Impedanc Matching Problem

Openhiemer · Mar 23, 2005

Hello,

I have a impedance matching problem that I need help with.

I have a 1Mb data transmission system that is designed to work with 100-150
ohm cable, We would like to use it on 55 ohm cable.

If anyone thas has a good understanding of transmission line theory and
would like to help out (I can probably pay for time of the problem is
solvable) please e-mail me.

To e-mail me please change the z characters to o characters and the hat
symbol to a at symbol. itmaster2001^yahzz.cz.uk (hopfully mailbots ar not
that bright yet)

Thanks

Reg Edwards · Mar 23, 2005

Try the 55-ohm cable. You will probably notice no difference.

Gareth · Mar 23, 2005

John Woodgate wrote:

I read in sci.electronics.design that Openhiemer <openhamer@dot.dot
wrote (in <Xns9622C01FD9184openhamermicrosoftco@158.152.254.254&gt about
'Impedanc Matching Problem', on Wed, 23 Mar 2005:

I have a impedance matching problem that I need help with.

I have a 1Mb data transmission system that is designed to work with
100-150 ohm cable, We would like to use it on 55 ohm cable.

If anyone thas has a good understanding of transmission line theory
and would like to help out (I can probably pay for time of the problem
is solvable) please e-mail me.

Do you have enough signal to use L-pads at each end of the 55 ohm cable?
Just two resistors in series, designed to look like 100 ohms one way and
55 ohms the other.

Use Courier font:

+----[R1]----+-----o o-----[R1]----+----+
| | | |
[100] [R2] <--55 ohms 100 ohms--> [R2] [55]
| | | |
+------------+-----o o-------------+----+

The calculation of R1 and R2 will keep you out of mischief for a few
seconds/ minutes/hours and will be a valuable addition to your little
black book of useful cribs.

I make R1 68 Ohms and R2 82 Ohms (nearest E12 preferred values), which
will give you just over 7 dB of loss at each end. Can you afford that
much attenuation?

Another option maybe to put an L pad (as above) at only the receive end.
That way you shouldn't get any reflections back down the cable. If
the system is bidirectional you may still get away with a pad at only
one end as the reflected signal going the other way will have been
through the pad and cable twice, and thus be attenuated significantly
and therefore less likely to cause an error in the wanted signal.

It may well work without any matching. I would try Reg's suggestion first.

--
-----------------------------------------------------------------------
To reply to me directly:

Replace privacy.net with: totalise DOT co DOT uk and replace me with
gareth.harris

Jim Thompson · Mar 23, 2005

On Wed, 23 Mar 2005 23:42:30 +0000, Gareth <me@privacy.net> wrote:

John Woodgate wrote:
I read in sci.electronics.design that Openhiemer <openhamer@dot.dot
wrote (in <Xns9622C01FD9184openhamermicrosoftco@158.152.254.254&gt about
'Impedanc Matching Problem', on Wed, 23 Mar 2005:

I have a impedance matching problem that I need help with.

I have a 1Mb data transmission system that is designed to work with
100-150 ohm cable, We would like to use it on 55 ohm cable.

If anyone thas has a good understanding of transmission line theory
and would like to help out (I can probably pay for time of the problem
is solvable) please e-mail me.

Do you have enough signal to use L-pads at each end of the 55 ohm cable?
Just two resistors in series, designed to look like 100 ohms one way and
55 ohms the other.

Use Courier font:

+----[R1]----+-----o o-----[R1]----+----+
| | | |
[100] [R2] <--55 ohms 100 ohms--> [R2] [55]
| | | |
+------------+-----o o-------------+----+

The calculation of R1 and R2 will keep you out of mischief for a few
seconds/ minutes/hours and will be a valuable addition to your little
black book of useful cribs.

I make R1 68 Ohms and R2 82 Ohms (nearest E12 preferred values), which
will give you just over 7 dB of loss at each end. Can you afford that
much attenuation?

Another option maybe to put an L pad (as above) at only the receive end.
That way you shouldn't get any reflections back down the cable. If
the system is bidirectional you may still get away with a pad at only
one end as the reflected signal going the other way will have been
through the pad and cable twice, and thus be attenuated significantly
and therefore less likely to cause an error in the wanted signal.

It may well work without any matching. I would try Reg's suggestion first.

480)460-2350 | |
| E-mail Address at Website Fax

Mark · Mar 24, 2005

Could depend on the data type. If the average value is zero (like
Manchester-encoded), use transformers.

...Jim Thompson

Agreed, the galvanic isolation is also very helpfull.

Mark

James Meyer · Mar 24, 2005

On Wed, 23 Mar 2005 18:53:08 +0000 (UTC), Openhiemer <openhamer@dot.dot> wroth:

Hello,

I have a impedance matching problem that I need help with.

I have a 1Mb data transmission system that is designed to work with 100-150
ohm cable, We would like to use it on 55 ohm cable.

Twisted pair cable is usually in the 100 to 150 Ohm realm. Coax cable
is usually in the 50 to 55 Ohm realm.

Is the cable suggested originally for the system twisted pair or
coaxial? There are some fundamental differences other than impedance to
consider.

Jim

Jim Thompson · Mar 24, 2005

On 23 Mar 2005 17:16:07 -0800, "Mark" <makolber@yahoo.com> wrote:

Could depend on the data type. If the average value is zero (like
Manchester-encoded), use transformers.

...Jim Thompson

Agreed, the galvanic isolation is also very helpfull.

Mark

480)460-2350 | |
| E-mail Address at Website Fax

Openhiemer · Mar 24, 2005

James Meyer <jmeyer@nowhere.net> wrote in
news:2e6441tp9pbf1d1fm6308tsoqndgbrckev@4ax.com:

Twisted pair cable is usually in the 100 to 150 Ohm realm. Coax
cable
is usually in the 50 to 55 Ohm realm.

Is the cable suggested originally for the system twisted pair or
coaxial? There are some fundamental differences other than impedance
to consider.

Jim

Apologies for not being more specific.

The devices are multidrop and are transformer coupled to (normally) 150
ohm twisted pair cable. The devices have a master/slave relationship so
there is no collission/multi transmission problems. The design limit of
the transmission system in this configuration is 1KM.

The cable that I have been asked to use is screened twisted pair with a
characteristic impedence of 55 ohm. I can get this working using a 100
ohm resistor at each end on a 40M cable. When I go above this length it
stops working. I assum that this is due to refections caused by the
impedence mismatch.

Andy

-------------------------------------------------------------------------
To e-mail me please change the z characters to o characters and the hat
symbol to a at symbol. itmaster2001^yahzz.cz.uk (hopfully mailbots ar not
that bright yet)

Openhiemer · Mar 24, 2005

Gareth <me@privacy.net> wrote in
news:qL6dneQSXZJXY9zfRVnygg@brightview.com:

John Woodgate wrote:
Do you have enough signal to use L-pads at each end of the 55 ohm
cable? Just two resistors in series, designed to look like 100 ohms
one way and 55 ohms the other.

I make R1 68 Ohms and R2 82 Ohms (nearest E12 preferred values), which
will give you just over 7 dB of loss at each end. Can you afford that
much attenuation?

Another option maybe to put an L pad (as above) at only the receive
end.
That way you shouldn't get any reflections back down the cable. If
the system is bidirectional you may still get away with a pad at only
one end as the reflected signal going the other way will have been
through the pad and cable twice, and thus be attenuated significantly
and therefore less likely to cause an error in the wanted signal.

It may well work without any matching. I would try Reg's suggestion
first.

Many thanks for that,unfortunately there are 20 devices to connect so the

power loss may be prohibitive.

I have been looking at transformer matching but I dont have a clue what I
need to buy/make or even where to get them.

Best wishes
Andy

-------------------------------------------------------------------------
To e-mail me please change the z characters to o characters and the hat
symbol to a at symbol. itmaster2001^yahzz.cz.uk (hopefully mailbots are
not that bright yet)

Mook Johnson · Mar 24, 2005

Verify that the characteristic impedance is from conductor to conductor and
not conductor to shield. Most twisted shielded pair I've seen is between
150 and 90 ohms. 120 is nominal.

Conductor to shield measuremnt will look more like a coax with and
appropriately low Zo.

Give the twisted pair a try as is. I've been able to transmit Manchester
data over 55kft on Brand X twisted pair line with surprisingly low BER.

"Openhiemer" <openhamer@dot.dot> wrote in message
news:Xns96237C7306B40herethereltduk@194.117.143.37...

James Meyer <jmeyer@nowhere.net> wrote in
news:2e6441tp9pbf1d1fm6308tsoqndgbrckev@4ax.com:

Twisted pair cable is usually in the 100 to 150 Ohm realm. Coax
cable
is usually in the 50 to 55 Ohm realm.

Is the cable suggested originally for the system twisted pair or
coaxial? There are some fundamental differences other than impedance
to consider.

Jim

Apologies for not being more specific.

The devices are multidrop and are transformer coupled to (normally) 150
ohm twisted pair cable. The devices have a master/slave relationship so
there is no collission/multi transmission problems. The design limit of
the transmission system in this configuration is 1KM.

The cable that I have been asked to use is screened twisted pair with a
characteristic impedence of 55 ohm. I can get this working using a 100
ohm resistor at each end on a 40M cable. When I go above this length it
stops working. I assum that this is due to refections caused by the
impedence mismatch.

Andy

-------------------------------------------------------------------------
To e-mail me please change the z characters to o characters and the hat
symbol to a at symbol. itmaster2001^yahzz.cz.uk (hopfully mailbots ar not
that bright yet)

James Meyer · Mar 24, 2005

On Thu, 24 Mar 2005 12:14:01 GMT, Openhiemer <openhamer@dot.dot> wroth:

The cable that I have been asked to use is screened twisted pair with a
characteristic impedence of 55 ohm. I can get this working using a 100
ohm resistor at each end on a 40M cable. When I go above this length it
stops working. I assum that this is due to refections caused by the
impedence mismatch.

Andy

Have you tried the screened cable with no resistors? If so, did it
work?

Multi-drop receivers are usually designed with a much higher input
impedance than the cable's impedance simply so that they won't load the cable
when many of them are connected in parallel to the cable.

Multi-drop transmitters are usually designed with an output impedance
much lower than the cable to enable them to drive multiple receivers.

In other words, multi-drop TX/RX devices don't inherently match the
cable at all.

Reflections, if they occur at all to a degree that might stop a
multi-drop system from working, are best handled by loading only the extreme
ends of the cable. A 50 Ohm resistor in parallel with the signal leads of the
cable at each extreme end of the cable should eliminate the majority of the
reflections. Make no connections with "matching" networks between the signal
leads and the screen of the cable. Only insure that the screen is continuous.

Jim

John Woodgate · Mar 24, 2005

I read in sci.electronics.design that Openhiemer <openhamer@dot.dot>
wrote (in <Xns96237C7306B40herethereltduk@194.117.143.37&gt

about
'Impedanc Matching Problem', on Thu, 24 Mar 2005:

The cable that I have been asked to use is screened twisted pair with a
characteristic impedence of 55 ohm.

That is an unusual cable, I think. Do you have a manufacturer's part
number or a specification including the conductor insulating material
and the dimensions of the cross-section? Most screened twisted pair is
around 100 ohms.
--
Regards, John Woodgate, OOO - Own Opinions Only.
There are two sides to every question, except
'What is a Moebius strip?'
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Openhiemer · Mar 24, 2005

"Mook Johnson" <mook@mook.net> wrote in
news:Vny0e.12879$ot.587@tornado.texas.rr.com:

Verify that the characteristic impedance is from conductor to
conductor and not conductor to shield. Most twisted shielded pair
I've seen is between 150 and 90 ohms. 120 is nominal.

Conductor to shield measuremnt will look more like a coax with and
appropriately low Zo.

Give the twisted pair a try as is. I've been able to transmit
Manchester data over 55kft on Brand X twisted pair line with
surprisingly low BER.

I have tested several lengths of the cable with a TDR between 1Mhz and 2

Mhz and the impedence is definitely 55 ohm between the pairs

I would agree that most transmission cables are between 100 and 150 ohm,
this one has taken me by supprise.

Best wishes
Andy

-------------------------------------------------------------------------
To e-mail me please change the z characters to o characters and the hat
symbol to a at symbol. itmaster2001^yahzz.cz.uk (hopefully mailbots are not
that bright yet)

Charles DH Williams · Mar 24, 2005

In article <Xns9623859E074CAherethereltduk@194.117.143.38>,
Openhiemer <openhamer@dot.dot> wrote:

I have tested several lengths of the cable with a TDR between 1Mhz and 2
Mhz and the impedence is definitely 55 ohm between the pairs

I would agree that most transmission cables are between 100 and 150 ohm,
this one has taken me by supprise.

Was the tdr coorectly configured to drive the pair with a true differential
signal (i.e zero common mode component) ? 55ohms is close to what I'd expect
if you had grounded one of the pairs...

Charles

Rob Gaddi · Mar 24, 2005

Openhiemer wrote:

Many thanks for that,unfortunately there are 20 devices to connect so the
power loss may be prohibitive.

I have been looking at transformer matching but I dont have a clue what I
need to buy/make or even where to get them.

Best wishes
Andy

I tend to have good luck with Coilcraft's stuff. Check out
http://www.coilcraft.com/ttwb.cfm . Those should have plenty enough
bandwidth for your signal, just take a 2:1 impedance ratio and call 50
ohms close enough to 55.

Also, Coilcraft is really good about sampling, which makes it fairly
easy to test whether the solution works.

Openhiemer · Mar 24, 2005

Charles DH Williams <C.D.H.Williams@exeter.ac.uk> wrote in
news:C.D.H.Williams-17ACBC.14571924032005@98.0/25.134.80.194.in-addr.arpa
:

Was the tdr coorectly configured to drive the pair with a true
differential signal (i.e zero common mode component) ? 55ohms is close
to what I'd expect if you had grounded one of the pairs...

Yes, It was configured correctly. I also purchased 500M of cable with
similar characteristics for test purposes and the transmission sytem
died at the same length as the installed cable (41M).

Best wishes
Andy

-----------------------------------------------------------
To e-mail me please change the z characters to o characters
and the hat symbol to a at symbol. itmaster2001^yahzz.cz.uk
(hopfully mailbots are not that bright yet)

Jim Thompson · Mar 24, 2005

On Thu, 24 Mar 2005 19:50:31 +0000 (UTC), Openhiemer
<openhamer@dot.dot> wrote:

Charles DH Williams <C.D.H.Williams@exeter.ac.uk> wrote in
news:C.D.H.Williams-17ACBC.14571924032005@98.0/25.134.80.194.in-addr.arpa
:

Was the tdr coorectly configured to drive the pair with a true
differential signal (i.e zero common mode component) ? 55ohms is close
to what I'd expect if you had grounded one of the pairs...

Yes, It was configured correctly. I also purchased 500M of cable with
similar characteristics for test purposes and the transmission sytem
died at the same length as the installed cable (41M).

Best wishes
Andy

Have you taken a look with a 'scope? Maybe you need to pre-comp?

It's been a LONG time (1980) since I did a long run, but IIRC I had to
pre-comp the signal, then do DC-restoration at the far end, to get it
to work error-free.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice

480)460-2350 | |
| E-mail Address at Website Fax

Jim Thompson · Mar 24, 2005

On Thu, 24 Mar 2005 20:05:04 +0000 (UTC), Openhiemer
<openhamer@dot.dot> wrote:

Jim Thompson <thegreatone@example.com> wrote in
news:6na441tr01htcq5eo3028tsen6dg5lo5e0@4ax.com:

Yep. One I set up between two GenRad (Phoenix) buildings had 18VAC
differential between "grounds" :-(

Whoo - I'm glad that I don't have that problem. If I did I could probably
recommend that they ditched the old cables and installed some nice
new fibre

Best wishes
Andy

480)460-2350 | |
| E-mail Address at Website Fax

Openhiemer · Mar 24, 2005

Jim Thompson <thegreatone@example.com> wrote in
news:6na441tr01htcq5eo3028tsen6dg5lo5e0@4ax.com:

Yep. One I set up between two GenRad (Phoenix) buildings had 18VAC
differential between "grounds" :-(

Whoo - I'm glad that I don't have that problem. If I did I could probably
recommend that they ditched the old cables and installed some nice
new fibre

Best wishes
Andy
-----------------------------------------------------------
To e-mail me please change the z characters to o characters
and the hat symbol to a at symbol. itmaster2001^yahzz.cz.uk
(hopfully mailbots are not that bright yet)

Pooh Bear · Mar 25, 2005

Openhiemer wrote:

Hello,

I have a impedance matching problem that I need help with.

I have a 1Mb data transmission system that is designed to work with 100-150
ohm cable, We would like to use it on 55 ohm cable.

Isn't it traditional at this point to suggest the use of 2 lengths of 55 ohm
cable in series ? ;-)

Graham

Impedanc Matching Problem

Openhiemer

Guest

Reg Edwards

Guest

Gareth

Guest

Jim Thompson

Guest

Mark

Guest

James Meyer

Guest

Jim Thompson

Guest

Openhiemer

Guest

Openhiemer

Guest

Mook Johnson

Guest

James Meyer

Guest

John Woodgate

Guest

Openhiemer

Guest

Charles DH Williams

Guest

Rob Gaddi

Guest

Openhiemer

Guest

Jim Thompson

Guest

Jim Thompson

Guest

Openhiemer

Guest

Pooh Bear

Guest

Log in

Welcome to EDABoard.com

Sponsor