Coax modelling question

I can model a coax cable in LT spice using the inbuilt model or a lumped 
collection of R C and L.

But the cable I want to model has no overall screen insulation, and 
somewhere between most and hardly any of it is very tightly wound onto a 
drum, so the outer on the drum is all 'shorted' together.  The outer is 
steel and has the same order of resistance as the copper inner.

That means there is part of the cable which looks like, I suppose, a 
mass of steel with an embedded coil.

I don't have a cable to hand.

How would you go about modelling this?  I suppose that for the wound 
section, the R reduces, the C remains the same, but I'm not sure what to 
think about the L.

Cheers
-- 
Syd

On 13/09/16 10:51, Syd Rumpo wrote:
> I can model a coax cable in LT spice using the inbuilt model or a lumped
> collection of R C and L.
>
> But the cable I want to model has no overall screen insulation, and
> somewhere between most and hardly any of it is very tightly wound onto a
> drum, so the outer on the drum is all 'shorted' together.  The outer is
> steel and has the same order of resistance as the copper inner.
>
> That means there is part of the cable which looks like, I suppose, a
> mass of steel with an embedded coil.
>
> I don't have a cable to hand.
>
> How would you go about modelling this?  I suppose that for the wound
> section, the R reduces, the C remains the same, but I'm not sure what to
> think about the L.
>
> Cheers

Transmission lines are usually modeled as a two-port. In a two-port, you 
get relations among V1, I1, V2 and I2 but you have no information on 
other variables, for instance VDB.

A------[     ]------C
   V1   [ box ]    V2
B------[     ]------D

You can have a situation where B and D are the same physical point (some 
baluns did that if memory is still ok) or you can have a situation where 
the coax shield forms a second transmission line with another conductor 
(perhaps "ground").

In your case, the differential signals V2 and V1 would be the same. The 
shorting would have an effect on the second transmission line. You would 
probably get a big capacitance between B and D and the true "ground"

Pere

On Sep 13, 2016, Syd Rumpo wrote
(in SED article <nr8eli$klu$1@dont-email.me>):

> I can model a coax cable in LT spice using the inbuilt model or a lumped
> collection of R C and L.
> 
> But the cable I want to model has no overall screen insulation, and
> somewhere between most and hardly any of it is very tightly wound onto a
> drum, so the outer on the drum is all 'shorted' together. The outer is
> steel and has the same order of resistance as the copper inner.
> 
> That means there is part of the cable which looks like, I suppose, a
> mass of steel with an embedded coil.
> 
> I don't have a cable to hand.
> 
> How would you go about modelling this? I suppose that for the wound
> section, the R reduces, the C remains the same, but I'm not sure what to
> think about the L.
Transformer and lumped-element models fail here. The issue is that above a 
critical frequency, outside and inside currents decouple due to skin effect. 
In other words, above that frequency, shorting the outsides of the shields by 
winding the coax on a metal rod has no effect on the currents on the inside 
of the shield, even if you solder coax shield to rod.

Joe Gwinn

On 09/13/2016 10:02 AM, Joseph Gwinn wrote:
> On Sep 13, 2016, Syd Rumpo wrote
> (in SED article <nr8eli$klu$1@dont-email.me>):
> 
>> I can model a coax cable in LT spice using the inbuilt model or a lumped
>> collection of R C and L.
>>
>> But the cable I want to model has no overall screen insulation, and
>> somewhere between most and hardly any of it is very tightly wound onto a
>> drum, so the outer on the drum is all 'shorted' together. The outer is
>> steel and has the same order of resistance as the copper inner.
>>
>> That means there is part of the cable which looks like, I suppose, a
>> mass of steel with an embedded coil.
>>
>> I don't have a cable to hand.
>>
>> How would you go about modelling this? I suppose that for the wound
>> section, the R reduces, the C remains the same, but I'm not sure what to
>> think about the L.
> Transformer and lumped-element models fail here. The issue is that above a 
> critical frequency, outside and inside currents decouple due to skin effect. 
> In other words, above that frequency, shorting the outsides of the shields by 
> winding the coax on a metal rod has no effect on the currents on the inside 
> of the shield, even if you solder coax shield to rod.
> 
> Joe Gwinn
> 

The LTspice transmission line model fails even worse, because iirc it
acts as a completely isolated, time-delayed source, as though it had a
perfect balun attached.

Cheers

Phil Hobbs

-- 
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

On Tuesday, September 13, 2016 at 10:16:22 AM UTC-4, Phil Hobbs wrote:
> On 09/13/2016 10:02 AM, Joseph Gwinn wrote:
> > On Sep 13, 2016, Syd Rumpo wrote
> > (in SED article <nr8eli$klu$1@dont-email.me>):
> > 
> >> I can model a coax cable in LT spice using the inbuilt model or a lumped
> >> collection of R C and L.
> >>
> >> But the cable I want to model has no overall screen insulation, and
> >> somewhere between most and hardly any of it is very tightly wound onto a
> >> drum, so the outer on the drum is all 'shorted' together. The outer is
> >> steel and has the same order of resistance as the copper inner.
> >>
> >> That means there is part of the cable which looks like, I suppose, a
> >> mass of steel with an embedded coil.
> >>
> >> I don't have a cable to hand.
> >>
> >> How would you go about modelling this? I suppose that for the wound
> >> section, the R reduces, the C remains the same, but I'm not sure what to
> >> think about the L.
> > Transformer and lumped-element models fail here. The issue is that above a 
> > critical frequency, outside and inside currents decouple due to skin effect. 
> > In other words, above that frequency, shorting the outsides of the shields by 
> > winding the coax on a metal rod has no effect on the currents on the inside 
> > of the shield, even if you solder coax shield to rod.
> > 
> > Joe Gwinn
> > 
> 
> The LTspice transmission line model fails even worse, because iirc it
> acts as a completely isolated, time-delayed source, as though it had a
> perfect balun attached.
> 
> Cheers
> 
> Phil Hobbs
> 
true

so the OP can use this model and add a resistor in series with the two output leads to simulate the wire resistance and then short together the input and output shield  ends to simulate the fact that it is would on a roll.

  m

On 09/13/2016 10:58 AM, makolber@yahoo.com wrote:
> On Tuesday, September 13, 2016 at 10:16:22 AM UTC-4, Phil Hobbs
> wrote:
>> On 09/13/2016 10:02 AM, Joseph Gwinn wrote:
>>> On Sep 13, 2016, Syd Rumpo wrote (in SED article
>>> <nr8eli$klu$1@dont-email.me>):
>>> 
>>>> I can model a coax cable in LT spice using the inbuilt model or
>>>> a lumped collection of R C and L.
>>>> 
>>>> But the cable I want to model has no overall screen insulation,
>>>> and somewhere between most and hardly any of it is very tightly
>>>> wound onto a drum, so the outer on the drum is all 'shorted'
>>>> together. The outer is steel and has the same order of
>>>> resistance as the copper inner.
>>>> 
>>>> That means there is part of the cable which looks like, I
>>>> suppose, a mass of steel with an embedded coil.
>>>> 
>>>> I don't have a cable to hand.
>>>> 
>>>> How would you go about modelling this? I suppose that for the
>>>> wound section, the R reduces, the C remains the same, but I'm
>>>> not sure what to think about the L.
>>> Transformer and lumped-element models fail here. The issue is
>>> that above a critical frequency, outside and inside currents
>>> decouple due to skin effect. In other words, above that
>>> frequency, shorting the outsides of the shields by winding the
>>> coax on a metal rod has no effect on the currents on the inside 
>>> of the shield, even if you solder coax shield to rod.
>>> 
>>> Joe Gwinn
>>> 
>> 
>> The LTspice transmission line model fails even worse, because iirc
>> it acts as a completely isolated, time-delayed source, as though it
>> had a perfect balun attached.
>> 
>> Cheers
>> 
>> Phil Hobbs
>> 
> true
> 
> so the OP can use this model and add a resistor in series with the
> two output leads to simulate the wire resistance and then short
> together the input and output shield  ends to simulate the fact that
> it is would on a roll.
> 

I don't think that works, because as Joe pointed out, the inner and
outer surfaces of the shield decouple at high frequency, and unbalanced
currents see a lot of inductance.  A steel shield will decouple sooner
than copper, because there's a factor of 1/sqrt(mu) in the equation for
skin depth. (Of course if it's nonmagnetic stainless, that doesn't help.)

At low frequency, much of the ground current would flow across the turns
of the coil because of the reduced resistance, but this would diminish a
lot at higher frequency.

Cheers

Phil Hobbs

-- 
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

On Tue, 13 Sep 2016 10:16:13 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

>On 09/13/2016 10:02 AM, Joseph Gwinn wrote:
>> On Sep 13, 2016, Syd Rumpo wrote
>> (in SED article <nr8eli$klu$1@dont-email.me>):
>> 
>>> I can model a coax cable in LT spice using the inbuilt model or a lumped
>>> collection of R C and L.
>>>
>>> But the cable I want to model has no overall screen insulation, and
>>> somewhere between most and hardly any of it is very tightly wound onto a
>>> drum, so the outer on the drum is all 'shorted' together. The outer is
>>> steel and has the same order of resistance as the copper inner.
>>>
>>> That means there is part of the cable which looks like, I suppose, a
>>> mass of steel with an embedded coil.
>>>
>>> I don't have a cable to hand.
>>>
>>> How would you go about modelling this? I suppose that for the wound
>>> section, the R reduces, the C remains the same, but I'm not sure what to
>>> think about the L.
>> Transformer and lumped-element models fail here. The issue is that above a 
>> critical frequency, outside and inside currents decouple due to skin effect. 
>> In other words, above that frequency, shorting the outsides of the shields by 
>> winding the coax on a metal rod has no effect on the currents on the inside 
>> of the shield, even if you solder coax shield to rod.
>> 
>> Joe Gwinn
>> 
>
>The LTspice transmission line model fails even worse, because iirc it
>acts as a completely isolated, time-delayed source, as though it had a
>perfect balun attached.

The LT Spice txline makes a nice 1:1 dc-daylight isolation
transformer.

I don't think it qualifies as fails. If they didn't do it their way,
they would have to separately model the shield-to-universe impedance,
which would be hard to do. Or make it a 3-wire device. Or something.
As-is, you can add your own shield model, or just ground the shield
pins on both ends.



-- 

John Larkin         Highland Technology, Inc

lunatic fringe electronics 

On Tue, 13 Sep 2016 09:51:04 +0100, Syd Rumpo <usenet@nononono.co.uk>
wrote:

>I can model a coax cable in LT spice using the inbuilt model or a lumped 
>collection of R C and L.
>
>But the cable I want to model has no overall screen insulation, and 
>somewhere between most and hardly any of it is very tightly wound onto a 
>drum, so the outer on the drum is all 'shorted' together.  The outer is 
>steel and has the same order of resistance as the copper inner.
>
>That means there is part of the cable which looks like, I suppose, a 
>mass of steel with an embedded coil.
>
>I don't have a cable to hand.
>
>How would you go about modelling this?  I suppose that for the wound 
>section, the R reduces, the C remains the same, but I'm not sure what to 
>think about the L.
>
>Cheers

Are you talking about a spool of insulated wire inside a metal bucket?

That would have various amounts of wire-wire coupling, which will
cause all sorts of nasty effects.

A sketch would help.

"somewhere between most and hardly any" makes modeling difficult.

What is this for?


-- 

John Larkin         Highland Technology, Inc

lunatic fringe electronics 

On 09/13/2016 11:31 AM, John Larkin wrote:
> On Tue, 13 Sep 2016 10:16:13 -0400, Phil Hobbs
> <pcdhSpamMeSenseless@electrooptical.net> wrote:
> 
>> On 09/13/2016 10:02 AM, Joseph Gwinn wrote:
>>> On Sep 13, 2016, Syd Rumpo wrote
>>> (in SED article <nr8eli$klu$1@dont-email.me>):
>>>
>>>> I can model a coax cable in LT spice using the inbuilt model or a lumped
>>>> collection of R C and L.
>>>>
>>>> But the cable I want to model has no overall screen insulation, and
>>>> somewhere between most and hardly any of it is very tightly wound onto a
>>>> drum, so the outer on the drum is all 'shorted' together. The outer is
>>>> steel and has the same order of resistance as the copper inner.
>>>>
>>>> That means there is part of the cable which looks like, I suppose, a
>>>> mass of steel with an embedded coil.
>>>>
>>>> I don't have a cable to hand.
>>>>
>>>> How would you go about modelling this? I suppose that for the wound
>>>> section, the R reduces, the C remains the same, but I'm not sure what to
>>>> think about the L.
>>> Transformer and lumped-element models fail here. The issue is that above a 
>>> critical frequency, outside and inside currents decouple due to skin effect. 
>>> In other words, above that frequency, shorting the outsides of the shields by 
>>> winding the coax on a metal rod has no effect on the currents on the inside 
>>> of the shield, even if you solder coax shield to rod.
>>>
>>> Joe Gwinn
>>>
>>
>> The LTspice transmission line model fails even worse, because iirc it
>> acts as a completely isolated, time-delayed source, as though it had a
>> perfect balun attached.
> 
> The LT Spice txline makes a nice 1:1 dc-daylight isolation
> transformer.
> 
> I don't think it qualifies as fails. If they didn't do it their way,
> they would have to separately model the shield-to-universe impedance,
> which would be hard to do. Or make it a 3-wire device. Or something.
> As-is, you can add your own shield model, or just ground the shield
> pins on both ends.

I didn't mean that it failed altogether, just that I don't think it will
work in the OP's application.  I don't immediately see how you could use
it to model a coax balun, even.

Cheers

Phil Hobbs


-- 
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

On Tue, 13 Sep 2016 11:37:37 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

>On 09/13/2016 11:31 AM, John Larkin wrote:
>> On Tue, 13 Sep 2016 10:16:13 -0400, Phil Hobbs
>> <pcdhSpamMeSenseless@electrooptical.net> wrote:
>> 
>>> On 09/13/2016 10:02 AM, Joseph Gwinn wrote:
>>>> On Sep 13, 2016, Syd Rumpo wrote
>>>> (in SED article <nr8eli$klu$1@dont-email.me>):
>>>>
>>>>> I can model a coax cable in LT spice using the inbuilt model or a lumped
>>>>> collection of R C and L.
>>>>>
>>>>> But the cable I want to model has no overall screen insulation, and
>>>>> somewhere between most and hardly any of it is very tightly wound onto a
>>>>> drum, so the outer on the drum is all 'shorted' together. The outer is
>>>>> steel and has the same order of resistance as the copper inner.
>>>>>
>>>>> That means there is part of the cable which looks like, I suppose, a
>>>>> mass of steel with an embedded coil.
>>>>>
>>>>> I don't have a cable to hand.
>>>>>
>>>>> How would you go about modelling this? I suppose that for the wound
>>>>> section, the R reduces, the C remains the same, but I'm not sure what to
>>>>> think about the L.
>>>> Transformer and lumped-element models fail here. The issue is that above a 
>>>> critical frequency, outside and inside currents decouple due to skin effect. 
>>>> In other words, above that frequency, shorting the outsides of the shields by 
>>>> winding the coax on a metal rod has no effect on the currents on the inside 
>>>> of the shield, even if you solder coax shield to rod.
>>>>
>>>> Joe Gwinn
>>>>
>>>
>>> The LTspice transmission line model fails even worse, because iirc it
>>> acts as a completely isolated, time-delayed source, as though it had a
>>> perfect balun attached.
>> 
>> The LT Spice txline makes a nice 1:1 dc-daylight isolation
>> transformer.
>> 
>> I don't think it qualifies as fails. If they didn't do it their way,
>> they would have to separately model the shield-to-universe impedance,
>> which would be hard to do. Or make it a 3-wire device. Or something.
>> As-is, you can add your own shield model, or just ground the shield
>> pins on both ends.
>
>I didn't mean that it failed altogether, just that I don't think it will
>work in the OP's application.  I don't immediately see how you could use
>it to model a coax balun, even.
>

It already models an ideal coax balun!

In the non-ideal case, one would connect something between the ground
pins of the two ends, to simulate the common-mode impedance. That
could be an inductor, or even another transmission line. And you could
add Z to ground on both ends, to sim coupling of the shield to the
universe. That is all far more complex than the simple Heaviside model
of a coax.

I have broken a txline up into a lot of short segments, so that I can
add distributed-to-ground capacitance along the line.

The LT Spice lines are fun, in that you can play games with the shield
currents. I could post a fun circuit, except that I'm still trying to
sell it.


-- 

John Larkin         Highland Technology, Inc

lunatic fringe electronics