Do you have the cap around the entirety of the second stage? At one point it sounded as if you were planning on connecting it between the collector and base of Q6, which wouldn't work well owing to the low impedance due to the cascode stage's emitter.
3 pF sounds suspiciously low.
Around the stage, base of Q6 to collector of Q7. As Self shows. I too am quite suspisious of the stability models. Got the P-Z to run. 97 poles and 54 zeros. Not sure I understand that at all. Only a few lines are different. It must geve me a reading on every termination point. You would think with almost 400 pages they would have more than a page on this. All it says is run the test, then build a lapace transform and plug iit in. Not much help. A nice feature would be a Bode plot.
Last edited:
Hi,
Could be. 5pF have been used in Goldmund Amp's.
You must remember you have two poles in this, one in the feedback network and second in the miller network.
The feedback network pole will primarily ensure sufficient phase margins and does so with minimal impact on the slew rate. The miller cap will help essentially to fine-tune the phase.
PERSONALLY I would select the minimal compensation the simulator suggests will be stable for miller, a bit more than is suggested for stability in the feedback loop and then increase the input stage degeneration until the real amplifier is stable using 10KHz squarewaves into low value capacitive loads (< 100nF)
I would aim for AT LEAST 200Khz -3dB, and match the input filter with the open loop bandwidth but no lower than 160Khz. 1MHz -3dB bandwidth for the Amp is not unreasonable, if stable.
I would say 26dB (gain of 20) is common and sensible for power amp's intended to be used with active Preamp's, for passive pre's 32dB - 36dB is more appropriate.
And what would these have to with the price of tea in china, where power cables are concerned? The issues relate to Ground/Earth connections, noise leakage into chassis/ground and currents circulating through interconnects and mains cables.
In most cases High End Audio Equipment is Class I (with Earth connection to chassis), in which anything involved stops at the local distribution strip.
Sure, but we are talking about effects that can easily degrade system SNR dramatically and cause added intermodulation. This is NOT less than one in the hundreds of millions.
Note, I do not advocate expensive mains cables or expensive interconnects. Mny "High End"mains cables are wanting on electrical safety, others are designed in ways that do not help, but often hinder the issue.
But understanding WHY cables make a difference, measurable and audible, can help to adjust our designs to avoid or reduce these issues.
Please reviews the diagrams posted earlier in this thread. Things will become clearer.
Ciao T
Could be. 5pF have been used in Goldmund Amp's.
You must remember you have two poles in this, one in the feedback network and second in the miller network.
The feedback network pole will primarily ensure sufficient phase margins and does so with minimal impact on the slew rate. The miller cap will help essentially to fine-tune the phase.
PERSONALLY I would select the minimal compensation the simulator suggests will be stable for miller, a bit more than is suggested for stability in the feedback loop and then increase the input stage degeneration until the real amplifier is stable using 10KHz squarewaves into low value capacitive loads (< 100nF)
I would aim for AT LEAST 200Khz -3dB, and match the input filter with the open loop bandwidth but no lower than 160Khz. 1MHz -3dB bandwidth for the Amp is not unreasonable, if stable.
I would say 26dB (gain of 20) is common and sensible for power amp's intended to be used with active Preamp's, for passive pre's 32dB - 36dB is more appropriate.
And what would these have to with the price of tea in china, where power cables are concerned? The issues relate to Ground/Earth connections, noise leakage into chassis/ground and currents circulating through interconnects and mains cables.
In most cases High End Audio Equipment is Class I (with Earth connection to chassis), in which anything involved stops at the local distribution strip.
Sure, but we are talking about effects that can easily degrade system SNR dramatically and cause added intermodulation. This is NOT less than one in the hundreds of millions.
Note, I do not advocate expensive mains cables or expensive interconnects. Mny "High End"mains cables are wanting on electrical safety, others are designed in ways that do not help, but often hinder the issue.
But understanding WHY cables make a difference, measurable and audible, can help to adjust our designs to avoid or reduce these issues.
Please reviews the diagrams posted earlier in this thread. Things will become clearer.
Ciao T
Hi,
NO. It is attend to the big ones first, before going after the small ones lest you look like a hypocrite who strains at the gnat but swallows the camel whole, or like that dude that was admonished to first remove the rafter from his own eye, before trying to remove a tiny splinter from his brothers eye.
So you do not judge facts, you are not even interested in facts, just in how they are presented and who they come from.
I recognise that there is a considerable amount of BS touted in all sorts of areas and markets and audio is no different. I also recognise that the ejit "debunkers" aid the matter immeasurably, as their agenda's are way too obvious and their unscientific experimentation an insult to intelligence.
The people who lack the education to understand their point see it as just another kind of advertising, but lacking the fun of real ad's, those sufficiently educated and intelligent are turned off by the obvious agenda, the manner and poor science displayed.
So the effect is rather the opposite one from that desired.
First to eradicate it would mean to eradicate deception everywhere and first in the large cases before even get bothered by small fry. To do that in turn would require universal serious education and changes to moral standards of society.
As for the business side, I do not make much money from audio, but much from IT consultancy. And what is good for my business is not BS, but that people who should know better actively do their best to avoid education on the issues for themselves and others, like you are doing in your post.
For example cables, mains and interconnects certainly, are subject to very real issues where R/L/C/G can make large differences, but many here flatly deny it.
When some members talk about it my "possibly as high as "-60dB" somehow becomes "how much -140dB stuff" (talk about disingenuous). Such behaviour does not help "the cause"
I rather doubt I get many customers here. And I do not accept accolytes. So your gripe seems based on false premises.
They are in the public domain, in many threads here at diy-audio (and elsewhere), so go look it up if you did not bother reading them first time around.
Ciao T
NO. It is attend to the big ones first, before going after the small ones lest you look like a hypocrite who strains at the gnat but swallows the camel whole, or like that dude that was admonished to first remove the rafter from his own eye, before trying to remove a tiny splinter from his brothers eye.
So you do not judge facts, you are not even interested in facts, just in how they are presented and who they come from.
I recognise that there is a considerable amount of BS touted in all sorts of areas and markets and audio is no different. I also recognise that the ejit "debunkers" aid the matter immeasurably, as their agenda's are way too obvious and their unscientific experimentation an insult to intelligence.
The people who lack the education to understand their point see it as just another kind of advertising, but lacking the fun of real ad's, those sufficiently educated and intelligent are turned off by the obvious agenda, the manner and poor science displayed.
So the effect is rather the opposite one from that desired.
First to eradicate it would mean to eradicate deception everywhere and first in the large cases before even get bothered by small fry. To do that in turn would require universal serious education and changes to moral standards of society.
As for the business side, I do not make much money from audio, but much from IT consultancy. And what is good for my business is not BS, but that people who should know better actively do their best to avoid education on the issues for themselves and others, like you are doing in your post.
For example cables, mains and interconnects certainly, are subject to very real issues where R/L/C/G can make large differences, but many here flatly deny it.
When some members talk about it my "possibly as high as "-60dB" somehow becomes "how much -140dB stuff" (talk about disingenuous). Such behaviour does not help "the cause"
I rather doubt I get many customers here. And I do not accept accolytes. So your gripe seems based on false premises.
They are in the public domain, in many threads here at diy-audio (and elsewhere), so go look it up if you did not bother reading them first time around.
Ciao T
Yes , but the sample they did send to stereophile has 47pF
compensations caps , as i mentionned it somewhere in this thread.
Indeed , with 5pF , caracteristics (do not read "quality" ) of the cables
become critical as the amp become marginaly stable with such
reduced compensation.
So yes , in this case cables make a difference but only due
to the amp s lack of tight compensation.
Hi,
You surmised from some simulations that this is what "They" did, this may or may not be reasonable, but to state it as an established fact is rather adventurous and unbecoming the truth.
You may state that you suspect this to be the case, but unless you accessed Stereophile's sample, opened it and actually verified the values you can not claim it as fact.
You may not my other notes on how to evaluate stability and what to adjust.
You may also note that the various GM Clone builds shown here appear to be stable enough using these values.
If the Amp is stable with 10pF, 100pF, 1nF, 10nF and 100nF load I normally consider it "stable".
Ciao T
You surmised from some simulations that this is what "They" did, this may or may not be reasonable, but to state it as an established fact is rather adventurous and unbecoming the truth.
You may state that you suspect this to be the case, but unless you accessed Stereophile's sample, opened it and actually verified the values you can not claim it as fact.
You may not my other notes on how to evaluate stability and what to adjust.
You may also note that the various GM Clone builds shown here appear to be stable enough using these values.
If the Amp is stable with 10pF, 100pF, 1nF, 10nF and 100nF load I normally consider it "stable".
Ciao T
The difference heard was a little more clarity, a little more detail. Nothing spectacular, to be sure, but obviously there, as 39/40 hits clearly prove.
Excatly how could the person wearing the headphones, with his back runed towards the man doing the switching, be influenced by the man doing the switching? Who is too busy keeping the log?
The cable is not a cable, rather it's the internal wiring of the unit, the part which carries the signal. In both cases, all other wiring was exactly the same, all from the same batch.
What positioning? Of the headphones?
I seriously doubt stray capacitance has anything to do with anything in this case, given that we made sure the cable runs were exactly the same. Any capacitance in play would have to come from the wiring itself, which is actually immaterial as such because it all adds up in the differences we both heard.
We were not trying to explain why the difference was there and where did it come from, we were simply trying to establish if it was there at all. It was.
INTERNAL SIGNAL WIRING, remember, not interconnect cables or speaker cables. We don't get into that, that's up to the customer to choose on his own and buy whatever he/she likes. And not from us, because we don't sell any.
I went out to get a selection of polystyrene caps. No longer in full production. Polypro not in small values. I guess in the 10 to 100p decade, we are back to silver mica. This project does not warrant "audio boutique" teflon caps for dollars each.
T,
Thanks again for the general topology advice. Trying to understand past the books, I see how one wants the slew rate sufficiently high within each stage. I see how loading without sufficient current as EB did is not good for symmetry. But, anything outside the audible range is by nature, distortion. It was not put on the CD intentionally. So, if it can be filtered without other negative effects, would not that be a good idea? Am I missing one of those other negative effects? I am making the assumption that by moving the gate resistors to the sockets, proper bypass to the heat sinks and ground returns to the center, I can keep the outputs under control.
T,
Thanks again for the general topology advice. Trying to understand past the books, I see how one wants the slew rate sufficiently high within each stage. I see how loading without sufficient current as EB did is not good for symmetry. But, anything outside the audible range is by nature, distortion. It was not put on the CD intentionally. So, if it can be filtered without other negative effects, would not that be a good idea? Am I missing one of those other negative effects? I am making the assumption that by moving the gate resistors to the sockets, proper bypass to the heat sinks and ground returns to the center, I can keep the outputs under control.
Hi,
CD is SO last Millenium. I gave up on these very early in this Millenium. Using Computer sources we have much > 44.1KHz sample rate and > 16 Bit material to play.
If you do intend to filter at 20KHz or so, it may be worth looki ng at how Naim do it, though I generally disagree with their approach...
Ciao T
CD is SO last Millenium. I gave up on these very early in this Millenium. Using Computer sources we have much > 44.1KHz sample rate and > 16 Bit material to play.
If you do intend to filter at 20KHz or so, it may be worth looki ng at how Naim do it, though I generally disagree with their approach...
Ciao T
Position of the wire being tested. A change in position can change inductive or capacitive coupling. Was it at a high impedance point? Was it screened? What I am trying to establish is what, if any, other differences were present. What function did the wire have in the unit?dvv said:
Let us assume for the moment that you reliably heard a genuine change, as you believe. Assume that the two wires were the same radius and mounted in the same way in the same place, so that the only difference between the two units is the metal in the wire. Then I would think that the most likely explanation is something connected with the solder joints, such as a thin semiconducting layer at the interface. I can believe that different metals can behave differently at interfaces; I find it harder to believe that two good conductors in a circuit which is relatively indifferent to conductance (provided it is adequate) can be distinguished.
If I read correctly, there were two entirely different units under test, one with a silver wire, one with a copper wire (dvv is welcome to correct me if I misread). This brings in quite a few more variables. The controls in a test like this are critical; the description of them sounded sloppy and incomplete, and the statistics were extremely suspicious. A good, rigorous test would not be that difficult to set up, and perhaps dvv might wish to consider it. See Linear Audio Volume 2 for a fuller description of some possible error sources and how a better test can be designed.
Hi,
CD is SO last Millenium. I gave up on these very early in this Millenium. Using Computer sources we have much > 44.1KHz sample rate and > 16 Bit material to play.
If you do intend to filter at 20KHz or so, it may be worth looki ng at how Naim do it, though I generally disagree with their approach...
Ciao T
OK, but I have about 1000 CD's still limited to the original bitrate and who knows how many were original Sony 14 bit masters with 2 bits of dithering. Not being in the music business, I have no direct access to higher quality even though any old basement band is probably working at 96 or 192 bits.
Generation gap: I make phone calls on my phone. Don't own an "i" anything. I can sit and do nothing but listen to music with a nice porter and cheese. That also limits my musical taste. I really don't care if hip-hop or country is recorded fantastically, I am not their intended audience.
Both types of wires were screened, and better yet, both come from the same manufacturer, Neotech from Taiwan.
Both were measured to the same length and are the same to a tolerance of +/- 1 mm.
Both were placed in EXACTLY the same way, as either one of them would be placed in the finished product. Ultimately, this is must, if both are the same length, unless we choose a completely different route - which would be ridiculous.
For the Nth time, the wire was used for signal conducting from the RCA Cinch on the back of the unit to the control board in the front of the unit (1st run), from the control board to the amplifier board (2nd run) and from the amplifier board to the headphone jack (6.3 mm banana plug, 3rd run), or, alternately to the back of the unit to the RCA Cinch ouputs.
The TOTAL length of the wiring, all 3 runs included, is 69 cm, or 27 inches. All contact points on all plugs are gold plated.
And I did forget to mention this - we got the same results from both versions of the amp, the all bipolar one and the FET input-MOSFET output version. I am not counting the all tube version because it is still under development, but even so, the difference was there.
I repeat - the difference in sound quality was NOT great, it was on the small side, but it was there in sufficent quantity and quality to enable 39 hits out of 40 attempts, effectively times 2 if we count both versions of the amp. You don't get that kind of hit rate from nothing.
DF96, I am not trying to convert anyone, I think the differences among cables, if any, are always on the small side, and the only exception to this rule, when the differences are easily detectable to a careful listener, is when you go from quality copper to quality silver wiring. This is not the first time I have witnessed this phenomenon.
You may believe this or not, but I can tell when a device uses Burr-Brown's FET op amps from the OPA 134/2134/4134 family. I find this op amp's sound to be particularly annoying, which is probably why I can pick it out, just as I can USUALLY (but not always) pick out Burr-Brown's 2604 op amp, for exactly the opposite reason - I just love the sound of that op amp. Those are the only two op amps I can pick out, BTW, normally I have no idea which one is being used under the hood.
You may have trouble believeing this, but that's your business. To me, this is a simple reality I can prove in vivo any time, anywhere - and I was taught that a good test is one which is completely repeatable and will yield VERY similar results every time if the conditions are kept the same.
Exactly how were these two completely different units, when the test was simply being double checked on two rather than one unit?
BOTH were first tested with copper wire, and then with silver wire. Both gave exactly the same results.
I would never engage in any such testing with just one-off unit, I want to be sure the effect is repetable.
Just to make sure there is no misunderstanding - we first tested the wiring in TWO bipolar units, and then repeated the same thing on TWO FET/MOSFET units, by first using copper wire in both, then silver wire in both, and then we went to the other type of unit and repeated the experience. I hope that this is clear enough.
Last edited:
Honestly, I'm having some trouble with your verbiage- your English is excellent, better than my Serbian (where I only know three sentences, all of which would get me shot), but when we're dealing with technical issues, precision is important. Can you describe exactly how the test was set up, maybe with a diagram as well? I'm still unclear on what was being switched and how, as well as controls.
I don't think I could see 39/40 even on a well-established audible change (e.g., 0.1dB level shift), implying that the difference was remarkably large. So it would be interesting if there's a solid way of demonstrating gross audibility of what would be a basically unmeasurable difference.
I don't think I could see 39/40 even on a well-established audible change (e.g., 0.1dB level shift), implying that the difference was remarkably large. So it would be interesting if there's a solid way of demonstrating gross audibility of what would be a basically unmeasurable difference.
Sorry, perhaps I wasn't concentrating, but I thought this is the first time you have told us exactly how the wire is used. Thank you for the details.dvv said:
Are the wires identical apart from metal? The dielectric around the inner, for example. Do they have identical capacitance? I ask because I believe that dielectrics are far more important for audio interconnects than metal.
I don't know about the fine details of both cables because unfortunately, the manufacturer was much less than forthcoming. As far as I can tell, they are identical save for the color of the external isolator, the copper one being red, and the silver one being translucent.
I tried to contact them for deatils, but all I got was a screenful of marketing babble from people who are obviously sales and equally obviously unwilling to go and ask the pertinent questions.
In essence, I'm as much in the dark as you are.
- Status
- Not open for further replies.