scott wurcer said:
What signal comparison occurs in that scenario? I don't catch the application here. For what it's worth it depends on era, in a modern radio studio a live broadcast could see as little as one or no caps before a conversion to to the digital domain, where stays right to the transmitter exciter input.
Both the info and man pages describe 'dd' as a plain bit copy utility. Did you mean use 'dd' to copy and compare the files with an 'oc' dump? I never heard the claim about differences remaining audible after a copy to HDD. What was the source?
GRollins, boiled down the rational is perilously close to correlation=causation.
rdf said:
It was long before digital anything. I was just trying to pick a senario where there were as many things in the signal path that should be "bad". For instance the output stage on my tuner was a cheap dual JRC op-amp with electrolytics on the input and output bypassed with two more. The sound was better than some listening sessions where great care was taken to eliminate tnis stuff. This case just sticks in my mind, I know it wasn't an A/B test but there were qualities that some of the best setups with reference quality inputs are missing (to me at least).
I didn't remember the exact utilities, but it suffices that bit by bit comparisons can be made.
Conrad, I think that you are thinking in the right direction. It is just that we have done some of our homework over the decades that jumps us ahead of many here. By the way, it might be ironic, but Charles and I have degrees in physics, not engineering. I think that is why we see things in the same way, besides personal experience.
I, too, avoid caps, when I can. Still, I have to use caps as final power supply bypass and as eq caps for RIAA stages. This is where I find the need for film caps to be critical. Then, WHICH film cap is also important.
For example, almost 10 years ago, my associate, Bob Crump suggested that we use Rel polystyrene caps as power supply bypass, and only sparingly. The amps that Parasound was building with my help, then used offshore film caps, just about everywhere.
We took a big Parasound power amp, reviewed by 'Stereophile' but not given any rating (and apparently, interest), and recapped it, and we changed some resistors, and some wire. It changed the sound of the amp, completely! Before we modified it, I personally measured it extensively. It was in fact my own personal power amp at the time.
The change was remarkable and obvious to everyone. Then, we could use the amp at the CES. Before, one my associates would not bring it, (the same amp) even as a backup after first auditioning it.
We were so happy with with the results, and our listening experience at CES, that we decided to buy power amps from Parasound and modify them ourselves. We did this for a few years, until the JC-1 (based on this same amp, but modified by us from the ground up) was released. Today we still use those very same Rel polystyrene caps as power supply bypasses. Well, recently, we had some problems, and we lost a few caps. We actually decided to upgrade the cap, but still keep it polystryrene. I could have changed to polypropylene in an instant, IF I thought that it would sound as good, but I have to compete with Charles, and until he uses Rel polypropylene as his power supply bypasses, I'm not going to stick my head out, and potentially change the sound quality of the unit.
You may not understand this, but Charles and I are ruthless competitors with each other. Just like drivers in an F1 auto race. Sometimes he wins, sometimes I win, but we are CONSTANTLY competing. We have done this for decades. We are also colleagues, who know how difficult it is to do what we do well, so cat-calls from would-be designers, just like would-be auto racers, are virtually laughed at.
I hope that this clarifies things a little.
I, too, avoid caps, when I can. Still, I have to use caps as final power supply bypass and as eq caps for RIAA stages. This is where I find the need for film caps to be critical. Then, WHICH film cap is also important.
For example, almost 10 years ago, my associate, Bob Crump suggested that we use Rel polystyrene caps as power supply bypass, and only sparingly. The amps that Parasound was building with my help, then used offshore film caps, just about everywhere.
We took a big Parasound power amp, reviewed by 'Stereophile' but not given any rating (and apparently, interest), and recapped it, and we changed some resistors, and some wire. It changed the sound of the amp, completely! Before we modified it, I personally measured it extensively. It was in fact my own personal power amp at the time.
The change was remarkable and obvious to everyone. Then, we could use the amp at the CES. Before, one my associates would not bring it, (the same amp) even as a backup after first auditioning it.
We were so happy with with the results, and our listening experience at CES, that we decided to buy power amps from Parasound and modify them ourselves. We did this for a few years, until the JC-1 (based on this same amp, but modified by us from the ground up) was released. Today we still use those very same Rel polystyrene caps as power supply bypasses. Well, recently, we had some problems, and we lost a few caps. We actually decided to upgrade the cap, but still keep it polystryrene. I could have changed to polypropylene in an instant, IF I thought that it would sound as good, but I have to compete with Charles, and until he uses Rel polypropylene as his power supply bypasses, I'm not going to stick my head out, and potentially change the sound quality of the unit.
You may not understand this, but Charles and I are ruthless competitors with each other. Just like drivers in an F1 auto race. Sometimes he wins, sometimes I win, but we are CONSTANTLY competing. We have done this for decades. We are also colleagues, who know how difficult it is to do what we do well, so cat-calls from would-be designers, just like would-be auto racers, are virtually laughed at.
I hope that this clarifies things a little.
john curl said:
John, my degree is also in physics. There is a certain trend there 🙂 .
Cheers
Alex
G.Kleinschmidt said:
Read Grey's post above. He nailed it on the head, and he wasn't even there when I was developing the amp.
I told you already, I was looking at some different types of distortion. It started off by looking at the "memory distortion" proposed by Lavardin. It then veered off into a different type of non-linearity that occurs with dynamic (ie, musical) signals. (I don't want to talk about this much more, as I consider it proprietary.) Addressing this dynamic linearity also happened to reduced the "THD+N as measured on a test bench with static signals into a resisitive load".
As Grey said, don't confuse cause with correlation. You keep jumping to unwarranted conclusions.
G.Kleinschmidt said:
Yes, that's true. No matter how linear you make a design, you can always make it more linear (as measured on a test bench with steady-state signals into a resistive load) by adding feedback.
But the question remains, how low of a level of distortion is audible? Back when the ICK-150 was released, the generally accepted figure for the threshold of distortion audibility was around 1% under most circumstances and perhaps 0.1% under certain conditions.
Now there are people that are saying unless you get down to 0.00003% by using the new National op-amp, then you are going to have audible problems. Please point me to any studies that support this assertion.
And while you are at it, please tell us what you think is the threshold of audibility. In other words, if one can attain a given level of THD+N without feedback, that there is no point to adding feedback as the level of perception has already been exceeded. (And by extension, all equipment that met that criterion would all sound exactly the same, as they are all equally perfect.)
One poster suggested a figure of -90 dB with real music. But John Curl has said that he can reliably hear differences between equipment that nulls down to -100 dB with real music. So what is your target number? And how do you propose to measure it? With THD+N on a test bench? With an input-output null?
Hi Charles,
Founder of Lavardin, Gerard Peirrot, known as Haphaistos has wrote a very long serie of articles about his research in French magazine L’Audiophile.
Years after, another searcher, Dr. Pierre Johannet has made a simple practical application.
As long as I understood, instead of No Global Feedback at all, a GFB network may be used, but injecting DC into input stage via this network is to be avoided.
For this, Mr. Johannet simply suggests to insert a capacitor to block the DC.
Another application, in his 50W Class A Amp (well known in French at that time), he reduce input stage’s Id current (differential 2SK30AGR) to about 250 uA.
This is the cross point of different temperatures curves in Id-vs-Vgs graph.
He said that the jfet will be more immune to “micro” change of temperature.
I don’t know if theoretically it make sense or not, but I’ve tried it and it works.
Mr. Johannet continues to work in this direction.
He controversially discovers a phenomenon that he calls MDI (Micro Decharge d’Interface) and makes some application to faith its effect.
His latest “gadget” called IONOSTAT.
Don’t ask me the details, since I really don’t know. Google it if you want more informations.
I just inform you in case of you don’t know it yet and hope this helps you in your research.
Cheers
Charles Hansen said:
Not in the literal sense perhaps, but you, John, Nelson, and I seem to be traveling the same road. I came to the rather abrupt realization about twenty-five years ago (when I was working retail and had access to numerous brands of equipment) that:
1) Amps that measured similarly sounded different.
2) Amps that had distortion specifications below the accepted threshold of audibility sounded different.
3) Decent quality tube amps sounded at least as good and frequently better than nominally "higher end" solid state circuits and top notch tube circuits were virtually untouchable by solid state pieces. (The gap has closed somewhat, fortunately.)
I built tube circuits for years, but tube prices eventually got to the point that I could no longer rationalize the expense.
What to do?
I reluctantly turned to solid state. Boy, was that a rude awakening! Almost none of what I had learned about tubes translated to solid state. I had to start almost from scratch. (Presumably the same thing applies to others; the number of manufacturers who build both good tube gear and good solid state can be numbered on one hand with fingers left over.) I've already summarized my two most important rules of thumb--wide bandwidth and low feedback. Is it due, as John believes, to TIM/SIM and the like? I'm not in a position to say, otherwise I'd be guilty of the same thing as G.Kleinschmidt. All I will say is that it's possible and that's as far as I will go at present.
I do have one or two oddities that I've carried over from the tube world. In solid state, many regard "high current is the ultimate goal" as graven in stone. While I agree that current is good, I also find that the highest practical voltage is beneficial. Heresy! As a practical example, I'm extremely uncomfortable with Borbely's practice of standing a cascode device directly on top of the gain device.
(Years ago I asked Nelson about this but apparently phrased my question badly--his response was that cascoding was "as easy as it looks." Which is true enough, but I was trying to pick his brain as to how much voltage to allow the gain device. Since Nelson is generous to a fault about answering even the most addled question, I accept that the fault was mine in not expressing myself clearly, and not an attempt on his part to jealously guard secret knowledge.)
I also still use caps in my circuits--on occasion even as coupling caps (horrors), although that happens more rarely nowadays as I learn ways to accomplish what I want to do without having to add ten or twenty active devices to the circuit. Do I like the sound of caps? No. But neither do I like the sound of five active devices added to a circuit to solve a problem that can be solved by a single cap. Like everything else, it's a question of tradeoffs. Obviously, the objective is to have a circuit with no unneeded devices and no caps. Sometimes the real world isn't so accommodating. Bummer.
I keep trying things and if they work I try to think of a way to push the same idea further. It works for me. Clearly others have entrenched positions that are diametrically opposed. Having been in those self-same trenches, and having found them to be ruts, I climbed out and began exploring the countryside. I like the view better from up here.
Grey
I majored in geology and psychology. I wanted to add physics, but they wouldn't let me triple major, the bums.
Grey
Grey
PMA said:
Actually, a decent foundation in physics is a pretty good place to begin. Is is necessary to know a lot of physics to do electronics? Obviously not. But I find it useful to think of semiconductor fabrication by analogy with crystal formation in nature; in particular, quartz (i.e. silicon dioxide, clear, conchoidal fracture, hexagonal crystal form, etc. etc. etc.). Inclusions (such as rutile [titanium dioxide]) are common enough, as are imperfections on the atomic scale, leading to twinning of crystals and disturbances of the basic hexagonal crystalline structure, not to mention various colored forms of quartz owing to replacement of silicon dioxide by iron compounds (rose quartz), manganese (amethyst), et. al. Okay, so where's this get me?
Well, deliberately introduced disturbances in the crystalline structure of semiconductors give us N and P semiconductors. Deliberately introduced inclusions give us resistors, capacitors, and sundry things like hookup pads in chips.
But there are limits as to how far you can stretch the basic silicon dioxide crystalline lattice before you get major dislocations. That limits what you can put into a semiconductor in much the same way that it introduces flaws into quartz crystals found in the field.
Obviously, I'm straddling the division between physics, chemistry, and geology here, but the basic principles remain if you trot on over into X-ray crystallography, for instance, which few will complain about being "pure" physics.
And this leads me, rather quickly I might say, back to the first post containing a quote from Charles Hansen's letter in Stereophile regarding dirty sand. I happen to know a thing or two about sand and its usual components (quartz playing a major role), and the things that might make it dirty in one sense or another.
See? It's all crystal clear...
Grey
scott wurcer said:
Hi Scott, thanks for joining this thread.
Although it has gotten a bit off track at points, one of the issues brought up early in this thread were Charles' comments in his letter about the limitations of audio components fabricated in IC form. He particularly mentioned resistors and capacitors that are available in an IC process.
I'd like to hear your views on this topic. I know linear IC technology has come a long way in the last 40 years, but I think you might be in a good position to share your opinion about the good and the bad aspects of what is available in current advanced linear IC processes.
For example, there are many different ways of making resistors. Are thin film resistors widely used on quality linear ICs and are they quite good or do they suffer some problems that might degrade their sonics?
Similarly, there are many ways to make small capacitors, like those used for compensation of an op amp. What are the choices available? What are some of the dielectrics used on these IC capacitors and are they more or less prone to exhibit DA or other properties that might degrade sonics? Are MIM capacitors sometimes used on quality linear ICs?
Charles also suggested that if capacitors fabricated on ICs were any good for audio, why can't we buy discrete capacitors made that way? Fair question. Could you shed some light on that?
Thanks,
Bob
Grey, you seem to have straddled the various subjects of study. I also have textbooks on chemistry, plastics, quantum mechanics, and crystal fabrication, as well as all my EE books. They are all necessary.
John,
As you know well, it's all interlaced. One informs the other. Could I do electronics without my formal education in geology (which necessarily includes heavy, heavy doses of physics and chemistry)? Well, yeah, I suppose so. Math is the only things that is truly, deeply necessary to do electronics. But it's just so damned much fun to know how the universe fits together that I can't resist.
Incidentally, I took as many electronics courses as I could slip in between my required courses, but by curious coincidence, the parts of the textbooks that were emphasized in class (e.g. negative feedback) have turned out to be the least useful. I've had to go back and re-read everything in order to fill in the gaps. And even then the good stuff (e.g. wide open loop bandwidth being good for sound quality) isn't even hinted at.
Like everyone else, I wish there was a good textbook that addressed precisely the concerns of those who want to build gear worthy of the moniker "high-end.'
Grey
P.S.: Oddly enough, one of the most useful books I've ever found for this branch of electronics is the ARRL Handbook. That and Horowitz and Hill will get any beginner on his/her feet, though every book I add to my collection tends to have one miniscule tidbit that is good to know.
As you know well, it's all interlaced. One informs the other. Could I do electronics without my formal education in geology (which necessarily includes heavy, heavy doses of physics and chemistry)? Well, yeah, I suppose so. Math is the only things that is truly, deeply necessary to do electronics. But it's just so damned much fun to know how the universe fits together that I can't resist.
Incidentally, I took as many electronics courses as I could slip in between my required courses, but by curious coincidence, the parts of the textbooks that were emphasized in class (e.g. negative feedback) have turned out to be the least useful. I've had to go back and re-read everything in order to fill in the gaps. And even then the good stuff (e.g. wide open loop bandwidth being good for sound quality) isn't even hinted at.
Like everyone else, I wish there was a good textbook that addressed precisely the concerns of those who want to build gear worthy of the moniker "high-end.'
Grey
P.S.: Oddly enough, one of the most useful books I've ever found for this branch of electronics is the ARRL Handbook. That and Horowitz and Hill will get any beginner on his/her feet, though every book I add to my collection tends to have one miniscule tidbit that is good to know.
Various thoughts-
There was a company that made custom (programmable?) analog ICs, and they also furnished the various internal structures in DIP packages so you could prototype your design using the exact same devices that would eventually be in your IC. I think that included the passives as well. Don't remember who or what happened to them.
After all these decades of intense competition between audio manufacturers, we seem to be about where we were at the beginning. There don't seem to be any amplifiers that can reproduce music with the output looking undetectably like the input. I haven't tried a "blameless", but am curious how well it would do. I realize it's a bit of a trick question, because if there really is program material below 20Hz, the BW will have to be extreme to meet my -90dB criteria, with no rise in THD at either of the limits, and no significant phase errors. IMO, this goal is far tougher than one might think using real interconnects and loads.
It wouldn't surprise me to find that any amps meeting this criteria were indistinguishable, and that they didn't sound as good as amps with measurable transfer errors that had caps and circuits tuned by ear. Does flawless signal replication equal the best sound? My guess is no, and at that point we leave engineering department and enter the art world. Nothing wrong with that, but there will new criteria for good and bad, and the answers won't come from calculations and measurements.
There was a company that made custom (programmable?) analog ICs, and they also furnished the various internal structures in DIP packages so you could prototype your design using the exact same devices that would eventually be in your IC. I think that included the passives as well. Don't remember who or what happened to them.
After all these decades of intense competition between audio manufacturers, we seem to be about where we were at the beginning. There don't seem to be any amplifiers that can reproduce music with the output looking undetectably like the input. I haven't tried a "blameless", but am curious how well it would do. I realize it's a bit of a trick question, because if there really is program material below 20Hz, the BW will have to be extreme to meet my -90dB criteria, with no rise in THD at either of the limits, and no significant phase errors. IMO, this goal is far tougher than one might think using real interconnects and loads.
It wouldn't surprise me to find that any amps meeting this criteria were indistinguishable, and that they didn't sound as good as amps with measurable transfer errors that had caps and circuits tuned by ear. Does flawless signal replication equal the best sound? My guess is no, and at that point we leave engineering department and enter the art world. Nothing wrong with that, but there will new criteria for good and bad, and the answers won't come from calculations and measurements.
One problem with opamps is that they strongly tend to be high gain and narrow bandwidth--exactly the opposite characteristics from the ones I'd want. Show me an opamp with something like .5Mhz bandwidth and, say, 25-30dB gain and I'll give it a shot.
Ironically, the caps in opamps would be less needed in an opamp such as I'm describing. There'd be less phase problems and hence less need to compensate. The ability to get rid of most if not all of the caps in an opamp would render their construction technique moot.
Scott,
I second Bob's questions earlier relating to resistor and cap fabrication, and related topics.
Grey
Ironically, the caps in opamps would be less needed in an opamp such as I'm describing. There'd be less phase problems and hence less need to compensate. The ability to get rid of most if not all of the caps in an opamp would render their construction technique moot.
Scott,
I second Bob's questions earlier relating to resistor and cap fabrication, and related topics.
Grey
- Status
- Not open for further replies.