Just a thought experiment…

[no1herenow]

Just a thought experiment…
I was thinking about this 20 years ago, and never had the time to solve it.

The problem is to design a configuration mechanism, that can control, let’s say 1 to N (N=16) 8-ohm speaker configurations, where, when driven from a single amplifier, keeps the load impedance between 4 and 16 ohms, and any permutation of speakers can be on at one time. No CV systems, no transformers.
Obviously a really simply way to do it is design a static network, switching in resistors to replace all the speakers that are off, but that suffers from terrible efficiency if only a few speakers are on.
So the goal is to use the minimum number of switches (relays), optimize an algorithm, and minimize inefficiency.

Next step will just be to write a program that does an exhaustive search of topologies, and correct interconnects (16 drivers is not exactly a big number).

Any ideas? (I know, just use a whole bunch of amplifiers, which is what I am doing, but the thought experiment is interesting).

[weltersys]

Originally Posted by no1herenow

The problem is to design a configuration mechanism, that can control, let’s say 1 to N (N=16) 8-ohm speaker configurations, where, when driven from a single amplifier, keeps the load impedance between 4 and 16 ohms, and any permutation of speakers can be on at one time. No CV systems, no transformers.

This could probably be accomplished with clever wiring of a pin matrix as used on old school lighting consoles.

Individual speakers would be wired to the pins (which contained a diode for use on the lighting console) and patched in various grid configurations.

Last time I priced a pin matrix, around 1990, it cost more than an entire digital console .

[Pano]

Yikes! Haven't see on of those in years.

[weltersys]

The matrix was simple in concept, but required gold plated contacts to be reliable, and very close tolerances, especially in the smaller size versions that were produced towards the end, like on the Leprecon boards.

With digital technology becoming so much cheaper, and doing so much more, those boards have been relegated to the dumpster.

Digital control in lighting took over much faster in lighting than audio, no question which worked better.

There are still anti- digital holdouts in audio, but the next console I buy will be digital.

[Pano]

Originally Posted by weltersys

Digital control in lighting took over much faster in lighting than audio, no question which worked better.

For sure! But lighting is easier, there's only one signal - 60Hz at 120V (or 50Hz 220). It has to go a lot of places, but it's all the same signal.

No1. You might want to turn that around and make a patch panel with 1/4" connectors, like the old telephone switchboards. Each amp would have its own line that could be plugged into any source on the panel. The amps remain connected to the speakers, full time. With the number of sources and amps you have, it would get expensive. That's a lot of 1/4" jacks and plugs.

[no1herenow]

Originally Posted by weltersys

This could probably be accomplished with clever wiring of a pin matrix as used on old school lighting consoles.

Individual speakers would be wired to the pins (which contained a diode for use on the lighting console) and patched in various grid configurations.

Last time I priced a pin matrix, around 1990, it cost more than an entire digital console .

---------------------------------------------------------------------

Ok.... I want this PC-controlled, so the changes just require a few hits on a keyboard.

(BTW, a long time ago, a guy at Bell Labs was grabbing the pin matrix boards, which had a nice walnut base, and he turned them upside down, to floor his attic). These things cost a bundle back in those days. Obviously he was terminated.

[weltersys]

Originally Posted by no1herenow

Ok.... I want this PC-controlled, so the changes just require a few hits on a keyboard.

(BTW, a long time ago, a guy at Bell Labs was grabbing the pin matrix boards, which had a nice walnut base, and he turned them upside down, to floor his attic). These things cost a bundle back in those days. Obviously he was terminated.

Wow, seems a 4" x 4" pin matrix was around $500 retail in the early 1980's, that was one expensive floor.

1 to 16 speakers, that is 32 connections, 32 x32=1024 solenoids to switch to allow all possible permutations.

I won’t be embarrassed if that number is incorrect, but it will be a lot of solenoids, and a lot of lines of code, then some weird computer to solenoid interface.

The only problem I have on any of my amplifiers are speaker switching solenoids making an intermittent connection.

The thought of implementing your thought experiment makes my head hurt...

[Speedskater]

Do all the connected speaker have to play at the same volume level?

[AllenB]

Does this mean you would like to put all of them in various combinations of series and parallel so that the combination can be driven by one amp?

I am not sure it would work. There are actually three factors involved, the impedance, the frequency response, and the sensitivity. Putting combinations in series, for example will affect all three.

Just to be dramatic, your permutations would go up to 4,096... but not, even, because the factors are continuously variable and there's no guarantee there will be a solution.

You could start by placing complete Zobel correction on each unit but.....

As I was suggesting on your other thread, there are two good reasons to use more than one amp. You could more easily retain the ability to tune the levels to overcome the sensitivity differences, pairing similar ones on a channel if you wanted. Then you could carefully pair them (in parallel, I'd expect) in a way that would give a more balanced looking impedance curve, for what that's worth.

[Pano]

I agree with you Allen. If it were me, I'd just pick an amp and a speaker(s) combo and then make the selections on the inputs. Think of them as powered speakers, if you will.