I quoted a few posts and wanted to add to some off-topic banter (at least the phrase "double-blind test" on this forum doesn't get replaced with a "censored!" smilie or get the poster banned), but I've wondered something for a lot of years now, and it's perhaps pertinent to ask now that tape recording is gone as a commercial activity, and before the idea becomes totally lost to the confines of dead trees (and PDF scans with "RDH4" in their filenames).
Magnetic tape recording uses (or used) an ultrasonic "AC bias" to linearize the tape magnetization curve and eliminate the horrible "zero crossing" distortion that results without it. Transformers aren't anywhere nearly as bad in this respect, but the magnetization curve is still there, even in the best audio transformers. An ultrasonic "bias," perhaps added by a third winding, could effectively make this distortion go away, or reduce it by one or two orders of magnitude. The bias signal could be isolated from the input and output by putting a parallel L-C trap in series with each winding.
Has this ever been done before? Are there any negatives other than the extra circuitry? I wonder if this might add some low-level wide-band noise from the domains switching at perhaps slightly different magnetization levels each time.
Have I ever asked about this before? It's been in the back of my mind for perhaps ten years or so.
Years ago I heard a recording - actually, two recordings from the same microphone at the same time, one was with the mic's output directly to a preamp input. The mic was also connected to a Jensen 1:1 audio transformer and then to the same model preamp for the other recording. It took me a few times of listening and comparing the two recordings, but once I knew what to "listen for" I could fairly easily hear the difference.
Magnetic tape recording uses (or used) an ultrasonic "AC bias" to linearize the tape magnetization curve and eliminate the horrible "zero crossing" distortion that results without it. Transformers aren't anywhere nearly as bad in this respect, but the magnetization curve is still there, even in the best audio transformers. An ultrasonic "bias," perhaps added by a third winding, could effectively make this distortion go away, or reduce it by one or two orders of magnitude. The bias signal could be isolated from the input and output by putting a parallel L-C trap in series with each winding.
Has this ever been done before? Are there any negatives other than the extra circuitry? I wonder if this might add some low-level wide-band noise from the domains switching at perhaps slightly different magnetization levels each time.
Have I ever asked about this before? It's been in the back of my mind for perhaps ten years or so.
Years ago I heard a recording - actually, two recordings from the same microphone at the same time, one was with the mic's output directly to a preamp input. The mic was also connected to a Jensen 1:1 audio transformer and then to the same model preamp for the other recording. It took me a few times of listening and comparing the two recordings, but once I knew what to "listen for" I could fairly easily hear the difference.
benb you will want to communicate with John Attwod here
ClariSonus - Clear and enjoyable sound
John has looked at your question and has an amplifier that he explored some of these territories with. Ask, he will be delighted someone cares.
As for the "sound" of core material. Well. Commercial manganese iron, the M series of grain and non grain oriented metals, is all done with power transform at about 440 Hz or thereabouts. Depends upon chemistry, thickness and annealing correctly. So, except for controlling the rise time involved in supporting fields between primary and secondary antennas, it is right out of the equation. This is not true of 45/55% nickle, which will still transform power out to approximately 3.5 kHz. This was the material of choice in the old days of telephones.... you know, the black plastic thingys in some museums? 80% Nickle is good to about 10 kHz and metallic glasses can do 28 kHz. That rise time issue will make even double blind testing defeat-able as it is audible as an increase "softness" to leading edge information, as you transit from amorphous to M50 right across the 20 to 20kHz range. Still, not a big deal.
The real difference in how audio transformers perform comes from the coil and it's construction and how the B Field antenna events are handled, to a degree. The real tell is how much of the capacitive coupling area, vs volume of distributed capacitance you can use with a given core material. You might take note of amorphous core output transformers not having concentric windings as a guide.
And the final areas of interest involve not assuming the coil is a lumped parameter of any sort and that core can be used in such a manner that it will passively self demagnetize. Thereby making the polarization caused by saturation a moot point. Changes in these parameters make rather large differences in perceived audio characteristics and will definitely defeat a double blind test.
Bud
ClariSonus - Clear and enjoyable sound
John has looked at your question and has an amplifier that he explored some of these territories with. Ask, he will be delighted someone cares.
As for the "sound" of core material. Well. Commercial manganese iron, the M series of grain and non grain oriented metals, is all done with power transform at about 440 Hz or thereabouts. Depends upon chemistry, thickness and annealing correctly. So, except for controlling the rise time involved in supporting fields between primary and secondary antennas, it is right out of the equation. This is not true of 45/55% nickle, which will still transform power out to approximately 3.5 kHz. This was the material of choice in the old days of telephones.... you know, the black plastic thingys in some museums? 80% Nickle is good to about 10 kHz and metallic glasses can do 28 kHz. That rise time issue will make even double blind testing defeat-able as it is audible as an increase "softness" to leading edge information, as you transit from amorphous to M50 right across the 20 to 20kHz range. Still, not a big deal.
The real difference in how audio transformers perform comes from the coil and it's construction and how the B Field antenna events are handled, to a degree. The real tell is how much of the capacitive coupling area, vs volume of distributed capacitance you can use with a given core material. You might take note of amorphous core output transformers not having concentric windings as a guide.
And the final areas of interest involve not assuming the coil is a lumped parameter of any sort and that core can be used in such a manner that it will passively self demagnetize. Thereby making the polarization caused by saturation a moot point. Changes in these parameters make rather large differences in perceived audio characteristics and will definitely defeat a double blind test.
Bud
Last edited:
Yes, it can. Push pull transformers are usually designed to sustain several mA of DC current (5 - 10mA). This is true for all materials, higher permeability -> lower tolerance to DC offset, and opposite.
Air gap, either fixed, distributed or quasi-distributed. is used to prevent saturation under DC offset.
Air gap, either fixed, distributed or quasi-distributed. is used to prevent saturation under DC offset.
Remember when talking about transformer cores and distortion that the core, to a first approximation, only sees the difference between primary (turns x current) and secondary (turns x current). This is why we can make a (fairly) linear transformer from a (fairly) non-linear core. This is quite different from a tape head or the tape itself.
I assumed that it is not a simple thing .....
so .... 'do not know' what to do if you connect or less
* my dac to the preamp input transformer superpermalloy
thanks luigi
DAC output should not have any DC offset. If it has, it should be fixed, not the transformer.
The purpose of the AC bias is to overcome the coercive force of the material, so that, as a given grain of magnetic material on the tape approaches the head, it is gradually exposed to a larger and larger, then smaller and smaller, loop on its B-H curve. Ideally, this loop should probably reach saturation, or close to it, forcing the loop to be centered around zero. (A number of materials can retain an average magnetization despite being exposed to an unbiased AC field greater than the coercivity -- in other words, degaussing doesn't always work!) The recording signal now looks like a DC offset, so as the tape material comes out of saturation, or the loop shrinks, it's forced off center to the desired direction.
To do the same thing with a transformer, you can't simply add a high frequency bias, implying the voltage is relatively small. In fact, the voltage has to be high, very high, because the frequency is also high. The reason laminated anything (iron or otherwise) is not used above a few kHz is because, even for the thinnest strips and laminations, eddy currents dominate, permeability drops and the core looks like a resistor. Applying HV HF AC will melt any iron-cored transformer. Still, you could do this with ferrite, which will increase size considerably for the same LF performance.
The audio signal will act to alternately saturate the transformer at HF, most likely introducing various IMD, which *will* be noticable (because as I said before, saturation affects signals greatly; minor variations in an already-huge permeability are negligible).
The problem with the tape analogy is, the tape comes out of the bias field as it passes. A transformer core is kind of just stuck there, and any attempt to "improve" it will only result in far more power, weight, cost, effort and distortion than, for instance, adding another tube or two and actively improving the transformer, for real, with negative feedback.
Tim
Great Topic !!
Funny I have been researching this stuff all week and some how I missed the thread!
I googled "THD vs core material".
Here are a few links that I have found that may be of interest,
Magnetic Metals Nickel Lamination, Electromechanical Assemblies and Custom Annealing Service
Audio Transformers By Cinemag Inc.
Audio Transformer Application Notes
http://cinemag.biz/application_notes/AN-104.pdf
The app note is particularly interesting.
Also I have not yet tried to contact any of the companys for prices yet.
jer
Funny I have been researching this stuff all week and some how I missed the thread!
I googled "THD vs core material".
Here are a few links that I have found that may be of interest,
Magnetic Metals Nickel Lamination, Electromechanical Assemblies and Custom Annealing Service
Audio Transformers By Cinemag Inc.
Audio Transformer Application Notes
http://cinemag.biz/application_notes/AN-104.pdf
The app note is particularly interesting.
Also I have not yet tried to contact any of the companys for prices yet.
jer
Last edited:
Radiometal......
Beacon Radio Ltd, Auckland, - now long gone - marketed many transformers. In a range of many P-P output transformers. This firm for the more popular designs provided three variations for each. The same winding construction was assembled with either silicon, super-silicon, or radio-metal laminations.
Beacon's catalogue disclosed primary inductance figures measured at 5 volts, 20hz - Examples:
10 watt/10,000 ohm 46, 110, 220 henries
15 watt/4000 ohm 18, 35, 70 henries
15 watt/10,000 ohm 50, 85, 180 henries
High inductance certainly provides extended bass response!! Distortion - ???
My last winding effort using a 60mm stack of radio-metal laminations (1.25inch wide) and intended to provide for 20+ watts at 3,000 ohms PP reveals a droop of 0.4db at 10hz and a HF droop of 1.15db at 40 khz. No resonances are apparent below 200 khz.
Both Bambini and Bernards published a "Coil and Construction Manual". There are no equations - just tables relating to a 1000 ohm example. Further tables provide arithmetical adjustments to modify.......
To use radio=metal laminations their figures suggest: (a substitution for silicon laminations):
Turns - divide by 1.3
Max power - multiply by 2.3
Mid-band losses - divide by 1.7
LF cut-off - equal
I am just a DIY winder upon a Chinese mandrel with an accurate counter. I don't even use the gearing provided - every turn by hand where I want it to be.....lovely.
ps - Follow Yves' section-layouts......
Graeme
Beacon Radio Ltd, Auckland, - now long gone - marketed many transformers. In a range of many P-P output transformers. This firm for the more popular designs provided three variations for each. The same winding construction was assembled with either silicon, super-silicon, or radio-metal laminations.
Beacon's catalogue disclosed primary inductance figures measured at 5 volts, 20hz - Examples:
10 watt/10,000 ohm 46, 110, 220 henries
15 watt/4000 ohm 18, 35, 70 henries
15 watt/10,000 ohm 50, 85, 180 henries
High inductance certainly provides extended bass response!! Distortion - ???
My last winding effort using a 60mm stack of radio-metal laminations (1.25inch wide) and intended to provide for 20+ watts at 3,000 ohms PP reveals a droop of 0.4db at 10hz and a HF droop of 1.15db at 40 khz. No resonances are apparent below 200 khz.
Both Bambini and Bernards published a "Coil and Construction Manual". There are no equations - just tables relating to a 1000 ohm example. Further tables provide arithmetical adjustments to modify.......
To use radio=metal laminations their figures suggest: (a substitution for silicon laminations):
Turns - divide by 1.3
Max power - multiply by 2.3
Mid-band losses - divide by 1.7
LF cut-off - equal
I am just a DIY winder upon a Chinese mandrel with an accurate counter. I don't even use the gearing provided - every turn by hand where I want it to be.....lovely.
ps - Follow Yves' section-layouts......
Graeme
The resistivity of Radiometal is 55 microhm cm. I infer that Audionote is using that to name their alloy rather than nickel composition.
BTW - The ad copy is mostly mumbo jumbo, but have you read ads for speaker wire, USB cables and interconnects lately? These guys are tame in comparison.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.