Friday 26 February 2016

Surround Sound Switch #2: Commutators

Previously:
Mother Of all Relay Boxes
Colouring In

We're on the trail of a switch to rotate the sound stage in a room with eight satellite speakers. I've modified the previous diagram to include this component, labelled "7P8T Switch" - it has 7 Poles or circuits to switch, and 8 Throws or settings to accommodate our target of eight compass point orientations. Notice that one of the eight outputs of this switch has now been attached to the previously orphaned speaker number 8.

I've also coloured the wires connecting the amplifier output stages to the switch inputs, and those connecting switch outputs to loudspeakers. When working with multi speaker systems, both ends of every wire should readily be identifiable at all times. I use the following colour scheme, which agrees with the applicable Consumer Electronics Association (CEA®) Standard where possible:

(-)Common NegativeBlack    
FL Front Left White     
FRFront RightRed    
CCentreGreen    
SLSurround LeftBlue    
SRSurround RightGrey    
SBLSurround Back LeftBrown    
SBRSurround Back RightOrange*    
SBSurround Back*Yellow    
SWSub-WooferViolet*    

CEA® uses "Tan" for SBR, "Purple" for SW, and has no SB channel distinct from SBL/SBR. ANSI/CEA-863-A uses "Khaki" for SBR. My wire stocks are limited to the palette of the electronic colour code.

Spot the error?
Looking back to the diagram, the wire colours indicate that when the switch is in its initial (default) position, the SBL output is connected to speaker 1 (brown wires), SL to speaker 2 (blue wires), and so on. There is still no connection to speaker 8 in this switch position, since there is no yellow wire coming out of the receiver. As the switch is operated, these default connections are broken and various others made.

If you use wire markers of the type shown here, you'll be painfully aware their numbers don't lend themselves to a logical speaker numbering. The electronic colour code doesn't correspond to the CEA® audio source colour scheme. Best to ignore these numbers and just concentrate on using the correct colour marker for each wire. And maybe that's just as well, given the manufacturing error evident in these French made markers. Noticed it yet? Hint: it's in the right hand set.

Rolling Commutators

Okay, so what form does this switch take? Well, it need not be a literal 7P8T switch. Such a thing would be able to connect any input to any output, for each given setting or position of the switch; whereas we have certain considerations of symmetry to satisfy. For example, wherever the Front Left & Right (FL/FR) signals go, we'll surely want the Centre signal (C) to end up between those two. And we want to assign every signal exclusively to its own, unique destination speaker: no two distinct signals should ever become connected together. Assuming that we're uninterested in "mirroring" left and right (maybe to watch an upside down TV while practising handstands?) the symmetry group we're aiming for is just the set of rotations - and not the reflections - of a regular octagon.

Electro-mechanically speaking, the simplest solution is a sliding contact arrangement known as a commutator. Picture a row (let's call it a comb) of seven or eight electrical fingers or brushes labelled abcdefgh, in sliding contact with a copper circuit board. Yes, a comb of brushes, a comb whose every tooth is a brush; stay with me here. We may take these to be the amplifier output signals, and may choose h to be either "no signal", or else the amplifier's central SB output, if such exists. On this board too is a pattern of contact pads, linked in chains to eight output pins, numbered 12345678 along the edge.
As recently as the first moon landing, an engineer in need of such a commutator might take a stroll down Glasgow's Stockwell Street on a Saturday afternoon, to a shop called Radio, Mechanical and Electrical (R.M.E.) Surplus Supplies Limited. Approaching one of the gentlemen serving behind the horseshoe shaped counter, he'd ask for a seven or eight pole switch. Momentarily sinking from view, that gentleman would presently resurface clutching a massive iron, copper and bakelite contraption which, he would explain, had just been ripped from a decommissioned submarine a few days earlier. It would weigh about two kilograms in new money. "One pound and five shillings", he'd venture, causing the hot ash to fall from his cigarette into the surrounding component trays. But it was always possible to make a deal.
This diagram illustrates such a linear switch, with its comb in the starting position (first row).
In the initial configuration shown here, it's easy to trace that input a is connected to output 1, b to 2, c to 3, and so on. Now, operate the switch once. That's to say, comb the inputs, by moving the row of brushes down by one step. Assuming your eyesight is much better than mine, you should see that input a now connects to output 2, b to 3, c to 4, and so on, up to input g, which now connects to (previously disconnected) output 8; and input h, which has now looped around to connect with output 1.

Having satisfied ourselves that the switching arrangement works as advertised, we can easily see how this design might be transformed into a rotary configuration, such as a flat disc or cylinder. Actually, there's one little niggle worth mentioning about that. Did you notice this switch was implemented using just a single layer of substrate? This is intimately connected to the fact that the underlying symmetry we're modelling (rotations of an octagon, remember?) contains no reflections or other crossovers. But if we want a commutator without an end stop, allowing the control dial to spin freely in all directions, then we'll have to introduce another topological layer to provide the multiple helical connections required between channels. Otherwise these will interfere with the comb wiper connection tracks.

The commutator design has one great advantage over most others: there is just a single point of sliding contact per channel, where the brush meets the copper. Most of the alternative designs we'll be examining later have multiple contact points along each constituent circuit, leading to increased contact resistance per channel, concomitant power loss, and more seriously, greatly increased non-linearity.

Nitpicker's Corner

"But isn't a commutator designed to keep electric flow direction constant under rotation?" - Okay, here's what to do. Dig up your house, cement it to the top of the switch rotor, and spin. Now whatever direction your house points, the music will always come from geographic North. You're welcome!

Next time: We travel to beautiful Bulgaria in the hunt for a 7P8T switch.

No comments:

Post a Comment