Transmission Lines as Applied to Audio

[J-ROB]

here are a few articles I can remember sorta related to the question David asks, at the speaker end:


Bob Cordell on Interface Intermodulation

https://pearl-hifi.com/06_Lit_Archive/1 ... ortion.pdf

and

Otala, Matt Intermodulation at the Amp-Speaker Interface, Wireless World 12/80, p42ff.

Feedback loops can serve as in input, so to speak.

But can we really terminate the cable in a way that solves the problem in the way we can with RF interconnections?

I'm just saying that transmission lines have nothing to do with it, as far as I understand the terminology.

"Iransmission line" is not a general synonym for "cable" or "interconnect," it is a particular case of interconnecting cables whose length approaches a wavelength of the signal.

[dberning]

Just to further clarify where I am coming from on this, consider high-frequency signals that leave the amp and pass to the speaker cable. These signals could be noise or left over unfiltered residue from D to A conversion, stuff outside the normal audio range of hearing. If the speaker cable and source impedance of the amp and the impedance of the speaker were all matched, this high-frequency energy would be dissipated as quickly as possible because 100% would be transmitted to the speaker (minus losses due to dc resistance and radiation). If the matching is less than ideal, some of that energy would be reflected at the interfaces, and would rattle around back and forth in the cable and some could potentially be back fed into the feedback loop of the amplifier and even fed to the speaker at a later time than the initial corresponding signal.

Now all of this may be of no consequence and I have no way of verifying whether this could present an issue in any particular audio system, but it is an issue in wide-band video systems. I recall working with some video amplifiers that would oscillate if the output cable was not terminated with 75 ohms at the far end of the cable.

David

[SoundMods]

Just to further clarify where I am coming from on this, consider high-frequency signals that leave the amp and pass to the speaker cable. These signals could be noise or left over unfiltered residue from D to A conversion, stuff outside the normal audio range of hearing. If the speaker cable and source impedance of the amp and the impedance of the speaker were all matched, this high-frequency energy would be dissipated as quickly as possible because 100% would be transmitted to the speaker (minus losses due to dc resistance and radiation). If the matching is less than ideal, some of that energy would be reflected at the interfaces, and would rattle around back and forth in the cable and some could potentially be back fed into the feedback loop of the amplifier and even fed to the speaker at a later time than the initial corresponding signal.

Now all of this may be of no consequence and I have no way of verifying whether this could present an issue in any particular audio system, but it is an issue in wide-band video systems. I recall working with some video amplifiers that would oscillate if the output cable was not terminated with 75 ohms at the far end of the cable.

David

To further support David -- this is one of the reasons a British engineer recommends connecting a 15-uH inductor in series with the hot feeds to the speakers to block return noise from entering the ampliifer and taking a ride on the feedback loop. Personally -- I have been using ferrite RF filters strapped to the hot feed to my speakers for decades now.

Is it audible you ask? YUP! A subtle fog is lifted.

[ratbagp]

So is the inductor installed at the amplifier end or the speaker end?

ray

[SoundMods]

So is the inductor installed at the amplifier end or the speaker end?

ray

Amplifier end -- as close as possible. It has to block or at least limit infiltration of reflected "noise" or RFI/EMI picked up from the airways.

[J-ROB]

Here are a couple photos of card wound resistors I found at the hamfest on Sunday FYI...about all I could find to buy!

The two on the nude phenolic formers look to be WE or Daven, And I found a weird rubber coated mica card resistor marked Tektronix. This construction is naturally low impedance. Zero impedance is basically impossible. Stability and low noise are the reasons for these rather than low inductance per se. There are some with special winding geometries for even lower inductance, but these kinds of resistors are usually in DC or low frequency circuits where the small inductance values have little effect.

--------------
Radio textbooks will say that power transfer is maximized and reflections minimized when a line is terminated in its characteristic impedance. So in David's example of a 75 ohm line, if you terminate it with 75 ohms to ground, it acts like it is infinitely long, so no reflections occur.

In audio, this approach of matching the line will result in input impedance values that might be tough to drive, unless you have one of Walt's 6 ohm Zout Levinsons (without the bobbin mod.)

I don't know how to make a cable with 10k-100kohm characteristic impedance. Twisted pair is ~100 ohms. Easier to go lower than higher with any cable one would want to use in hifi.

However characteristic impedance of a cable is not the same notion as a transmission line. In a transmission line, the instantaneous values of voltage and current vary along the line and this phenomenon has pronounced effects when the line is at least significant fraction of a wavelength.

Think of a transmission line speaker. The length of the passage is a specified fraction of a wavelength and the phase rotation of this defined length is used for impedance transformation. 1/4 wave being the commonest implementation. a quarter wave transformer transforms an open circuit into a dead short. Complete inversion. Xl becomes equal and opposite Xc and so on.

In my current speaker project, I measured a 425hz suckout with my phone. lo and behold, the front to back dimension is a half wave at that frequency! Aha! ReflectionZzzz!! I put in that piece of rock wool board I was looking for to break up that path inside the cab and no more suckout. This is a reflection, coming back 180 degrees out of phase to the cone, but I wouldn't call this a transmission line design, using acoustic path length as a transformer, its just a box that had a destructive reflection.

I'd be careful about using any reactive components to terminate lines because while they will work at one frequency, they may PEAK the reflection at another frequency. Audio is very wide bandwidth phenomenon, covering many octaves. You can't cover that range with a tuned circuit. For the same reason, transmsission line principles seem to have limited relevance in audio interconnections.

Anyway, with speakers the back EMF and the speaker acting as a microphone sending signals back to the mothership would probably far outweigh reflections due to impedance mismatch at the speaker terminals.

[SoundMods]

Wait a minute!!!! What Hamfest!!!!

[J-ROB]

..._ .. . _. _. ._

[SoundMods]

..._ .. . _. _. ._

HUH?!

[pboser]

Morse code?