line level signal splitter?

[eleanor296]

hey guys!
Haven't been here in a while, busy with school... about to start building again though 
Here's a question:
How could I split a line level signal into, let's say, 4 parts?
I'm aware of the several splitter designs out there, but I doubt that they would work well at line level signals?
I basically want to use my amp's serial loop, split the signal into 4 parts, route them through 4 rack units, then mix the signals back together.
That way I can have several parallel options, and if I don't want any fx at all, I can put the volume controls on 3 units to 0, and have the fourth unit (which has a hardwired bypass, no converters in the signal path) set to bypass...
I'd be one happy camper.

If somebody could point me to an article or two, that'd be GREAT!  Or even a project somebody else already finished.

Thanks
Andy

[flo]

The Send output of your amp is probably a low-impedance output while the 4 inputs you want to route to are high-impedance inputs.
If so, you can just passively split the signal by connecting 4 jacks in parallel. Perhaps you can make a simple box with 1 jack IN and 4 jacks OUT all wired up with each other.

[Morocotopo]

http://www.muzique.com/lab/superbuff.htm

Should work with line level if you feed it with 18V, two batteries or regulated power supply, might run out of headroom if fed with 9V, duplicate R7, C3 and C2 for each opamp output and you have a 4 way splitter. Somebody correct me if I´m wrong...

http://www.muzique.com/schem/mixer.gif

Idem for the power.

Combine both in one box and you have a 4 bus FX loop to add to your amp.

Hope it helps

[eleanor296]

Morocotopo, you are the MAN!
I'll have to try that out.
Anybody can verify this?
If not, I'll probably be able to breadboard it within the next two weeks... once I get done with that stupid psychology project... 

http://www.muzique.com/lab/superbuff.htm

Should work with line level if you feed it with 18V, two batteries or regulated power supply, might run out of headroom if fed with 9V, duplicate R7, C3 and C2 for each opamp output and you have a 4 way splitter. Somebody correct me if I´m wrong...

http://www.muzique.com/schem/mixer.gif

Idem for the power.

Combine both in one box and you have a 4 bus FX loop to add to your amp.

Hope it helps

[eleanor296]

sorry, forgot something....
Do you think that I'll have to worry about isolating the different paths?
It'll all be running through rack fx, so I guess there's at least a "potential" for ground loops?
I'm still such a newbie when it comes to signal grounds....

Andy

[eleanor296]

I'm ordering the parts tonight to see if it works...

More when the parts arrive!

Andy

[trixdropd]

I wouldn't think you'd need to isolate them if they are not amps. Just make sure they are terminated from one source. Try it, if not then add them.

[R.G.]

Let's talk about splitters.

The name is deceiving. It implies that incoming signals need to be literally split into several pieces. That may or may not be true.
Every signal source has a signal impedance. This is an internal resistance that's part of the signal source itself and cannot be removed. In passing audio signal round, we usually want to read a voltage, not a current. So to drive several loads, we want the loads not to diminish the signal's voltage appreciably, and we want them not to interact much, if at all. Not interacting means that one or the other does not hog the available signal current and drag down the signal level for all of the loads.

A guitar pickup has an impedance that's composed of the 4K to 18K of simple resistance of the wire wound in the pickups, the inductance of the pickup itself, and the resistance/capacitance of the guitar controls. Of these, the inductance is by far the worst, because the 2 to 5 henries of inductance easily reach a value of 100K to 200K ohms at treble frequencies. This is bad because (a) it's high and (b) it's higher at some frequencies than others. To prevent treble from being really noticeably lost, the total amount of loading on an unamplified, unbuffered guitar pickup needs to be over ten times the pickup impedance at its worst (highest) point. Tube guitar amps settled on 1M as the right impedance.

Now we get to splitters. Hook one guitar to two amps and you get (a) some treble loss because now it's only half a megohm with both inputs in parallel, and probably (b) some hum because the AC power line sides of both amps are not identical. A simple splitter may simply be a buffer, an amplifier with a large input impedance and a small output impedance. A buffer for guitar might have a 1M input impedance but a 10 ohm output impedance. So it can drive -gulp- ten thousand guitar amps in parallel with no detectable signal loss. If we didn't have that ugly hum problem in (b) we could just use any opamp buffer and be done with it. The input impedance of guitar amps is so high compared to the buffer that they are practically non-interfering.

But back at line level signals. First you have to know your enemy. What's a line level signal? It turns out that there are at least two definitions. One is the pro audio/studio one of 1milliwatt into 600 ohms, differential and balanced about ground. Doing a quick stretch through Ohm's law give the voltage level: 0.775Vrms into 600 ohms is 1mW. The other is the consumer audio definition, where the equipment is not bothered with niceties like keeping signals differential and balanced, or keeping the impedance down to 600 ohms to keep hum and noise pickup down, so the consumer standard is best thought of as 775mVrms, unbalanced and into an unspecified impedance of generally around 10K.

If you have to go into a mixing board which needs balanced 600 ohm sources, it's a different job than getting 775mV unbalanced into 10K. In the first place, you need almost twenty times as much current and you have to make it balanced. But it does get rid of that pesky hum thing. In the case of the consumer "standard", all you have to do is provide enough headroom so that you can deliver 775mV plus some headroom over that to a 10K resistor. There are long arguments in hifi land about how much headroom is needed. Probably you need something like 6 to 12db. +12db gets you to a bit over 3V rms and about 4.38V peak, and that's about all you can even theoretically get out of a 9V source.

If you're going to provide pro audio line level, you probably need something like +/-9V to +/-15V power supplies, with lots of current available for each output, but aren't as worried about hum. A splitter for pro audio needs a line driver for each output to supply the current, but needs to make it balanced and this can be coupled into the loads directly. The old standard is the NE5532 opamp for 600 ohm loads, but National also makes the LM833 specifically for situations like this. So the splitter becomes an input buffer plus perhaps one driver per load to provide the current.

If you're splitting to consumer line loads, you may be able to get by with one buffer and a resistor in series to each load. The resistor keeps the loads from loading down the signal source if one load goes mad. But you have that hum problem. Sometimes you can simply connect them up and get away with it. The "splitter" is then not so much a real splitter as a parallel driver. If the hum is too bad, you have to provide isolation of the loads' AC hum by either transformer isolating all but one, or by using some kind of differential driver to make the loads each think they are getting an hum-unaffected signal. This is usually done with one opamp per load driving either a transformer isolator or diffamp setup. Again, one driver per load, but primarily to get around hum.

Splitters are more "custom drivers" than splitters. They don't really split anything, just provide safe and non-interactive driving.

[punkin]

Nice wright up there R.G.

[eleanor296]

Yeah!  Thanks for the info.

I have an effects processor that I'd like to use, but it colors the tone too much (a/d d/a converters)
So what I wanted to do is "split" the signal before the processor, only have part of the signal go into it,
then mix both signals back together at the end.
All this in the FX loop of my amp.
So yes, I only need the "consumer" line level.
Later, I would like to integrate a relay into the mixer, to be able to defeat the signal that passes by the processor, if that makes sense?
So that I can run 100% for tremolo, synth, etc.
Or I can just do it the way I described it in my first post, using the processor that has a hardwire bypass.

I'm not confined to 9V, since it'll be rack-mounted, I can also go 18V without problems.  Oh, and current shouldn't be a problem either.

Your write-ups are great, RG!  The difference between you and other "experts" is that I can actually understand what you're saying. 

Thanks!
Andy

[R.G.]

So what I wanted to do is "split" the signal before the processor, only have part of the signal go into it,
then mix both signals back together at the end.
All this in the FX loop of my amp.
So yes, I only need the "consumer" line level.
Later, I would like to integrate a relay into the mixer, to be able to defeat the signal that passes by the processor, if that makes sense?

Have a look at the "Multi-Vol" at GEO. It does most of what you want. You may have to pervert the return buffer into a mixer, but that's easy.

Good luck with it!

[eleanor296]

So what I wanted to do is "split" the signal before the processor, only have part of the signal go into it,
then mix both signals back together at the end.
All this in the FX loop of my amp.
So yes, I only need the "consumer" line level.
Later, I would like to integrate a relay into the mixer, to be able to defeat the signal that passes by the processor, if that makes sense?

Have a look at the "Multi-Vol" at GEO. It does most of what you want. You may have to pervert the return buffer into a mixer, but that's easy.

Good luck with it!

Thanks!
The buffer I'm using is the one Morocotopo pointed out,
http://www.muzique.com/schem/mixer.gif

Would his suggestion for using the Super Buffer not work as well/better?
Maybe I just think it would because it involves more opamps?  Wouldn't that "drive" the outputs better if I, let's say, had an output for each single buffer, than 4 outputs for 1 buffer like in your suggestion?

Thanks
Andy

[R.G.]

The buffer I'm using is the one Morocotopo pointed out,
http://www.muzique.com/schem/mixer.gif

Would his suggestion for using the Super Buffer not work as well/better?
Maybe I just think it would because it involves more opamps?  Wouldn't that "drive" the outputs better if I, let's say, had an output for each single buffer, than 4 outputs for 1 buffer like in your suggestion?

The devil is always in the details in questions like this. First, there's nothing particularly super about that buffer. A "building maintenance engineer" still carries a mop.

Humans with no other way to judge the merits of some situation often guess that something that weighs more must be better - no fooling, there are scholarly experiments in this - as well as thinking that something with more parts, costing more, etc. etc. must be better. It's been proven that you can cause sales of some widgie to drop by lowering the price as long as people don't really understand what's inside - they think it's cheaper, so it must not be as good. Similarly, Boutique Pedal Madness seems to make people think that someone who just learned to solder last week and who buys one part at a time, uses hissy resistors and doesn't yet know where to get circuit boards made is justified in his claim that he should be paid several hundred bucks for his version of whatever. If the hypothetical hand-wiring guru asked $49.95 for it, he'd sell maybe one. Asking $500 each gets him a hundred person waiting list.

Back at your question. It depends on what you're driving. If you're driving 600 ohm balanced lines to a 600 ohm balanced load at +20dbm, you'd better have a pretty solid driver per line. An LM833 or NE5532 ...might... barely... work. The pros would use a discrete line driver like the Jensen 990 discrete opamp or some other exotica. If you're driving a 10K resistor from an output that is not inductive in nature, and has an impedance of ten to a hundred ohms, you might lose a fraction of a db in volume, but you won't see the huge losses in treble that are the hallmark of tone sucking on a guitar signal.

So - would one buffer work better than another? Absolutely, definitely, positively yes. But for what purpose? The winner changes when you change the rules. Does a hand made Ferrari beat a flat-bed two ton Chevy truck? Maybe. Depends on whether you're trying to pick up girls, convince an IRS agent you're honest, talk your way out of a traffic ticket, make it to the next town in an hour, or get a ton of hay out to the farm before dark. The details matter. If they didn't, manufacturers would simply spray a smooth, shiny layer of "goodness" on each product at the end of the assembly line. Two coats for the premium customers. 

[eleanor296]

hmmmm... thanks.
I get your analogy.  But.. which buffer's the Ferrari and which one's the Truck? 
I guess I'll try them both and see what I like better.
Of course I'd like for it to be as transparent as possible.
Losing a little bit of overall volume is not a problem there, my rack processors are VERY hot, and with remarkable s/n ratio, so that really won't be an issue at all.
I'll put the parts on my Mouser order tonight.
I've got a pretty sizable order by now... I like it 
I just wish Mouser had nice enclosures as well.

Andy

[R.G.]

Of course I'd like for it to be as transparent as possible.

Let me put it another way. If you simply want 600 ohm balanced line driving, put in an LM833 driving each transformer, or two LM833s in a balanced-output setup and put a +/-15V supply on them. Cheap, effective, and designed by the manufacturer to do the job. If you have to have +24dbm levels at 600ohms balanced, go make some Jensen 990 discrete opamps.

Opamps are ...designed... to use excess gain to hide any internal nonlinearities. In general, after the first couple of generations of opamps which had obvious flaws for audio, an opamp driver is pretty much as transparent as you get, leaving aside the hifi tweako mumbo jumbo for a moment. "More transparent" differences between one opamp and another are very, very hard to prove **as long as you never let the opamps clip**.

I just wish Mouser had nice enclosures as well.

MIddle Atlantic makes some nice 1U and 2U blank rack enclosures for about $50. It's about as good as I've found for bare rack stuff. You could check out SESCOM, too, if they're still around.