What happens when a signal is split... and...

[Ethan]

What happens when a line-level balanced signal output is "Y'ed" or split, driving two different inputs at the next stage? For this let's say both loads are the same.

Then, what happens when the 2 load impedances are different? Does the driving source output just see the total parallel impedance of the two different loads?

 

[NewYorkDave]

Ethan, either you've been really slacking off in your studies, or you already know the answer to this. :wink:

Draw an equivalent circuit with a voltage generator and source and load impedances as resistors. Then the answer become obvious.

By the way, we almost never "mult" in broadcast. That's considered a hack job or at best, a quick kludge in an emergency. We use distribution amps.

 

[rodabod]

[quote author="NewYorkDave"] That's considered a hack job or at best, a quick kludge in an emergency. We use distribution amps.[/quote]

For driving two identical loads in parallel, is it really that bad?

Assuming the source is capable driving double the current, should it often prove a problem?

 

[Ethan]

Dave, I have absolutely no confidence when it comes to "applying what I've learned". :wink:

Thinking out load...
If the two parallel input impedances were say 600 ohms, the load it presents to the source output is 300 ohms--the lower load would mean that it would suck more current (in this case double). This could be anywhere from OK to bad depending on the situation. If the output Z of the source is 300, then we'd get the most power and voltage transfer (which if this were 1950 we might care, but now we're more concerned with voltage transfer--uh, right?), if the load goes lower than 300 (still assuming our output Z is 300) then we start to get problems. Most modern equipment line level inputs have an input Z of >1K so one MAY be able to split once and get away with little ill-effects provided the driving output Z is pretty low?

Dave, I'm ready for pay for some humble pie :grin:

 

[CJ]

Making up for a drop in level is no big deal. It's making up for high end loss or phase shift that is the problem. And then there are grounding issues. The easiest way to find out what happens sonically is just to hook two things up to the signal and see what happens.

I think a transformer with two 600 ohm secondaries will do the trick in most cases.

There is a ton of information on splitting on the Jensen site.

 

[PRR]

> For driving two identical loads in parallel, is it really that bad?

Maybe not. But things change. What happens when one of your "identical loads" turns into a SHORT? (Bad wire, pinched cable, mis-patch....) It goes dead, can't be helped if you gonna short things. But also the other, not-shorted, feed goes dead. Use a distribution amp, short one output, that DA output runs a little warm, the other outputs drop in level 0.01dB, the show goes on and nobody notices except the guy on the end of the shorted line.

 

[NewYorkDave]

Exactly!

Imagine a program buss in a television station, and it's simply "multed" to its various destinations. You're on the air, and some bozo messing around in master control makes a wrong patch... Or the clumsy engineer (named Dave, probably) has just finished working behind the racks and he pinches a cable while he's dropping the heavy raised-floor tile back in place...

Interestingly, most broadcast DAs do not use an "amplifier-per-output" approach as you might expect. The usual topology is a miniature push-pull power amp (or a pair of them in balanced systems) driving a network of build-out resistors. A truly worthy DA can work, without deleterious effects, with up to one-third of its outputs running into dead shorts. This is a recommendation first put forth (to the best of my knowledge) in this paper.

Multing CAN work in a small self-contained (single room) environment where there's no "on air" patching going on. But just be aware that there is no truly "bridging" load in the real world; when you mult something, it has an effect (however small) on your signal.

Most modern line-level stuff has an input Z of 10K or greater, and source impedances are usually less than 1K, so there's not much of a theoretical objection to "Y"ing an output in such a circumstance. Just be aware of the potential real-world consequences, that's all.

 

[fum]

I'm presently doing this for my headphone distribution system, but had no idea that this was done in broadcast as well. Cool.

The resistor network for this was pretty darn simple. Kev had one drawn up on his old site at one point, for headphone distribution.

Regards

ju