I stand corrected and apologize for missing that you matched the caps carefully; 90'ish seconds is much better than it initially sounded (i.e. I read it as desoldering one cap, resoldering another), and I was going off that assumption, which would mean the gap between X and Y would be tens of minutes to hours. Either I missed you writing that detail, or that's new information forthcoming.
Backwards: don't you need to claim that memory, via whatever model you choose is, in fact, still robust with your test? 70-90s is well past the usual test gap length, and how much does the visual/tactical stimulus of the cap change procedure affect your results? Can you characterize that? How about the biases of non-blinding?
My world is built up on: PLOS Medicine: Why Most Published Research Findings Are False
What effect does having a small N and a small difference between the two test arms (electrically speaking) have on the plausibility of your test? And that's before we get to the fact that the test itself is poorly conducted and reliant on increasingly long(er) term memory function. I'm not saying there's *no* difference in sound between said capacitors, just that the test you described isn't going to tell us one way or the other.
I don't remember that list of arguments, but as to the latter: pot, please meet kettle. Evidence has not been forthcoming. Usually the one asserting conclusions from a test is the one who was to prove his findings.
Did we read the same paper? Let me grab a couple particularly salient paragraphs from the paper:
...
Translation, we expected at d' < 1 that the tetrad test's lower sensitivity would overcome its theoretical power differences, but that's not what we found out: tetrad did better across the spectrum.
Now this is one test, and I don't know how well it tracks to the rest of the field, much less audio preference testing. But it sure looks like, at least for this test, (and the rest of the paper's data supports this statement) that the tetrad test managed to be the more robust test.
Can you understand why I'm digging in my heels when this is the thrust of the paper?
What am I supposed to admit I'm wrong about, at least within this specific scope? (We needn't go down that road...)
Now, yes, my research tends to use more reliable measurements than human taste/human hearing (fluorescent tagged particle are subject to plenty of vagaries, but it's still pretty easy to say they did/didn't stick where you wanted them), and with different protocols, but I'm just hitting at the classical experimental mistakes and how even very well meaning people can fish out a "positive" from a dataset that yields a null.
Last edited:
This is the second time you have displayed that graph, first in a link to a writeup, now naked.
Do we again repeat the list of errors within that author's textual description of the effect?
Oh, and btw. You do realize that graph has absolutely nothing to do with skin effect, right? It is describing the long line impedance change as a result of the conductor resistance. Not anything to do with skin effect at all. Skin effect would show up on that graph as the frequency goes up, as the inherent 15 nH per foot of the core conductor will go to zero, and the effective core resistance would climb. Since neither were mentioned nor graphed, the author of that graph considered neither.
It would be great if everybody refrained from using that graph for something it does not apply to. Skinning is complex enough without tossing up inherently bad information to confuse everybody.
John
Last edited:
While researching the "skin effect" and going back into the journals, texts and archives, I did come across a little known paper. Fascinating. It refers to a common condition where something called the "thin skin effect" was noted... if you search enough, perhaps it will come up.
Even my mid-fi allows me to 'hear into the mix' with certain recordings. It is one of those revelations when you first hear it. well revelatory on how awful pop records can sound on a good system. Esp when you end up with a 'patchwork quilt' of sound.*
Shame people can't keep on track or would be an interesting training exercise to take a track everyone should have and list all the places you should be able to hear something happening. People can listen and learn then, and who knows maybe find something new.
Shame people can't keep on track or would be an interesting training exercise to take a track everyone should have and list all the places you should be able to hear something happening. People can listen and learn then, and who knows maybe find something new.
That's a matter of opinion.
They were brought to the market as cheap home "bookshelf" speakers. Accent on cheap.
The xover components were also cheap.
The drivers were cheap, and had small cheap magnets.
They blow up all the time.
They don't "reveal" very much other than what a cheap speaker sounds like - which was exactly the entire point.
I suppose placed under a forklift, they will stop on a dime. Beyond that, not. There isn't really any absorption inside the box.
Guess we're operating in a different realm. 🙁
_-_-bear
Skin depth in copper at 50/60 Hz is on the order of a centimeter. This will matter in certain applications (extremely high current bus bars) but for *most* of us, not so much.
Now we differ on the meaning of broken! 🙂 We may actually agree on things like beeswax paper and foil capacitors. They measure amazingly bad and really do seem to influence my perception of the sound going through a single one.
I have done the experiment on how moisture affects capacitor distortion and a very small bit does increase the distortion by an amount that would seem out of place. Now I don't have hard figures on the exact chemical composition, but I would expect enough impurities in beeswax to be responsible for their performance. Distortion products from my single new sample as high as -25 dB IM 18&20 kHz.
As to opamp rolling, I have no problem changing older ones to newer designs with similar parameters. When the NE5534 was new rolling it in often caused large changes due to the wider bandwidth causing it to oscillate. But in general it is a fun hobby pursuit that on some occasions may reveal something. (Like the rest of the circuit was improperly optimized.)
OK.... I am dragging out my Network/Z analyzers and will test a variety of cables myself and see. I have to find my high Z probe and CT, too.
I dont doubt your physics on the subject. I need electrical models, though.
THx-RNMarsh
Last edited:
JN,
The consultant driven specifications often call for 10 gauge wire for all the drivers including the compression driver. For some strange reason I find four pairs of 20 gauge with the same twist and very thin 300 volt insulation work much better.
My prior specification compliant design used 110 strand 10 gauge copper wire with a very thin insulation. The new stuff works better. It also requires less equalization for air induced high frequency losses. Wonder why? 🙂
ES
The consultant driven specifications often call for 10 gauge wire for all the drivers including the compression driver. For some strange reason I find four pairs of 20 gauge with the same twist and very thin 300 volt insulation work much better.
My prior specification compliant design used 110 strand 10 gauge copper wire with a very thin insulation. The new stuff works better. It also requires less equalization for air induced high frequency losses. Wonder why? 🙂
ES
This is quite bold. You have to back this up or I will assume you have been smoking herbs from that nursery you frequent. Post your system specifics and schematics.
https://www.omicron-lab.com/fileadmin/assets/customer_examples/Article_Skin_Effect_v1_0.pdf
I can then see if there are any common cable useage where skin effect occurs in the audio range.
THx-RNMarsh
They're not so new, 4 years old in their current configuration.
At least two of the things I heard in mixes were validated by talking to the guys who did the mixing (it helps to know your favorite musicians well enough 😀). "Really? You picked that up???" And it was fun having one of them listening to his latest mix before sending off to the mastering guys and having him notice stuff that escaped him before. But mixing, panning, compression, EQ are all things that are well-established to be audible.
If I want to make an extraordinary claim that contradicts established physics or engineering, I would feel pretty stupid or dishonest not bringing the evidence.
That really is bunk. Too much chemistry maybe.
There is no way you can make that claim - for a start it contains at least one undefined subjective factor.
Or, alternatively, I can *guarantee* that my system has a better "groove factor" than yours. So can everyone here, should they so choose.
Yes. For 60 hz distribution, copper conductors less than 4 inches diameter are solid. When the conductor is larger than 4 inches diameter, power companies will go hollow as the copper deeper than 2 inches (industry accepted practice) will add only to weight and cost but not current carrying capability.
Close enough to what I larned back in 74.
No problemo. Of course, a true analytical model is beyond simple explanation.
Why do you wonder?😕
The standard 10 gauge will have about 200 nH per foot. #20 with thin insulation can get down to 180 to 170 nH per foot. You have paralleled 4 pair, so you have an equivalent inductance of 45 nH per foot. At the lengths you typically mention, the inductance comes into play, no?
Well, let's just say that you will get "some" measurement result.
The good things I see:
The first thing they say is any measurement is bogus if the calculated exponential depth is not much much smaller than the diameter. This is inherent in using the planar wave exponential depth model even if they do not state that openly. This is exactly as I stated prior, that the curvature of the surface affects the penetration depth such that the planar model does not work properly.
They also say that the measurements are bogus at 10Khz and below.
I like they seem to have eliminated inductive effect within textual discussion even though they did not really elaborate on it. They seem fairly competent in their understanding.
The issues I see:
1. That meter is incapable of separating out the in phase resistive losses caused by proximity effect from the in phase losses caused by skin effect. Had they used a true loop instead of a flattened parallel wire construct, it would be better, the send and return conductors are communicating magnetically. (see below for proximity effect elimination).
2. For that length of wire loop, there will be magfield enhancement caused by the length to spacing ratio, this will also modify the inductance vs frequency, therefore the proximity effect losses vs frequency will be a function of frequency. (this enhancement is noted as one of the three components of the terman equation.)
3. A truly scientific evaluation (which was NOT the purpose of that specific writeup) would have used several different gauges of copper, several aluminum, entirely round loops, and variation of loop diameter. That would have allowed a rigorous verification of the accuracy of the measurement.
All that said, it was a very good writeup nonetheless. Thanks for it.
If I were doing that test again, I would made a coaxial cable using a #10awg core conductor and 3mil thick tefzel heatshrink insulation, short it at the end, perhaps 10 foot long.
Did it so long ago, I'm not sure where that wire is anymore. Probably tossed it.
Edit: Forgot to mention: Since the outer braid of a coax has no internal magnetic field caused by the outer braid, the coaxial construction will eliminate proximity effect altogether on the core wire. Also, they did not mention that the loop CANNOT be near any metal at all, magnetic and non-magnetic. Eddy currents in any nearby conductive object will be interpreted by the meter as an increase in the DUT resistance.
John
Last edited:
I must admit that statement confuses me. When I adjust a pan control, everybody can easily hear the sound go from one speaker to the other.
Compression, especially when done poorly, is also quite evident.
Eq the same.
Are you simply playing devils/s advocate or is this part of an ongoing battle I've missed?
John
I think there's a healthy bit of summer doldrums. I certainly haven't helped, but I'm getting a lot of grumpy from a lot of folks.
- Status
- Not open for further replies.