Output Transformer "Using All Its Potential"

[nashvillebill]

Okay, here's one from over on an audio forum. A newbie has bought a tube amp with OT's with multiple taps:, 2, 4, 8, 16 ohms. He has not yet bought speakers. What should he buy? The advise from a senior member with many posts was "Buy 16 ohm speakers (and put them on the 16 ohm taps), that way you will be using all of the OT."

Huh? That one baffled me, unless there's some arcane theory I'm missing. So using 8 ohm speakers on the 8 ohm tap somehow won't use all the OT?

Of course, these are the folks who raise their speaker cables off the floor, and buy cryo treated fuses....

[g1]

Some guys think there is some advantage to having current running through the whole of the OT secondary winding, as opposed to part of it.

[J M Fahey]

Nonsense.
Besides he does not know how the transformer is wound, he can't even open his mouth until he knows what gauge wire was used for each section.

But that's not the point anyway.

When you "use" a transformer you use its core, **all** of it, for the very good reason that all turns, no matter if few or a lot, go all around it.

Thinking that some turns only use part of a core surface or section and you need all to use it fully indicates not having the slightest clue about how a transformer works.

[GainFreak]

I've also read that running a 16 Ohm speaker provides better damping as opposed to 8 or 4 Ohm speakers.
Also usually the 16 Ohm secondary is fully interleaved which results in brighter sound (less leakage inductance) and better low frequency response.

[Enzo]

I'd say that is largely rationalization on the part of those claimants. Fully interleaved? How does anyone know that unless they took apart that particular transformer and looked? And why would a lower Z winding not be interleaved then?

[R.G.]

Well, the answer is two-fold:
(1) Maybe. It depends on how the internal windings were done, including sectioning, interleaving, and tapping.
(2) It's likely to be too small a difference to hear.

A 4 ohm secondary has half the number of turns that a 16 ohm winding does - that old impedance is the square of the turns business. An 8 ohm winding has 1.414 (theoretically!) times the number of turns in a 16 ohm winding. It is easy-ish to make one secondary winding with taps at 0.5 and 0.7 times the total turns to get 4, 8, and 16 ohms. But that doesn't get you interleaving for better frequency response and lower crossover distortion at all. So you need to at least do two of these in parallel to get any interleaving. Better would be what the Williamson OT did: many sections, interleaved all over the primaries, and connected series/parallel external to the transformer to get the correct impedance. But that's not a simple tap.

Every turn has some leakage inductance from every other turn, depending on how far away from that other turn it is in the winding window. You would like to (as a transformer designer) have every winding completely interspersed with every other winding for low leakage, but that's impossible to do. So you compromise.

Having some portion of the windings not carrying current does in fact mean that it's not contributing to the coupling. And since it theoretically could be providing power, the portion of its winding space that could be producing heat and dissipating it is not being used. But the effect is not very big compared to the overall power capability, and not terribly significant in a transformer that's distinctly mid-fi by stereo standards.

This is kind of one of those grain-of-truth things that gets you to swallow a bigger conclusion that may not necessarily follow, or be that significant.

[R.G.]

I'd say that is largely rationalization on the part of those claimants. Fully interleaved? How does anyone know that unless they took apart that particular transformer and looked? And why would a lower Z winding not be interleaved then?

Yep, rationalization. You got that one right. No way to tell without the maker's specs in more detail than they're usually available or un-winding one.

A lower Z winding is usually not interleaved, for quick-and-dirty reasons. It's much faster and cheaper in labor to wind a small number of thick wire once than to do it the deluxe, superior way of winding many same-turns sections interleaved all over the place then paralleling them up, or doing the series/paralleling thing I mentioned with the Williamson.

There's really no way that a generalization about a particular impedance being interleaved or not can be settled without tearing one down, and hoping all of them are the same - which is another story on its own.

[nashvillebill]

Thanks, that's sorta what I expected!

[GainFreak]

From Fryette users forum:

A while ago I read Steve's explanation of how the Vintage 30 has evolved, and at the end he pointed this out:

"When you change the impedance rating of the speaker, you also change the inductance of the coil which changes the resonant frequency among other things. Also an 8 ohm speaker will be driven by the amp set at 8 ohms which lowers the amplifier impedance (compared to the 16 ohm setup), and thereby lowers the amplifier damping factor, which flattens out the system (amp plus cab system) frequency response and resonant behavior. So if you have 2 identical amps and 2 identical cabs - one cab loaded with 8 ohm speakers and amp set to 8 next to one loaded with 16 ohm speakers and amp set to 16, they will sound noticably different. The 16 ohm rig will usually sound brighter and deeper."

So I've always wondered something. That applies when you're talking about 4x12 cabs, but what about the way a 2x12 is wired?
What is going to achieve that wider frequency/more resonant/brighter/deeper feel?

Two 8 ohm speakers wired in series for a 16 ohm load?
Or two 16 ohm speakers wired in parallel for an 8 ohm load?

What is more important in the equation for a 2x12 cab...the individual speaker impedance, or the overall cab impedance?

This is the answer from Steve:

What I was referring to was the construction of the speaker itself. Taking a 16 ohm vintage 30 for example, if you simply decrease the speaker voice coil turns to make it an 8 ohm speaker, the chances that it would sound the same in all other respects are slim. By playing with other factors such as wire gauge and doping, they can me made to sound similar, but not identical.

So, it's important to know that when you are comparing cabs. Comparing two identical cabs operated by identical amplifiers, but one cab loaded with 8 ohm V30s and the amp set to 8 ohms vs one cab loaded with 16 ohm V30s and the amp set to 16 ohms, you are going to hear a pronounced difference in sound and performance. The differences will be partly the difference in electrical characteristics of the speakers and partly the resulting change in reactance of the amplifier.

Even so, I think you'll find that two 8 ohm speakers wired in series for a 16 ohm load will produce a deeper wider sound.

[g1]

At this point someone usually nods their head knowingly and responds (to Steve, or whoever) with "yep, all the vintage Marshall cabs were 16 ohms, that's why". But they don't consider that the classic Marshall stack was 2 cabs with the head set to 8 ohms.

[Justin Thomas]

I think the only real consideration by Marshall (& Fender, I bet) was, how can we make these things available to AS MANY PEOPLE AS POSSIBLE, FOR a competitive price point, using parts that we can get STEADILY, that will allow the most VERSATILITY for AS MANY PEOPLE as possible, not all blow up, AND make great music? I can hear Jim & Ken now... "hmmm, if we use these parts, we can sell more, it's more modular, amd our customers have choices and options, so everybody wins."

Reasoning behind it all? $, $, $! It's only recently where manufacturers got into all the minutae & esoterica of tramsformer rations & resistor compositions.

Justin

[J M Fahey]

From Fryette users forum:

"When you change the impedance rating of the speaker, you also change the inductance of the coil which changes the resonant frequency among other things.

Sorry but no.
The voice coil inductance raises impedance at higher frequencies (usually starting >1KHz), because it's a small inductor in series with the DC resistance of the wire.
Resonant frequency is mechanical and usually lies between 60 and 100Hz.
The "inductor" is moving mass and the "capacitor" is compliance (elasticity).

Also an 8 ohm speaker will be driven by the amp set at 8 ohms which lowers the amplifier impedance (compared to the 16 ohm setup), and thereby lowers the amplifier damping factor,

No because the transformer works both ways, it turns speaker impedance into what the tubes like and viceversa.
The 8 ohms tap shows a certain internal impedance given by turns ratio and that combined with speaker 8 ohms gives a certain damping.

The 16 ohms tap, shows a different (higher) internal impedance because it has a higher # of turns ... but the 16 ohms speaker has also a higher impedance.
The damping factor, which is a ratio, does not change.

which flattens out the system (amp plus cab system) frequency response and resonant behavior. So if you have 2 identical amps and 2 identical cabs - one cab loaded with 8 ohm speakers and amp set to 8 next to one loaded with 16 ohm speakers and amp set to 16, they will sound noticably different.

May sound different, but not because of the explanation offered by Steve.

The 16 ohm rig will usually sound brighter and deeper."

May sound brighter and deeper, but not because of the explanation offered by Steve.

What I was referring to was the construction of the speaker itself. Taking a 16 ohm vintage 30 for example, if you simply decrease the speaker voice coil turns to make it an 8 ohm speaker, the chances that it would sound the same in all other respects are slim.

Oh, but NOBODY does that
If you have a guitar 16 ohms speaker , remember that in such speakers the VC is wound to a length that equals top plate thickness (usually 8 mm) or maybe 1 mm longer, and you pull half the turns to make it 8 ohms, your coil now is half as long as before and you are wasting 50% of your magnetic strength
What you actually do is to use slightly less turns (79% as many, definitely not 50%) and a slightly thicker wire (26% thicker), which combined give you a half impedance but same length as before.
By the same token, and knowing that inductance changes with the square of turns, the 8 ohms speaker will have half impedance but 62% the inductance, so it will have *slightly* less treble.
Barely audible, definitely not the marked difference described.

Comparing two identical cabs operated by identical amplifiers, but one cab loaded with 8 ohm V30s and the amp set to 8 ohms vs one cab loaded with 16 ohm V30s and the amp set to 16 ohms, you are going to hear a pronounced difference in sound and performance.

Not pronounced, but slight.

The differences will be partly the difference in electrical characteristics of the speakers and partly the resulting change in reactance of the amplifier.

The "reactance" (impedance/damping?) non change has been explained above, the transformer different taps take care of that.
The change in electrical characteristics results in slight (quite less than 1 dB) treble loss.
The actual change, which exists and although still small is more important than above mentioned, is mechanical and so obvious once somebody points at it that it clicks instantly:
It's pure Geometry, forget Ohms and milliHenries: same lenght winding but thicker wire means more copper is in that gap meaning it weighs more. Period.
So the same coil in 8 ohms instead of 16 ohms is slightly (26%) heavier, will accelerate and stop slower than an equivalent 16 ohms one.
The net result is that the 16 ohms one has slightly more attack, bite, you name it.
But it's a purely mechanical effect, not an electrical one.

Even so, I think you'll find that two 8 ohm speakers wired in series for a 16 ohm load will produce a deeper wider sound.

Deeper wider than what?

If compared to (different coil) 16 ohms speakers, it's the other way round.

If compared to same speakers, wired in parallel and connected to proper 4 ohms tap, there's no change .

FWIW I actually design voice coils for different sounds on a regular basis an am painfully aware of the compromises which have to be reached.

With due respect, I think Mr Steve Fryette designs great amps, but must order speakers from Celestion (or other suppliers) , probably stating what he wants, but let the fine details to them.
Or listen to many samples and pick what he likes best.

Oh !!! and where did I pull the 26% and 79% ratios from, thin air?
No, the factor is the cubic root of 2 .

[GainFreak]

At this point someone usually nods their head knowingly and responds (to Steve, or whoever) with "yep, all the vintage Marshall cabs were 16 ohms, that's why".

Considering how the early Marshall output transformers were wound (almost no interleaving, except for the 16Ohm tap) I guess that it's due to the OT not the cab. The better (more) a secondary is coupled to the primary, the less leakage inductance which results in better high frequency response.

If compared to same speakers, wired in parallel and connected to proper 4 ohms tap, there's no change .

I got a little confused so let's keep it simple and practical. Which combination of speakers (for a 2x12" cab) will have better low frequency response:

1/ Two 4 Ohm speakers in series
or
2/ Two 16 Ohm speakers in parallel.

[teemuk]

Let's pause for a moment to think...

These are guitar amplifiers we're talking about. They are NOT going to feature frequency response that extends from 20Hz to 20 kHz, let alone the fact that a generic guitar cabinet couldn't even reproduce such response.

On a contrary, they are usually designed to have a quite limited bandwidth, mainly in the mid-range region, because too extended low frequency response would mean "farting" and too extended high frequency response would mean "icepicks stuck to your ears".

So do we really care if the frequency response expands to resemble a good HiFi amp when we connect the speaker to some magical tap in the transformer...?

Just saying.

[sgelectric]

As I recall this idea was initially expounded in one of Gerald Weber's books.
IIRC Kevin O'Connor called it BS in one of his books and suggested that perhaps the fact that GW was (is) in the speaker business might have something to do with it.
That or technical ignorance.

SG