[Squalish]

Is fully balanced crossfeed attainable through merely haveing two crossfeed board, a plus and a minus(presumeable each L/R grouping passes AC relative to each other), or is it more complicated?

 

[stereth]

Isn't "balanced crossfeed" an oxymoron?

 

[Squalish]

Balanced bridge has little to do with channel seperation except for the seperated grounds. 4 conductors are used for balanced stereo, a +right, -right, +left, and -left. Rather than an AC signal being pushed/pulled from one conductor to each driver, and the electricity coming back on a common ground, two identical but reversed polarity(inverted) DC signals go into the drivers, ideally cancelling out completely there. Aside from the mostly doubled specs on the amp(which is mostly what I'm after, several have mentioned much better bass handling and increased midrange detail IIRC on the BB gilmore), this makes crossfeed purely voluntary, as opposed to the chaotic inductive/capacitive/whateveritive type induced by the cable.

If my methodology is wrong, please correct me, I'm still learning.

I assume that as long as you do it identically to + and - sections, the simplest form of crossfeed, a fixed resistence from left to right(whether it be zero ohms for mono or more for normal crossfeed), would work. What I'm unsure about is the various adjustments made, such as bass boost for psychoacoustic compensation and delays, that Linkwitz, Moy, and Meier have added.

Tangent, weren't you working on something like this, with a multiloop gain stage to adjust gain to 1(or is it 0?)?

 

[tangent]

Quote:

Tangent, weren't you working on something like this, with a multiloop gain stage to adjust gain to 1(or is it 0?)?

The person who asked me for that wanted the META42 sound with crossfeed but for a Stax balanced headphone system. He didn't need amplification (the Stax system provides that) but he did want the flavor of the META42 mixed into his system. I tell you all this so you can see where my idea came from.

My plan was to have a balanced receiver which converts the signal to unbalanced to go through the amp and the crossfeed, and then use a balanced driver to go out to the headphone system. Thus gain is 1 through the system, and you don't have to get clever about the bits in the middle. Since the crossfeed/amp path is short, a lot of the benefits of balanced operation aren't very significant. Also, crossfeed circuits like to be buffered, which the gain and balun stages provide. Probably the balanced receiver and the amp could be blended a bit, saving some components.

I did consider doing what you're talking about at one point. My objection to that was that most caps have pretty poor tolerances, so you'd have to do a lot of tight matching, because the errors will appear like "signals" at the amp. 1% mismatch is like a -40dB signal -- plainly audible. Now realize that you have to match the balanced crossfeed halves to 0.1% to get tolerable error levels (-60dB) . Sounds like a mess to me, with unclear benefits.

 

[Dusty Chalk]

Quote:

Originally posted by Squalish Is fully balanced crossfeed attainable through merely haveing two crossfeed board, a plus and a minus(presumeable each L/R grouping passes AC relative to each other), or is it more complicated?

As tangent hinted at, no, this would not provide crossfeed, it would "double" the crossfeed. You would want to provide half as much crossfeed on each of the two directions (+ and -) to get singular crossfeed. Don't think of the signal as being only on the + in an unbalanced system, it's still + and - (the electrons gotta go somewhere, right?).

You DIY'er's correct me if I'm wrong, I'm still lurning, too.

 

[tangent]

I think what you'd end up doing is running the + signals through one crossfeed circuit and the - signals through the other. The error signal issue I brought up is because R+ isn't modified in exactly the opposite way as R-, so this appears as a differential signal to the next stage. Same for L+ and L-. Unless you match those two circuits very closely, these error signals can easily be too high to be tolerated. Matching caps in particular is a big problem.

 

[Squalish]

Okay, so caps are a no-no, as tolerences are rarely under 5%.

Would this error signal be analogous to a DC offset, silently killing your cans, or would it have an effect on the sound(I'm guessing it would bring the + and - out of phase or something?)? I would think one might be able to live with a tiny DC offset that a highly matched pair might bring, but if there is an audible effect, that obviously wouldn't be the way to go.

I'm re-reading Headwize's documentation on crossfeed. I have a renewed level of respect for some of the guys that figured out/designed this stuff in the first place.

Next question: What problems would a very simple resistor-based channel-mixing filter, without time delays or filters or anything, have?

 

[Dusty Chalk]

Quote:

Originally posted by tangent I think what you'd end up doing is running the + signals through one crossfeed circuit and the - signals through the other. The error signal issue I brought up is because R+ isn't modified in exactly the opposite way as R-, so this appears as a differential signal to the next stage. Same for L+ and L-. Unless you match those two circuits very closely, these error signals can easily be too high to be tolerated. Matching caps in particular is a big problem.

No, I meant when you said that you basically just implemented an unbalanced crossfeed so that "...you don't have to get clever about the bits in the middle..."

I still think if you're going to do anything, you put a regular crossfeed circuit on the +, and leave the minus alone. Or would that cause some sort of phasing problem? Yes, I think it would.