No, it's classic established physics. In this case TL theory is not even really an approximation, it's so close to reality and the lumped model as to be negligible. One can verify this by running simulation, comparing TL model with lumped model for a typical system at hand - doing this I obtain the following amplitude deviations between lumped and TL models :
1KHz 8.3 E-9 dB
10kHz 8.3 E -7 dB
100kHz 8.3 E-5 dB
Essentially no deviation in frequency or phase response nor any group delay in the audioband. So simulation confirms classic theory, I can hardly feign surprise..........there's no reason to think it would be any other way, certainly not heard one on this thread !
Actually you haven't detailed anything, jneutron.jneutron said:
What latency? There is none beyond the effect of lumped parameters, this is trivial to simulate, measure, and confirm with classic theory.
Are you really suggesting an 'effect' which apparently defies description cannot be observed either....... ?! It might as well not exist then, this seems like nonsense. In reality, the cable, in the context of the audio system at hand, simulates and measures as a plain piece of wire with lumped parameters. The reason nothing else can be observed is simply because there is nothing else to observe, isn't that what one might expect ?
It's impossible to mismatch such a short line as this, over the audio spectrum. Matching does not have a meaning, characteristic impedance does not have a meaning here. It's impossible for a delay of anything like 20 or 30uS to arise from a line such as this in this context, and unsurprisingly neither simulation nor measurement shows it.jneutron said:
5uS is half a geological era in this context, and for sure it should be readily observable on the line if present. But it isn't. I can only imagine you're confused between the speaker load impedance, which is complex (in the maths sense of embracing a time element) and the wire which isn't ? Makes no sense that the cable should interact whilst conveying an audio programme spectrum in any event, lumped or TL model, and IME it doesn't.
Last edited:
This makes no sense to me, for reasons I won't go over again many of the terms don't mean anything in context here. Thanks for the links, I'll take a look and try to decipher what it really might be you are measuring/simulating. I don't obtain anything like this. I strongly suspect something other than what you think is going on.
Last edited:
Actually, it requires something on the order of 200 lumped elements to get close to reality and the t-line. The article scott and I are referencing details that, unfortunately it is unpublished as far as I know.
There's your misunderstanding.
Who is discussing amplitude?
You assume no delays from amplitude data? Wow..
Sure I have, and you mention it in your next post..
again, the next post..
Nope. I suggest nothing.
What I state, is that you are incapable of measuring a 5 uSec delay of a 1Khz current signal in a low impedance loop.
And it's not your talent that is the problem. It's the equipment at your disposal.
That is incorrect on both counts. It has been measured by me, it has been simulated by scott. The only details needed are the cable parametrics.
You're not understanding this, are you?
You've never tried this, have you? In the context of a 20 foot long t-line, it is indeed a geologic era, but in the context of a 500 hz sine within a 4 ohm loop, it is a PITA to measure.
Have you tried to simulate the step response of a 20 foot 100 ohm line with an EDC of 10 into various load resistances? How did you do it?
jn
Last edited:
Yeah, aluminum only has the advantage where weight and conductivity matter equally (like overhead transmission lines and hf voice coils etc)
A few pointers for simulation of what I've done:
L is nH per foot. C is pf per foot.
EDC is the equivalent (relative) dielectric coefficient (permittivity).
A coax cable has the relationship LC = 1034 DC. It is a constrained system, the e field and the m field are entirely constrained to within the outer coaxial conductor. As such, DC is the permittivity of the dielectric. (it is actually epsilon relative times mu relative, we are assuming free space permeability.
A parallel pair has the relationship LC = 1034 EDC. Because it is an unconstrained system, the permittivity of the insulation is not the only thing affecting the line. The geometry of the conductors will determine how far externally the magnetic and electric fields will splay out of the system. The consequence of this unconstrained field behavior is an increase in inductance, and a decrease in capacitance.
The prop velocity is C/sqr(EDC) for the unconstrained, C/sqr(DC) for the constrained, and for generic, C/sqr(epsilonR MuR).
The line Z is of course, Z = sqr(L/C)
For my plot:
Z = 100
Load is from 1 to 20.
Source Z = 0
Vprop =C/sqr(10)
To duplicate the plots using LRC, solve for L and C using:
LC = 10340.
and SQR(L/C) = 100
Also of note is:
For a parallel pair, everybody ignores the fact that round conductors in the audio band have 15 nH per foot internal inductance.
And for coax, the inner core wire has 15 nH per foot as well in the audio band.
You can consider that as a constant across the audio band, as the skin effect approximation equation is totally bogus at these frequencies, so the internal inductance does NOT collapse below 20 Khz.
jn
Last edited:
And remember, be careful of the units. Many people end up with errors of a factor of 31.6. (square root of 1000)
jn
jn
Last edited:
All this math, gah!
Let's just use plate amplifiers with dsp, certainly we don't have to worry about the wire inside the cabinet, right?
+1 active!
🙂
Let's just use plate amplifiers with dsp, certainly we don't have to worry about the wire inside the cabinet, right?
+1 active!
🙂
The last plate amplifier I used had two plates at the output stage..
6L6GC plates...
And yah, some people do rant on about the wire inside the cabinet...sigh..
jn
Think I can piece together what's going on here. You are measuring load current and correctly observing various latencies versus the time reference of the driven voltage. If so, this is correctly an artefact of energy stored in the drive unit/cabinet. Impedance of real speakers is complex, ie has real part and a j time variable which can be large. The impedance of speakers is dynamic ie varies in real time because they can store energy - so instantaneous speaker impedance depends on recent history......it's how speaker impedance can 'dip' below the resistance of the voice coil which must involve negative contribution from stored energy BTW.....
So load current can definitely lag 'line' voltage, but that has nothing to do with cables or TLs. I know this because I design audio current drive amplifiers, so am thoroughly familiar with sensing load current and its time relation to original signals in the context of real speaker loads.
I will look up your stuff, jneutron but expect to confirm that what you are measuring is an artifact of the speaker box rather than the cable - at least that makes sense and is known. In the current domain, it's plain common sense that short lines don't delay current for 5uS.........
So suspect a Laithwaite gyroscope moment may have befallen you, jneutron? Classic physics is always unlikely to be wrong. But that's not bad company to keep in my book.
Actually, the risetime at the load. I never measured at the source, although it will only be one transit behind the load, which is of no importance.
If I had measured using a speaker, then the complexity of the load would really affect the measurement reliability. This is a pure resistance discussion so far.
The discussion has not come anywhere close to being at the level of using a cabinet yet, you haven't understood the theory yet.
Ah, that would probably explain the difference between my measurements of my 18 incher in free air vs locked VC.. 😱 Your expounding off on a tangent...
Ah good. Once you understand, you will gain. What are you using as a CVR? Caddocks? Dale NI's?
I no longer play with such terrible resistors. They have horrible b dot performance, even ignoring their self inductance and the caddock's proximity based current redistribution.
AGain, no speaker box. Pure resistance with self inductance below a nanohenry and no bdot error. BTW, where are you going to "look up" this stuff? Whenever I look for mismatch waveforms on small lines, I always get waveforms for loads greater than line. I've never, ever, found one with load<<line, nor with z source = 0.
You will eventually learn..perhaps someday you will actually try to measure real hardware? You do understand that your argument is claiming faster than light energy transmission, right?
I have a few items in the queue, but then I'll be setting up my unimat to drill some new resistor arrays, both single pass as well as pass through for my EDM work. Perhaps you are in a position to collaborate? I'm afraid the bulk of my previous measurements may involve a COI. I've given a few of my arrays away, but they were made with clad perf, so do not have very good thermal characteristics and at high slew rates, the spreading resistance tends to cause asymmetrical current sharing.
You crack me up. What I speak of is currently being used within the feedback system of a ~10 power 9 dollar machine, and another is in the works.
I'm not employed because of my good looks..much as I'd like to think so..🙁
PS..you do understand that this behavior has also been measured with a speaker consistent load, no? Cyril Bateman used a reflection bridge to view the reflection from a complex speaker load and a 4 meter cable.
I'll try to dig up the article if you're interested..
jn
Last edited:
To the OP:
You hear more of everything probably because you are paying attention. The blocks made you think "what do I hear now?" and your attention is more focussed.
The good news is that your system actually does sound as good as you think.
The bad news is that the lifters only triggered you to pay attention to it... or is that good news?
You hear more of everything probably because you are paying attention. The blocks made you think "what do I hear now?" and your attention is more focussed.
The good news is that your system actually does sound as good as you think.
The bad news is that the lifters only triggered you to pay attention to it... or is that good news?
What you're outlining simply can't happen with a resistive load, there will be some mistake. And yes please link the article.
Given that I've measured it, "can't happen" is entirely incorrect.
Given that I've modelled it, and it matches measurement, "can't happen" is entirely incorrect.
Given that scott independently verified that the equivalent LCR model ALSO matches reality, matches my T-line model, and matches my measurements, "can't happen" is entirely incorrect.
With a dynamic driver load as you've mentioned you do, you will be unable to discern a line delay of 5, 10 20, or 30 uSec from the reaction of the load. You do understand your argument falls apart, as you cannot separate the line's energy storage mechanism from that of the load.
I'm afraid I can't link the article, I don't know where it is. I have a copy on one of my computers, so I'll try to find it. I have a hard copy..I like paper..old school.
The article is by Cyril Bateman, the title of it is "Cables, Amplifiers and Speaker interactions, part 1. You'll know you have the right one when you look on page 7, figure 8. He has a scope waveform showing a delay of about 4 uSec between the incident and reflected wave of a 10-Khz signal into an "ESP replica crossover, a Kef T27 tweeter, and some generic bass driver. He used an HP 8721A IIRC.
Ah, googled it..
http://www.waynekirkwood.com/images/pdf/Cyril_Bateman/Bateman_Speaker_Amp_Interaction.pdf
jn
Last edited:
I claim the same about my own modelling and measurements, plus mine is backed by convention and indisputable established physics. The truth will out in these things, just ask Laithwaite...
Thanks for the link.
PS obviously will take some time to read, but as suspected prima facie based on a skim read the 'reflections' measured seem consistent with the simulated speaker load as per the effect I outlined - not an artefact of the cable.
Last edited:
You are saying there is a 5uS lag between current and applied potential in a resistive load.......
seems implausible that the measurement can be too hard for prosumer sound card AND audibly significant
I have posted group delay sims going back and forth from 16/44 audio .wav and shown fft can easily resolve ns of group delay
http://www.diyaudio.com/forums/analogue-source/245555-temporal-resolution-6.html#post3697702
and of course you are free to use 96 or 192k, 24 bits
seem like microseconds of audio frequency group delay deltas should be falling of a log easy to see
I have posted group delay sims going back and forth from 16/44 audio .wav and shown fft can easily resolve ns of group delay
http://www.diyaudio.com/forums/analogue-source/245555-temporal-resolution-6.html#post3697702
and of course you are free to use 96 or 192k, 24 bits
seem like microseconds of audio frequency group delay deltas should be falling of a log easy to see
Last edited:
Actually, I've no idea what you were modelling. You did say you measure current output of an amp, but you've not detailed anything which would indicate that you can measure what we discuss.
My pleasure. There is also a part two. I'm not sure if that one describes the modelling he did to get really close to actual. But it was 200 elements he had to go to for a realistic LCR model of a cable.
A reflection is energy being returned to the source. You cannot measure that using a resistor. You need to really read the article to understand the discussion.
At that, I still have my doubts about you understanding the concepts..But that's ok, we have time..
No silly. You still don't understand the discussion, do you. Your question exposes that.
Listen carefully. When the line and load are matched, there is NO settling time, no delay.
When the line and load are terribly mismatched, there will be a settling time which is dependent on the level of mismatch.
This is not rocket science.
jn
- Status
- Not open for further replies.