Spicing twisted pair

[John Larkin]

I may have posted this before, but the issue of modeling a shielded
twisted pair has come up again.

https://www.dropbox.com/s/62soxgyi8zxd3it/Twist_1.jpg?raw=1

It needs three txlines. This model checks well against a TDR of the
real cable.


--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Phil Hobbs]

On 7/11/19 4:50 PM, John Larkin wrote:

On Thu, 11 Jul 2019 13:42:44 -0700, John Larkin
jjlarkin@highland_snip_technology.com> wrote:


I may have posted this before, but the issue of modeling a shielded
twisted pair has come up again.

https://www.dropbox.com/s/62soxgyi8zxd3it/Twist_1.jpg?raw=1

It needs three txlines. This model checks well against a TDR of the
real cable.

It is probably not true, in a typical cable, that the differential and
commom-mode velocities are equal. That makes things a lot messier.

It's a good thing, though, because if they were the crosstalk would be a
lot worse.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[John Larkin]

On Thu, 11 Jul 2019 13:42:44 -0700, John Larkin
<jjlarkin@highland_snip_technology.com> wrote:

I may have posted this before, but the issue of modeling a shielded
twisted pair has come up again.

https://www.dropbox.com/s/62soxgyi8zxd3it/Twist_1.jpg?raw=1

It needs three txlines. This model checks well against a TDR of the
real cable.

It is probably not true, in a typical cable, that the differential and
commom-mode velocities are equal. That makes things a lot messier.


--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[John Larkin]

On Thu, 11 Jul 2019 17:20:47 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 7/11/19 4:50 PM, John Larkin wrote:
On Thu, 11 Jul 2019 13:42:44 -0700, John Larkin
jjlarkin@highland_snip_technology.com> wrote:


I may have posted this before, but the issue of modeling a shielded
twisted pair has come up again.

https://www.dropbox.com/s/62soxgyi8zxd3it/Twist_1.jpg?raw=1

It needs three txlines. This model checks well against a TDR of the
real cable.

It is probably not true, in a typical cable, that the differential and
commom-mode velocities are equal. That makes things a lot messier.


It's a good thing, though, because if they were the crosstalk would be a
lot worse.

Cheers

Phil Hobbs

In a multipair cable, like CAT5 UTP, the pairs generally have
different twists to scatter the velocities and minimize crosstalk.

https://www.dropbox.com/s/m74vvxknufqvs5l/Hitachi_Pairs.JPG?raw=1

But envision putting a fast photodiode at one end of a shielded pair,
and doing a single-ended bootstrapped drive at the other end. I
basically refused to do that.


--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Clifford Heath]

On 12/7/19 6:42 am, John Larkin wrote:

I may have posted this before, but the issue of modeling a shielded
twisted pair has come up again.

https://www.dropbox.com/s/62soxgyi8zxd3it/Twist_1.jpg?raw=1

It needs three txlines. This model checks well against a TDR of the
real cable.

Did you need those 1 milli-ohm resistors? Why? Did you mean Meg?

 

[Phil Hobbs]

On 7/11/19 8:51 PM, John Larkin wrote:

On Thu, 11 Jul 2019 17:20:47 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 7/11/19 4:50 PM, John Larkin wrote:
On Thu, 11 Jul 2019 13:42:44 -0700, John Larkin
jjlarkin@highland_snip_technology.com> wrote:


I may have posted this before, but the issue of modeling a shielded
twisted pair has come up again.

https://www.dropbox.com/s/62soxgyi8zxd3it/Twist_1.jpg?raw=1

It needs three txlines. This model checks well against a TDR of the
real cable.

It is probably not true, in a typical cable, that the differential and
commom-mode velocities are equal. That makes things a lot messier.


It's a good thing, though, because if they were the crosstalk would be a
lot worse.

Cheers

Phil Hobbs

In a multipair cable, like CAT5 UTP, the pairs generally have
different twists to scatter the velocities and minimize crosstalk.

https://www.dropbox.com/s/m74vvxknufqvs5l/Hitachi_Pairs.JPG?raw=1

But envision putting a fast photodiode at one end of a shielded pair,
and doing a single-ended bootstrapped drive at the other end. I
basically refused to do that.

Wise. Putting photodiodes on cables is so tempting to some folks, but
is one of the many roads to perdition.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[John Larkin]

On Fri, 12 Jul 2019 11:34:47 +1000, Clifford Heath
<no.spam@please.net> wrote:

On 12/7/19 6:42 am, John Larkin wrote:

I may have posted this before, but the issue of modeling a shielded
twisted pair has come up again.

https://www.dropbox.com/s/62soxgyi8zxd3it/Twist_1.jpg?raw=1

It needs three txlines. This model checks well against a TDR of the
real cable.

Did you need those 1 milli-ohm resistors? Why? Did you mean Meg?

Those are just there for ease of snooping currents.

The 1M things keep LT Spice from complaining about a floating node.
But the latest version doesn't seem to mind.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

On Friday, 12 July 2019 03:03:59 UTC+1, Phil Hobbs wrote:

On 7/11/19 8:51 PM, John Larkin wrote:
On Thu, 11 Jul 2019 17:20:47 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:
On 7/11/19 4:50 PM, John Larkin wrote:
On Thu, 11 Jul 2019 13:42:44 -0700, John Larkin
jjlarkin@highland_snip_technology.com> wrote:


I may have posted this before, but the issue of modeling a shielded
twisted pair has come up again.

https://www.dropbox.com/s/62soxgyi8zxd3it/Twist_1.jpg?raw=1

It needs three txlines. This model checks well against a TDR of the
real cable.

It is probably not true, in a typical cable, that the differential and
commom-mode velocities are equal. That makes things a lot messier.


It's a good thing, though, because if they were the crosstalk would be a
lot worse.

Cheers

Phil Hobbs

In a multipair cable, like CAT5 UTP, the pairs generally have
different twists to scatter the velocities and minimize crosstalk.

https://www.dropbox.com/s/m74vvxknufqvs5l/Hitachi_Pairs.JPG?raw=1

But envision putting a fast photodiode at one end of a shielded pair,
and doing a single-ended bootstrapped drive at the other end. I
basically refused to do that.

Wise. Putting photodiodes on cables is so tempting to some folks, but
is one of the many roads to perdition.

Cheers

Phil Hobbs

I think that's where Perdita got lost.


NT

 

[Bill Sloman]

On Friday, July 12, 2019 at 4:03:59 AM UTC+2, Phil Hobbs wrote:

On 7/11/19 8:51 PM, John Larkin wrote:
On Thu, 11 Jul 2019 17:20:47 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 7/11/19 4:50 PM, John Larkin wrote:
On Thu, 11 Jul 2019 13:42:44 -0700, John Larkin
jjlarkin@highland_snip_technology.com> wrote:


I may have posted this before, but the issue of modeling a shielded
twisted pair has come up again.

https://www.dropbox.com/s/62soxgyi8zxd3it/Twist_1.jpg?raw=1

It needs three txlines. This model checks well against a TDR of the
real cable.

It is probably not true, in a typical cable, that the differential and
commom-mode velocities are equal. That makes things a lot messier.


It's a good thing, though, because if they were the crosstalk would be a
lot worse.

Cheers

Phil Hobbs

In a multipair cable, like CAT5 UTP, the pairs generally have
different twists to scatter the velocities and minimize crosstalk.

https://www.dropbox.com/s/m74vvxknufqvs5l/Hitachi_Pairs.JPG?raw=1

But envision putting a fast photodiode at one end of a shielded pair,
and doing a single-ended bootstrapped drive at the other end. I
basically refused to do that.

Wise. Putting photodiodes on cables is so tempting to some folks, but
is one of the many roads to perdition.

When I did it, I put common base transistors on the end of the cable, and hooked the photo-diodes up to the collectors.

The write-up talked about pseudo-terminating the cable. The currents and voltages differed by the right amount between the on- and off-states to look as if the cable had been terminated with it's characteristic impedance, but the impedance didn't look resistive during the transition.

We were trying to get 10MHz data transfers over 18 metres of shielded ribbon cable, and got it. The previous arrangement had been designed for 5MHz and 6 metres of cable, and hadn't been doing well at the higher rates (for a variety of reasons).

The boss got my set-up to work at 12MHz, which scared me silly - it had been toleranced for 10MHz, and I hadn't done the work to be sure that the worst case devices would still work at 12MHz.

One did fail in the field, but it turned out that one of the HP optocouplers involved wasn't meeting it's data-sheet spec. Replacing that single device solved the problem, but the (slightly psychopathic) manager involved got another engineer to repeat my tolerancing, and he came to the same conclusions that I had.

--
Bill Sloman, Sydney