How close to the real thing?

Maybe y'all will be interested in this- it seems relevant to the thread...

Recently I was in the recording studio, we were trying to make a recording of my band. Our bass player is leaving town for a while, and we didn't want to forget some new material, so we were making a simple recording to help us with that.

I set up a pair of Neumann U-67s in a figure-8 pattern. They went into a cheap Mackey mixer, from there into an Alesis Masterlink, recording at 88.2KHz with 24 bits.

On playback, I used a set of Grado SR325 headphones. This is an open style of headphone. All the time, I had to look at the bass player, or our drummer, to see what they were doing. That is to say, the sound was so real I could not tell that they were not doing something while I was trying to listen to the recording. So I had to look at them to see that they were actually *not* playing! I could not turn my back on them, because if I did so, my brain was telling me right away that they were playing their instruments, rather than me listening to a recording.

In previous recording experiences (again with headphones) I have experienced the same thing.

So how far are we? The microphones and headphones have been there for quite some time. Microphone preamps and a lot of the other intervening electronics have been too.

So where is the technology weak? That seems to be a better question. Just from playing in the studio, that question is easy to answer. Speakers, power amplifiers and the actual media itself are the problems. Anyone who has released a CD knows that the biggest degradation in the sound from the master to playback occurs in the process of making the CD itself. Also I have grave doubts about 44.1 KHz 16 bits and always have.

Some amps and some speakers are so realistic that if you give them a direct microphone feed, they can easily fool you into thinking that the music is really happening. Others are not so nimble. If we had recordings in which the media was not damaged by mass reproduction (and in specific not damaged by a CD player, which if it is using Redbook standards has no hope of extracting all the data off the disk), we would also be closer. But to really do it right, you need a higher scan frequency and more bits. Or analog.

So IMO we are not far away at all. 90% of what I am talking about here are the issues with media, the remaining 10% is the difference between actual state of the art amps and speakers and those that purport to be.

We do have another issue- one which might explain why the headphones work so well:

The original performance is always in a space with its own acoustic signature. The mics are always in that space too.

Your room is not.

So the idea that the musicians are going to be 'in the room' is tricky. IMO/IME, the model to use is that your stereo and the room its in is a sort of 'space/time machine' that has the ability to graft itself *onto* the original space of the music- but with the acoustic signature of your room included.

With headphones you have no acoustic signature of your room- only that of the mics. I don't like headphones, as I feel like the sounds I hear are coming from behind me sort of. So I prefer the presentation of speakers, even though they often take things down a notch. Of course, headphone can't make you viscerally feel the music the way live and speakers can...

Everyone have a good holiday, y'hear? :)

Shadorne has hit the nail on the head- commercial recordings are the primary barrier to making it sound real.

If you were to take a set of good studio microphones and put them in a different room, away from your system, you might be quite surprised to find out how lifelike the system can be if the mics are set up right. Nothing (so far) compares to a direct mic feed...

Onhwy61, that may be so. When I got the U67s nearly 30 years ago, I knew they were great mics, but had no idea how really great they actually are.

As our equipment for recording and playback has improved, I've really come to realize that microphone technology may well be one of the areas that was well advanced beyond the rest of audio world by several decades. That simple fact is, if you have mics of this quality, you don't need 'signature' or 'reference' as descriptors :) They just work.

One time I did an on-location recording with a pair of RCA ribbon mics. I had a couple of audiophile friends with me that wanted to see it being done. At one point, I had to move the mics a couple of feet. One of my friends was wearing headphones, listening to the live mic feed. Since ribbons can be fairly sensitive, when I got to the mic stand, I said 'I'm going to move the mics now' so he would be prepared for some noise.

When I was done and got backstage again, he was in a state of shock. He had seen me go through the stage door, and then a few seconds later, he heard me *behind* him (at the point where I was ready to move the mic stand). On wheeling around, wondering how I got back behind him without him knowing it, he saw I was not there!!

Now this was a jaundiced audiophile, and training to be a conductor (currently conducting in Moscow) and *knew* that audio equipment could never sound that real. I'm telling you, he nearly had a heart attack.

That's true but they are listening for the notes, not the sound. Its an entirely different portion of the brain that is used. Try sitting in front of a keyboard and see if you can work out the notes of a favorite melody and you will see what I mean.

If you want a recording of a piano in the home to sound as real as an actual piano in the home, you have a real challenge. The problem is that the sound of the piano exists in the room- if you try to record and play back in the same room, you will have double the room signature. It won't sound right. That's why I stress that understanding the model of stereo is the first step to appreciating how real it can really sound, because if it really is sounding real, it will sound like the musical event, spliced on to the end of your room.

I can hit 110 db at my listening chair with ease, and the system does not strain to do it (me speakers are 98 db and the amplifiers have 140 watts in class A). In fact at that volume it sounds quite relaxed- you have no idea its making that sound pressure until you try to talk to someone sitting beside you.

So its not 'ear-splitting' bit OTOH the equipment was built specifically to not distort the odd harmonics. That type of distortion (and IM) leads to 'ear-splitting' behaviors.

FWIW, low ordered distortion is pleasant, but it does obscure detail, and the ear tends to hear it as a fatness or warmth in the sound. Lower orders are the 2nd, 3rd and 4th.

Transistor amps have almost none of these distortions, but they are common with SET amplifiers. P-P tube amps on the other hand tend towards the 3rd and 5th.

Now it is the higher odd orders that our ears use as loudness cues. Anytime they are distorted, the amp will sound louder than it really is. This is why SETs will sound very 'dynamic' for their power, as normally they don't make much distortion, but if you push them (which is what transients do) then the loudness cues appear **but only on the transients**.

Transistor amps tend to have these higher odd orders all the time. This is one of the reasons they tend to sound hard or bright. Now its important to note that these harmonics do not have to be very distorted, usually 100th of a percent are audible, simply because these harmonics are so important to the human ear.

This, BTW, is why two amps can seem to have such different tonal characteristics even though they both measure flat frequency response on the bench. The addition of global negative feedback to any amplifier will increase the odd-ordered distortion slightly, which is why any amp with GNF will tend to sound brighter even though frequency response is unaltered.

(The trick, IMO, is to build an amplifier that does not use feedback, and use other means to eliminate distortion.)

The bottom line here is that distortion is always audible and to nearly anyone. Its just that it does not *sound* like distortion to us, often it sounds like a tonal aberration ('bright' in the case of many transistor amps, 'caramel' in the case of many tube amps).

Timlub, It looks to me like Fas42 places a lot of importance on soundstage and imaging. There are different types of 'distortion' that affect that- primarily of bandwidth. The better it is the more intact the phase relationships, which are what defines image location.

The ability of the speakers to 'disappear' is equally important to a system's ability to convince as is detail and tonality.

When feedback is applied to an amplifier, low level harmonic noise is injected into the output of the amplifier. This is nearly all high-frequency information. Now it happens that room ambiance in recordings is often also high frequency and is also low level. The result is that by adding feedback, the low level detail associated with room size can be masked by the harmonic noise floor that is present in the amplifier. In fact *all* detail below this level will be masked. That is why amplifiers that run little or no feedback often seem to have bigger, wider and deeper soundstages.

Tubes and transistors seem to act very similar to feedback. I've less experience with class D, but from what I am seeing there are several techniques of using feedback and some work better than others, so it would be unsafe to generalize that feedback works in all class D amps without increasing odd ordered harmonics. In some it seems to work well though.

Not heard one yet that images as well as I am used to but I've not heard them all either.

Hi Kirkus, that one is easy! Its the same issue of when you hear a really good digital recording, one that is really convincing. What would it sound like it had been done analog?

The whole experience here comes down to one of expectation and intent. If you *know* that the system is underpar (maybe its only a table radio) *then* you don't have much expectation of it. But if its a real high end system- it had better bring home the bacon, because it has some high expectation attached.

I heard a system across the hall from me at THE Show about 10 years ago, maybe a little more. It was Joni Mitchell, singing with a full orchestra. It had been recorded by Sony, digitally, in 6 channels. The system was entirely solid state. The format was a digital reel to reel, 1/2" wide. I really did not hear the system, it was only the experience of the music.

So what would it have been like if it was 2" analog with my amps on each speaker? Hey, maybe it would have been better. We'll never know. I can say this though- if that system/recording does the job for you, if somehow it manages to sound *real* than that is worth paying attention to.

I have never said that feedback should be eschewed- all I have done is point out what it does. Now if you are reading between the lines, you will also see that I have also been pointing to how things can be dealt with. I'll give you a couple of examples. FWIW, these are all on the cutting edge of the art.

1) since feedback is bad, build tube or transistor amps without,
2) since the evil of feedback has a particular cause (propagation delay), build circuits that gets around this problem. Then you can have all the feedback you want.

There are people who are doing both, or at least are approaching both. Nelson Pass is one, Spectron is another. I think there are some class D amps out there that do some timing things to get around this issue too. And of course there are all the zero feedback tube amps out there...

Yes, feedback is somewhere in there. But there is correct feedback, and "bad" feedback! If feedback was inherently not good then no piece of recorded music could ever be made to sound good, since every recording device and studio is riddled with feedback design techniques and circuitry, going back to almost the very start of electronic recording.

This is incorrect. The Ampex 351 tape machine, used by both RCA and Mercury (and a host of others) has a zero-feedback recording circuit. Neumann microphones use small tube preamps which are zero feedback. I can go on but you get the point.

there is no way that well-designed solid state amplifiers "tend to have these higher odd orders all the time. This is one of the reasons they tend to sound hard or bright. Now its important to note that these harmonics do not have to be very distorted, usually 100th of a percent are audible, simply because these harmonics are so important to the human ear." Total harmonic distortion, meaning any and all spurious frequencies away from the fundamental, is normally at about -70db or better, usually better, and there's no way that such low-level distortions could possibly cause amps to sound hard and bright.

OK- you obviously understand how low the distortion levels are we are talking about. I think I did express that 100th of a percent is audible- seems like that needs more emphasis. Since we humans use the odd orders (5th, 7th and 9th) in order to determine how loud a sound is, obviously while the ear is not sensitive to *some* things, this is one thing that the ear is *very* sensitive to. BTW all of this has been known since the 60s and Norman Crowhurst was writing about this subject in the 1950s.

So this is indeed a way that explains why transistor amps can sound harsh while having otherwise flat frequency response. Note also that with many transistor amplifiers, as power output decreases there is a dip in distortion and then it rises again as power output continues to decrease. This is one of the reasons that low level detail is challenging for transistor designs.

It also points to the way to make transistors work as well as tubes, FWIW. I can point to several SS amp manufacturers that have been exploring zero feedback designs and some of them are as good as some of the best tube amps I have heard. Mind you this coming from a tube amplifier manufacturer...

Hi Kirkus, that was not what I recalled so I looked up the Ampex schematic... Its not the record head that has feedback- prior to that the record calibration circuit does use a feedback mechanism. The rest of the circuit has none.

I did think about the playback after posting :/

Of course LP lathes do use feedback, primarily used to control resonance. You can't get channel separation in a Westerex 3D, for example, if you don't use feedback. However there is work being done to this day to try and find a way around that. Apparently feedback is not popular with mastering engineers and for good reason: if the electronics even turn on in the wrong order at power-up, the cutting head can be destroyed.

I have a set of Western Electric mic preamps that are zero feedback. I pulled them out of a dumpster about 30 years ago and boy am I glad I did. They are really transparent, after being updated with Jensen transformers and otherwise rebuilt.

Irvrobinson, in a nutshell, frequency response variation is not why tubes sound different! I've heard that idea expressed before, but its hard to find real world examples so I have to chalk it up to mythology. This is easily proven by using a speaker with a flat impedance curve.

The fact of the matter is that the ear interprets non-clipping harmonic distortion as tonality. With tubes, quite often we see a great deal of lower ordered harmonics, which the ear hears as warmth or 'bloom', IOW because the lower orders are seen by the ear as musical, humans are more tolerant of their presence although such will mask detail.

In the case of transistor amplifier audible distortions we are indeed talking about -70, -80 db phenomena. General Electric proved how sensitive humans are to odd orders back in the mid 60s- its not like this is rocket science, but OTOH if you don't know about this quality of human hearing there may be nothing I can say.

So I think quoting research references just clouds your argument.

So when making a point of fact, one should never point to basic long-standing research?? Don't confuse the situation with facts??

However, it is readily proven, and here is how it is done. Take any amplifier and speaker, you will also need a VU meter and a sine/square wave generator. This is very simple test equipment. Set the sine wave to 1KHz 0VU into the speaker.

Now cover the meter and turn the signal all the way down. Set to square wave (odd ordered harmonics). Turn up the volume until you perceive the same sound pressure. Uncover the meter. You will find that it is reading between 20 and 30 db less.

The human ear is very sensitive to odd orders because it uses them to tell how loud sounds are. To claim that we cannot hear something that is 70 db down is beside the point- we are not talking about something that is being masked. I think you must be thinking that these harmonic distortions are somehow going to always be buried and they are not. This is one of the most basic rules of human hearing.

Now negative feedback is well-known to inject odd ordered harmonics into the output of the amplifier although at low level. Its easy to hear too- but best done on an amplifier that is functional operating open-loop (transistor amplifiers that meet that requirement are rare but they do exist). Norman Crowhurst pointed out in the mid-50s that the addition of negative feedback injects harmonics up to the 81st into the output of the amplifier- this stuff is not imagined by any stretch. Take a look at Nelson Pass' article on distortion:

Audio, Distortion and Feedback at
http://www.passdiy.com/projects.htm

if you have trouble believing that I do my homework. Nelson Pass is one of the leading designers alive today.

Chaos Theory says that an amplifier that has loop feedback is an example of a chaotic system that exhibits several stable states. Interestingly and not by coincidence, the formulae you see to express feedback in an amplifier are *identical* to the classic formulae for basic chaotic systems.

Now Chaos uses the term 'bifurcation' to refer to distortion and what it says confirms what Norman Crowhurst pointed out decades earlier, that the addition of feedback will destabilize the amplifier and inject low level harmonic and inharmonic noise (the inharmonic noise is the result of intermodulations at the feedback node).

The way this happens is that the amplifier has a time period, called propagation delay, which is a finite time in which it takes the input signal to propagate to the output. It is nowhere near the speed of light! In fact it is so slow that at high frequencies by the time the feedback gets back to the input of the amp to do its work, the input signal will be seen to have changed. For this reason the negative feedback is always lagging behind and so is unable to correct the signal it was supposed to. As frequency goes up, the phenomena becomes more pronounced.

With steady-state signals the damage is not too severe, but with a constantly-changing waveform (music) the resulting distortion is much higher than -70 db.

Now most transistor amplifiers are push-pull and so they have even-ordered harmonic cancellation at the output. If the amplifier is balanced throughout (and many of them are) then this even ordered cancellation will occur throughout the amplifier (we do this in our amplifiers for the same reason). So really, the main distortion components of a transistor amplifier are going to be almost entirely odd orders!

Now you may think I am a big tube proponent, but if that is the case you may not have read some of my earlier comments. It is *easier* to make tubes sound more like music because it is easier to build tube amps that work without feedback. But go back and look at what I said about transistor amps without feedback.

Hi Frank, opamps usually have much lower propagation delays than power amps do! So feedback is more effective with them.

To improve the effect of feedback and decrease the resulting odd ordered harmonic generation, decrease the propagation delay of the circuit.

Irvrobinson, while our amplifiers tend to have higher output impedances, that impedance curve is nearly identical to the frequency response curve of the amplifier, IOW linear from 2Hz-100KHz.

Kirkus, I think you missed my point! The quote you put up is edited and not what I said. Try re-reading my post, without the idea that I am trying to make you wrong- that was not my intent at all.

Irv, maybe you were joking but Sound Labs have anything but a flat impedance curve. Just because a speaker has a variable impedance curve does not mean that an amplifier with a high output impedance cannot drive it well, without tonal anomalies. It is all in the intention of the designer, as the article that Al linked points out.

A vital point here is that distortion in amplifiers and speakers is perceived by the ear as a tonality, and without this understanding that tonality won't get measured. This is close to the heart of the subjectivist/objectivist debate. Once you understand how the ear/brain perceives things, a lot of this debate goes away.

The problem here is there is little that is convenient that will quantify the subjective experience. However, that is not to say that the subjective experience *cannot* be quantified, it can and has by Dr. Herbert Melcher.

In his recent work (unpublished so far) he has shown that if music is delivered intact to the brain, it is processed in the limbic system. He has also shown that as a stereo system violates human perceptual rules, the processing is moved to the cerebral cortex.

So while we can argue about what works and what does not, our brain is working it out anyway, whether we like it or not.

Kirkus, your 'quote' of me was not verbatim and thus the meaning and demeanor was altered.

It's my impression from many of your past postings that there are a handful of conceptual errors in your understanding of the traditional application of negative feedback and its Nyquist stability criteria . . . to the point that a discussion of the associated theory and measurement performance is moot.

Yes, I imagine when one has a different viewpoint, it is convenient to use such an argument. I *am* familiar with Fourier, Shannon and Nyquist, FWIW. However I see a lot of their 'relevant' theorem as being misapplied in audio. The problem here is that while theorem is supposed, there are real-world phenomena that do not care about the theorem. When you realize that the real world isn't going to go away, often it is more pragmatic to observe it and accept that it exists.

Now I would exhort you to take a look at Chaos Theory as well,

http://en.wikipedia.org/wiki/Chaos_theory

and having done that read Norman Crowhurst's book on negative feedback and amplifiers, called Basic Audio. You can download volume 3 as a pdf from
http://www.pmillett.com/tubebooks/technical_books_online.htm

In volume three, page 26 Crowhurst graphed the behavior of an amplifier with feedback (Nyquist diagram) that years later Choas Theory identifies as a 'strange attractor' (strange attractors are used to predict the behavior of a chaotic system). You will also see that the formula for feedback and chaotic systems are pretty much the same thing.

But I (very respectfully) remain curious as to whether or not you've ever had an auditory experience that pegs the needle on your own personal, internal "sonic truthiness" scale? And has it ever been delivered by equipment that has a design approach that's incongruous with your own? (Please note that the phrase "pegs the needle" is an important one, meaning that at the time of the experience, an experience closer to the truth cannot be imagined and/or is simply irrelevant).

Of course, and I mentioned exactly this rather early on. So I am one who maintains that we are closer to 90% than 5%, insofar as microphones, headphones and simple audio electronics (no power amps or speakers) are concerned. I made the point at that time that the recording/playback media is arguably the biggest failing.

I have had multiple experiences like this in the studio, and I have had a few like this at home with my stereo (but they don't qualify due to your criteria). My comment about the ramifications of that has already been posted and misquoted.

Fas42, on that page you will see a pattern diagram. I pointed this out as it is an example of a strange attractor.

Now what Chaos Theory has to say about this confirms what Crowhurst has pointed out in various places in his writings:

By the use of feedback in an amplifier there will be a harmonic noise floor (bifurcation) injected into the output of the amplifier. The amplifier will thus exhibit stable and chaotic behaviors.

The harmonic noise floor is inherently different from that of a noise floor composed of hiss. Our ears can hear about 20 db into the latter but none into the former. So the effect in an amplifier with loop feedback is that low level detail will be truncated and this is readily audible (IME) as a loss of ambient and soundstage information.

So its my opinion that you want the amplifier to behave in a way to more closely adhere to the rules of human hearing. I am adamant that the rules of human hearing trump all other considerations. In this case the masking rule of human hearing is where the problem is: we can't hear into that harmonic noise floor. Ridding the amplifier of negative feedback takes care of this and also rids you of the problem of making the slight amount of odd-ordered distortion that is part of that noise floor.

So you kill two birds with one stone, but you introduce another problem- how to get rid of lower-ordered harmonic distortion, which can also mask detail. The ear will hear this as a warmth, bloom or fatness in the lower registers and some people do find it annoying because the coloration can be obvious. BTW, this is something tubes are very prone to.

So IMO/IME, you have to do everything you can do eliminate distortion without feedback. That can be a bit of a trick and there is no one single design panacea for that.

Vernneal, consider that at a concert imaging in the PA is not important. You may well be listening to mono.

Horns have no trouble doing imaging, that's for sure. There are other threads that have covered this subject.

Fas42, I agree completely about the power supplies. In fact I run a separate power supply with its own power transformer for the driver section of our amplifiers. The idea is to prevent any sort of noise that might occur in the output section from having any influence on the driver. This is one of the ways to really reduce IM distortion, as the power supply noise issues are usually modulation issues, but the effect is less pronounced for THD.

And I also agree that the engineering has to be right- you are absolutely correct that there are a ton of variables that affect any design and its easy for a problem in just one of those variables to completely overshadow other design parameters. IMO right here is where you encounter the human element in design.

Y'all have a nice holiday!!