Topics

Coax BNC Cable Termination


james89es@...
 

A few months ago, Dean Huster had included the following in a post:

"You don´t HAVE to use a termination on coax if it´s a low-frequency
signal. For high-frequency or fast-pulses, yes, you must terminate".

Can someone comment on how low is low frequency? What happens if you
try to use no termination at too high a fequency?

-Jim


John Rehwinkel <spam@...>
 

Jim asks:

"You don´t HAVE to use a termination on coax if it´s a
low-frequency signal. For high-frequency or fast-pulses, yes, you
must terminate".

Can someone comment on how low is low frequency? What happens if you
try to use no termination at too high a fequency?
Basically, the phenomenon you're trying to avoid here is signal
reflection. What coax does is provide a path with a known
impedance to carry the signal from point to point without
changing it (essentially). In order to do this, the coax
(which has its own characteristic impedance, such as 50 ohms)
wants to look into the same impedance, both for maximum
signal transfer, and more importantly, to avoid reflections.
A signal bounces off of a sudden change in impedance, so if
you have a piece of 50 ohm coax connected to a 1 megohm 'scope
input, you'll get a reflection. These reflections change the
shape of the waveform and smear it out.

A generally accepted rule of thumb is that if a section of
mismatched impedance is less than a tenth of a wavelength
long (at the highest frequency of interest), the reflections
will be minor enough not to matter much.

A 100MHz signal has a wavelength of about 3 meters, so if
you run it through an unterminated piece of coax that is
less than 0.3 meters long, the signal shouldn't be affected
much. Note that if you're viewing a 100MHz square wave, it
has components that are much higher than 100MHz, so figure
accordingly.

Similarly, a 10MHz signal should be okay through three meters
of mismatched coax (long enough for most use). 1MHz signals
should tolerate a 30 meter long mismatch, and so on. Naturally,
if you really need accurate waveform measurements, you should
avoid mismatched feedlines entirely. But for most use, under
10MHz or so, any lashup will basically work.

Note also that terminating a signal is pretty easy, a 51 ohm
resistor to ground will do a passible job (or pick up one of
those fancy 50 ohm feedthru BNC terminators and put it on your
'scope input). Two 100 ohm noninductive resistors in parallel
and opposite each other will work even better. A quick way
is to use a BNC tee and a ThinNet terminator -- I have these
in my junk box, other folks might too. Works just fine for
most situations.

-- John Rehwinkel KG4L
spam@...


John Miles <jmiles@...>
 

A quick way
is to use a BNC tee and a ThinNet terminator -- I have these
in my junk box, other folks might too. Works just fine for
most situations.
Be careful using "network-grade" coax terminators for high-frequency RF.
The ones I've seen become reactive as heck above 50 - 100 MHz or so.

Unless you have a way to check their return loss at the frequencies you're
concerned with, I would steer far, far away from 10Base2/ThinNet
terminators in RF lab work.

-- jm


F F <ferfons@...>
 

From: "John Rehwinkel" <spam@...>
Reply-To: TekScopes@...
To: TekScopes@...
Subject: Re: [TekScopes] Coax BNC Cable Termination
Date: Thu, 5 Jul 2001 17:39:29 -0400
(snip)

1MHz signals
should tolerate a 30 meter long mismatch, and so on.
My humble twopence: Crappy rule of thumb! Few years ago while integrating an ATE we had a 1MHz square signal we thought it wouldn't need matching pretty much for the same reasons. Oh we were so bloody wrong and it took a while to figure it out. We did have to match that particular stretch of cable - 3 or 4 "lousy" meters of coax - down to 50 ohms (TDS420 alterative Zin) and through line drivers, keeping the signal below the scope's max input power rating. That's how we got it sorted.

Fernando
Portsmouth, UK




_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.


Miroslav Pokorni <mpokorni@...>
 

That rule of thumb is not crappy, but there is a caveat: an unterminated
cable appears as a nasty, predominantly capacitive impedance; for your
run-of-the-mill, polyethylene cable, capacitance is 30 pF/ft, what would
amount to 3000 pF for 30 meter piece. If your driver can handle that kind of
load, you are OK, but most drivers, without special measures, would
oscillate, if not fall flat on their face. Special measures are usually
adding a resistor between output and load, sometimes trying to compensate
phase, whatever does it take to survive.

So, when you transfer 1 MHz signal on a 30 meter cable you want see much of
reflection effects. Rule of thumb says significantly smaller that ¼
wavelength; 1 MHz in air is 300 meters, in polyethylene around 600 meters
what yields 1/20 of wavelength; that is better than ¼ of ¼ wavelength.
However, cable driver might be a problem.



Regards

Miroslav Pokorni

-----Original Message-----
From: F F [mailto:ferfons@...]
Sent: Friday, July 06, 2001 1:42 AM
To: TekScopes@...
Subject: Re: [TekScopes] Coax BNC Cable Termination

>From: "John Rehwinkel" <spam@...
<mailto:spam@...> >
>Reply-To: TekScopes@...
<mailto:TekScopes@...>
>To: TekScopes@...
<mailto:TekScopes@...>
>Subject: Re: [TekScopes] Coax BNC Cable
Termination
>Date: Thu, 5 Jul 2001 17:39:29 -0400

(snip)
1MHz signals
>should tolerate a 30 meter long mismatch, and so
on.

My humble twopence: Crappy rule of thumb! Few years ago
while integrating an ATE we had a 1MHz square signal we thought it wouldn't
need matching pretty much for the same reasons. Oh we were so bloody wrong
and it took a while to figure it out. We did have to match that particular
stretch of cable - 3 or 4 "lousy" meters of coax - down to 50 ohms (TDS420
alterative Zin) and through line drivers, keeping the signal below the
scope's max input power rating. That's how we got it sorted.
Fernando
Portsmouth, UK





_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at
http://www.hotmail.com <http://www.hotmail.com> .

To unsubscribe from this group, send an email to:
TekScopes-unsubscribe@...
<mailto:TekScopes-unsubscribe@...>


Your use of Yahoo! Groups is subject to
http://docs.yahoo.com/info/terms/ <http://docs.yahoo.com/info/terms/>


F F <ferfons@...>
 

Your thesis is entirely based on a 30-metre stretch of cable. However if you had carefully read my email, the lenght in question was quite inferior. A mere 5-metre chunk of cable if that much.

Regards

Fernando
Portsmouth, UK

From: Miroslav Pokorni <mpokorni@...>
Reply-To: TekScopes@...
To: "'TekScopes@...'" <TekScopes@...>
Subject: RE: [TekScopes] Coax BNC Cable Termination
Date: Mon, 9 Jul 2001 17:24:55 -0700

That rule of thumb is not crappy, but there is a caveat: an unterminated
cable appears as a nasty, predominantly capacitive impedance; for your
run-of-the-mill, polyethylene cable, capacitance is 30 pF/ft, what would
amount to 3000 pF for 30 meter piece. If your driver can handle that kind of
load, you are OK, but most drivers, without special measures, would
oscillate, if not fall flat on their face. Special measures are usually
adding a resistor between output and load, sometimes trying to compensate
phase, whatever does it take to survive.

So, when you transfer 1 MHz signal on a 30 meter cable you want see much of
reflection effects. Rule of thumb says significantly smaller that �
wavelength; 1 MHz in air is 300 meters, in polyethylene around 600 meters
what yields 1/20 of wavelength; that is better than � of � wavelength.
However, cable driver might be a problem.



Regards

Miroslav Pokorni



-----Original Message-----
From: F F [mailto:ferfons@...]
Sent: Friday, July 06, 2001 1:42 AM
To: TekScopes@...
Subject: Re: [TekScopes] Coax BNC Cable Termination

>From: "John Rehwinkel" <spam@...
<mailto:spam@...> >
>Reply-To: TekScopes@...
<mailto:TekScopes@...>
>To: TekScopes@...
<mailto:TekScopes@...>
>Subject: Re: [TekScopes] Coax BNC Cable
Termination
>Date: Thu, 5 Jul 2001 17:39:29 -0400

(snip)
1MHz signals
>should tolerate a 30 meter long mismatch, and so
on.

My humble twopence: Crappy rule of thumb! Few years ago
while integrating an ATE we had a 1MHz square signal we thought it wouldn't
need matching pretty much for the same reasons. Oh we were so bloody wrong
and it took a while to figure it out. We did have to match that particular
stretch of cable - 3 or 4 "lousy" meters of coax - down to 50 ohms (TDS420
alterative Zin) and through line drivers, keeping the signal below the
scope's max input power rating. That's how we got it sorted.
Fernando
Portsmouth, UK





_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at
http://www.hotmail.com <http://www.hotmail.com> .

To unsubscribe from this group, send an email to:
TekScopes-unsubscribe@...
<mailto:TekScopes-unsubscribe@...>


Your use of Yahoo! Groups is subject to
http://docs.yahoo.com/info/terms/ <http://docs.yahoo.com/info/terms/>
_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.


Miroslav Pokorni <mpokorni@...>
 

The 5 meters of cable is still 450 pF, still quite load for most drivers.

Regards

Miroslav Pokorni

-----Original Message-----
From: F F [mailto:ferfons@...]
Sent: Tuesday, July 10, 2001 1:06 AM
To: TekScopes@...
Subject: RE: [TekScopes] Coax BNC Cable Termination

Your thesis is entirely based on a 30-metre stretch of
cable. However if you
had carefully read my email, the lenght in question was
quite inferior. A
mere 5-metre chunk of cable if that much.

Regards

Fernando
Portsmouth, UK

>From: Miroslav Pokorni <mpokorni@...>
>Reply-To: TekScopes@...
>To: "'TekScopes@...'"
<TekScopes@...>
>Subject: RE: [TekScopes] Coax BNC Cable Termination
>Date: Mon, 9 Jul 2001 17:24:55 -0700
>
>That rule of thumb is not crappy, but there is a caveat: an
unterminated
>cable appears as a nasty, predominantly capacitive
impedance; for your
>run-of-the-mill, polyethylene cable, capacitance is 30
pF/ft, what would
>amount to 3000 pF for 30 meter piece. If your driver can
handle that kind
>of
>load, you are OK, but most drivers, without special
measures, would
>oscillate, if not fall flat on their face. Special measures
are usually
>adding a resistor between output and load, sometimes trying
to compensate
>phase, whatever does it take to survive.
>
>So, when you transfer 1 MHz signal on a 30 meter cable you
want see much of
>reflection effects. Rule of thumb says significantly
smaller that ¼
>wavelength; 1 MHz in air is 300 meters, in polyethylene
around 600 meters
>what yields 1/20 of wavelength; that is better than ¼ of ¼
wavelength.
>However, cable driver might be a problem.
>
>
>
>Regards
>
>Miroslav Pokorni
>
>
>
> -----Original Message-----
> From: F F [mailto:ferfons@...]
> Sent: Friday, July 06, 2001 1:42 AM
> To: TekScopes@...
> Subject: Re: [TekScopes] Coax BNC
Cable Termination
>
> >From: "John Rehwinkel"
<spam@...
><mailto:spam@...> >
> >Reply-To:
TekScopes@...
><mailto:TekScopes@...>
> >To: TekScopes@...
><mailto:TekScopes@...>
> >Subject: Re: [TekScopes] Coax
BNC Cable
>Termination
> >Date: Thu, 5 Jul 2001 17:39:29
-0400
>
> (snip)
> 1MHz signals
> >should tolerate a 30 meter long
mismatch, and so
>on.
>
> My humble twopence: Crappy rule of thumb!
Few years ago
>while integrating an ATE we had a 1MHz square signal we
thought it wouldn't
>need matching pretty much for the same reasons. Oh we were
so bloody wrong
>and it took a while to figure it out. We did have to match
that particular
>stretch of cable - 3 or 4 "lousy" meters of coax - down to
50 ohms (TDS420
>alterative Zin) and through line drivers, keeping the
signal below the
>scope's max input power rating. That's how we got it
sorted.
> Fernando
> Portsmouth, UK
>
>
>
>
>

_________________________________________________________________________
> Get Your Private, Free E-mail from MSN
Hotmail at
>http://www.hotmail.com <http://www.hotmail.com> .
>
> To unsubscribe from this group, send an
email to:
> TekScopes-unsubscribe@...
><mailto:TekScopes-unsubscribe@...>
>
>
> Your use of Yahoo! Groups is subject to
>http://docs.yahoo.com/info/terms/
<http://docs.yahoo.com/info/terms/>
>


_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at
http://www.hotmail.com.


To unsubscribe from this group, send an email to:
TekScopes-unsubscribe@...



Your use of Yahoo! Groups is subject to
http://docs.yahoo.com/info/terms/


John Miles <jmiles@...>
 

Also keep in mind that a 1 MHz square wave needs at least 10 MHz of
error-free bandwidth, preferably more, to come out looking right. With a
sine wave, you might never have noticed the problem.

-- jm

----- Original Message -----
From: "F F" <ferfons@...>
To: <TekScopes@...>
Sent: Tuesday, July 10, 2001 1:05 AM
Subject: RE: [TekScopes] Coax BNC Cable Termination


Your thesis is entirely based on a 30-metre stretch of cable. However if
you
had carefully read my email, the lenght in question was quite inferior. A
mere 5-metre chunk of cable if that much.

Regards

Fernando
Portsmouth, UK


John Rehwinkel <spam@...>
 

That rule of thumb is not crappy, but there is a caveat: an
unterminated cable appears as a nasty, predominantly capacitive
impedance
The we're usually setting things up, yes. Coax less than 1/4
wave long, terminated in more than its characteristic impedance,
appears capacitive. If it's terminated in less than its
characteristic impedance, it appears inductive. At 1/4 wave
long, it reflects the opposite of its termination (a shorted
1/4 wave segment looks open, an open 1/4 wave segment appears
shorted). At more than 1/4 wave, things reverse, and a line
terminated in more than its characteristic impedance appears
inductive, and so on.

Naturally, wavelengths only apply to pure (i.e. sine wave)
frequencies, and most signals have more than one frequency
component, so things get complicated.

But as Miroslav said, most lines are relatively short, and
terminate in 1 megohm scope inputs, and therefore appear
capacitive.

-- John Rehwinkel KG4L
spam@...


F F <ferfons@...>
 

Re John's comments, I agree and I'm pretty sure a sinewave would had spared me the midnight oil ;-)

Re Miroslav comments, I can't remember the cable's cap spec - this took place some ten odd years). What I can tell you though is in fact it wasn't easy to pick a suitable driver for this application.

Fernando
Portsmouth, UK


From: "John Miles" <jmiles@...>
(snip...)

Also keep in mind that a 1 MHz square wave needs at least 10 MHz of
error-free bandwidth, preferably more, to come out looking right. With a
sine wave, you might never have noticed the problem.

From: Miroslav Pokorni <mpokorni@...>
(snip...)
The 5 meters of cable is still 450 pF, still quite load for most drivers.
_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.


Miroslav Pokorni <mpokorni@...>
 

The midnight oil that would have been spared if you used sinewave would have
been spent on debugging problem coming from jitter of sinewave clock, so I
do not think that you have anything to regret.

As for coax cable, unless you used a ½ inch rope (RG6) or foamed dielectric
(what was very expensive 10 years ago), your standard ¼ inch cable would
have been around 30 pF/ft, 5 pF give or take, so you were dealing with 450
pF load. A driver that comes to mind is National's DS0026, which is
characterized for 500 and 1000 pF load (I had to actually go and check for
what loads is driver characterized), but that is a driver that was not
widely known or used, so its availability might have been poor.

Regards

Miroslav Pokorni

-----Original Message-----
From: F F [mailto:ferfons@...]
Sent: Wednesday, July 11, 2001 1:42 AM
To: TekScopes@...
Subject: Re: [TekScopes] Coax BNC Cable Termination

Re John's comments, I agree and I'm pretty sure a sinewave
would had spared
me the midnight oil ;-)

Re Miroslav comments, I can't remember the cable's cap spec
- this took
place some ten odd years). What I can tell you though is in
fact it wasn't
easy to pick a suitable driver for this application.

Fernando
Portsmouth, UK


>From: "John Miles" <jmiles@...>
(snip...)

Also keep in mind that a 1 MHz square wave needs at least 10
MHz of
error-free bandwidth, preferably more, to come out looking
right. With a
sine wave, you might never have noticed the problem.

>From: Miroslav Pokorni <mpokorni@...>
(snip...)
>The 5 meters of cable is still 450 pF, still quite load for
most drivers.


_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at
http://www.hotmail.com.


To unsubscribe from this group, send an email to:
TekScopes-unsubscribe@...



Your use of Yahoo! Groups is subject to
http://docs.yahoo.com/info/terms/