balanced and unbalanced inputs/outputs

Forum newcomer grok gave me an idea this afternoon. With the announcement of the Model 990, we've had a lot of discussion about some of its numerous features - including the balanced outputs. In fact, the debate evolved to a very technical level, and there is a ton of good info both for and against the concept of balanced connections. I suspect, however, that most potential 990 owners are going to be like me and hook it up to existing amps or buy amps like the 7100 or 770 to go with it, making the whole issue of balanced outputs moot, so it seems that we might try splitting things out. With that in mind, I'm starting this thread with the idea of letting folks compare their first-hand experience with balanced and unbalanced connections. What about it, folks? It's been a regular topic of discussion here, even before the 990 sprang up over the weekend. A central thread on the topic could be a chance to really kick some things around. (Just try to limit the kicking to ideas, not other gunslingers. )

Gonk, thanks for starting a new thread. Here is a good link to get the ball rolling.

http://www.audioholics.com/techtips/audioprinciples/interconnects/Balancedvsunbalanced.php

This is a very interesting topic that generates a wide variety of opinions from audiophiles. My personal experience with balanced signals has generally been favorable, particularly in regard to hum and noise rejection. The sonic differences are often quite subtle, and equipment must be designed appropriately to provide the full benefit. Many units offering "balanced " outputs simply use an additional op amp to create an inverting output, and this can actually degrade the sound. True balanced differential operation for 8 (7.1) discrete channels would require 16 separate paths for amplification, volume control, and (hopefully) D-to-A conversion. Another possibility would be the use of output transformers to galvanically isolate the signal, but that would also be an expensive solution requiring 8 discrete transformers. It would be distinctly unusual to find either of these approaches utilized in an inexpensive preamp/processor, but maybe the Outlaws have pulled it off. The Outlaw M200 amplifier is an example of true balanced differential operation and is among the least expensive quality amplifiers on the market. It is a shame that an XLR input wasn't provided to take full advantage of the design. As for the 990, perhaps someone in the know can provide the answer?

I’ve worked in many professional and enthusiast situations with both audio and video signals. As to transferring a signal from point A to point B:

The properties of the cables are more important than the cost and much more important than hype. I think it sad when people spend $25 or more per foot to move a signal a few feet in their house when the professionals that recorded the signals in the first place used great-quality cables at $2.50 per foot or less to move a signal over even larger distances.

In a system where various types of signals pass among several pieces of equipment, there are various potential sources of signal interference. Changing from unbalanced short runs to balanced short runs in just one portion of the signal path will not eliminate problems of a more fundamental nature. If you have only one problem area and that problem is due to long unbalanced interconnect runs, then balanced runs will help. Otherwise, the benefits of balanced lines are offset by the inclusion of additional electronics in the signal path to deal with a balanced I/O situation, and you’ll still have to deal with the other problems, such a ground loops, if you have them.

If my power amp(s) were on the order of 10 feet from my pre/pro, I’d use good unbalanced cables. If that distance were on the order of 100 feet, I’d use balanced runs.

No matter which option you choose, use quality cables but don’t pay for hype.

IMHO, Outlaw included balanced connections on the 990 and 790 more because of what some consumers have been led to believe and are likely to buy at that price point rather than the actual audio engineering needs of the majority of users. If you are one of the rare users that would benefit from balanced lines, then by all means run balanced lines.

For me, the balanced out will be an advantage. I use a tri-amped system and the electronic xover (which would is driven by the pre/pro) has balanced inputs. At the moment, I use a cable and short the - in of the channel. I would have preferred an unbalanced input for the xover, but I liked the unit otherwise, so I built the cables. I'll probably get a 990 and use the balanced outputs. My 2 cents...

So, for most of us, would using a balanced connection (if it were possible) make sense between the pre/pro and a sub woofer, given its typically a longer run than between pre/pro and amp? Is it possible and advisable to mix balanced and unbalance connections in a system this way?

When using a balanced connection, are the lost beneficial distortion aspects of any consequence given the restricted material found in the "dot one" channel?

I'm with Stabie. I also use pro active xovers and the balanced to unbalanced thing is a pain. It would be nice to have both to make it as flexible as possible.

Balanced is also nice since my gear is in another room and operated by RF remote, I can still place my amps very close to the speakers without worrying about spurious noise picked up from long interconnect runs.

Originally posted by Kevin C. Brown in the 990 forum:

SH: I thought balanced outputs (i.e., balanced interconnects) remove *all* noise on both sides of the circuit? Why would odd order be any different from even order? The "order" is just the relationship of the added components to the original signal. But out of phase is out of phase, and they should be removed just as well. ??

Just reposting it here for him.

Owl's- Thanks. I'll have to keep an eye on this thread too. I'm still making my way through that Audioholics post...

No problem. Frankly, I'm curious to the answer myself.

I asked this question, balanced OR unbalanced, of a well known equipment evaluator and review writer for Widescreen Review about this and here are his responses, paraphrased into one quote.

Is it ready for shipping?

LQQK

The Model 990? The first batch of production units is apparently crossing the Pacific on a freighter at this very moment.

it would be nice to be able to run a balanced line for those of us who have balanced inputs on our subs... i am assuming the 990 has a balanced sub pre-out...

there are no ill effects from having a balanced cable carrying your lfe signal.

Well, I count eight balanced connections on the rear panel . That leads me to assume that one of those is a subwoofer connection.

The even order distortion cancellation that Soundhound mentions is a characteristic of the balanced stage itself. In other words, it cancels the even order distortion products that the stage would otherwise produce itself were it not a balanced design. All circuits produce some distortion products and it is the ratio of even to odd harmonics that he is concerned with. With many off the even order (euphonic) harmonics cancelled, the more harsh and offensive odd order harmonics are unmasked and become the dominant sonic characteristic of the amplifier. This is a separate issue from the noise rejection abilities of the balanced input circuit.

Apologies to Soundhound if I have mischaracterized his opinion, which he is more than capable of defending himself.

Balanced does not remove noise. Noise is random and in fact as soundhound mentioned, noise will actually go up (3 db I think) in a balanced configuration. This is because you have 2 input stages producing uncorrelated noise from the stage. What balanced does do is remove common mode signals from the signal pair. A differential input has a figure of merit referred to as CMRR, or commom mode rejection ratio. If the differential signal coming in had a 60Hz signal injected onto both +/- inputs from wire coupling, then the signal would be reduced by the CMRR. CMRR's are usually frequency dependant with low frequencies getting rejected better. This is good for audio as mostly your trying to get rid of hum. Since hum is often injected by power cables running near audio cables, the balanced can save some cable placement issues. I suspect there are cases where ground loop problems will be minimized by balanced too.

Hope this helps
Mike

This still makes no sense to me. A balanced circuit cancels out *all* of the *common* components that are out of phase with respect to the other line. There is no difference between even or odd order.

This might not be 100% correct, but here's how I think about it: If I start with a simple 20 Hz tone, 2nd order harmonic is 40 Hz, 3rd order is 60 Hz, 4th order is 80 Hz, n * 20 Hz, etc. I don't think it matters what the order is.

I have read *a lot* about balanced circuits in my lifetime (and forgotten a fair bit too ), and I've never seen any discrimination mentioned before between even and odd order harmonics.

??

You are completely confusing the action of a balanced connection with a balanced design. As has been stated in this thread by Tekdredger and Stabie, a balanced connection only cancels what noise is picked up by the cable connecting two balanced components (these can be unbalanced designs but have a balanced receiver and transmitter on each end). The distortion of the other component(s) is unaffected except for the distortion actually generated by the transmitting and receiving electronics.

A design that is completely balanced internally will cancel all even order harmonic distoriton that is generated within that component only, and pass through the odd order ones. If you have a power amplifier that is balanced all the way through, the only distortion components that that amplifier produces that will be present on it's output will be the odd order ones.

Any distortion that is generated by other components upstream will pass through that power amplifier unaffected because obviously the balanced amplifier doesn't know what is music and what is distortion in material that is presented to it's input.

The balanced amplifier however will add only odd order harmonic distortion to the signal.

One of the major reasons that vacuum tube amplifiers sound so musical is the fact that the majority of what distortion they produce is even order. What odd order distortion that they produce is essentially non-existant beyond the 3rd order.

Designing a component that is balanced all the way through from input to output is a bad design decision. There is absolutely no advantage sonically or theoritically to this type of design. There is however the very real and very audible downside of this type of design which is a result of the cancellation of all of the even order harmonic distortion components that that amplifier produces.

If an amplifier must have balanced inputs, the least soncally degrading way to accomplish this is to design a standard unbalanced circuit and add balanced-to-unbalanced circuitry at the amplifiers input which can be bypassed if the user is only using an unbalanced connection.

Do you have any references for this? A lot of higher priced amplifiers *do* have fully balanced designs (one of the more cost effective ones being ATI, actually), and I have never heard this mentioned before.


Plus, the 990 will not be a fully balanced design anyway. It will have balanced outputs, but I doubt it will even have dual differential DACs.

this topic has delved into the pointless domain. people are gonna use balanced if they want to and unbalanced if they dont want to use balanced... when you upgrade to balanced, ill buy all your other cables for $1 (total).

Here are some good refs I've found (I think the audioholics link, while talking about balanced connections, still does a nice job in explaining what's going on; ref'ed in the 990 thread I think):

http://www.audioholics.com/techtips/audioprinciples/interconnects/Balancedvsunbalanced.php
http://www.tubecad.com/march99/page4.html
http://www.hometoys.com/htinews/feb05/articles/ati/purebalance.htm
http://ftp.agilent.com/pub/semiconductor/morpheus/docs/diff_design.pdf pg 4 and 5
http://www.mcintoshlabs.com/mcprod/..%5Cdata%5Cbrochures%5CMC402br.pdf

There used to be something on Bryston's site too, but I can't find it at the moment.

And, if a balanced connection (interconnects) doesn't discriminate between even order and odd order distortion, why would a balanced design?

A balanced connection does not discriminate against even and odd order distortion presented to it's input, but the balanced driver on the sending end and the differential ampifier at the receiving end as a loop will themselves only generate odd order distortion. This has nothing to do with anything outside of the sending and receiving loop. Any distortion of any kind that is present at the input of a balanced driver or a fully balanced amplifier will be passed through unaltered since it doesn't know what is distoriton in the signal and what is the music.

Fully balanced designs (as opposed to balanced connections) were originally developed as a way to reduce distortion. This they do! However they do it by cancelling the amplifier's own even order distortion components and leaving the odd ones intact.

This is exactly like somebody saying that they have "lowered their cholesterol" by eliminating the "good" cholestrol and leaving the "bad" cholesterol!

If you do a Google search under wording like "even order distortion in balanced amplifiers" you will get several references. Many that I got were concerned with push pull stages in tube amplifiers, but you should be able to piece together a picture of when and how this process takes place.

There are designers of course who have differing ideas about the best way to design an amplifier. What I am presenting are some objective facts about the behavior of these types of designs, and their sound quality which has been confirmed in listening tests by myself and those who I've known over the years.

The readers here can do the research if they wish, and draw their own conclusions. I've said about as much as I can about the subject without endlessly repeating myself.

Allow a naive question to enter the discussion please:

If the effect of fully balanced designs is so sonically poor, then why would ANY amplifier designers ever do it at all?

Again, I know this is a low-level question, but I am getting really curious about this...

It's almost entirely marketing. One manufacturer does it and others feel obligated to follow suit because they obviously want to sell product. They don't want to get caught without a feature that is perceived as desirable by the consumer, even if it is ultimately detremental to the best sound quality.

A good example is the infamous "spec wars" of the 1970s where Japanese vied with each other to have the lowest distortion and the highest damping factor. The problem is that this was done with massive amounts of negative feedback which caused these amplifiers to sound hard and sterile, giving Japanese amplifiers a bad reputation that has not entirely been restored. But they just had to have the best specifications on paper, sound quality be damned.

The consumer audio field is rife with fads that come and go, along with massive amounts of mis-information and downright snake oil.

Outlaw is commendable for it's lack of hype and attempts to misguide consumers, but the fact is that even they have to bow to market trends in order to sell product.

There are several reasons that could be given, but one not to be overlooked is marketing.

Marketing, as in, “Wow, this has balanced outputs and the matching amplifier has balanced inputs, therefore this is ‘serious’ equipment and must be better.” Also, conversely, should the Outlaw new flagship pre/pro not have balanced outputs, and balanced outputs become more prevalent in the near high-end market, consumers will think the units without balanced outputs are lacking something that’s needed, whether it’s really needed or not. Therefore it’s not necessarily what’s best that drives the market and influences design, but what is perceived to be “best” by the customer and what is perceived to be “needed” by the customer that becomes helpful in the competition of the marketplace.

Tell me you can’t think of features that started with a couple manufacturers and within a year just had to be a part of every mass market receiver/processor and are now considered near useless by most people once the hype of the fad has moved on to a different feature set?

Now, there are genuinely good reasons to use balanced connections, and balanced connections are not just a passing fad used to help sales, but don’t discount a manufacturers’ need to both lead consumer perception and then meet that perception with products.

I’d buy a pre/pro with balanced outputs if there were a dozen other good reasons to buy that pre/pro, but I wouldn’t turn my world upside-down just to have balanced connections. That being said, for the success of Outlaw and for those that will find balanced connections a definite benefit, I’m glad the 990 has balanced outputs.

(I must have been writing this while Soundhound was posting!)

Don't get me wrong, I'm not against having balanced inputs/outputs as an option, but rather the way that they are implemented in some cases. If the balancing circuitry can be easily bypassd, leaving a simpler and purer signal path, that is great and a win-win for everybody. If however a design is completely balanced from input to output, those people who are not interested in this feature are locked into the sonic penalty that this type of design can entail, which includes more noise from the added active electronics and a harsher distortion spectra.

Of course, if you read everything available on this harshly debated topic, you might also think that this statement is pure and simple marketing BS too... depending on what you want to sell. You can't hear even, or odd, distortion noise if it is at a level too low to hear...

As with most things regarding audio playback... each person needs to try it out for themselves in their own system and see what sounds best for them.

Part 1:

I asked ATI.

Their gist is this: yes, even order is reduced but odd order isn't. But a balanced amplifier design still also reduces any common mode noise introduced into the signal external to the amp itself that ends up on both legs inside the amp. So they say, 2 out of 3 sources of noise are reduced. They also added, which kind of makes the whole argument a little moot, that we are talking about a level of 0.003% distortion here and listening tests have shown that it can take upwards of 1% to be audible, so it certainly isn't true that leaving behind odd order vs even order at this level will make a component sound any worse. They do say that one of the reasons why they offer balanced amps is because other amp-makers do too, and that balanaced designs are commonly perceived as superior to single ended designs. So there is some marketing in there.

So I do have to admit, that hey, I learned something. I didn't know about the even vs odd order thing. One question though: my impression is then that you still have a choice. In most (if not all) fully balanced designs, you can still just use RCA connections to use half of each circuit, right? And even if you want to use balanced connections but not the balanced part of the amp (why not just buy a single ended design then...), I have seen those converter jack type things in pro catalogs. ??

Part 2 will be McIntosh.

I pointed out earlier in this thread that with a completely balanced design you are locked into this circuit and it's disadvantages, no matter what. I also mentioned that the best way to get around this state of affairs is to have an unbalanced amplifier, but offer a balanced input only which can be bypassed if desired.

This gives all the advantages of noise cancellation offered by a balanced interconnection, but provides a more pure signal path at the same time. If somebody does not want to use the balanced feature, they can simply bypass it completely. You can't do this if the entire circuit is balanced.

The fact is, once the cable is terminated into a summing amplifier as in the input to a balanced amplifier, the noise picked up by that cable is nulled out - having the rest of the amplifier's circuitry balanced will not cancel noise any more than when the input only is balanced.

It is the function of the amplifier's chassis and it's shielding to reject noise which is trying to make it's way into the inside the amplifier. In any event, the signal levels within a power amplifier are far, far higher than as is the case in interconnects so that the signal will be that much higher above any potential interferrence.

I am suprised by ATI's candidness on this issue, but what they say about the audibility of distortion is off base. Yes it is true that distortion is not generally audible until it reaches 1%, but this is only true of the even order ones!

Odd order distortion is audible at far lower levels, especially ones above the 5th order as a hardening, sterility and harshness of the sound.

I am talking about levels of odd order harmonic distoriton in the .00X% range here - this is not a moot point!

Additionally, the .003% distoriton rating that ATI cites is taken at the point in the amplifier's power curve where distoriton is lowest - in the case of a 100 watt amplifier, this would be roughly 60 watts. If you look at any solid state power amplifier's power verses distortion curve, you will see that distortion rises rapidly at low and high power levels.

The distortion at the power levels you typically listen at, and the levels where the amplifier operates 90% of the time is around 1 watt.

Any distortion present at this level is critical, as the ratio of distortion to signal (music, film soundtrack) is much less. This is precisely where you want any distortion present to be benign. In other words you want any distortion at normal operating levels to be predominately even order, not harsher high order/odd order distortion.

It is also fact that the presence of the even order harmonic distortion elements, especially the 2nd, will mask the higher order ones and render them more benign. Of all the distortion elements (and you must have distortion in all electronics), you certainly do not want to get rid of the ones that are helping to reduce the audibility of the harsher distoriton elements!

I've brought up the example of tube amplifiers before. They typically produce almost exclusively even order distortion, most of it of the 2nd order - the most benign of all. What presence of the 3rd harmonic there is, it's audibility is completely swamped by the presence of the strong 2nd order distortion component.

This is a major reason why tube amps sound so natural.

I disagree with this. If it takes upwards of 1% distortion of both even and odd order components to be audible, ain't no way I believe that odd order alone hundreds of times lower is going to be audible. I do believe that odd order is "worse" than even. But over a hundred times lower is still over a hundred times lower.

Fully balanced components have been around a while. They have been reviewed by many publications and many reviewers. If a fully balanced design is so detrimental to good sound quality, why haven't more reviewers commented on this with measurements to support it? I simply have not seen that. What I have seen, is an awful lot of people who do tend to prefer the sound quality of fully balanced components.

You seem pretty sure of this. Would you care to tell us about the double blind listening tests you've performed between balanced and unbalanced amplifiers to back up your position? (I'm not interested in what manufactruers or reviewers say - just the results of your own tests.)

After all, you didn't believe me about the fact that balanced designs cancel even order distortion, did you?

Actually, I have done 2 sighted tests in the last 12 months or so:

1)


At one time, I had Proceed (fully balanced with balanced connectors), Bryston (not balanced with balanced connectors), and Acurus (not balanced with just RCA jacks) amps all on hand at one time. I did extensinve listening tests over about 6 months. 2 channel music and movies. I ended up keeping the Proceeds. They just sounded smoother on top with more mid range detail and clarity, to me.

2) Drove 160 miles last summer to go listen to a Pass Labs X150.5. At the dealer, I also compared that amp to an Aragon 8008 Mk II. (The Pass Labs is fully balanced, the Aragon isn't.) I preferred the Pass Labs amp. This is the most musical sounding amplifier that I've ever heard.

Now, I personally don't think there are large differences between amps. I agonized over the decision to replace the Acurus amps, that in one form or another, I had had in my system for about 8 years with the Proceeds. The Proceeds cost a fair amount more, and even though I did prefer them, the differences were subtle at best. But to me, they were there. The point is, is that I had no ulterior motive to preferring the fully balanced design in this case, and in both comparisons, the fully balanced amp certainly did not sound *worse* than the single ended designs.

So there.

I said blind testing, not sighted testing. It is well documented that sighted testing is essentially useless in determining subtle differences.

When I was a test engineer at Altec Lansing, we performed literally hundereds of blind tests. During the design phase of one line of amplifiers, two amps were designed to be as identical as possible except that one was fully complementary (balanced) and one was conventional. The conventional amplifier sounded much better, and more like a tube reference amp than the balanced design. The balanced design was abandoned in favor of the conventional one.

Useless to you maybe.

From Jim Austin in the latest Stereophile.

Tell you what- you keep believing what you want to believe, and I'll go enjoy my system some more.

I hope you don't believe everything you read in Stereophile - the "audiophile community" is hardly the ultimate technical resource!

For those of us that don't quite understand the difference between odd and even order distortion and the difference it makes I have included a piece of an article that will help explain some of it.

As you probably know, harmonics (also called overtones) are what make up the timbre of
each instrument so that the A440 of a clarinet sounds different from the A440 of a
saxophone. The sax tone is rich in harmonics while the clarinet sound is almost pure
fundamental without overtones. Certain harmonics occur naturally in music -- both in the
overtones of timbre and as component notes in the harmony of chords. These overtones are
often called the "low order" or "even order" harmonics such as the 2nd, 3rd and 4th. These
overtones relate to the fundamental note by being one octave above (double the frequency);
an octave and a fifth above (called the "quint" in pipe organs, an octave and a half above)
and two octaves above for the 4th.

The ear has a great tolerance for these overtones - because they relate in a musically
consonant way and are are already present in the original signal, as either timbre overtones,
upper harmony or both. For example, it is often possible to add up to 20% second harmonic
to a single note before any change is audible. Then all that happens is that the ear hears the
timbre change.. it does not hear "distortion trash." If you're a guitar player or keyboardist try this simple experiment. Play two notes, play the low one loudly and the high one very soft at
first, then increasingly loud, an octave apart at the same time. Notice how loudly you can
play the upper octave note before you can even hear it... and that what you do eventually
hear sounds mainly like a timbre change in the lower note -- not two separate notes. Now try
playing the same original note but this time also play a note two octaves plus one step
above. No matter how softly you play this note, it is distinctly audible and separate from the
original.

What we've done with this experiment is demonstrate the effects of different harmonics in
contrasting distortion types. You've heard how inoffensive the low order (sometimes called
"even order") harmonics can be... especially compared to the prominence of the upper order
(often called "odd order"). And as you might guess by now, tube amp distortion is primarily
low order while that of most transistor units is odd-higher order... and hence very
objectionable. Clearly then, a tube power amp with a higher percentage of Total Harmonic
Distortion actually can sound cleaner and clearer and much less harsh than a solid state
amp which carries a lower total distortion rating. It's not the total distortion that counts as much as the distribution of distortion components: It's all but impossible to hear even a few percent of 2nd harmonic distortion whereas a fraction of "high order" type becomes audible
as annoying hash.