OK, so your definition of 'non-zero' is 'zero'. Doesn't that make arithmetic somewhat difficult for you?
Some charts with Si steel initial permeability (1955). I would guess that more recent Si steels perform better.
Seeing the 3% Si steel perm. drop approx. linearly with excitation, this would indicate that the magnetizing current stays about the same regardless of signal amplitude.
The voltage transfer of the signal itself through the core is still guaranteed by Faraday's law. So if some circuit is losing the signal, then it obviously can't handle the steady magnetizing current. That's what happens with high Zout. Other than lowering Zout (the most practical solution), one could inject a small high freq. AC carrier (above the audio band) into the amplifier to improve the low level response. (like tape recorders used to use)
Seeing the 3% Si steel perm. drop approx. linearly with excitation, this would indicate that the magnetizing current stays about the same regardless of signal amplitude.
The voltage transfer of the signal itself through the core is still guaranteed by Faraday's law. So if some circuit is losing the signal, then it obviously can't handle the steady magnetizing current. That's what happens with high Zout. Other than lowering Zout (the most practical solution), one could inject a small high freq. AC carrier (above the audio band) into the amplifier to improve the low level response. (like tape recorders used to use)
Last edited:
So to less induction because there is to less permeability.
I don't know what kind of material the manufacure of the transformer in post 21 is, just M4 EI laminations. My experiance with most EI transformers is that they are less good then the c-core versions.
Measurements at a SE transformer with HiB material shows also a drop in induction at low level low frequency signals but far more less then pp versions.
Hysteresis/ permeability problems
I don't know what kind of material the manufacure of the transformer in post 21 is, just M4 EI laminations. My experiance with most EI transformers is that they are less good then the c-core versions.
Measurements at a SE transformer with HiB material shows also a drop in induction at low level low frequency signals but far more less then pp versions.
Hysteresis/ permeability problems
"So to less induction because there is to less permeability."
The point I have been trying to make is that the induction is independent of the core permeability. One could pull the steel core out and (except for the big leakage inductance factor then) the voltage induced in the secondary would stay the same relative to the voltage placed on the primary (if driven from a low enough Z source that could handle the then HUGE magnetizing current).
Practically speaking, a low Z source should be able to handle the magnetizing current from the 100X core permeability variation. Obviously, less perm variation IS better, but expensive = NiFe, (or Ferrite could work, both are Bmax limited). The P-P case starts out with 10X less magnetizing current than the SE case, so is easier to fix in practice. The C core or even toroid works better since the material is 100% correctly grain oriented, giving higher permeability and lower magnetizing current.
So what are you using to drive the core when your LF signal disappears? A pentode or high Rp triode will not work obviously. To use an Xmit tube (high Rp) will require a HF carrier add-on for example.
A small SE design is great for circuit simplicity, learning, easy construction .... even sounds good if you like 2nd H. But when it gets taken to mainstream HiFi with Xmit tubes and huge high Z OTs, with dangerous B+, the circuitry needs to "man up" some on the drive capability (local Fdbk). The bigger the core, the more the magnetizing current. The air gap puts in another 10X. The extra turns and turn length puts in more distributed shunt capacitance.
The point I have been trying to make is that the induction is independent of the core permeability. One could pull the steel core out and (except for the big leakage inductance factor then) the voltage induced in the secondary would stay the same relative to the voltage placed on the primary (if driven from a low enough Z source that could handle the then HUGE magnetizing current).
Practically speaking, a low Z source should be able to handle the magnetizing current from the 100X core permeability variation. Obviously, less perm variation IS better, but expensive = NiFe, (or Ferrite could work, both are Bmax limited). The P-P case starts out with 10X less magnetizing current than the SE case, so is easier to fix in practice. The C core or even toroid works better since the material is 100% correctly grain oriented, giving higher permeability and lower magnetizing current.
So what are you using to drive the core when your LF signal disappears? A pentode or high Rp triode will not work obviously. To use an Xmit tube (high Rp) will require a HF carrier add-on for example.
A small SE design is great for circuit simplicity, learning, easy construction .... even sounds good if you like 2nd H. But when it gets taken to mainstream HiFi with Xmit tubes and huge high Z OTs, with dangerous B+, the circuitry needs to "man up" some on the drive capability (local Fdbk). The bigger the core, the more the magnetizing current. The air gap puts in another 10X. The extra turns and turn length puts in more distributed shunt capacitance.
Last edited:
If i test a 15kOhm / 500Ohm line output transformer (post 21) i test with a tube that people actually use for such transformer. So i tested it with a 6sn7.
It makes no sense to use a low output impedance wich people not use normally with tube (output)transformers
The se transformers i test with a ccs and serie resistor ( 800 Ohm if i test a 300B transformer)
It makes no sense to use a low output impedance wich people not use normally with tube (output)transformers
The se transformers i test with a ccs and serie resistor ( 800 Ohm if i test a 300B transformer)
I see i made a "small" mistake by using the word induction while i had to use the word inductance.... Sorry for that.
In 'classA' the bias is modulated by the a.c. and never reverse. I can't understand why oriented atoms would reverse the polarity when the bias is never reversed. Only the intensity of the magnetization force changes without reversing the polarity, what happens in the secondary is a reverse of polarity but is not the force causing the signal. The secondary loss is not hysteresis.
There is loss of the signal due to hysteresis because it takes some signal energy to reverse the polarity, it is the Remanence. If you stop a positive signal mmf the immediate effect is that the transformer will give you back more energy and stay magnetized a little, to produce a negative polarity you will have to 'spend' waste some negative signal to bring to 0 that Remanence first, which doesn't happen in S.e.t.
I am trying to understand where you learned or concluded of hysteresis when the mmf is only positive and never stops or reverse.
This is the article I found and it agrees with what I am saying : http://www.hawthorneaudio.com/photos/albums/userpics/11473/SE-v-PP-Part2.pdf (www.hawthorneaudio.com)
Last edited:
NO! Not even wrong. You are confusing current effects with voltage effects. Your model of field variation is non-physical.
I just looked at the hawthorneaudio pdf link, and where it starts with "The Hidden Glory of the SE OPT", it is pretty much ALL wrong. This is what I mean about mis-information being spread everywhere in the audio realm.
The electrons that make the internal (permeability enhanced) magnetic field in magnetic materials act as fixed strength magnetic dipoles. They do not become stronger and weaker dipoles pointed in the same direction. How can the magnetic field change strength if they do not change directions? This is what magnetic domains and domain wall motion entail, electron spin flips. The energy to overcome local anisotropic electron environments to flip the spins causes hysteresis.
You should consider any information on magnetics in the audio internet realm as highly suspect, especially SET stuff, without first checking with some 1st rate magnetics or physics references.
I'm not going to repeat the explanations any more. It is obviously nearly impossible to straighten out audio brainwashed individuals.
I just looked at the hawthorneaudio pdf link, and where it starts with "The Hidden Glory of the SE OPT", it is pretty much ALL wrong. This is what I mean about mis-information being spread everywhere in the audio realm.
The electrons that make the internal (permeability enhanced) magnetic field in magnetic materials act as fixed strength magnetic dipoles. They do not become stronger and weaker dipoles pointed in the same direction. How can the magnetic field change strength if they do not change directions? This is what magnetic domains and domain wall motion entail, electron spin flips. The energy to overcome local anisotropic electron environments to flip the spins causes hysteresis.
You should consider any information on magnetics in the audio internet realm as highly suspect, especially SET stuff, without first checking with some 1st rate magnetics or physics references.
I'm not going to repeat the explanations any more. It is obviously nearly impossible to straighten out audio brainwashed individuals.
Last edited:
Is there already a transmission line transformer design for an 2500/8 Ohm pp ( or se ) ?
With so many knowledge here we could expect it soon i guess.......
With so many knowledge here we could expect it soon i guess.......
Thanks for staying with us poor brainwashed individuals.
You expressed it very well, the energy to overcome local anisotropic electron environments to flip the spins causes hysteresis. I take my information from a university text book but it is very basic to apply to audio transformers.
What I learned I learned it with distortion meters, arbitrary waveforms and high resolution equipments, not with internet links, but as a starting point, with the conviction that even if the orientation is bad I can correct the path and learn painfully from the mistakes.
My problem was to call the phenomenon legitimately hysteresis, when the phenomenon is encountered in a non reversal of the magnetic field of the primary. This was not clear to me.
For you the S.e.t. is a return to the stone age right?
You expressed it very well, the energy to overcome local anisotropic electron environments to flip the spins causes hysteresis. I take my information from a university text book but it is very basic to apply to audio transformers.
What I learned I learned it with distortion meters, arbitrary waveforms and high resolution equipments, not with internet links, but as a starting point, with the conviction that even if the orientation is bad I can correct the path and learn painfully from the mistakes.
My problem was to call the phenomenon legitimately hysteresis, when the phenomenon is encountered in a non reversal of the magnetic field of the primary. This was not clear to me.
For you the S.e.t. is a return to the stone age right?
Thanks for hanging on! 🙂
Yes, usual SET = Stone Age, at least for me.
But it DOES have its place for learning, simplicity (circuit), hobby use, and interesting sound for some. I just get annoyed when I see it being pushed as the "ultimate" with extremely expensive exotic xfmrs.
By the way, to make a super "SET", just use a P-P class A design with a triode on one (output) side, and a pentode on the other side. Can leave the pentode as an undriven CCS, -OR- for twice the power output, drive the pentode (as P-P mode) too.
The triode sets the output voltage by means of its low Rp running the show.
---------------------------------------------------------------
On that transmission line OT. Here is the derivation starting from the conventional RF transmission line xfmr and reducing to an audio version using 1:1 bifilar common mode chokes. These can all be put on the same core for best efficiency (best input inductance) and economy. Suggest winding as sections on a sectioned bobbin. Takes two of these in parallel to make P-P. Does NOT like DC, so use P-P for current balance.
As configured, it is NON DC isolating (like an autoformer). Bandwidth will seriously blow away ANY audio OT by orders of magnitude. (Still requires sufficient total input inductance for the LF end like usual OT. Inductance figured across all 4, or N, sections in series.,) Wire should increase in diameter from the HV input to the LV output by section to handle the increased current. (could use multiple parallel strands of same size wire, for both 1:1 windings, ie: 1,2,3,4 strands...left to right) You might notice that the final reduced version resembles a forced voltage divider. Each section is strapped to be equal voltage to the others. (ordinary autoformers just HOPE for this to happen at HF)
More sections can be chained on for higher turns ratios (higher impedance ratios)
note:
Strictly designed, it would take 4 sections (instead of 3 shown) for 4:1, with the load ACROSS the 4th section.
Oops: correction! The design shown is a 4 to 1 winding ratio. That would actually be 16 to 1 impedance ratio.
Yes, usual SET = Stone Age, at least for me.
But it DOES have its place for learning, simplicity (circuit), hobby use, and interesting sound for some. I just get annoyed when I see it being pushed as the "ultimate" with extremely expensive exotic xfmrs.
By the way, to make a super "SET", just use a P-P class A design with a triode on one (output) side, and a pentode on the other side. Can leave the pentode as an undriven CCS, -OR- for twice the power output, drive the pentode (as P-P mode) too.
The triode sets the output voltage by means of its low Rp running the show.
---------------------------------------------------------------
On that transmission line OT. Here is the derivation starting from the conventional RF transmission line xfmr and reducing to an audio version using 1:1 bifilar common mode chokes. These can all be put on the same core for best efficiency (best input inductance) and economy. Suggest winding as sections on a sectioned bobbin. Takes two of these in parallel to make P-P. Does NOT like DC, so use P-P for current balance.
As configured, it is NON DC isolating (like an autoformer). Bandwidth will seriously blow away ANY audio OT by orders of magnitude. (Still requires sufficient total input inductance for the LF end like usual OT. Inductance figured across all 4, or N, sections in series.,) Wire should increase in diameter from the HV input to the LV output by section to handle the increased current. (could use multiple parallel strands of same size wire, for both 1:1 windings, ie: 1,2,3,4 strands...left to right) You might notice that the final reduced version resembles a forced voltage divider. Each section is strapped to be equal voltage to the others. (ordinary autoformers just HOPE for this to happen at HF)
More sections can be chained on for higher turns ratios (higher impedance ratios)
note:
Strictly designed, it would take 4 sections (instead of 3 shown) for 4:1, with the load ACROSS the 4th section.
Oops: correction! The design shown is a 4 to 1 winding ratio. That would actually be 16 to 1 impedance ratio.
Last edited:
Late edit deleted the images for some reason....
Fixed here.
(Oh, and can ignore my above comment about strictly designed version with 4 sections. Load is already across a section.)
The transmission line OT derivation:
Fixed here.
(Oh, and can ignore my above comment about strictly designed version with 4 sections. Load is already across a section.)
The transmission line OT derivation:
Last edited:
Build both SE and PP - only then will you know what is best for you.
IME its not the absolute level of harmonic distortion WRT 0%, but the balance between odd and even distortions and the way in which they reduce with increasing harmonics of the fundamental frequency.
As soon as you cancel 2H, 3H becomes more prominent. But, there are more than one ways to skin a cat. Although, invariably the cat ends up skinned and someone wears a new coat.
Also, to say that with a decent interstage (probably not Hammond), SE to PP grid drive captures more of the best of both worlds than either alone, but SE does do some stuff better.
Build both, otherwise you will be reading at forums like this and forever wondering if the grass is really greener OTOS.
LH/S
IME its not the absolute level of harmonic distortion WRT 0%, but the balance between odd and even distortions and the way in which they reduce with increasing harmonics of the fundamental frequency.
As soon as you cancel 2H, 3H becomes more prominent. But, there are more than one ways to skin a cat. Although, invariably the cat ends up skinned and someone wears a new coat.
Also, to say that with a decent interstage (probably not Hammond), SE to PP grid drive captures more of the best of both worlds than either alone, but SE does do some stuff better.
Build both, otherwise you will be reading at forums like this and forever wondering if the grass is really greener OTOS.
LH/S
I already found this information on the internet but this is completely useless for tube output transformers.
To less winding ratio and no dc isolation.
To less winding ratio and no dc isolation.
OK. You don't understand hysteresis.gabdx said:
Hysteresis simple means that when I do a change, then do the opposite change, the path taken (in this case by flux - but it could be something else) in the second case is not simply the reverse of the path in the first case.
SET is what people used to do when they hadn't thought of PP or wanted something superficially simple (and, sometimes, cheap). It lived on in the form of SEP as the output stage in most domestic radios and gramophones, where fidelity was not a requirement but 'nice tone' was a requirement. I suspect it still lives on where 'nice tone' is a requirement, but now people are more sophisticated and have learnt to deny this.
I have never wondered about SET, as far as I can remember. I have wondered about SET fans.Ludwig Haus said:
SET is superficially simpler than PP, but in reality is more complex because the PSU is more tightly coupled to the output stage. In a SET the whole signal current goes through the last PSU cap. In Class A PP only distortion products (much lower level) go through the cap so it has less work to do. In Class B PP the cap sees a full-wave rectified version of the signal, but at least this doubles the frequencies so again easier for the cap to look like a low impedance.
Most set lovers already had those "0" distortion transistor amps and probebly pp tube amps befor. For some reason they like the set more. Nothing wrong with that i guess.
For my esl i need a pp amp just because the esl is pp and not se. I can make a pp transformer for this amp or even 2 se in pp mode.
For my esl i need a pp amp just because the esl is pp and not se. I can make a pp transformer for this amp or even 2 se in pp mode.
"this is completely useless for tube output transformers"
You can always use a bifilar 1:1 transformer on the secondary (with good HV wire insulation) to isolate.
In the case of the reduced transmission line xfmr, one just uses a trifilar winding on the last (low Z) common mode choke. (with good HV wire insulation for the isolated secondary)
However, isolation is not even necessary for tube output if done right. Use 0V (ground) at the center tap position of the P-P xfmr, with the output tube cathodes at -B voltage. Heaters and bias V have to be floated at -B too. And the grids get the usual cap coupling from their drivers. (clamp diodes across the grid-cathode for turn-on transient)
Circlotron or Totem-pole outputs can be used also.
There are other optimizations of the transmission line OT as well, but similar issues with isolation exist there.
You can always use a bifilar 1:1 transformer on the secondary (with good HV wire insulation) to isolate.
In the case of the reduced transmission line xfmr, one just uses a trifilar winding on the last (low Z) common mode choke. (with good HV wire insulation for the isolated secondary)
However, isolation is not even necessary for tube output if done right. Use 0V (ground) at the center tap position of the P-P xfmr, with the output tube cathodes at -B voltage. Heaters and bias V have to be floated at -B too. And the grids get the usual cap coupling from their drivers. (clamp diodes across the grid-cathode for turn-on transient)
Circlotron or Totem-pole outputs can be used also.
There are other optimizations of the transmission line OT as well, but similar issues with isolation exist there.
Last edited:
Agreed. Problems only come when people try to claim that SET distortion is somehow closer to the original sound than less distortion from a PP, or invent spurious arguments that try to claim thet SET has less distortion than PP.esltransformer said:
If there was a relation between soundquality vs distortion then all people would buy the amplifier with the lowest distorion.
- Status
- Not open for further replies.