Lets check this out, shall we? I have stated something contrary to this.
edit(I have hilited in red the final parameters of the cables, sorry for the color Bob..)
Andrews 2.25 inch dia heliax: data from catalog
C = 21.8 pf/ft
Z = 50 ohms
L from catalog us 1.87 uH per foot )totally incorrect of course, sigh...
V = 93%
Vmax = 13kV
Calc new inductance per foot
Z = Sqr(L/C)
50 = sqr(L/C)
2500 =L/C
2500*21.8 = L
L = 54500 pH, or 54.5 nH per foot. (actual inductance per foot)
Calculation of dielectric permittivity.
v = 1/sqr(epsilon)
vsquared =1/epsilon: epsilon =1/v squared
epsilon from manu prop speed: =1/v squared =1/(.93*.93) =1.15
Cross check...LC = 1034 epsilon..
epsilon =LC/1034 = (54.5 * 21.8 )/1034 = 1.149 good.
next cable, MIL M17/22-RG27
Z = 50 ohms
L = 2500 * C = 2500 * 50 = 125000 pH/ft = 125 nH/ft.
permittivity:
LC = 1034 epsilon
epsilon = LC/1034 = (125 * 50)/1034 = 6250 / 1034 = 6.04
V =1/sqr(epsilon) =1/sqr(6.04) =1/2.45 = .406....40% lightspeed.
So, lets make equivalent 10 element models.
For the heliax, 10 elements using L of 54.5nH and C of 21.8 this is a coax ten feet long.
To equate the RG27, scale the capacitance to match.
21.8/50 = .436... RG27 will have 21.8 pf per .436 feet.
RG27 has 125 nH per foot, times .436 feet equals 54.5 nH per .436 feet.
The exact same 10 lumped elements of RG27 will reflect a cable 4.36 feet long.
If we assume v is 1 nS/foot..
Heliax has a prop velocity of .93, RG27 has .4
10 feet of heliax will have a prop delay of 10 feet/.93 feet/nS, or 10.75 nSec end to end.
4.36 feet of RG27 will have a prop delay of 4.36 feet/.4 feet/ns, or 10.9 nSec end to end.
The error in calculation is certainly within expectations for back of napkin (bar napkin) calculations.
The statement:
Is absolutely born out by the calculations. I stand corrected in regard to my assertion that identical lumped models do not reflect (changes in permittivity of the dielectric) differences in prop v or cable length difference.
This does not address however, my concern that the lumped element analysis does not consider the physical propagation delay adequately, it still stands. It may be consistent with jcx's 28 nSec fudge factor..
jn.
Last edited:
Thanks for doing the check. The reason I was so certain is that a transmission line at a given frequency is defined simply by its propagation constant and characteristic impedance. The propagation constant always appears with the physical length; it translates from physical length to electrical length. It also includes the attenuation. There are no other parameters, so therefore the line is fully described by these two. Therefore any two lines with the same parameters will behave in the same way; electrically they are identical.
Of course, things can change as you change frequency but in most cases they change smoothly until the normal TEM model no longer applies.
A lumped line may or may not be a good approximation. That depends on how electrically long the line is in wavelengths.
Of course, things can change as you change frequency but in most cases they change smoothly until the normal TEM model no longer applies.
A lumped line may or may not be a good approximation. That depends on how electrically long the line is in wavelengths.
No prob. Accuracy is most important.
My concern is with behaviour far below wavelengths, the group delay jcx modelled.
Cyril Bateman shows about a 5 uSec delay (actual hardware, not sim) voltage delay at the source end for a cable driven at at 10Khz. He used an HP 8721 reflection bridge, and the scope waveform clearly shows the reflection of a 10 Khz sine as a result of line to load mismatch. I've no idea how the 8721 does it, but you can be sure I'll be downloading the manuals for that puppy.
Of note, the scope is showing the voltage reflection delay, it is 3 orders of magnitude larger than the end to end prop delay of the cable would indicate.
It is not the current going into the line by the way. Just the reflection of voltage.
Electrodynamic drivers with voice coils are current conversion devices. What is the current?
jn
ps. Thank you (unnamed individual) very much for the articles. It is great reading..the oscillation bursts at the peak of the waveform is also very interesting..
pps. I coulda just said I was mistaken, but where's the fun in that??
Last edited:
Sometimes protection diodes are from the input nodes to substrate. That is what I've seen on the physical chips under a metallurgical scope. They were to clamp inputs one diode drop past substrate potential, sometimes used to prevent latchup. I've never seen back to backs between the inputs, so can't tell you.
Sometimes the E/B junction is incapable of supporting more than a coupla hundred microamps reverse avalanche without altering permanently their parameters. Don't ask me the physical details, I've already a headache doin those t-line calcs..
Seriously, I don't think they worry too much about current profile during avalanche of an E/B junction, probably just worried about base thickness...don't wanna fail to "early", eh??.**It's one of those things like "hay doc it hurts when I do this".....well, then don't do that...
jn
** dont worry, I'll keep my day job..
Last edited:
As jcx said there are actually quite a few (we call them "H" bridge input) op-amps. They only gain you the slew rate advantage. Extra pins beyond shutdown or compensation are rare. We have some amps with a kelvin to the output.
JNeutron,
? Voltage or current devices, take your pick.....
By electrodynamic driver, I mean one in which the force induced within the voice coil is a function of the current flowing through it times the magnetic field (total flux times current conductor length) normal to the current of the coil.
jn
Mr. Wurcer,
Sounds formal doesn't it.... What determines when one of these opamps or even a discrete device is discontinued? Are we bound by the same rules that are applying to computer CPU speeds and the every increasing speeds of logic circuits or is there something else that determines the lifespan of one of these devices? If a device is meeting the purpose that it is designed for what changes are driving the obsolescence of what seem like perfectly capable devices. I only ask you because you seem to be intimately involved in their design.
Sounds formal doesn't it.... What determines when one of these opamps or even a discrete device is discontinued? Are we bound by the same rules that are applying to computer CPU speeds and the every increasing speeds of logic circuits or is there something else that determines the lifespan of one of these devices? If a device is meeting the purpose that it is designed for what changes are driving the obsolescence of what seem like perfectly capable devices. I only ask you because you seem to be intimately involved in their design.
This current vs voltage argument is pointless. Of course, it is BLI, so what?
Almost all loudspeaker systems are DESIGNED to operate properly with voltage drive.
This can go back to Rice and Kellogg who INVENTED the direct radiator speaker in about 1925. That is the only way that it can work optimally.
Individual drivers, of course can be current driven, just like servo motors usually are, but that is a special case.
Almost all loudspeaker systems are DESIGNED to operate properly with voltage drive.
This can go back to Rice and Kellogg who INVENTED the direct radiator speaker in about 1925. That is the only way that it can work optimally.
Individual drivers, of course can be current driven, just like servo motors usually are, but that is a special case.
Last edited:
Calling it pointless certainly doesn't make it so..
You state the obvious.
Focus, John.
When a 10Khz sine is sent from an amp into a wire which has an impedance different from line, there will be a reflection. This can be easily viewed using a directional bridge, like the 8721A unit. This measurement does not require a line anywhere near a wavelength, a few meters is easy.
As a result, the load will not be responding to the amplifier voltage as it is sent down the line, but something else.
How the cable alters it, that is the question. Measurement of the current within the line will be more indicative of what the speaker is responding to.
The current vs time will be impacted by the line. My primary concern is transient response and it's impact on soundstage. I am not interested per se in the steady state response you speak of. I generally become bored very quickly when listening to pure sines.
jn
Sheesh, don't you know anything about Agilent 8721A directional bridges? I've known about it for something upwards of .....10 minutes now...get with the program..
I'm interested by the response too, while i suspect the answer to be purely commercial.
So would the easy solution be the Mogami wire with 8 ohm impedance as shown in Nelson's article, perhaps terminated as also shown in article? Seems easier and more "attractive" than dressing 8 pairs of CAT5/6 cable, even though only subtle changes, sometimes, were heard between the various cables.
I'm actually facing to a choice, in order to mod an existing device composed of severall opamps. There are so many solution that my question is more philosophical than else and i had never made real test about various combinations.
On a sonic point of view, is-it better to use (when possible) the same OPA(with the same sonic character and, may-be some compromises ) or different ones (with different character but may-be, more adapted) when they are one after the others ?
On a sonic point of view, is-it better to use (when possible) the same OPA(with the same sonic character and, may-be some compromises ) or different ones (with different character but may-be, more adapted) when they are one after the others ?
Last edited:
Make what good?
I'd center the Z to minimize the difference in the delay. 20 ohms would seem ok to me. 4 pairs of #18 gets you down near #12awg. No need to go nuts.
jn
- Status
- Not open for further replies.