DIYHiFi.org

Uh........I dunno. Maybe. Probably. Probably about as important as all the other metrics that we have.

I would rather this be a separate thread from the "new-age" design thread. Probably will be a bit more provocative, possibly in an assertive manner. Everyone who wanders into my area should know it is a mine field. I would rather keep the general area less contentious.

As I mentioned in that thread, we really don't have the metrics to gauge sonic performance. We may never will. But what we have done follows something along these lines, and here is where we ended up:

In the 70s, we made amps with tons of global feedback. They had THD values that had so many zeros in front of the number that you had to count them in order to recite it.

But they sounded like crap. (Yeah, I know.........anecdotal evidence.) So, some guys set about to come up with schemes to explain why. Their methods differed a bit between them, but they all seemed to zero in on amount of global feedback vs open-loop pole frequency vs slew rate vs front-end linearity vs.............you get the idea.

Where any of them right? Well, maybe not 100%, but they were on to something. Those of us who did not follow the orthodoxy of the AES and were willing to try new ideas were generally rewarded for our efforts. (Yeah, I know, even more anecdotal evidence.)

But cross the powers that be at the AES............proceed at your own risk!!!!!! Even something simple as asking why a lot of the preprints that are presented at the Convention never make into the JAES, and you can not imagine the hostile response that generated.

I don't think that I would have received a worse response than if I had opened a potato latke stand outside the Nuremberg rallies.

I think that even some of our critics will admit that the AES has stomped on them at one time or another. For doing pretty much the same: questioning the powers that be.

This may seem like a petty issue, but it is a fact that the personalities involved in this has led to the depth of the rift. There are folks on both sides of the debate that are motivated by politics. Sad, but true. One side will claim that they are above personal attacks, but they do it all the time. Maybe in closed quarters, but it does happen. The other side, with probably the same percentage of political attack dogs, will claim that they are only acting in response to how they are treated. Since 99% of the folks don't know who threw the first stone, they tend to believe that the other side has all the bad guys, and their heroes are the good guys.

Smart folks on both sides of the issue, spending too much time defending their stance because they feel they are under personal attack from the other guys.

Yeah, sounds like the AES behaves pretty much the same as forums do! Human nature, I guess.

OK, so much for dubious context. But, you have to accept that, and then clear that out of your mind in order to be open to moving onward. If you aren't willing to listen, learn, try new stuff, and be willing to fail more than once, then just flip a coin, take a side, and watch these guys fight each other for another 30 years.

So, what is important in audio design? Well, I have to admit that I don't know. I do know what I have tried, what I liked, what its strong and weak points are, and why I think the way that I do.

Let's look over some of the stuff that has shown a glimmer of hope:

High loop feedback vs low loop feedback vs no loop feedback. I have tried all 3. I found that the folks who bought the stuff that I built liked stuff more as I went from lots to little to none. (Yeah, I know, a lot more anecdotal evidence. Sorry, but $$$$$ in my wallet speaks louder than DBTs and controlled tests. Didn't put a gun to anyone's head. They had no idea what I did, why I did it, or why building an amp with poor THD and high output Z shouldn't sound good. At least by AES standards.)

But, when you stop and think about it, how important can THD be???????????? I think we have all heard tube gear that sounds great. Lots of distortion there, right? SS amps with lots of feedback that sounded like total crap. Have any of you measured the THD of your speakers? We all have tons of source material that was recorded on open-reel tape decks. Tape decks with heads that have 1% third-order distortion.

So, it can't really be THD can it?

Nope, probably not.

OK, so then we come up with IMD, S/N, S/N+THD, SID, TIM, PIM, and on from there. None of them really explain it either.

Some clever guys come up with theories that show, at least on paper, that the cause is where the open-loop pole is. Especially when there is tons of NFB.

Other guys theorise that the problem is TIM, and the way to fix that is to increase the amount of degeneration in the front-end. Only to get the usual academics in the JAES to respond that you don't need more than 50R, and possibly not more than 100R in the extreme case.

But, those of us who were open-minded found out that linearising the front-end, or raising the open-loop pole generally sounded better. Even though it has to measure worse! So, what gives?

We try op-amps that have lower OLG, and open-loop poles at 10 kHz, and we like them. Hey, it fits our preconceived notion of audio design, but is that the answer? To be honest, no, probably not.

But, it has to be part of the story. It doesn't always work, it is worse on paper, but yet it almost always yields good results.

Yet, there are lots of amps that use tons of feedback. A lot of folks like them, and some don't. But enough of them sell, to folks who think that they sound good to them. (Hey, that anecdotal evidence is ok when it works to the advantage of our critics. Funny, isn't that?)

The bottom line is............I've said it before..........I'll probably say it lots of times more...........we have no way of predicting how a given topology will affect a musical note, and how it will be perceived by the listener.

And this is just topology. An amp is nothing more than a modulated power supply. What about the supply?

Diode noise? Sure, it generates crud that probably gets in the circuitry.

What about a transformer that has a 4 degree conduction angle? Trying to charge those huge filter caps, with only its intrinsic resistance and the ESR of the cap to serve as current limiters. Ever take a 500 VA transformer to make a preamp supply, when 5 VA is probably enough? What happens? Does the transformer let all sorts of crud through in those 4 degrees, when maybe, just maybe that little 5 VA jobbie struggles to charge that 15,000 uF cap. Does it go into some odd uncharted non-linear mode?

I dunno. Well, maybe it does something weird. Something has to 'splain why it sounds better. And it usually does sound better. (ARGH!!!!!!!! Anecdotal........!) But ask anyone who makes power amps for a living. Most of them, if they are honest, will admit that the power supply is probably 80% of the sound.

Imagine that. All the topology tricks in the world, and the supply makes a bigger difference. How can that be? Again, I dunno. But I do know that if you design any piece of audio gear, just by your engineering class notes, or what your simulator tells you, you may be disappointed.

Now, I am not telling you to completely ignore all your book learnin', and to burn your simulation s/w (tempting, isn't it?), but in a sense you have to flush your brain clear of all of that stuff.

Don't get me wrong: a solid educational background is beneficial. Knowledge of simulation can be a powerful design tool.

But that is the point: they are just tools. Tools, just like any test equipment. They are tools: not rigid guidelines that have to be followed at all costs.

The most successful designers that I know have a much better education than I do. But they don't let that get in the way of their quest to make better equipment. They use that as a tool to allow their mind to venture into areas that their critics will say is la-la land. They are not afraid to address any issue, try any solution, and discard the ones that don't work. (Since they are better educated than I am, they will look for ways to 'splain why a ferrite bead over a power cord sounds good when you first put it on, but then sounds like doo-doo a week later. I just see if it works or not, and blow off the reason why.)

And we haven't even attempted to address why wire, resistors, caps, power cords(!) all sound different. How can passive parts that have absolutely nothing t do with the audio path, or at most, play a much less significant role to the topology, have any effect on the sound. But, those of us who have done this a long time, and are honest (much to the disdain of our critics) know that these things play a role.

Admittedly, a very, very minor role. But it is those minor roles, when melded together, are what makes an audio product one that stands out above the others.

How can that be?

"Don't use that 'I dunno' excuse again, bub!"

Well, if the shoe fits.................

OK, I'm typed out for a while. Will probably take y'all time to read it and digest it. Hopefully, it will give you some new insight. Well, maybe not new per se. But new in the sense that it reinforces what you suspected.

But were afraid to ask the AES!!!

OK, time for you guys to add you thoughts. Rambling, or otherwise.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

Yes, I've learned that the hard way. Although I don't make amps for a living.
At the Pub one of those hardcore objectivists found very strange what I was doing and accused me of trying to equalize the sound of my amp through the PSU and local decoupling.
Oh ignorance... call it EQ, call it what you want.
The PSU has a big impact on the sound, and don't give me lessons about high PSRR circuits and how my circuit is probably flawed - it its irrelevant, the PSU has always a sonic signature. It is on the signal path.

_________________
Carlos Filipe

"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius - and a lot of courage - to move in the opposite direction." Albert Einstein

I missed the AM-PM part.

Kidding aside, nobody that I know of ever looks at any form of PM added or changed by their audio product. The only stuff I've found has been by Paul Klipsch with regard to horn loudspeakers and some guys who looked into the problems of wow and flutter in phonographs. Some of that did make it into the JAES. You have to wonder why they aren't at least consistent by discussing that with regard to electronics. Wait; what did I just say? Forget that.

You have to think that PM on a sound would be something recognized by the ear-brain aural system. How else would an animal know whether that larger faster animal with really big teeth and an empty stomach is running toward him or away from him?

As I babbled one about before, the front end of an amplifier, unless it is perfectly linear, acts like a phase detector. If you feed a delayed version of the original signal back into one of the ports, what do you get at the output?


Has anybody here looked at the high frequency current through a power transformer for an amplifier or preamp while it's actually playing some music, even a note? Is there any published information on that? Or through the supply filter cap? I'd be very curious. I wonder what it would take to make such a measurement...

Here's yet more anecdotal information.

I have a headphone amplifier. Big power transformer, big filter caps for what it is.

One day I decided to see what I could do to make it "sound" better, since I got some new headphones. So, I thought I'd start with the power supply.

First, I took my usual approach and replaced the diodes and did the transformer damping thing. The usual difference between seveal shades of gray.

Next, I wondered if the line cord made a difference. As it turned out, yeah the line cord made some difference. A minor shade of gray difference. (Oh, btw, the source was a modified iPod, so the only connection to the wall was at the headphone amp - no common mode connection between components or any of that.)

Then a funny thing happened.

While changing line cords, I got lazy and left the source running and left the headphones on.

When the line cord was unplugged altogether the sound was *way* better. I won't say black and white better - who knows what that is - but really, really light gray to almost black kind of better. Of course this only lasted as long as the caps stayed charged. With the tiny load they supplied and their size, that was maybe 15 seconds or so.

What's that say?

Jocko!



Let me stop for a microsecond at this little particular - only because I had spent time looking at it. It's not DC R and ESR that limits the current; the role of these is less then secondary. The main limiter is the time constant of the resonant circuit formed by transformer secondary leakage inductance - storage capacitance. That characteristic "gaussian" like shape of the current pulses is due to this, and the perceived DC R - ESR has a role only as damping at the series resonance.

So the conduction angle is depending on: first of all the current load across the storage capacitors; then the transformer secondary leakage inductance. Conduction angle is always higher for an EI transformer, and maybe this is why I hear them less agressive? Conduction angle also gets higher with higher storage capacity! (contrary to the beliefs) Another important factor, for high load currents is the core saturation characteristics, because in saturation the resonance is not valid any more. (because of the generally low cond. angle, the charge pulses are at about 7-10 times higher than the load current, so saturation arrives sooner)

My little 2 cents.
As regards the rest, very true words, nice reassumption, thanks!

Ciao, George

It says that you actually took time to listen, and weren't afraid that things might not be as they seem.

Yep, I remember all the "rubberised snubberised" food fights. Again, when we try it, we are heretics. When they try it, they call it their "research". (One day I need to measure some of those on the network analyser.)

But it all gets back to the same basic problem: people who think that they can will something to sound good, simply by the sheer force of their intellect. Their education becomes a dogma, as opposed to a foundation to build upon. The Catholic Church could not have done a better job suppressing really audio design than the AES. The fact that lesser beings are successful goads them.

Just like one of our critics admitted yesterday: the name "John Curl" is worth more than 4 PhD's. They just can't stand that.

Yet, "we" are successful.

Before I made the bone-headed decision to start making audio gear to sell to punters, I was re-working various amps. (I got so tired correcting such major goof-ups, that should have been obvious to the designers, I figured that I could not possibly do worse on my own. What the hell was I thinking?) Anyway, my last series of re-works, after I had done enough other stuff to keep them from blowing up all the time, was to futz around with the OLG, with the goal of getting that open-loop pole close to the top of the audio band.

In theory, it should have sounded worse. Less loop gain = more distortion and higher output Z. Yet, everyone, and I mean everyone thought that it sounded better than the stock configuration. Even people as stubborn, cantankerous and disagreeable as Phred thought that it sounded better. ("There you go again......more anecdotal evidence.") Anecdotal or not, money in my pocket is all the proof that I need.

"So, tell us what this great mod was."

Uh, call it "Miller feedback" Only took one resistor. On paper, it made the simplified schematic look like a cascoded gain stage. Whatever it did, it worked. Yeah, the bass probably wasn't as tight, but it didn't matter. It made things sound better. At least it was perceived as sounding better, and that is the important part.


Speaking of things that shouldn't sound good, but did........

A buddy made an amplifier that had no feedback loops. All MOSFET. Because of all the device capacitances, it had a really low -3 dB point. I mean a really low -3 dB point. So low that you needed to pair it up with a speaker that was so bright that no one could stand to listen to it.

And damping factor.............I forget, but it wasn't that great.

One of the best amps I had heard. Especially in the bass.

HUH?????????????? How can that be? Had no right to sound that good. At least not if you asked the usual suspects.

The public said something else. Parlayed that into one of the most successful audio companies around these days. A perfect example of someone with a good education (degree in physics) that doesn't let his book learnin' get in the way of real learning.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

Good input, George.

Yep, that area that approaches saturation has to be part of the answer. Toroids have much different leakage inductance values compared to similar E-I trannies. Much in the same way the larger ones would be over the tiny ones that are all that we really only need.

Also could 'splain why putting series R before the cap helps. Not only acts as a good LPF for the crud coming through, but significantly damps the resonant energy as well.

See, if you were so closed-minded as to never try it, you would not have thought to try to find out why it worked.

Or, maybe it wouldn't have. You just have to be willing to try.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

Exactly! And the effect of the first several ohms [or even below] is really strong on Q! Also here gets in the picture the rectifier forward dynamic impedance - diode sound, anybody?

Ciao, George

I saw Jocko’s thread this morning and got to thinking on the way to work. Always a bad idea I know. I have a long commute.

Some time ago I was asked to put together a DI box by a buddy of mine. After talking we thought it might be fun to have one box with different flavors. To be able to switch between circuits. This, thought I, would be a good place to play with different opamps as unity gain buffers. In the studio it’s not always about the best sound it’s about different sounds. I’ll try anything once just to try it. Worst I can be is wrong.

This experiment takes Jocko’s comments about THD, feedback, etc. to one extreme. 100% feedback.

I put together a UGB with the following; .1u Wima MSK4 input C, 1M input R to common, 10U Polycarbonate output C. Supply is a pair of LM3xx. Pretty basic stuff.

Caveats/qualifiers/conditions
1 No food fights about opamp UGB being junk, not the purpose of this.
2 No questions/comments as to what’s better.

I tried; AD845, OPA2134, AD829, LM356, LM411 and others I don’t remember. This was going on 2 years ago.

Result
They all sound different.
I changed the supply from the LM3xx to a pair of shunts. They sound different again.
Start adding local bypass Cs, snubbers, A c across the rails. All stuff that Carlos has posted about for years. The sound changes again. Change the snubber and rail to rail C, again the sound changes.

Now in reading this I’m sure you’re all going; “Yea, so what. We all know this, nothing new.”

But why? Why do they all sound different? Why does changing the local bypassing change the sound so drastically? Why does one sound “better” with a 3.3n rail to rail but the next one sounds “better” with a 220u RR? (Rhetorical questions)

According to theory they’re all “supposed” to be the same or close to it. Is the theory wrong or is something missing from the theory?

Bypassing/decoupling for HF isn’t supposed to make any difference at audio. Changing the supply Z that the amp see’s by a few milliohms isn’t supposed to make a difference at audio. But they do.

We can argue/discuss specs till the end of time; PSRR, CMRR, open loop response, blah blah. But the theory tells us, and so does the AES many times, that at audio none of this matters or, it’s supposed to matter so little that it’s irrelevant.

How do we measure what’s taking place? Can we measure what’s taking place? I’m a test engineer, I test things for a living so I want to use some type of measurement to compare items. If a motor shaft breaks during a test at work and it isn’t supposed to I look closer and ask questions. I get a microscope. If need be I’ll send items to an electron microscope. What’s the overhung load? What’s the vibration harmonics? How strong are the harmonics? What’s the material and rating for the material? What’s the finish of the material? And more. No one gets excited about doing this, it’s expected that this occurs. If the answer isn’t obvious then I have to look even closer and/or develop a test to reproduce the failure to get to the bottom of the issue.

But for audio I’m more of a subjectivist. I do test but I don’t get hung up on them. Huh? Completely against everything I believe or have been trained to do. I’ve asked myself the questions; Am I a subjectivist because I don’t have $100K of gear in my shop to test stuff? Is it because I’ve seen too many instances where things sound great but the “theory” says something else is supposed to sound better but doesn't? Because person A says opamps suck but I’ve heard some great opamp circuits. Because person B says tubes suck but I’ve drooled over a few tube amps.

What about the guys that say the tests we have/do are good enough. If I said that at work they’d fire my butt so why does that argument work in the audio world? Why in the audio world is there an argument when someone develops a new test to find something no one thought was there? (jitter anyone)

Jocko started this thread and has asked some good questions. I’m asking more questions. Is it possible to get the answers, what do you think we are missing to get the answers?

Mike

21 Khz


Output Z ~ 1/3 Ohm at 1 KHz, > 1 Ohm at 20 KHz


More anecdotal evidence. To an awful lot of paying customers...

I think that Jocko's comparison to the Middle Ages and after (and probably before, too...) is apt. It all comes down to who has what to gain or lose.

One of the key things is that much of the publicized research about the aural process has been done by people who were trying to find out what was "good enough" at some point for a commercial reason. "Good enough" is a relative concept. In those same Middle Ages, the average life span was 35 years old. Now it's much longer, so we can all weigh on the economic system in our old age.

When somebody decided that THD was a "good enough" test, that's pretty much all that could be done. Now, that's not so. But we don't have a lot of information about what really affects the listener's response to sound.

Is PM important? How about 7th harmonic distortion?

Some people have been clever enough to be able to find their way pretty much, based on observations they've made about their experiments. But it's not all nicely packaged in a way that satisfies people who rely on credentials of some sort for their livelihood and/or self-esteem.

As I said earlier, I find it rather amazing that well documented measurements made a couple decades ago on the listener effects of wow and flutter aren't even considered relevant today. Have people evolved that quickly? I'd think that if anything more refinement is needed now, since people's expectations have evolved.

Maybe that's dinosaur thinking.

Steg

Lets bet.

My take: Ayre V3

Right or wrong, Charles Hansen's no nonsense approach and his clear (even to me, go figure) posts are most impressive to me. No, I do not say that he is alone in that, there are quite some others around here I admire for crystal clear thinking-and writing....errr....what's this forum niche about anyway? ;-)

Ciao, Tino