Secondaries, impedance and speakers - some observations...

[EFK]

Maybe someone with a lot more knowledge than I can answer this for me. There are countless threads dealing with impedance matching etc., and various theories regarding the use of "the whole winding" of the OT. That being said.

Is there any science or math behind the concept of a different impedance - all else being equal - affecting the tone/output of the amp? I'll use an example. Lets assume a single amp used as a test subject. No changes. Now let's assume you have two identical speakers, save that one is 8 ohm and one is 3.2 ohm. Now lets assume you have two identical output transformers wound, each identical save the turns ratio is modified to reflect the required indentical impedance to the output section. Would you expect them to sound different, or the same, or ???

I guess what I'm asking is, speaker and OT differences aside, are there proven differences inherent to a low impedance load (3.2 ohm or 4 ohm) vs. a higher impedance load (8 ohm or 16 ohm)?

[Jazz P Bass]

I think I see what you are asking but the test can't be setup because the output wattage would change if it is the same amp.

[EFK]

This is getting at what I'm asking. Why would output watts change? Same amp, 7K to 3.2 ohm vs. same amp, 7K to 8 ohm for example. Maybe I'm not explaining myself properly!

[Jazz P Bass]

I may have to take that back,
Still, as far as the physics go, the power transfer function would be identical.

[J M Fahey]

The amp would work the same.
As à side note, same speaker different impedance coil: the higher inmpedance one will be subtly brighter, because copper weight will be less.

[g1]

If I can change your example around a little, you are talking about proper impedance matching in both cases? Like an amp with an impedance switch set properly connected to a 4 ohm load, then set properly for connection to an 8 ohm load?
The proper impedance match should ensure maximum power transfer, so the output wattage should be the same in both cases.
The only difference should be due to the speakers, as JM noted, (frequency response curves, coil weight etc.).
Also there may be some slight differences due to feedback network if the output impedance tap is not compensated for. Not so sure about that one though.
So I think the question probably boils down to: for a given speaker model, is there any audible difference between different impedance versions?

[R.G.]

For the very limited case of
(1) two speakers being identical except for being wound for different impedances
(2) two amps, designed to be as identical as possible except for the output transformers having different ratios
(3) all else as identical as possible

No, there is no necessary difference in response to a first order. There are second-order, smaller effects due to it being impossible in a practical sense to actually make identical speakers except for impedance and identical output transformers except for ratio. You could get very close if you were making these parts to be as identical as possible, but in practice no one ever does.

Getting speakers to be acoustically identical is a matter of making the mechanical setup and electromagnetic setups identical. Using matching cones, spiders, and voice coils as well as adhesives, etc, for the construction gets you most of the way there. Then as JM says, you have to match the ampere-turns product on the voice coil *and* the weight of the copper at the same time to get equally-massive voice coils at the same time as you get different impedances. It may not be possible to select these arbitrarily as in 8.0000 ohms and 4.000 ohms or 3.2000 ohms. And of course, the frame and magnet have to be matched. But you could do a decent job.

Making otherwise-identical output transformers with different output impedances is harder. The response of an OT depends on literally everything in it, right down to how well it's jogged together and varnished. Leaving the primary windings and interleavings identical and changing the secondary windings to have half/double or square-root-of-two more/less turns in the same spaces and get the winding resistances to be in the same proportion as the change of ratios may or may not be possible, or may be possible only for certain "magic" ratios. But you could get close-ish, and possibly use external parts to correct what you got in the transformer to be more identical.

All that being said, there are physical processes that make frequency response in speaker/transformer setups change that ARE a function of impedance. Mostly these are issues of trying to wind different numbers of turns of wire into the same physical spaces and get scaling of resistance, inductance, and self-capacitance (and for speakers, mass and weight, which are different) to scale linearly. These things fundamentally are not linear, so the can't be scaled linearly. However, for small changes, in the range of ratios of 2, 4, perhaps 8, the failure of scaling is small.

It turns out that this is the reason we don't have speakers with impedances of 2.2K ohms or 30milli-ohms. For the range of commercially practical magnets, and frame/cone/voice coil materials, you reach a loosely-defined peak of effciency and broad audio response in the range from about 2 to 32 ohms. This is mostly due to the fact that today's speakers are largely electromagnetic motors.

There are other ways to do this. There are electrostatic loudspeakers. Look up "magneplanar". There are ribbon tweeters. And there are drivers specialized for driving the acoustic impedance at the throat of a horn. Those involve different trade-offs.

But in the limited case I think you're talking about, no.

[EFK]

Thanks everyone for sounding off. I was thinking about this because I remember reading somewhere that there was an inherent difference of some type in amps working into an impedance matched 4 ohm load for example vs. a 16 ohm load. Can't remember where I saw that but since it was somewhere on the internet it must be true.

I was thinking about it, because I've been getting a bunch of speakers reconed lately by a very competent and experienced reconer. Primarily, these have been old Rola speakers for Supro amps, which means 3.2 ohm speakers which are practically impossible to find unless you have the original speaker. Anyway, I lucked into a couple of alnico Rolas which are EXACTLY the same speakers used in a bunch of Valco amps, however these were 8 ohm. So, I sent them out and asked if they could be reconed in 3.2 ohm. No problem, and here's where it gets interesting, as he said there was excess space for the 8 ohm voice coils and that they would sound much better at 3.2 ohm, which apparently due to heavier wire is a bigger coil. Now, I'm no expert on speaker function, but what's funny is that this actually seems to match up with some of my experience with these old Rolas. I like old Rola alnico speakers, and I've been through a bunch of them. And what I've always found interesting is that if I'm using them with a new made amp with multiple OT taps, the same amp using a 3.2 ohm Rola sounds clearer, brighter and louder than if I run it through the identical type of Rola that happens to be in 8 ohm. So while this obviously doesn't apply to all speakers, it's happened to me more than once specifically with these Rolas and I was never able to explain it until the recone guy made the comment he did.

[R.G.]

So, I sent them out and asked if they could be reconed in 3.2 ohm. No problem, and here's where it gets interesting, as he said there was excess space for the 8 ohm voice coils and that they would sound much better at 3.2 ohm, which apparently due to heavier wire is a bigger coil. Now, I'm no expert on speaker function, but what's funny is that this actually seems to match up with some of my experience with these old Rolas.

That makes sense. The fit of more/less turns of thinner/thicker wire onto a voice coil is one of those places where everything else can't be equal without really strenuous effort.

In speaker designs, it would be best for the linearity of the speaker if the voice coil could be exactly one turn of wire of negligible size. This would mean the entire coil/turns saw the same magnetic field all the time, no matter what. And it would be best if the speaker magnet could make an equal-intensity magnetic field uniformly in the entire volume of space the voice coil moved through.

As a practical matter, neither of these are possible. You can't make magnets produce a completely uniform field, only come close, and even then there is leakage and edge effects where the strength varies. You obviously can't make a zero-width, zero-resistance voice coil. The opposite is more practical - a small-length cylinder of magnetic field, with a very long voice coil. Only part of the voice coil is inside the M-field at any time, and what there is inside the M-field is the same no matter what movement the voice coil does... but even then there is some limit to the length of the voice coil, so there are end effects.

And there are limits to the linear movement of the speaker cone and spider. So it makes sense - changing the length/turns/wire thickness/wire mass changes the frequency response. Many of these factors either contradict one another or interact in complex ways.

I'm convinced that people who design speakers are one subtle form of masochist.