Cool.
I do not claim that they will be better or worse. All I say is, there are clear electrical differences that may or may not capable of raising any delays to the level of audibility. Utter dismissal of any effect based on a crude model of the wires and the load combined with an ignorance of human discernment capability is not what I call compete engineering.
However, the bog standard LCR model doesn't really address in a clear fashion how a varying impedance load will affect sonics. That's why I like the T-line, simplicity. I could model varying loads with different cable LCR bulk values, but there will be no clarity of understanding in comparison with T-line.
Falling magnitude and phase shift are of course better modelled with LCR, I would not recommend T-line for that.
In essence, I'm trying to pull the best of each model. Kinda like light with wave vs particles, each has practical aspects that are not readily useful for all applications.
jn
Last edited:
Since this is diyAudio forum, what would be useful is audible aspect of cable lifters. Have you done a double blind test of those lifters? Please share if you have. If not, why do you bother posting?
I disagree with many of the things John has said here, but they're always interesting and intelligent and make me think hard. So, I for one am quite glad that he's posting.
So now, you require that only people who have "done a double blind test" are qualified to engage in a discussion?? And if they haven't you are going to challenge them as "why do you bother posting"?
What next? When we discuss ohms law, you still going to require a "double blind test"?
jn
If only things were that easy...
Many many years ago (like early 90s?) the seating arrangement in my tiny apartment was such that I couldn't get to a nice sweet spot between the stereo speakers, so had to compensate with the balance control. At some point I acquired a little Sony Dolby Surround decoder box, which had among its features for some reason the ability to set it up as a simple full-bandwidth (well, 16/44 anyway) stereo delay. Cool, I thought - now I can properly address my seating offset. So I set about adjusting the delay on the near channel with a mono source until the direct sound from both speakers was once again correlated at my noggin. But now something else sounded wrong! I quickly realized that the reflected sound in the room was now decorrelated, which sounded noticeably not-right in its own unique way. Worse yet, this new problem of course existed at any listening location. (sigh) So back to balance control for me, and back to the closet for the Sony gizmo.
Since then I've never understood all the emphasis on exact interchannel delay settings with surround systems. I don't use it myself - it seems to solve one problem only to create a worse one.
Sorry to stray from the topic.
-- Jim
Last edited:
No, definitely not. My view has been that full TL treatment is unnecessary at audio frequencies. Correctly applied, it must give the same result as the more appropriate lumped version.jneutron said:
I'm not certain that "patiently" is a word I would use in this context. Perhaps others can judge this better than me.
Use a pi cell (or a T cell) instead of an L cell and this problem disappears.
Given human ITD capability, your view is incomplete.
The amp driving a capacitance directly is not consistent with reality.
The speaker shunted by a lumped capacitance is also inconsistent. T at least doesn't do that. But again, it doesn't come close to a 200 cell or a t line in the 2 to 5 uSEc domain.
For mag/phase, lumped is perfectly adequate.
For impedance mismatch and resultant delay issues vs load impedance, the lumped is insufficient.
jn
ps...an engineer pushing for an exact solution, a physicist settling for approximations... what's the world coming to..😀
Last edited:
You are missing the point. With audio products, such as the one mentioned in the title of this thread, when it's all said and done, it's got to answer that one question, the audibility. Without listening, how do you answer that?
Not a worry, extremely relevant in fact. Why I make such a fuss about what I call "convincing" sound is that then all this headachy stuff is taken care of, automatically, by the ear/brain. It's had a lifetime of getting on top of understanding what the subtle cues in the sounds that it picks up means, and can decipher precisely where the positioning of a sound source is, given enough information.
Where all this electronic fiddling fails, is in that it's an extremely crude crutch trying to mimic what our organic hearing does gracefully every second while we're awake, with all the normal sounds occurring around us, all the time. It sorts it all out, and neatly organises all the input. But our feeble attempts to parrot that electronically are bound to fail unless everything in the situation is "perfect" - hence the inglorious "sweet spot", etc.
Extremely clean reproduction ensures enough low level cues get through to the ear/brain, and then "it all makes sense". No crutches are required, and the "sweet spot" is everywhere and anywhere - no matter how jumbled up the sound is by the time it reaches our ears, enough correct information registers, and the hearing system decodes it, and it sounds 'right' ...
Much as I hate to be suckered back in, this is of course rubbish. The simulation is based on a model and parameters Scott provided for a cable - which not only seem reasonable, but which parameters and model JN has endorsed.
Audioband delays are trivial to validate using lumped models - it's acknowledged lumped and TL models produce identical results for audioband risetimes, not least by Davis and Greiner.
The notion that 5uS of ITD, said to be the threshold of audibility, might occur from cables in conjunction with load is just total fantasy to me. At 8R load 10kHz c 120nS delay, and even at 1R load only c 1uS.
If one mistakenly applies rf risetime signals to the cable, overshoot and ringing results and my previous post shows correct simulation of TL behaviour in this respect, and reality.
Attached are fresh lumped model simulations for 8R and 1R loads for audioband risetimes using Scott's model and parameters. It's easy to see there is no significant delay in the context of audible ITD IMO. Blue trace is applied signal, red is load current. X time, Y level.
Attachments
Last edited:
Probably because understanding how a signal travels from an amp to speaker would be critical in understanding the effect if any of cable lifters, its called education by debate.....
It can't be incomplete because we are dealing with band-limited signals, in the frequency region where zero or one lumped cells fully model the cable behaviour. Your 5us settling is just a filter response.jneutron said:
Band-limited signal. If shunt caps worry you then use a T cell.
That's good, because mag/phase is all we have. Mr. Fourier assures us of this.
Physicists are always taught to use the appropriate level of abstraction, the appropriate degree of approximation. My experience is that some engineers are not, so they can use inappropriate methods - either too complex (and potentially confuse themselves) or too simple (where the approximation is no longer valid - but in some cases they were never taught that it was an approximation!).
Yes, a simple filter response based on cable parasitics and load impedances. But where on earth is 5uS supposed to come from ? With the best will the time constant is not even close, it's far smaller. Time constant L/R where L is cable inductance R is the load..........surely this is a goose chase at best.
Last edited:
Don't forget that if you want '95% settling' then you need to wait for a few time constants to elapse. I can't remember the exact details, but let us suppose a cable with RF impedance of 100 ohms feeding a speaker which has dipped to 2 ohms. Cable length 5m.
Inductance will be somewhere around 0.4uH/m, so 2uH. Low impedance load means inductance dominates over capacitance. 2uH and 2ohms gives 1us time constant. Three time constants is 3us.
Inductance will be somewhere around 0.4uH/m, so 2uH. Low impedance load means inductance dominates over capacitance. 2uH and 2ohms gives 1us time constant. Three time constants is 3us.
Hmmm, but if one works with c 25uS audioband risetime, the 'delay at 95% point' is far far shorter, see my latest sim for 1R and 8R, albeit for c 1uH cables. 'Settle time' depends on rise time, and this makes a big difference here. Perhaps delay is c 0.1uS typ.
Plus, as you posted earlier on this thread, |2R| should be the description of any speaker impedance dip, for sure such load is complex and the effect of real speaker loads on current phase totally swamps any contribution from cables. It amuses me when speaker impedance is said to dip below the resistance of the voice coil.............
When working with audioband risetime and nominal speaker loads, any delay due to cable LR filter seems trivial and negligible.
Last edited:
- Status
- Not open for further replies.