Output Transformer Saturation

[raiken]

Excellent information in this thread!

Having said that, I'd still like to see someone show me a scope photo of an output transformer saturating in a guitar amp...

[ThermionicScott]

Having said that, I'd still like to see someone show me a scope photo of an output transformer saturating in a guitar amp...

One of those itty-bitty Hammond 125A transformers would make a good candidate. Who'd like to sacrifice one in the name of science?

[Steve Conner]

1) Here's the wiki illustration:

That illustration is wrong (although the explanation in the wiki is right). The circuit in that picture would have the asymmetrical behaviour that RG describes, but that's not how a mag-amp works. The whole point of the saturable reactor for AC control is that it has two separate cores, and each one saturates on alternate half-cycles. This is what gives it its symmetry. It also has to be that way so that there can be two control windings, so you can cancel the induced AC voltage in them.

It's also untrue that the load flux needs to be small compared to the control flux. They're both equal, as required by amp-turn balance. If you want 10 amp-turns to flow through to the load, you need to put 10 amp-turns through the control winding. To get current gain in a mag-amp, the control winding must have many more turns than the load winding, and the two halves of it can get thousands of volts of AC induced, which is why it's so important to cancel it out.

Really the saturable reactor works in a similar way to the flux-gate magnetometer, or the "DC current transformers" that were popular before Hall effect ones came out.

I've seen OPTs saturating quite often, and will post some scope shots soon. They all do it (well PP ones at least) as soon as you try to make the rated power below the lowest frequency they're rated for. There's no need to talk of "sacrifices in the name of science" as it doesn't harm the transformer in the least.

[raiken]

I've seen OPTs saturating quite often, and will post some scope shots soon. They all do it (well PP ones at least) as soon as you try to make the rated power below the lowest frequency they're rated for.

Of course you can saturate a transformer by running it outside the ratings, but have you seen them saturating in production amps, and is that what gives them the "magic" tone? That is the question...

[defaced]

It might be possible to mechanically modify the magnetic path by mechanically sliding a plug of iron in and out to form a smaller area of magnetic path that would cause lower saturation for the same MMF by reducing the effective area of the iron path. This could get mechanically tricky to do, as well as being not all that reliable.

I'm fairly certain I've seen welding power supplies (old machines we had in high school) that worked this way. Crank the knob one way to increase current, crank it the other way to decrease current. Basically the knob was tied mechanically to a chunk of metal that moved up and down through the core of a transformer. I always assumed this just changed the percent of air core to iron core transformer thereby changing it's operating characteristics. Then again, I could just be mis-remembering, it has been a wee bit over a decade since I cleaned out that machine.

[Steve Conner]

Oh, well, it's a good question.

I know that it's an issue in bass amps. The third harmonic distortion from saturation makes them sound fatter. Ampeg are on record as saying that they made the OTs in their tube bass amps too big, and had to go back and undersize them to get more saturation.

And on the schematics for the Fender 300PS, the OT is only specced down to 50Hz, whereas even a 4-string bass goes down to 42Hz. It's one of the bigger bass OTs I've seen, too, at about 10lbs.

But in a guitar amp I wouldn't bet on it. Most of the output power of classic tube amps is in the midrange, the bass is attenuated because they tend to have some treble boost or undersized coupling cap somewhere. If you get a guitar amp with a flat response and full bandwidth, people think it sounds dull and farty.

And the low E on a guitar is 84Hz, so it's not even that low.

I don't own any production tube amps to test, mine are all homebuilt.

[raiken]

Oh, well, it's a good question.

I know that it's an issue in bass amps. The third harmonic distortion from saturation makes them sound fatter. Ampeg are on record as saying that they made the OTs in their tube bass amps too big, and had to go back and undersize them to get more saturation.

And on the schematics for the Fender 300PS, the OT is only specced down to 50Hz, whereas even a 4-string bass goes down to 42Hz. It's one of the bigger bass OTs I've seen, too, at about 10lbs.

But in a guitar amp I wouldn't bet on it. Most of the output power of classic tube amps is in the midrange, the bass is attenuated because they tend to have some treble boost or undersized coupling cap somewhere. If you get a guitar amp with a flat response and full bandwidth, people think it sounds dull and farty.

And the low E on a guitar is 84Hz, so it's not even that low.

I don't own any production tube amps to test, mine are all homebuilt.

That's what I get for asking a bass player.

Regardless, it would still be nice to see some pics of actual output transformer saturation in an audio context. I've done it with switchmode magnetics, and the result wasn't very pretty.

[loudthud]

The whole idea of "Output Transformer Saturation" is right up there with "Tube Watts vs. Solid State Watts" in the myths made up by salesmen at Guitar Center. These myths were born in the late 60's to explain the difference between the sound of Tube amps and Solid State amps. Actually, the myth pre-dates Guitar Center but I couldn't resist poking fun. It's technical mumbo-jumbo used to sell amps to non-technical customers.

The primary mechanism at work is the soft output power vs impedance curve of tube amps with little or no feedback and how such amps interact with the complex impedance of a speaker when the amp clips. If you want to increase it, try increasing screen resistors and/or making a soft screen supply. On a solid state amp, add a resistance in series with the speaker and 5 or 10 dB of feedback around it.

[Steve Conner]

Here are the results of my mini investigation of OT saturation. I took two tube amps and drove them to full output at various frequencies.

The first one I tried was my Crown SXA, a hi-fi amp similar to the Williamson. It's 30w per channel with an ultralinear output stage.

I measured 36w before clipping at 100Hz, and tried to maintain the same peak voltage as the frequency was turned down. You can see that the waveform starts to get misshapen below about 30Hz, and by 20Hz it's a real mess. The heavy negative feedback in the Crown probably makes it look a lot worse than the transformer itself.

The second one I tried was my "Ninja Deluxe" which is the guts of a 1970s model Selmer Treble'N'Bass 50 transplanted into a Mojo 5E3 cabinet. This did remarkably well. Not only does it crank out a 5E3-slaying 42 watts at 100Hz, but the output stays looking good down to a cone-rupturing 40Hz, and below that the saturation is more gradual than the Crown's.

Verdict: Full output down to low E on the bass, so maybe Selmer were serious about the "N'Bass" part of the name! And possibly not the best OPT choice for an open-backed cabinet. In any case, there's no saturation to be seen at 84Hz, low E on the guitar. And the OPT in the Selmer is a similar physical size to other classic 50-watt amps.

However, as the frequency is lowered to 20Hz, you can eventually see the distinctive "bite" out of the falling edge of the waveform, caused by the transformer core running out of volt-seconds. Because this amp has less NFB, the distortion is more gentle, and more like what the transformer itself would produce without NFB around it.

An important thing to realize is that saturation isn't caused by high power. It's caused by a combination of high power AND low frequency. Hence, it only affects the bass register. Think of it like excursion on a woofer.

[Steve Conner]

Second set of pics.

[Diablo]

So, it looks like OT saturation really isn't part of the tone/character for guitar amps. So, how significant is the OT design in the overall sound of the amp? Is any tonal variation from the OT mostly due to the reflected impedance into the power tubes, and changes in the impedance from OT to OT? Everyone seems to agonize over the choice of the OT, paper vs plastic, interleaving, and type of steel etc. It sure doesn't look like they influence saturation at low frequencies, so how can they contribute to the tone at various frequencies?

[R.G.]

So, it looks like OT saturation really isn't part of the tone/character for guitar amps.

Certainly not the way it gets bandied about in advertising, it isn't. OT saturation is a distortion mechanism for the lowest frequencies.

So, how significant is the OT design in the overall sound of the amp?

From somewhat to very. Just because it won't saturate doesn't mean it has no effects. Signal transformers cause very definite changes to signals. But they are (a) measurable and (b) non-magical changes.

Is any tonal variation from the OT mostly due to the reflected impedance into the power tubes, and changes in the impedance from OT to OT?

The real bottom line is that transformers do predictable changes. EE research long ago developed a series of transformer models to help predict transformer performance to a desired degree of accuracy. The more accurately you want to predict what happens, the more accurate - and complex - a model you use.

A signal transformer is, among other things, a set of frequency-dependent components all bundled up where you can't necessarily change one thing without changing another. Transformer design is almost pure compromise, because the things you change to make one thing better make another thing worse.

In the middle of a transformer's frequency and magnetic drive range, a transformer is nearly ideal. That is, it converts voltage and current up/down by its turns ratio(s), and the effects of the filtering components are small enough to neglect.

At the low end, bass is limited by the inductance of the primary. The primary winding must magnetize the core to get signal to go across to the secondary. There are both resistive losses and reactive losses to do this. The reactive loss is the amount of signal from the source that simply leaks through the primary inductance and is not available to the secondary. The effect of this inductance can be measured, designed, and otherwise manipulated.

The signal source driving the transformer will have some source impedance, through which it drives the primary. The reflected secondary load appears in parallel with the primary inductance. So right off the bat, you have a high-pass filter determined by the source impedance, the secondary loading, the wire resistances, and the primary inductance. If the secondary load is a resistor, it's a simple filter. If the secondary load is a reactive load like a loudspeaker, then the filter is no longer simple.

Above the mid-band where everything is kind of ideal-ish, the leakage inductance, parasitic capacitances, and core losses of the transformer come into play. The impedance of the primary inductance gets so big by mid-band that it can usually be ignored for the high end.

Leakage flux in the transformer appears to the electronic side of things as a small inductor in series with the primary and secondary. Parasitic capacitance appears as a self-capacitance in parallel with the primary, in bridge to the secondary itself, in parallel with the secondaries, and in general, everywhere. So the high-end filtering looks like a source with its now-complex impedance (if needed) driving a parallel capacitor, series leakage inductance, shunt capacitor, reflected complex secondary load, and a bridge capacitor from the primary to the output. All of the complexities of filters with multiple reactances come into play for this. The complexity of this filter is one thing that limits the application of feedback in amps with output transformers - the phase shift from the filter at high frequencies prevents you from using all the gain you'd otherwise need.

It is a daunting task to design an OT which has big enough power handling, big enough primary inductance, small enough leakage inductance, small enough (and balanced enough, at a higher level of accurate design) shunt and bridged capacitances.

Can all this filtering affect tone? Oh, yeah!

Everyone seems to agonize over the choice of the OT, paper vs plastic, interleaving, and type of steel etc.

Paper versus plastic bobbins/windings is superstition and myth. Both paper and nylon are transparent to magnetic field at these frequencies. I have an article on this at geofex, which I expanded into one of my monthly articles while I was doing the column for Premier Guitar. An unqualified statement that paper has better tone than plastic is a good indicator that the person making the statement does not understand magnetics to any significant degree.

Random *winding* on a bobbin makes a difference, because leakage inductance and capacitance depend on where each wire is with respect to its neighbors. Correct winding in sectionalized, interleaved layers is easier to do on a cardboard bobbin, but it's the winding technique, not the bobbin material that makes a difference. You can interleave/sectionalize on a bobbin, although it's not common.

Interleaving windings is the way to reduce leakage inductance. Interleaving *the correct way and amount* is how leakage is controlled and high frequency response extended. Interleaving is not binary, where you would either have it or not. Interleaving is a matter of degree: how many turns/layers of which winding and how they are interspersed and interconnected.

Type of steel and its thickness does make a difference. However, almost no guitar amps ever use the more expensive and higher performing fancy-alloy steels. They're almost always plain old silicon-iron, same as power laminations, or perhaps a slightly thinner lamination. Thickness matters at the high end, because the core is laminated in the first place to eliminate eddy current losses in the iron. Thinner lams reduce eddy current losses more. The steel determines the B-H curve, which I've hammered on a lot. Choice of a different core material can affect the primary inductance and turns a lot. But again, almost no guitar amps (none that I've ever seem) use anything but ordinary power-transformer lams. Maybe fancy nickel-iron or cobalt-iron lams are used somewhere. Maybe those Plitron toroidal outputs use fancy core materials. Maybe.

It sure doesn't look like they influence saturation at low frequencies, so how can they contribute to the tone at various frequencies?

Filters. See filters, above.

[Diablo]

Thanks RG for taking the time to post that good explanation. As to the choice of steel for the OT, my understanding is that some (most?) use M6 grain oriented steel, while others use M19 which is a non-oriented steel. Some of the folks cloning Trainwreck amps have had transformers produced with the non-oriented steel in order to duplicate the original OT, which they believe used non-oriented steel. Being a steel guy, and a tube amp hack, I'm just trying to understand why the lossier (higher core loss) steel would have some extra special sonic "goodness". I think you've explained it - everything (ferromagnetic material + winding pattern) influences the filtering.