Mr.Curl makes a good case for the use of a buffer at the output of most intrgrated circuits. I have found that adding a buffer inside the feedback loop of an integrated circuit can produce huge sonic dividens. So by the time you get the ic working for best sonics you might as well built a discreet circuit . Some current Levinson products are a good example of this.
With a discreet circuit you have almost total control of the circuit parameters and not dealing with a black box(ic).
Not to take away forom the great IC designs that Mr.Wurcer and ohers have created, but for the present time, if you are looking for the ultimate in sonics, discreet is the way to go. One day I am sure that IC's will get there, not today. Now on the question of transformers..............
Jam
With a discreet circuit you have almost total control of the circuit parameters and not dealing with a black box(ic).
Not to take away forom the great IC designs that Mr.Wurcer and ohers have created, but for the present time, if you are looking for the ultimate in sonics, discreet is the way to go. One day I am sure that IC's will get there, not today. Now on the question of transformers..............
Jam
Jam, you may be well intended, or not, but the REAL solution available to me is whether I can sink a class A current by using an added fet connected to the output stage and the - power rail, and REDUCE the 7th harmonic that I know is in the crossover transistion. I have asked Scott repeatedly about this, to no avail. What kind of extra current would be optimum? I ask once again. Don't give me that $1 million dollar jazz, just answer the question.
john curl said:
John, It's pretty easy to hook up a spectrum analyser and try it. I figure your're just being a little difficult. Seriously though, I though maybe you commercial high-end guys could justify an all out IC op-amp and maybe someone like THAT would be interested. I would gladly donate my time gratis to the project, Les Tyler is an old friend. That's not an idle offer, I'm serious.
scott wurcer said:
You wouldn't also happen to know Liv Tyler would you? I'd gladly donate my time gratis to help out with her. 😀
se
Steve Eddy said:
I see the thread has moved on, but I want to respond to Steves comments.
This is the basis for another circular conversation so I'll bracket my comments with a couple of thoughts: First this is DIY audio and I'm not being forced to use any safety ground at all. When I'm experimenting I do it to see what performance improvements I can achieve for my personal enjoyment and, as you pointed out, using multiple safety grounds can cause issues with components with which you have no control over their internal grounding.
Second, I do allow one such ground in my system and that is in my CD player. I see no need to disconnect it, as it by itself is doing no sonic harm and does provide the whole system with the intended path to the neutral in case Heck does freeze over and the worst occurs. Experience, work habits, and assembly techniques have kept this from being an issue.
That said, a more realistic problem is noise induced into the ground networks (chassis, etc.) of the various system components (preamp, cd, etc.) that are allowed to circulate between components on the cable shields as well as interference that is picked up by these cables. These are more to the point of my comments. These have a defined loop destination and through thoughtful layout can be routed to bypass the signal path. Immunity to grounding issues (I'm not addressing multiple safety grounds here) in associated system components is achieved in the preamp.
By the way, I'm not taking a side in the transformer debate, I'm just commenting on the need for galvanic isolation.
Mike.
john curl said:
Not using those exact words, no. But the smarmy comments you have been making imply nothing less.
Fine. I have no problem at all with what someone else's may prefer to use or not use as the case may be. My problem is with your impugning those who would prefer to use them.
Who exactly are these "practical" manufacturers you mention? You've used this term several times so far. But who exactly are they?
I asked this previously but it went unanswered. Let me try again:
What is perfect, John? And exactly what metric is used to define perfection or exactly how close to perfection you have come? And how close to perfection does something have to be perfect before it no longer matters except except as a collection of specs and measurements?
se
Steve Eddy said:
Put the shoe on the other foot you're a drop dead babe and you're choices are Keith Richards or Iggy Pop. Door number three.
I have just reduced, in simulation, bias of my output buffer to some 300uA. Load is 600R. I made a distortion analysis, that is pretty nasty. Then I added current sink at the output. It helps, but not completely. I get all the harmonics, including the 7th, with decaying amplitude according to their order. No way it can be compared to a buffer with 10mA bias. No hi-end output stage at all.
Just BandAid
Just BandAid
I'd really like to see the outcome of what Mister Curl would come up with if he were to design a speaker....Steve Eddy said:
MikeBettinger said:
No, but then just because someone DIYs doesn't mean they've DIYed everything in their system.
Actually you'd probably be more safe if you did disconnect it.
The ground conductors in interconnects and the ground wiring and circuit board traces in the various components were never designed to handle AC mains fault currents. Expecting them to do so not only creates a potential shock hazard, but a potential fire hazard as well.
NOT a good idea.
Interference picked up by the cables isn't a "grounding" issue and "thoughtful layout" won't bypass it. Either you sufficiently shield the cable so it doesn't pick up any interference, or you have an input stage with good common-mode rejection to keep from passing the noise along.
And I never said or intended to imply that there was any inherent, universal need for galvanic isolation. Only that it comes along with the transformer territory. If it's not needed, it doesn't hurt. But it's there for those who don't expect circuit board traces and their interconnects to handle AC mains fault currents.
se
Thanks, PMA, you have done me a great service. I am poised to measure the distortion, but I can't find any 797's around here to test. I have been originally set-up with a 5534, just for getting the bugs out. Now I am ready to finalize the design. No extra buffer stage for this design, just the minimum that I can get away with, WITHOUT adding 7th harmonic to the mix, if I can help it. I can't even find a LT1028 at the moment. Darn, I will have to order some.
I had hoped to bypass actual measurements, but to no avail. Thanks everyone.
I had hoped to bypass actual measurements, but to no avail. Thanks everyone.
x-pro said:
The main reason to use an IC is to save money. If the IC costs $25, it is probably less expensive to build a discrete circuit. So in this case the IC would only save space and potentially some assembly labor (one part instead of 10 or 20 parts).
scott wurcer said:
Scott, could you please clarify what you are offering?
There is no high-end company in the world that can afford a $1,000,000 NRE for a custom IC.
On the other hand, there are many designs that would only need a small amount of modifications to make them much more usable for true high-performance audio.
I think the onus has to be on the IC manufacturer to make the part that is attractive enough to sell to the audio market. Obviously this is what National is working on right now, but in my opinion they are taking the wrong approach. The audio market does not need another "zero" in front of the THD spec. We need something that is actually innovative and will sound better.
The opportunity is there to "own" that market segment. For example, Burr-Brown pretty much "owns" the audio DAC segment, Cirrus pretty much "owns" the S/PDIF receiver/transmitter segment (AKM is making a go at this one), and ADI "owns" the market for video DACs (usually called encoders).
If you want, I would be glad to consult with you on this. I have done this before with the video division of ADI in Scotland. Our DVD player was using 14-bit ADI DAC chips designed for satellite communications and such. The design group called us up and asked us what was needed in terms of features and performance to make true high performance video DAC. It took them three generations before they finally hit the targets, but now the level of performance we established in 1999 is (kind of) available in mass-market DVD players from Denon, Pioneer, et cetera.
I figure that the only companies that could do this properly would be ADI, National, and Burr-Brown (TI), as this part should have FET inputs. It would be trivial to make this so it could be used either with or without feedback, and also as a single-ended or balanced design, so that you could reach as broad a market segment as possible.
But I can pretty much guarantee that we could come up with a chip that would be universally recognized as the best sounding IC on the market. As long as the price was somewhere between $5 and $10, there wouldn't really be any reason for people not to use it. If the marketing was done properly, whoever made this chip could pretty much "own" the audio IC market segment.
Scott, this is a follow-up to my previous posting. I finally realized that you meant "That Corp." when you said "THAT". Obviously they are focused on the audio segment, which is good. But the big question I see with regards to them is whether or not they have the expertise to make low-noise complementary JFETs.
I kind of doubt it, but would love to be proven wrong. Right now even we discrete guys are in a world of hurt because Toshiba recently discontinued the last audio-grade, low-noise P-channel JFET in the world. It appears that it is quite possible that *nobody* has the expertise build these. We placed a large EOL order that should last us for at least 50 years, but I don't know what other designers are going to do. I guess they will have to either use bipolars and input coupling caps, or else non-complementary designs with their higher distortion. (Remember Pavel's simulations a couple of dozen pages ago of JC's complementary input stage?)
So to make an IC with low-noise JFET inputs, I figure that there are only a handful of companies on the planet that even stand a chance of being successful....
I kind of doubt it, but would love to be proven wrong. Right now even we discrete guys are in a world of hurt because Toshiba recently discontinued the last audio-grade, low-noise P-channel JFET in the world. It appears that it is quite possible that *nobody* has the expertise build these. We placed a large EOL order that should last us for at least 50 years, but I don't know what other designers are going to do. I guess they will have to either use bipolars and input coupling caps, or else non-complementary designs with their higher distortion. (Remember Pavel's simulations a couple of dozen pages ago of JC's complementary input stage?)
So to make an IC with low-noise JFET inputs, I figure that there are only a handful of companies on the planet that even stand a chance of being successful....
Good input, Charles. We will someday succeed in making an 'adequate' p ch j-fet , but it will cost real money. So what?
High feedback and high quality just don't go hand in hand, I'm afraid. How about those transformers, anyway?
High feedback and high quality just don't go hand in hand, I'm afraid. How about those transformers, anyway?
john curl said:High feedback and high quality just don't go hand in hand, I'm afraid. How about those transformers, anyway?
Zero feedback. Not even emitter degeneration. 😀
se
Steve Eddy said:
Innaresting response. My posts were nothing more than a lighthearted, end of workweek ""here's my take on this based on my experience" sort of comments. No part of it was dogmatic or judgmental. Every part of it is legitimate and based on working components or many experiments; nothing is based on hear-say. Just trying to spark some discussion. Once again I must be speaking in another lanquage.
So in response, I guess:
1.) I thought I touched on this.
For your concerns: Run as many safety grounds as you have wire for. 🙂
For the real world problem of noise circulating on the interconnect shields between components and ending up as part of the signal, this can be dealt with by proper grounding. 🙂 (What is he talking about? Don’t bother asking!) 🙂 🙂
2.) If you want to live in fear of that transformer shorting to the core or case then by all means, ground the Cr#p out of everything. I don't run any power ground through PCB traces nor do I use lightweight shields/connectors on my cables. I work around high power RF generators and electronics with power cords that rival fire hoses in size and understand the reasons for the way safety grounding is implemented. Translating this down to 110v in the US is more based on stupid people and lawyers than necessity. 🙂
3.) I wasn't talking about poorly shielded cables bringing interference in through the front door, I was talking about (see 1. above) noise on the grounds (shields) finding a path back through the signal path to the supply and becoming, for the very reasons you quoted in one of your posts, part of the ground reference. This, by the way, is my understanding of the need for galvanic isolation. You don't have to look very hard to see many examples of this proudly posted. 🙂 Grounding and layout I do understand. 🙂
Poor interpretation on my part of what was being discussed. 🙄
Back to the background for me.
- Status
- Not open for further replies.