• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Output Transformer - Silicon Steel or Amorphous?

Hi, Tube folks,

Currently there are 2 major flavors of audio output transformers (both SE and PP) - with silicon steel and amorphous cores.

Amorphous cores are ferromagnetic alloys based on iron (73.5%), silicon (13.5%), boron (9%), niobium (3%) and copper (1%)). Depending on the type of heat treatment after quenching they may have a rectangular (Br/Bs = 0,2-0,9), linear or circular hysteresis loop. They allow to build smaller transformers with given power rating with lower losses, high saturation flux (up to 1.56 Tesla), etc.

The question is - which one would you suggest? Different people state different things - some say only silicon steel (because of higher distortions of amorphous cores - IMHO a somewhat unproven statement), other say "I DON’T EVER WANT TO GO BACK to silicon steel".

Any suggestion(s) are greatly appreciated.
 
Last edited:
I don't want to appear as being too biased on the subject, but here's a link from the US distributor's site who imports the Hashimoto line of transformers:

http://tube-amps.net/Technical_Aspects.htm

They purposely avoid permalloy and amorphous cores and have some explanations as to why they use Orient Core Hi-B steel. The company has a long history (aka Sansui) and I would note that the quality of their current product line in excellent in every detail.

Regards, KM
 
The issue is not just the steel.

Amorphous core has excellent characteristics. The BH curve, saturation induced polarized core moment, until H goes opposite phase and the mag field crashes through zero to catch up, think zero crossing distortion here, is greatly reduced. Unfortunately you end up having to limit the E Field coupling to allow enough leakage inductance and distributed capacitance to keep the frequency response from forming a high Q resonant peak at 36 kHz or so. The core is also active out to about 18 kHz in a direct power transform function.

M6 or M3 commercial grade silicon steel does not have the BH curve, saturation induced polarization release of amorphous core. This causes a higher level of cross over distortion. This can be mediated with proper construction techniques that are the antithesis of those used for power transformers. Once implemented the two cores have very similar back wave performance.

The commercial core does benefit from allowing a far more efficient capacitive coupling scheme to be used, without encountering the resonance peak problems than beset amorphous core. This state of affairs also allows a lower dielectric constant material to be used as the dielectric barrier across all antenna events in the coil, meaning all events.

The difference in sonic’s is interesting. An amorphous core OPT for PP will be very clear, have great transient response and excellent tonality. However, they are somewhat one dimensional in that most of the fine gradient detail that provides internal note structure and makes transient noises into musical events, has been stripped away by the lack of E Field coupling and the higher dielectric constant materials used in the coils, essentially half of the information must be lost to avoid frequency response peaks.

An M3 commercial core, properly constructed for audio purposes, will sound slightly softer than an Amorphous core transformer. There is after all more distortion, and when comparing 99.997% correct signals (Amorphous) with 99.993% correct signals (M3), this is what you should expect. However, there can be a great deal more internal information made available, for all musical signals, because you do not have to throw away half of the signal and the antenna event is actually encouraged during it's E Field portion.

In addition the M3 and M6 core are basically only providing a ferrous bounding box for the coil antenna event above 400 Hz. The material still responds but the losses are so high, with extended frequencies, that it neither adds nor subtracts to the signal passing from coil to coil, so long as that core construction issue has been dealt with. Neither amorphous core not M3 core is "better", they have differing strengths, and when both are used properly either will provide superb audio, within the limitations found above, and their pricing will be about the same too. An M6 core unit, with more distortion and only slightly less information, without quite the refinement of either of the other materials, is a much less expensive alternative, with very few drawbacks.

Bud
 
The issue is not just the steel.

Unfortunately you end up having to limit the E Field coupling to allow enough leakage inductance and distributed capacitance to keep the frequency response from forming a high Q resonant peak at 36 kHz or so. The core is also active out to about 18 kHz in a direct power transform function.

An amorphous core OPT for PP will be very clear, have great transient response and excellent tonality. However, they are somewhat one dimensional in that most of the fine gradient detail that provides internal note structure and makes transient noises into musical events, has been stripped away by the lack of E Field coupling and the higher dielectric constant materials used in the coils, essentially half of the information must be lost to avoid frequency response peaks.

Bud

Thanks a LOT, Bud! I never expected SO comprehensive reply. I did not understood entirely mechanism of loss of information in amorphous core. So, your suggestion is to use M3 silicon steel? Do you have any info about Orient Core Hi-B silicon steel used by Japanese companies?

I found that some Luxman audio transformers have frequency response 10Hz - 100,000 KHz, not 100% sure, it may be a seller mistake. What kind of material they used?
 
Published specs can be misleading unless they have some detail around the spec, meaning... what was the power level used, how they drove the transformer (dc current/imbalance for PP or dc current for SE), etc. I'm using a Hashimoto H-507S with a 45 DHT design. Published spec is out to 100KHz within 3dB. The amp does measure -1dB at 50KHz at 1-watt output and about 35ma of current in the primary winding. The circuit is zero-feedback and I use the 5K tap for the primary winding and the 16-ohm tap on the secondary. Measured distortion for 50KHz @ 1-watt is actually lower than 1KHz @ 1-watt.

Regards, KM
 
Not entirely no. B/H curve saturation occurs even in SE OPT's, though not to the degree that it occurs in PP, due to the typical gap.

Core gap in E/I transformers is also the problem with an E/I SE OPT, the larger the gap the longer it will take for core polarity to catch up with the back wave of the driving signal. C core gaps are driven on both sides of the gap thus eliminating most of this latency, their drawback being the nonlinear flux created in the window, by having a gap in the center of the winding

For the most part, core is your enemy in signal transformers, above 400 Hz, even with amorphous core. If you construct an E/I core with the same distributed gap that works so well in PP E/I core, the core will actually help by passively demagnetizing on the back wave, after a saturation peak has polarized the core. This method actually works so well that it is dangerous to use it without a gap, in a SE OPT.

The same sonic values hold true, as in PP, between amorphous core and M3 E/I core, though i don't find them as noticeable.

Probably the two best providers of custom designed amorphous core SE OPT's are Intact Audio and Tribute Audio. O-Netics, the company I work for, is the only provider of partial self demagnetizing E/I core structures that I know of. I am sure there will be others in time, we do tend to pirate from one another as occasion permits.

Bud
 
But in a single-ended transformer, the core never demagnetizes (i.e., the field never collapses) because of the ever constant DC component. The AC signal manifests itself as minor AC loops along the DC magnetization curve. The AC loops may breech the the upper left hand quadrant of the BH graph with a large signal but the incremental permeability remains constant, so there is very little distortion and no "crossover" distortion to speak of.

John
 
jlsem,

You are correct, the field never collapses. However the AC current and DC current are not a combined event. The DC current does not "swamp" the AC signal current.

A SE OPT does not suffer from the zero crossing distortion of a PP, it does suffer from lack of wave form tracking, on the settling side of a wave form. Core polarization occurs at the gap, with AC saturation peaks and the gap does add a time to null point distortion. It is neither as pervasive as that from poorly designed PP core transformers, nor of the same audible type. Rather it is the "fattness", most especially to lower frequency signals, that is common to typical E/I SE OPT's.

Amorphous core SE OPT's have almost none of this problem. They do have a disruption to the planar magnetic fields arising within the window, that support the 1K to 100k increase in the transform of the antenna signal from primary to secondary. This is audible at the leading edge of the signals and is often mistaken for detail. This is a minor fault in most cases.

Distributed gap E/I transformers also have gap induced distortion, but it does not show up until relatively high levels of power are induced. Again, a minor fault in most cases.

In a recent ad hoc experiment, in audible characteristics, between SE and PP OPT's, both with distributed gaps and in prototypes from Wright sound, the only real difference between SE and PP 300 B amplifiers was the resolution of what can only be called musical beauty, from the SE amplifiers. They had noticeably more odd order distortion than the PP amplifiers, but they simply provide an emotional benefit missing from the deep class A PP 300 B units. Load lines were 3.3 kZ for the SE and 6.6 kZ from the PP.

Bud
 
Not entirely no. B/H curve saturation occurs even in SE OPT's, though not to the degree that it occurs in PP, due to the typical gap.

Core gap in E/I transformers is also the problem with an E/I SE OPT, the larger the gap the longer it will take for core polarity to catch up with the back wave of the driving signal. C core gaps are driven on both sides of the gap thus eliminating most of this latency, their drawback being the nonlinear flux created in the window, by having a gap in the center of the winding

For the most part, core is your enemy in signal transformers, above 400 Hz, even with amorphous core. If you construct an E/I core with the same distributed gap that works so well in PP E/I core, the core will actually help by passively demagnetizing on the back wave, after a saturation peak has polarized the core. This method actually works so well that it is dangerous to use it without a gap, in a SE OPT.

The same sonic values hold true, as in PP, between amorphous core and M3 E/I core, though i don't find them as noticeable.

Probably the two best providers of custom designed amorphous core SE OPT's are Intact Audio and Tribute Audio. O-Netics, the company I work for, is the only provider of partial self demagnetizing E/I core structures that I know of. I am sure there will be others in time, we do tend to pirate from one another as occasion permits.

Bud

Can you please explain what is meant by "back wave"? In the case of a sine
wave would it be the part from positive crest to negative crest? For a square
wave it's the falling edge?

Thanks
 
Yes, exactly those portions, the ones beset with "settling time" issues of storage and resonance. I am after the same change in laggardly performance with the EnABL speaker treatment process and the Electron Pools. As far as am concerned, the front wave / rise time performance of just about everything is just fine.

Bud
 
All tube OPT's have DC current flowing in their primary wires. The AC current flowing is the only one that causes change. And the only current that flows in the secondary is AC, unless we refer to an auto former

As far as a SE transformer is concerned, DC current offset is not important, other than it holds the permeability of the core, hopefully, at a point just below 2K gauss or so.

Push Pull transformers have DC current flowing also, but it is not important unless it is holding the core at some level of permeability other than zero.

You were not specific about which condition you were referring to. I will point out that the only signal that is of importance is that found on the secondary and a well designed PP OPT and SE OPT will be very close to one another at this point.

It is quite likely you have never dealt with a PP OPT that does not have crossover distortion and, as might I, accept this condition as necessary to the breed. It is not. Just as the saw that SE OPT's are always weak in high frequencies and sloppy in bass is not truth either. Both of these attributes are easily remedied in SE OPT's and are just a function of treating a signal device the same way a power device is treated. The same is true with PP OPT's.

Not picking an argument here. We just come from different places, with differing experiences.

Bud
 
Hello Bud

nice explanations and I like tom read it, thanks!

Please let me have a question from the other side... you mentioned that all the problems of amorphous core OPT is described by the audiophiles with "detailed sound".
So why are the most reputated amps out there build with amorphous or permalloy core and beloved from the crow? Do they listen all so bad? Maybe we are talking on the order of parts per thousand.

Anyway, let's follow the theory discussions. Can you explain your view of a "perfect" transformer for single ended amps?

Joao
 
Not all of the well reputed SE amps have amorphous core, just most of them. Wright Sound's excellent Model 7 mono blocks have an E/I core SE OPT and are very well thought of. Unfortunately it looks like they will disappear with the owners death. They are also in need of two changes, a replacement of the interstage photo flash capacitor with a multicap/reliacap PPFX tin and poly cap and the bridge resistor between the CCS and driver halves of the 6sn7 with an extremely good non inductive foil resistor, like the Rn60D Dale/Vishay brand. These small alterations remove the last vestiges of "sharpness" in these amps.

They are not considered to be in the "class" of the huge, plate choke fed behemoths, but this is more a status thing than any lack of performance.

Other than Wright Sound, O/Netics has not been aggressive about getting involved with the Audio manufacturing community, due more to the unstable financial environment than anything else. And, we do think that Intact and Tribute deserve to be in business and so are not interested in a more direct fight using amorphous core, which we have experience with. There are a number of satisfied DIY customers with our E/I products however, and they also have amorphous core units from other manufacturers.

A negative example comes to mind with Romy the Cat. His SE OPT's all have amorphous core. He has had mine and does not like them. Not because they lack detail or musical characteristics, but due to their reduced angle of acceleration on the leading edge of transients. His amps also have power supplies with 30K mf of high voltage capacitance, or more. This just might have something to do with his choices.

The false detail comes from leading edge overshoot. With an amorphous core you must be very careful with your coil construction. All dielectric materials need to be of a fairly high dielectric constant so that the release into B Field, from E Field activity, does not excite the core into half wave, uninhibited overshoot. Lundal does this very successfully with mylar, as layer to layer insulation, and separation of primary and secondary coils from each other. Another way to provide some "ballasting" is to deliberately increase wire size, by utilizing Amorphous core permeability, and thus add distributed capacitance.

However, as you say, these are really minute differences and without actually hearing a direct A/B comparison, the owner of either basic core format will not know there might be something they are missing.

So, choice of type of core for the OPT might be best done first and then the rest of the system built to suit that choice.

Bud
 
Last edited:
Quote:
"It is quite likely you have never dealt with a PP OPT that does not have crossover distortion and, as might I, accept this condition as necessary to the breed. It is not."

Hello Bud,
How does a PP OPT avoid crossover distortion? Increased air gap or bifilar winding? Pinstriping? HF bias? Or something new?

Don
 
Last edited:
The false detail comes from leading edge overshoot. With an amorphous core you must be very careful with your coil construction. All dielectric materials need to be of a fairly high dielectric constant so that the release into B Field, from E Field activity, does not excite the core into half wave, uninhibited overshoot. Lundal does this very successfully with mylar, as layer to layer insulation, and separation of primary and secondary coils from each other. Another way to provide some "ballasting" is to deliberately increase wire size, by utilizing Amorphous core permeability, and thus add distributed capacitance.

Bud, what kind of test protocol do you use at O-Netics that allows you to observe these phenomena? I'm interested in learning more about the latest procedures for measuring and predicting the behavior of transformer cores.

John