Line impedance measuring & matching -- techniques

classic Classic list List threaded Threaded
12 messages Options
Reply | Threaded
Open this post in threaded view
|

Line impedance measuring & matching -- techniques

Dinesh Guleria

Hi ,

Please can anyone on list suggest , how to do Line impedance measuring & matching. For RS485 & CAN bus cables ?

Regards,
Dinesh

Reply | Threaded
Open this post in threaded view
|

RE: Line impedance measuring & matching -- techniques

John Dammeyer
I’m not sure I understand your question.
As an example.  RG-58U coax cable has a 50 Ohm impedance.  If you attach that cable to a TV with a 75 Ohm output impedance and an antenna rated at 50 ohms you will have a mismatch and some of the signal value will be lost or reflections will affect the signal quality.

Are you asking how cable is characterized to be 50 ohms?

If you purchase Beldon Thick DeviceNet cable the blue and white wires have a characteristic impedance of 120 ohms at the frequencies that show up with CAN messaging.  To prevent reflections the ends of the transmission line need to be terminated with that same value of 120 Ohms.

The CAN bus drivers are designed to drive a load.  That load is 60 Ohms in parallel with 120 devices that each have an input impedance high enough to not exceed the driver.

For example.  Say your driver has a max output current of 70mA.  A CAN_H signal level of 4.5V would drive current out to the two 120 ohm resistors and also into each CAN receiver.  Most of the current enters the CAN_L pin which is held at about 1.0V.  So the voltage across the bus at worst is 3.5V

If the receiver has an input resistance of 100K then the parallel combination of 120 resistors, each 100K in value, is 833 ohms.  That’s in parallel with the two 120 Ohm bus termination resistors.  Total resistance seen by the driver during a dominant is then 52 Ohms or 67 mA which is within specification of the driver.  That’s the DC side.

The AC side is also interesting.  The cable design is 120 Ohms impedance.  With 120 ohm resistors at each end the transmission line is matched and all the AC energy is absorbed by the resistors.  But if this were a 45 Ohm impedance cable you now have a mismatch and you get reflections which could mean the signal is degraded or bits are even changed.  

So you think, well I’ll terminate the cable with 45 Ohm resistors to once again correctly terminate the transmission line.  The parallel combination of those two resistors now means the drivers have to supply 156mA without even considering the parallel combination of the receivers.  That’s too high.

What if you ran 300 ohm twinax cable like the stuff used for FM antennas.    Well first of all the CAN driver isn’t a balanced driver so that’s not an ideal cable even if you terminate the ends with 300 ohm resistors.  The lack of twists means the signal is susceptible to both magnetic and electrical interference.

So could you run CAN with 300 ohm cable?  Of course you could.  Might even work really well.    I’ve run 1Mbps CAN between two ends of 100’ of Z wire.  That’s the pink 4 conductor unshielded telephone cable.  It worked.  But that’s two nodes and a block of messages from one end to the other.  No idea if some parts of the cable had  ‘null’ spots where the signals might totally cancel out.  Not even sure what Z wire impedance is.  But I wouldn’t want to fly in an airplane that used fly by wire control signals that way.

John Dammeyer







From: [hidden email] [mailto:[hidden email]] On Behalf Of Dinesh Guleria
Sent: February-02-16 10:22 PM
To: CANLIST
Subject: [CANLIST] Line impedance measuring & matching -- techniques

Hi ,
Please can anyone on list suggest , how to do Line impedance measuring & matching. For RS485 & CAN bus cables ?
Regards,
Dinesh

--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: Line impedance measuring & matching -- techniques

Dinesh Guleria
Thanks John for your reply.

>> I’m not sure I understand your question.
I mean to say that if CAN network is having problem in communication, and i need to investigate what is causing mismatching of impedance. Then for that :--
1> I will have to measure the line impedance in loaded & unloaded condition
2> Then check what is causing impedance mismatch.
3> correct these factors.

So how to measure line impedance in loaded & unloaded condition ?
Also How to correct this ?

Regards,
Dinesh

On Wed, Feb 3, 2016 at 12:40 PM, John Dammeyer <[hidden email]> wrote:
I’m not sure I understand your question.
As an example.  RG-58U coax cable has a 50 Ohm impedance.  If you attach that cable to a TV with a 75 Ohm output impedance and an antenna rated at 50 ohms you will have a mismatch and some of the signal value will be lost or reflections will affect the signal quality.

Are you asking how cable is characterized to be 50 ohms?

If you purchase Beldon Thick DeviceNet cable the blue and white wires have a characteristic impedance of 120 ohms at the frequencies that show up with CAN messaging.  To prevent reflections the ends of the transmission line need to be terminated with that same value of 120 Ohms.

The CAN bus drivers are designed to drive a load.  That load is 60 Ohms in parallel with 120 devices that each have an input impedance high enough to not exceed the driver.

For example.  Say your driver has a max output current of 70mA.  A CAN_H signal level of 4.5V would drive current out to the two 120 ohm resistors and also into each CAN receiver.  Most of the current enters the CAN_L pin which is held at about 1.0V.  So the voltage across the bus at worst is 3.5V

If the receiver has an input resistance of 100K then the parallel combination of 120 resistors, each 100K in value, is 833 ohms.  That’s in parallel with the two 120 Ohm bus termination resistors.  Total resistance seen by the driver during a dominant is then 52 Ohms or 67 mA which is within specification of the driver.  That’s the DC side.

The AC side is also interesting.  The cable design is 120 Ohms impedance.  With 120 ohm resistors at each end the transmission line is matched and all the AC energy is absorbed by the resistors.  But if this were a 45 Ohm impedance cable you now have a mismatch and you get reflections which could mean the signal is degraded or bits are even changed.

So you think, well I’ll terminate the cable with 45 Ohm resistors to once again correctly terminate the transmission line.  The parallel combination of those two resistors now means the drivers have to supply 156mA without even considering the parallel combination of the receivers.  That’s too high.

What if you ran 300 ohm twinax cable like the stuff used for FM antennas.    Well first of all the CAN driver isn’t a balanced driver so that’s not an ideal cable even if you terminate the ends with 300 ohm resistors.  The lack of twists means the signal is susceptible to both magnetic and electrical interference.

So could you run CAN with 300 ohm cable?  Of course you could.  Might even work really well.    I’ve run 1Mbps CAN between two ends of 100’ of Z wire.  That’s the pink 4 conductor unshielded telephone cable.  It worked.  But that’s two nodes and a block of messages from one end to the other.  No idea if some parts of the cable had  ‘null’ spots where the signals might totally cancel out.  Not even sure what Z wire impedance is.  But I wouldn’t want to fly in an airplane that used fly by wire control signals that way.

John Dammeyer







From: [hidden email] [mailto:[hidden email]] On Behalf Of Dinesh Guleria
Sent: February-02-16 10:22 PM
To: CANLIST
Subject: [CANLIST] Line impedance measuring & matching -- techniques

Hi ,
Please can anyone on list suggest , how to do Line impedance measuring & matching. For RS485 & CAN bus cables ?
Regards,
Dinesh

--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>

Reply | Threaded
Open this post in threaded view
|

RE: Line impedance measuring & matching -- techniques

John Dammeyer
How do you know that the communication problems are due to mismatched impedance?
That’s like saying. “I know the bananas aren’t ripe yet because they are green so perhaps that’s why the bread has green mold on it”

Is this a school class assignment?
John

From: [hidden email] [mailto:[hidden email]] On Behalf Of Dinesh Guleria
Sent: February-02-16 11:29 PM
Cc: CANLIST
Subject: Re: [CANLIST] Line impedance measuring & matching -- techniques

Thanks John for your reply.

>> I’m not sure I understand your question.
I mean to say that if CAN network is having problem in communication, and i need to investigate what is causing mismatching of impedance. Then for that :--
1> I will have to measure the line impedance in loaded & unloaded condition
2> Then check what is causing impedance mismatch.
3> correct these factors.

So how to measure line impedance in loaded & unloaded condition ?
Also How to correct this ?

Regards,
Dinesh

--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: Line impedance measuring & matching -- techniques

Dinesh Guleria
>> How do you know that the communication problems are due to mismatched impedance?
I think you did not got from my mail, I am asking case in which one have to investigate what is causing mismatching of impedance on line.
Then how one use to proceed.

Regards,
Dinesh

On Wed, Feb 3, 2016 at 1:24 PM, John Dammeyer <[hidden email]> wrote:
How do you know that the communication problems are due to mismatched impedance?
That’s like saying. “I know the bananas aren’t ripe yet because they are green so perhaps that’s why the bread has green mold on it”

Is this a school class assignment?
John

From: [hidden email] [mailto:[hidden email]] On Behalf Of Dinesh Guleria
Sent: February-02-16 11:29 PM
Cc: CANLIST
Subject: Re: [CANLIST] Line impedance measuring & matching -- techniques

Thanks John for your reply.

>> I’m not sure I understand your question.
I mean to say that if CAN network is having problem in communication, and i need to investigate what is causing mismatching of impedance. Then for that :--
1> I will have to measure the line impedance in loaded & unloaded condition
2> Then check what is causing impedance mismatch.
3> correct these factors.

So how to measure line impedance in loaded & unloaded condition ?
Also How to correct this ?

Regards,
Dinesh

--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>

Reply | Threaded
Open this post in threaded view
|

Re: Line impedance measuring & matching -- techniques

Dinesh Guleria
In reply to this post by John Dammeyer
>> How do you know that the communication problems are due to mismatched impedance?
I think you did not got from my mail, I am asking case in which one have to investigate what is causing mismatching of impedance on line.
Then how one use to proceed.

Regards,
Dinesh

On Wed, Feb 3, 2016 at 1:24 PM, John Dammeyer <[hidden email]> wrote:
How do you know that the communication problems are due to mismatched impedance?
That’s like saying. “I know the bananas aren’t ripe yet because they are green so perhaps that’s why the bread has green mold on it”

Is this a school class assignment?
John

From: [hidden email] [mailto:[hidden email]] On Behalf Of Dinesh Guleria
Sent: February-02-16 11:29 PM
Cc: CANLIST
Subject: Re: [CANLIST] Line impedance measuring & matching -- techniques

Thanks John for your reply.

>> I’m not sure I understand your question.
I mean to say that if CAN network is having problem in communication, and i need to investigate what is causing mismatching of impedance. Then for that :--
1> I will have to measure the line impedance in loaded & unloaded condition
2> Then check what is causing impedance mismatch.
3> correct these factors.

So how to measure line impedance in loaded & unloaded condition ?
Also How to correct this ?

Regards,
Dinesh

--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>

Reply | Threaded
Open this post in threaded view
|

Re: Line impedance measuring & matching -- techniques

Dinesh Guleria
Ok got one good specs by Steve Corrigan .

Regards,
Dinesh

On Wed, Feb 3, 2016 at 2:37 PM, Dinesh Guleria <[hidden email]> wrote:
>> How do you know that the communication problems are due to mismatched impedance?
I think you did not got from my mail, I am asking case in which one have to investigate what is causing mismatching of impedance on line.
Then how one use to proceed.

Regards,
Dinesh

On Wed, Feb 3, 2016 at 1:24 PM, John Dammeyer <[hidden email]> wrote:
How do you know that the communication problems are due to mismatched impedance?
That’s like saying. “I know the bananas aren’t ripe yet because they are green so perhaps that’s why the bread has green mold on it”

Is this a school class assignment?
John

From: [hidden email] [mailto:[hidden email]] On Behalf Of Dinesh Guleria
Sent: February-02-16 11:29 PM
Cc: CANLIST
Subject: Re: [CANLIST] Line impedance measuring & matching -- techniques

Thanks John for your reply.

>> I’m not sure I understand your question.
I mean to say that if CAN network is having problem in communication, and i need to investigate what is causing mismatching of impedance. Then for that :--
1> I will have to measure the line impedance in loaded & unloaded condition
2> Then check what is causing impedance mismatch.
3> correct these factors.

So how to measure line impedance in loaded & unloaded condition ?
Also How to correct this ?

Regards,
Dinesh

--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>


Reply | Threaded
Open this post in threaded view
|

Re: Line impedance measuring & matching -- techniques

Bertil Bäck-2
In reply to this post by Dinesh Guleria
Hi,
I once was asked to evaluate if it was possible to use a Industrial Ethernet cable for CAN highspeed.

So first we have the 120ohm impedance. So this is at a rated Mhz, CAN have them listed for 1Mhz and Ethernet usually for 10 or 100Mhz

So one need to find out what the Ethernet cable impedance is at 1Mhz.  I just did as a article on the Internet said. I put a trim potentiometer in one end of the cable and a Frequency generator in the other end and put in on 1Mhz and square wave. I have a scope on both ends of the cable and trim the potentiometer so I get the best square signal. Then I measure the resistance of the trim potentiometer. This is a very crude way of getting the impedance of the cable.

Then if one think of  impedance problems in networks. We once was issued with finding out problems with a CAN network. That time we saw sporadic error frames on the network. This boiled down to some nodes was only 10cm away from each other. 1Mhz is a high speed signal and one have different impedance in parts of the network. When nodes are to close to each other they can be touch of as being in parallel. Messing up the impedance in that part of the network.

Ref.
Minimum distance between CAN nodes
http://v2.can-newsletter.org/uploads/media/raw/9fa400cd22515f17f2c9732bc52c68fd.pdf.

Quick & dirty cable length & impedance measurement
http://www.edn.com/electronics-blogs/benchtalk/4431974/Quick---dirty-cable-length---impedance-measurement

Br,
Bertil Bäck
R&D Manager, Hardware
+358505886895

TK Engineering solves your CAN, CANopen, J1939 and NMEA2000 needs by providing the hardware, software and know-how you need to design, manufacture and troubleshoot your industrial- marine- and heavy machine CAN networks. Call +358 6 357 6300 or email [hidden email] to learn how we can help you solve your CAN related tasks.


From: "Dinesh Guleria" <[hidden email]>
Cc: "CANLIST" <[hidden email]>
Sent: Wednesday, February 3, 2016 11:07:44 AM
Subject: Re: [CANLIST] Line impedance measuring & matching -- techniques

>> How do you know that the communication problems are due to mismatched impedance?
I think you did not got from my mail, I am asking case in which one have to investigate what is causing mismatching of impedance on line.
Then how one use to proceed.

Regards,
Dinesh

On Wed, Feb 3, 2016 at 1:24 PM, John Dammeyer <[hidden email]> wrote:
How do you know that the communication problems are due to mismatched impedance?
That’s like saying. “I know the bananas aren’t ripe yet because they are green so perhaps that’s why the bread has green mold on it”

Is this a school class assignment?
John

From: [hidden email] [mailto:[hidden email]] On Behalf Of Dinesh Guleria
Sent: February-02-16 11:29 PM
Cc: CANLIST
Subject: Re: [CANLIST] Line impedance measuring & matching -- techniques

Thanks John for your reply.

>> I’m not sure I understand your question.
I mean to say that if CAN network is having problem in communication, and i need to investigate what is causing mismatching of impedance. Then for that :--
1> I will have to measure the line impedance in loaded & unloaded condition
2> Then check what is causing impedance mismatch.
3> correct these factors.

So how to measure line impedance in loaded & unloaded condition ?
Also How to correct this ?

Regards,
Dinesh

--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>

Reply | Threaded
Open this post in threaded view
|

RE: Line impedance measuring & matching -- techniques

Al Thomason

From: [hidden email] [mailto:[hidden email]] On Behalf Of Bertil Bäck:

 

“…So one need to find out what the Ethernet cable impedance is at 1Mhz.  I just did as a article on the Internet said. I put a trim potentiometer in one end of the cable and a Frequency generator in the other end and put in on 1Mhz and square wave. I have a scope on both ends of the cable and trim the potentiometer so I get the best square signal. Then I measure the resistance of the trim potentiometer. This is a very crude way of getting the impedance of the cable.”

 

Am wondering what the results were / what conclusion you came to with regard of using Ethernet type cable? 

(And was it CAT-5, or something else???)

 

-al-

 

 

Reply | Threaded
Open this post in threaded view
|

RE: Line impedance measuring & matching -- techniques

John Dammeyer
In reply to this post by Dinesh Guleria

Hi all,

I haven’t got any Ethernet cable hanging around but I do have this 46.5’ length of Thick DeviceNet cable rated for CAN bus and 120 Ohm  impedance.

Driven at one end by an Agilent 33220A Arbitrary Waveform generator set for 5V out and 120 Ohm impedance along with 1uS pulse every 5uS.

Scope at the other end is Tek MSO3034 with probe impedance set to 1M

Here’s the result without termination at the scope end.

http://www.autoartisans.com/CAN/NoTerminationThickDeviceNet.jpg

 

And here’s the result with termination at the scope end.

http://www.autoartisans.com/CAN/TerminationThickDeviceNet.jpg

 

And just for general interest the next photo is from a 36m long cable where there are 50 nodes are on the last 10m and the other end has one CAN node.  It’s also Thick DeviceNet terminated at both ends with 120 Ohms.  Each node is tapped of the main cable about 19cm apart with a stub length of 30cm.

 

It’s in this photo you can see the reason why there are bus length limitations on the controller area network bus.  The scope triggers on the ACK, which follows the CRC Delimiter, but is delayed by 300nS due to the to/from distance over the length of the cable. 

 

That little bump in each bit (about 200ns into the bit) is described perfectly in Steve Corrigan’s article on node to node spacing.  And the shifting of the bus levels after the ACK is due to ground bounce as the load goes from minimal current draw to rated cable current draw over the length of the bus on the power wires which share the ground with CAN signals.

 

http://www.autoartisans.com/CAN/SmplACKDel36m-RGB-2-Photo3.jpg

 

The green trace is the differential sum of the CAN_H and CAN_L.  I only had a two trace scope at that time.  The TTL side of the node receiver would have been more interesting.

 

John Dammeyer

 

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Al Thomason
Sent: February-09-16 8:34 AM
To: [hidden email]
Subject: RE: [CANLIST] Line impedance measuring & matching -- techniques

 

From: [hidden email] [[hidden email]] On Behalf Of Bertil Bäck:

 

“…So one need to find out what the Ethernet cable impedance is at 1Mhz.  I just did as a article on the Internet said. I put a trim potentiometer in one end of the cable and a Frequency generator in the other end and put in on 1Mhz and square wave. I have a scope on both ends of the cable and trim the potentiometer so I get the best square signal. Then I measure the resistance of the trim potentiometer. This is a very crude way of getting the impedance of the cable.”

 

Am wondering what the results were / what conclusion you came to with regard of using Ethernet type cable? 

(And was it CAT-5, or something else???)

 

-al-

 

 

Reply | Threaded
Open this post in threaded view
|

RE: Line impedance measuring & matching -- techniques

John Dammeyer
Hi guys/gals,
I found a coil of CAT5 (blue) Ethernet cable.  I think it's about 100'.
Same experiment as before using the blue and blue/white twisted pair.
http://www.autoartisans.com/CAN/NoTerminationEtherNet.jpg

http://www.autoartisans.com/CAN/TerminationEtherNet-120.jpg

BTW.  Neither of these tests show how the cable will behave with multiple nodes on various length stubs nor common mode ground bounce.  But it's a starting point.

John Dammeyer

Am wondering what the results were / what conclusion you came to with regard of using Ethernet type cable?  
(And was it CAT-5, or something else???)

-al-



--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: Line impedance measuring & matching -- techniques

Bertil Bäck-2
In reply to this post by Al Thomason
Hi,
so the use case for this is that we need to have both CAN (dual busses) and Ethernet installed. We have industrial Ethernet cable that have 4 wires and quite large cross section. We try to optimize production costs and BOM.

The Ethernet was listed with 100ohm at 100Mhz. I got the same at 1Mhz :)

I could prove that it was quite bang on 100Mhz. If I put 150ohm the signal is all distorted.
We went ahead and installed the cable on a full system for next level of testing. It worked fine there also. We still use 120ohm for termination. To not mess with the transceivers.

One improvement still would be to change to split terminators, but we have not went ahead with that yet.

Br,
Bertil


From: "Al Thomason" <[hidden email]>
To: [hidden email]
Sent: Tuesday, February 9, 2016 6:34:29 PM
Subject: RE: [CANLIST] Line impedance measuring & matching -- techniques

From: [hidden email] [mailto:[hidden email]] On Behalf Of Bertil Bäck:

 

“…So one need to find out what the Ethernet cable impedance is at 1Mhz.  I just did as a article on the Internet said. I put a trim potentiometer in one end of the cable and a Frequency generator in the other end and put in on 1Mhz and square wave. I have a scope on both ends of the cable and trim the potentiometer so I get the best square signal. Then I measure the resistance of the trim potentiometer. This is a very crude way of getting the impedance of the cable.”

 

Am wondering what the results were / what conclusion you came to with regard of using Ethernet type cable? 

(And was it CAT-5, or something else???)

 

-al-