A Toroidal Output Transformer: is it too much ?

[fikri]

Well im just wondering what kind of advantages do we get if we use toroidal for output transformer (poweramp). I have seen many amp manufacturer uses toroidal for power transformer, but i havent found one that use toroidal for output transformer.

[zachomega]

Toroidal transformers don't saturate the same as conventional transformers.  In many ways they are more perfect.  You can get much closer to the maximum output load on the transformer before it will saturate...however, once it does saturate it is completely saturated as opposed to a standard transformer which will saturate slowly but much earlier. 

I've personally wanted to try this for some time myself.  Just never got around to it.  I'd be interested in any "real world" results you come up with.

-Zach Omega

[brett]

Hi

In many ways they are more perfect.  You can get much closer to the maximum output load on the transformer before it will saturate...however, once it does saturate it is completely saturated as opposed to a standard transformer which will saturate slowly but much earlier. 

I'm no expert, but I think there's a mix of fact and fiction in there. 

I've used toroids in inductors and for small coupling transformers with mixed success.  But toroids are unlikly to be suitable for output transformers because small amounts (sometimes TINY amounts) of DC cause them to saturate.  This is because they have no "gap" or "air gap" to limit the magnetic flux caused by DC.  You might be able to find toroids that include information about their ability to handle DC.  Often the limits are a few mW.

A few clever people will be saying "Ah-ha!  If the two primaries are counter-wound, then the magnetic flux will cancel out."  Yes, in an ideal world this would happen, but in the real world tubes and bias currents and primary DC impedances are never equal and so balancing to within a percent or two isn't easily achieved.  (I didn't say impossible, though) 

Regarding saturation: This is because there is no definitive "maximum output" from transformers.  There are only general guidelines.  For typical silicon steel (it's NOT iron) used in E-I transformers, it is wise not to go beyond about 1.2 Teslas of magnetic fluss density (or 14000 Gauss IIRC).  That's where the B-H curve often starts to change and "saturation" is approached.  Small commercial power transformers seem to run at maximum flux densities (minimum* output) from about 0.5 to 1.5 T.  The same is true of output transformers.  Some have a conservative power rating and operate at a low percentage of saturation at maximum output (e.g. 40%), while others have a power rating that reflects close to saturation of the core (e.g. 80%).

Note that saturation is often considered a good thing, but at saturation there is no increase in transformer output power with increased input power (the transformer cannot transfer any extra power).  Where does the extra power go?  I'm can't say for sure, but I suspect most of it is dissipated as heat in the output tubes and transformer. 

(Of course, once the output tubes and/or transformer are cooked, the output goes to zero and balance is again returned to the universe.  )

* that not a mistake.  Maximum flux density occurs in transformers when the input is at a maximum and the output is at a minimum.  For power transformers, this is when no current is being drawn from the secondary.

[MKB]

There are some high end audio amp manufacturers that use toroid OT's to good results.  But they don't seem to be favored for guitar amps for the reasons Zach mentioned.  The saturation of the core is a very important part of the tonal equation, OT's sometimes use different types of laminations in the core to change the tone of the part. 

Another problem with toroid OT's is that they are very intolerant of tube mismatching.  You have to do a better job of current balance than is necessary in a standard EI core OT.

There are advantages to the toroid OT though; less radiated field, smaller size, and lighter weight.

[rocket]

Why should torroidal transformers react badly to tube missmatches.

Another problem is that the winding cannot be as neatly structured as in conventional trnasformer - thus the capacity of the winding is higher.

A compromise between toroidal and M or EI cores are C type cores, which are also used for outputransformers.

Anyway a topic like this is better discussed in a tubeamp forum like ampage.org

[R.G.]

Brett is correct. What you've heard is partly truth, partly fiction.

Toroidal transformers are nearer ideal than EIs because they have continuous cores, no air gap. This makes for much higher primary inductance and hence lower exciting losses per unit of iron (which I use as jargon - it is indeed 4% silicon steel these days). But the lack of an air gap means that it will saturate much more easily. Actually, transformer iron/steel always saturates at the same flux density, which depends on the exact mix of the alloy. But the air gaps require much more magnetic push to let the iron get up to that density. In a transformer, the vast majority of MagnetoMotiveForce (MMF) is spent pushing flux through the air gaps, not the iron.

The maximum output from any transformer depends almost solely on how hot you let it get. Transformers will work happily at temperatures that will sizzle your skin if touched as long as the internal insulation will withstand the heat. Toroids are better than EIs at this because they have more surface area per unit of volume, and can dissipate more heat. The same weight of iron and copper in a toroid will transfer more power than in an EI transformer for this reason.

Toroids are indeed much less tolerant of tube mismatches. This is because push-pull output transformers are always wound with the primaries opposing and the DC bias current flowing through both and cancelling. If the cancellation is not accurate, the imbalance can easily saturate a toroid. This usually needs an active bias balancing circuit to force the balance with tube aging.

Very few technical people, even EEs, have an accurate idea of what saturation is and how it happens in a transformer. Saturation means that you have aligned all of the available magnetic domains in the iron and no more input force will make any more domains align. Therefore the magnetic field in the iron cannot change any more, and it can neither oppose further increases in primary current nor transfer any more energy to any secondaries, as these depend on field change to happen. Since in an inductor (which is what a primary is) the current is proportional to the volt-time integral impressed on the winding, the volt-time integral is what forces saturation. If you want to saturate a transformer, you have to provide more volt-time to its primary, by either increasing the voltage or the time (that is, lowering the input frequency).

Saturation is a primary-winding phenomena. You cannot saturate a transformer from the secondary. The exciting current which flows in a transformer at no load is all of the current that ever goes into the iron to set up magnetic field. The field opposes further increases in current and self limits if it is of less than saturation volt-seconds. Loading a secondary subtracts volt-seconds from the core; the lost volt-seconds can no longer oppose primary current flow, so the primary lets in more current. The power actually does flow through the transformer. As it does, the core magnetic field does not change. Only the primary and secondary currents change. So you can't saturate from the secondary side. You can overload it and burn the transformer out by ohmic heating of the wires, but that is a purely current-flow effect. The core is fine.

Saturation of an output transformer is more problematical. You certainly have the elements at hand to do it. There is a DC source feeding the transfomer and two DC-controlling devices on the transformer. But I think that OT saturation is much less common that we think. The DC voltage feeding the OT is fixed at design time - it's the DC power supply. The frequency is fixed, it's the lowest note you feed the transformer, or the lowest frequency the coupling capacitors let through. It's certainly possible to feed the transformer a saturation volt-second mixture, but I think it's rare. And think about this - it's only on bass notes.  A transformer's saturation resistance goes up linearly with frequency because as seconds goes down, volts goes up. At low , 82 Hz, an OT will saturate at some value of exciting voltage. One octave up it will withstand twice the voltage, and two octaves up it will withstand four times the voltage. Saturation is a low frequency phenomena. So it doesn't act the way common musical folk knowledge says it does. Feeding a bass into your tube amp will saturate the output transformer much more easily than turning up the volume. As to designing the OT to saturate for "tone"; it's possible to set the saturation such that it will saturate on bass notes in the lower octave, but only if you know reasonably exactly what tubes and power supply will be used with the OT. This is much more information than most OT transformer designers have available when they start fitting turns. IMHO, good tonal saturation of OTs is a happy accident when it happens. Very, very few designers have the technical skills needed to make it happen and even fewer of them have the ability to design both the OT and the amp circuit. I did have this opportunity when I designed the Workhorse amps, but it is remarkably rare to have the technical backgrounds in the same person who designs the OT and the amp circuit, as these are usually done by different people, sometimes in different decades of time.

On winding format:

Another problem is that the winding cannot be as neatly structured as in conventional trnasformer - thus the capacity of the winding is higher.

This is a non-sequiteur. Toroidal windings can be *very*neatly structured. The supplier for the toroidal power transformers in the Workhorse line makes them very neatly indeed. I've watched them being wound. But neatness does not equal capacitance at all.

The winding on a toroid is the equivalent of a two long skinny windings, one on top of the other. The winding on a shell type (EI) transformer is two short, fat windings. The long skinny winding is the best possible winding for low leakage inductance, but has a higher inter-winding capacitance, although a low self capacitance. On balance, the toroid style winding is very good for audio coupling. That same "lower radiated field" that you hear about toroids means that more of the field stays in side the iron and couples both windings. The short, fat winding on an EI is... a compromise. It's very easy to wind and stack, and that's largely why it's common. The short, fat winding is bad for letting magnetic flux leak between windings so that not all of the flux which couples one winding couples the other. This is leakage and is modeled as leakage inductance. The worst shape for leakage is a very short, but high build winding. EI windings are moderately bad. This is why EI output transformers have all that interleaving and special formats - to try to force the leakage to be smaller. The EI is about the same in self capacitance, but has lower inter-winding capacitance than the toroid. This has an effect on the very high end, but it's usually outside of the audio range at the top end. It affects power transformer stability with feedback but not audio.

[JonFrum]

Plitron makes toroidal audio output transformers - I can't get on the web site right now. I seem to remember him having technical articles as well.

[puretube]

THIS is the toroidal OT guy...

[brett]

Hi

That same "lower radiated field" that you hear about toroids means that more of the field stays in side the iron and couples both windings. The short, fat winding on an EI is... a compromise.

My limited reading also indicates that on E-I transformers the "sharp" corners on the laminations, as well as the "corners" on the windings are sites for increased flux density and magnetic flux loss.  I don't have any direct evidence of this, but it seems likely given the accepted concepts about magnetic flux and paths.

Given that small toroids are readily available (e.g. out of PC power supplies), I'll have a go at making a small push-pull amp with a toroidal output transformer, and compare the results with an E-I transformer (For simplicity, and to give the toroid the best possible chance of working, I'll give it no DC bias current.  e.g. by using it to "bridge" two LM386s).  With any luck, one of them will sound quite good. 

cheers

[MetalGuy]

AFAIK some manufacturers introduce a very tiny air gap to their toroidal output transformers, even to their PTs. Since the inductance of a guitar OT is not high maybe this can solve the problem with saturation.

[R.G.]

AFAIK some manufacturers introduce a very tiny air gap to their toroidal output transformers, even to their PTs. Since the inductance of a guitar OT is not high maybe this can solve the problem with saturation.

This is a classical illustration of contradictory requirements.

One of the reasons that toroids have such good magnetic properties - low radiation, high inductance, good response, and so on is that they have no air gap. It's the lack of an air gap that enhances a lot of the things that are good about them. Introduce an air gap to make saturation resistance go up and you increase radiation, lower inductance, etc.

The lack of an air gap is both the bad news and the good news at the same time.

The air-gapped toroids are probably cut-C cores. This is the technique used for many of the Golden Era hifi output transformers.

The inductance of a guitar OT is low by hifi standards only because the low frequency response is not very low. Also, the high frequency requirements are not very high.

A hifi output transformer has a figure of merit of the primary inductance divided by the primary to secondary leakage inductance. A good hifi OT will have an inductance ratio of over 10,000. Really good ones top 100,000. Since it's quite difficult to deal with leakage inductance in a toroid, introducing a gap to lower the primary inductance lowers the figure of merit directly.

It's probably easier to figure out a circuit to avoid saturation.