a small C-core on the core iron tape thickness of 0.25 mm. primary inductance - 9H. For example, Japanese Transformer SE XE-20-600S for 6C33C had primary inductance - 2.2H.
In Russia, there is a choice of any modern magnetic materials, HI-B 0.08mm-0.1mm - 0.15mm. Amorphous nanocrystalline and permalloy too. But permalloy is very expensive. It has a low demand and is considered obsolete, not because of poor sound quality, but because of its high price. Mix different magnetic materials in the magnetic circuit, it makes no sense. A material which has large losses will degrade the advantages of other materials combined in this magnetic circuit. The most preferred use of nanocrystalline material. The only problem is that you need to choose the right for each particular case, as well as pipes, the best type of material.
I think we are saying the same thing.
Hysteresis is insignificant compared to magnetizing current, and low Zout from the tubes can handle the magnetizing current well. Either big high current tubes or local N Fdbk for fixing that. And the magnetizing current does not directly impose non-linearity onto the xfmr voltage tranfer function, only by corrupting the voltage output of the power tubes by additional loading. (relatively small additional unless approaching total core saturation)
On the mixed materials, your point about losses is correct, agreed. The high Mu material will saturate well before the hi flux material, causing some losses. However, only a small amount of the hi-Mu material is generally used for linearizing the low signal amplitude region (where the Mu drops off for the high flux material). The nickel alloy pin-striping OTs for example only use a few % nickel laminations. Copper losses are probably more significant as long as full power is not at the high frequency end. (so audio OK)
The ferrite alternative for "striping" a steel laminate toroid acts similarly, only a small amount (thin toroid) of ferrite would be used relative to the steel toroid.
Even 100% steel toroids (M-6 say) deal with some saturation in the inner ID region (due to shorter magnetic path). The saturation spreads out to bigger radius as the excitation V * t increases. If one is driving the core to anything over 50% saturation (for core area), typically the inner ID material of the core is beginning to saturate. (for 2 to 1 OD to ID ratio) Toroids generally get driven to much higher flux than the E-I cores, due to the lack of lamination butt end gaps, which saturate prematurely at 50% max flux. So toroid ID saturation is nearly standard operating procedure.
Low source impedance, or low Zout again, can fix the Mu droop problem as well (for the low amplitude signal region) without the "striping" mod. So "striping" the core in either form is something of a "gold plating" type addition. Paid more, must be better....old story. The real problem is that few traditional VT Amplifiers are designed specifically with low Zout in mind. You have to put this into the design from the get-go.
There is also that LEM (linearity enhancement) Module approach, where a small HF bias signal (above audio) is injected into the amplifier, similar to that used on tape recorders, to overcome magnetic coercive force (or so called "stiction"), primarily at low signal amplitude. Cheap if you add it yourself, may not be needed much for modern xfmr steels. Low Zout tubes would fix the problem also.
Hysteresis is insignificant compared to magnetizing current, and low Zout from the tubes can handle the magnetizing current well. Either big high current tubes or local N Fdbk for fixing that. And the magnetizing current does not directly impose non-linearity onto the xfmr voltage tranfer function, only by corrupting the voltage output of the power tubes by additional loading. (relatively small additional unless approaching total core saturation)
On the mixed materials, your point about losses is correct, agreed. The high Mu material will saturate well before the hi flux material, causing some losses. However, only a small amount of the hi-Mu material is generally used for linearizing the low signal amplitude region (where the Mu drops off for the high flux material). The nickel alloy pin-striping OTs for example only use a few % nickel laminations. Copper losses are probably more significant as long as full power is not at the high frequency end. (so audio OK)
The ferrite alternative for "striping" a steel laminate toroid acts similarly, only a small amount (thin toroid) of ferrite would be used relative to the steel toroid.
Even 100% steel toroids (M-6 say) deal with some saturation in the inner ID region (due to shorter magnetic path). The saturation spreads out to bigger radius as the excitation V * t increases. If one is driving the core to anything over 50% saturation (for core area), typically the inner ID material of the core is beginning to saturate. (for 2 to 1 OD to ID ratio) Toroids generally get driven to much higher flux than the E-I cores, due to the lack of lamination butt end gaps, which saturate prematurely at 50% max flux. So toroid ID saturation is nearly standard operating procedure.
Low source impedance, or low Zout again, can fix the Mu droop problem as well (for the low amplitude signal region) without the "striping" mod. So "striping" the core in either form is something of a "gold plating" type addition. Paid more, must be better....old story. The real problem is that few traditional VT Amplifiers are designed specifically with low Zout in mind. You have to put this into the design from the get-go.
There is also that LEM (linearity enhancement) Module approach, where a small HF bias signal (above audio) is injected into the amplifier, similar to that used on tape recorders, to overcome magnetic coercive force (or so called "stiction"), primarily at low signal amplitude. Cheap if you add it yourself, may not be needed much for modern xfmr steels. Low Zout tubes would fix the problem also.
Last edited:
Yes, at the expense of a slight influence of the hysteresis loop it looks like our opinions coincide.
On account of the use of combined magnetic cores, all the materials that you have listed steel, permalloy and ferrite, have approximately the same loss, and rejoin them to the magnetic circuit may be the case in practice. The only example in the case of application of toroids, rings of different materials and should be more accurately of the same diameter to be able to fold them together. This practice is inconvenient because it is necessary also to look for these same ring.
My thought was that if the amorphous alloy used is not necessary to mix it with steel rings e.g. due to savings in costs. The overall effect will be as pure steel. A overpay for the use of amorphous alloy is still significant. Therefore, in this case advantageous to use 100% amorphous magnetic core, not combined. Yes, its price will be higher. But the result will be much better, too.
On account of the fact that the inner layer to saturate the tori faster external is a myth. Because these two layers are inside one of their resistance to the magnetic flux of the winding aligned. Now, if separate layers for these would be used its own individual winding so that in this case would be saturated with a smaller diameter than the larger diameter faster in this case because the resistance of the magnetic flux would be different.
Returning to the subject of the alleged big distortion in the transformer due to the alleged hysteresis curve, I want to note that due to the load current is very non-linear and has a large amplitude compared to the magnetizing current. So in terms of reducing distortion in the transformer is most advantageous to direct the basic resources to reduce the losses in the magnetic core and windings. For example use of silver wire and an amorphous alloy with the smallest losses possible today. And optimized through the use of computer models of the proportions between the diameter, height and cross-section of the magnetic circuit.
On account of the use of combined magnetic cores, all the materials that you have listed steel, permalloy and ferrite, have approximately the same loss, and rejoin them to the magnetic circuit may be the case in practice. The only example in the case of application of toroids, rings of different materials and should be more accurately of the same diameter to be able to fold them together. This practice is inconvenient because it is necessary also to look for these same ring.
My thought was that if the amorphous alloy used is not necessary to mix it with steel rings e.g. due to savings in costs. The overall effect will be as pure steel. A overpay for the use of amorphous alloy is still significant. Therefore, in this case advantageous to use 100% amorphous magnetic core, not combined. Yes, its price will be higher. But the result will be much better, too.
On account of the fact that the inner layer to saturate the tori faster external is a myth. Because these two layers are inside one of their resistance to the magnetic flux of the winding aligned. Now, if separate layers for these would be used its own individual winding so that in this case would be saturated with a smaller diameter than the larger diameter faster in this case because the resistance of the magnetic flux would be different.
Returning to the subject of the alleged big distortion in the transformer due to the alleged hysteresis curve
, I want to note that due to the load current is very non-linear and has a large amplitude compared to the magnetizing current. So in terms of reducing distortion in the transformer is most advantageous to direct the basic resources to reduce the losses in the magnetic core and windings. For example use of silver wire and an amorphous alloy with the smallest losses possible today. And optimized through the use of computer models of the proportions between the diameter, height and cross-section of the magnetic circuit.
My boss became interested in the subject of making audio toroidal transformers, but is not interested in their marketing. Our company produces microprocessor equipment for testing microprocessor protection in power engineering. There are many different machines for winding toroidal transformers. From very small to power 10kW and even higher. Magnetic cores made of HI-B steel are 0.01mm thick; there are many available sizes available. In addition, there are many amorphous rings D = 204mm, d = 140mm, h = 35mm, weight 3.85kg. They can be used one by one or even stacked two for powerful tubes such as GU-72, GMI-11, GMI-83. The quality of winding can be controlled personally.
i’ve been researching winding my own torroidal for power and the idea of perhaps switching the circlotron OTL to torroid output and winding my own.
Nice progressive winding - and i have a question, was each section wound straight or stacked diagonally?
also the design steps i took from a TI guide show designing for the lowest 3dB down point for selecting the cores. Did you design for a 20Hz -3dB point?