Free Power Attenuation?

[ashcatlt]

Part of me is saying that this idea is so simple that if it was actually effective at all, everybody'd be talking about.

Another part of me vaguely recalls having read about this fairly recently.

Now, I was thinking specifically about a Super Reverb, but if it works, it would work for any four speaker cabinet.

Ready?

Reverse the wires on one speaker.

That's it. Quick, easy, completely reversible.

The idea is that this one speaker will pull while the others push, and will cancel - and be cancelled by - the contribution of one other speaker's worth of sound. So you'll only hear the contribution of two speakers, and the perceived volume will be halved.

This might not be enough to allow you to crank up your Marshall while the baby's sleeping, but it might help control stage volume, so you're not drowning out the rest of the band and pissing off the soundguy.

Seems like this method would overcome a number of the problems of dedicated attenuators.

1). Like I said, it's free and reversible.

2). The amp still sees the same load from the cabinet, with all of its frquency dependent reactivity.

3). The speakers are still moving the same amount. If we accept that speakers must get moving to some extent to sound their best, or if we like the sound of speaker distortion, we still get all that.

Anybody heard of or tried this. Does it make enough difference to be worth it? Is it potentially dangerous? Specifically I wonder if it would cause some acoustical loading or unloading that could damage the amp.

[newey]

Never heard of this or tried it. My armchair speculation would be "not much difference", as it doesn't seem to me that acoustic phase cancellation works in the same way as electrical phase cancellation.

Each speaker can be thought of as a point source, with the sound waves interfering with each other, and potentially cancelling, only along the margins of the dispersion. Or, at least, that's my off the cuff thought.

If switching one speaker effected a meaningful partial cancellation, switching 2 should then give almost a full cancellation, and I don't think that would happen. There's also the issue of differential resonances from the cabinet, given differing placement of each speaker- seems like a whole lot of variables involved here.

I can't see how this would be dangerous, at least not in the short term. Perhaps the different loading on the out of phase cone from the other 3 would, over time, degrade that cone- but again, that's just speculation.

I think some testing may be in order. I'd do it, but I don't own a 4-speaker cab.

[sumgai]

ash,

Interesting, but for all the reasons pointed out by newey, it doesn't work. As you yourself guessed, if it were that easy, then a lot of players would be doing it now.

In point of fact, you can emulate the (bad) effects of speakers-out-of-phase by simply reversing the wires to one channel of your stereo - right or left, it won't matter. Notice the weird tonality? Equally notable, as you move around the room, the effect seems to change somewhat. That's the sound-wave interaction that newey was talking about. As the point-sources intermingle, and bounce off the walls, it becomes a collection of standing waves. What we percieve as normal is what could be called, loosely, the base standard set of standing waves. Our ears can easily discern any discrepencies in such, and our brains are "wired" (taught) to think of these as less than desirable.

One might actually hit on a combination of tonalities that actually do sound good, or at least better than all four in phase, that's certainly possible..... a small possibility, to be sure, but nonetheless, it's not impossible.

If it were me, I'd go the distance as newey suggests, and see what happens with either pair of diagonal speakers connected out-of-phase. That should offset any cabinet resonance factors, or so I believe. And since it's reversible, it's a cheap experiment - just the kind we Nutz love! ;D



sumgai

[ashcatlt]

newey - Wave math is the same no matter the medium - be it an elecromagnetic field or your favorite fishing pond. Whether we're talking about an electron, an air molecule, or a weight hanging from a spring, if it's being pushed in one direction and pulled in the opposite direction with exactly the same force, it ain't gonna move.

My thought is that the proximity of the speakers one to another in this cabinet is much closer than what you'd find in most home stereos, and so the frequency where phase cancellation starts (or stops, in this case) to become apparent will be much higher up than what sg is talking about. I'm pretty sure that's part of the point of having all these speakers in one cabinet like this.

We'll see what Dave thinks about my experimenting with his antique. I hope he lets me. In the meantime, I decided to fake it with Sound Forge. Just for fun, and to try to illustrate a point re: phase relationships.

Let's start by ignoring the contribution of the room. This theoretical experiment takes place either in a completely dead space or hanging in the middle of the air somewhere.

Next, we'll take 4 perfect point source speakers and place them on a plane together in a square configuration where the non-diagonal pairs are 12" on center. This should be pretty close to a Super, with 10" speakers and about 2" in between. We'll call the speakers A, B, C, and D as per my pretty picture below.

Now we'll take a perfect omnidirectional microphone and place it on a line a perpendicular to the plane of the speakers, going through the center of speaker A, and 3' out. This is 3' from speaker A, 3.16' from speakers B and C, and 3.32' from speaker D.

We play white noise through them, gradually adding more speakers, then finally reversing the polarity on speaker D and capture the frequency spectrum analysis of each configuration.

Here's what it looks like:


The first thing I notice here is that the A+B+C+D trace is significantly louder than the A+B+C-D trace across most of the frequency range. In fact, it looks like about 5db difference, which is pretty darn close to the half volume I had talked about.

Then, I notice that the curve with the -D is actually alot flatter, closer to the curve of A alone than with any other combination.

The next thing I notice is that the area where these two curves devaite is centered around 4KHz. Most speakers of this type are going to have a pretty steep rolloff right around this area, which may tend to mask the differences here.

I find it interesting to note that the biggest dip comes with the even numbered "in-phase" speakers.

Now, moving around the cabinet will change this a bit, by changing the difference in distance between each speaker and the microphone, which will move the center of that phase-cancellation dip. Moving toward the cabinet (or around to the side of it) will increase the difference (ask Pythagoras), and lower the frequency of cancellation. Moving away from the cabinet (or toward the center of it) will decrease the difference, quickly raising the frequency to a point where it becomes meaningless.

That is, the further from the amp you are, the more closely the shape of the -D will resemble that of the +D. Or, more correctly, the more all of the traces will resemble A alone.

Of course, as mentioned above, the real world is far more interesting with all its variables and dependencies.

I'm encouraged by the fact that sg didn't come in waving his arms and screaming "FOR THE LOVE OF [expletive deleted] NO!!!" Leads me to believe it ain't gonna explode the amp.

[newey]

Nice analysis, Ash. +1 for the thought process and the graphing.

My point was, even regardless of speaker placement or room acoustics, if we treat each speaker as a point source (which it isn't in the real world), the sound emanates from that point in an ever-widening cone, and the 2 adjacent OOP cones won't interact to produce cancellation all across each dispersing cone, but only at the margins, where the waves interfere.

Like pluncking 2 stones into a pond.

But I don't see how this is destructive either, so test away.

I'm not sure sg's analogy to stereo speakers wired OOP is much help here, since with 2 separate channels, the sound sources are not congruent waves (wouldn't be much point to having "stereo" if they were!). We wouldn't expect much cancellation to occur if the sources differ.