Optimal location of receiver on a transmission line

[Mike Vines]

I am measuring a communications bus with an oscilloscope at various points along the line. I understand that there will be reflections at each stub as well as at the ends of the bus if the impedance doesn't match. I also understand that these reflected signals will add/subtract with the incident signal as well as change its phase. If I am interested in observing the resultant wave, does it matter where my oscilloscope is located in relation to the transmitter or ends of the lines?

For example, if I measure a 150-foot bus at 3 locations:
1) Immediately next to the transmitter at one extreme end of the bus
2) In the middle of the bus (at the 75 foot location)
3) At the extreme end of the bus opposite the transmitter

Will the waveforms look the same? The symbol rate is 1 MHz.

Thanks for any help.

 

[Svein]

Will the waveforms look the same? The symbol rate is 1 MHz.

No. The usual behavior on a non-terminated line is a positive reflection (bigger amplitude) at the far end and a negative reflection (lower amplitude) at the near end. You will get "ringing" at both ends.

Some calculations:

• 150 feet is 45m
• The signal speed in a copper line is 0.2m/ns
• Thus the roundtrip delay (back and forth) is (90/0.2)ns = 450ns
• If your symbol rate is 1MHz, you will have a lot of reflections in each symbol, since the symbol time is 1000ns

 

[Mike Vines]

... behavior on a non-terminated line ..

Thank you for your reply. I should have mentioned that the line is terminated on both ends with its characteristic impedance of 78 $\Omega$. I suppose that there will still be some slight reflections though, since it's unlikely to be perfectly matched. Would it then be a good idea to place my oscilloscope in the middle of the bus (away from the ends) to avoid seeing the ringing in the signal?

 

[CWatters]

How is it terminated? Series or parallel can make a difference.

 

[Svein]

Thank you for your reply. I should have mentioned that the line is terminated on both ends with its characteristic impedance of 78 $\Omega$. I suppose that there will still be some slight reflections though, since it's unlikely to be perfectly matched. Would it then be a good idea to place my oscilloscope in the middle of the bus (away from the ends) to avoid seeing the ringing in the signal?

No. If you have a standard transmission line (not multidrop), the state of the signal at the end of the line is what matters. If you cannot decipher the signal with an oscilloscope there, you cannot expect a line receiver to do it.

I am somewhat surprised about 78Ω, however. I know about coaxial cable at 50Ω (RG-58) and 75Ω (RG-59) and twisted pair (100Ω to 120Ω). What kind of cable do you use?

Twisted pair setup

 

[CWatters]

If I am interested in observing the resultant wave, does it matter where my oscilloscope is located in relation to the transmitter or ends of the lines?

The wave form you see can be different at the source, middle and destination depending on how it's terminated.

There are some good articles on the web on why different termination strategies are appropriate in different situations.

http://web.cecs.pdx.edu/~greenwd/xmsnLine_notes.pdf

http://www.ti.com/lit/an/snla034b/snla034b.pdf

 

[Baluncore]

Place the oscilloscope at the far end of the line.

If the oscilloscope has a 50 ohm input impedance, connect it through a 27 ohm series resistor, (close enough to 28 ohm), to the end of the line. That will terminate the line by making it look like 50 + 27 = 77 ohms. There will be a 50/77 signal attenuation seen on the oscilloscope due to the termination.

If the oscilloscope is high impedance, use a T connector on the front of the oscilloscope, with the line on one side and the termination on the other. That will minimise reflections.
Do you have a 75 ohm T connector?
Take care when mixing 50 and 75 ohm connectors that look much the same. They can damage each other due to slight differences.
If possible use a 'T', 'Pi' or 'L' attenuator to convert between the 50 and 78 ohm worlds.

 

[CWatters]

Just noticed this bit in bold..

I am measuring a communications bus with an oscilloscope at various points along the line. I understand that there will be reflections at each stub as well as at the ends of the bus if the impedance doesn't match.

Does that mean you have multiple receivers along the line?

 

[berkeman]

I am measuring a communications bus with an oscilloscope

Are you using a low-capacitance differential probe to make these measurements?

 

[Mike Vines]

Thanks to all who replied. Here is some more information.

What kind of cable do you use?

Here is the cable: goo.gl/Sbrsgi

There are some good articles on the web on why different termination strategies are appropriate.../QUOTE]

Thank you for pointing me to those references. I will take a look at them.

Do you have a 75 ohm T connector?

The line is terminated with a 78 ohm resistor when I measure at the stubs along the line. When I measure at the end of the line, I remove the terminating resistor and plug it directly into the 50 ohm scope. Would this (goo.gl/YlikNS) 75-to-50 ohm connector be a good choice to interface to match the line/scope impedances?

Does that mean you have multiple receivers along the line?

Yes, I have multiple receivers along the line. I would like to look at the signal at each receiver/stub and at the end of the line to see if they are all receiving the same signal. I thought the reflections might affect the signal integrity depending on where I measured.

Are you using a low-capacitance differential probe to make these measurements?

I initially was using another piece of twinaxial cable to make a single-ended measurement. Upon realizing that I was grounding the negative conductor, I picked up a test cable that split the positive and negative conductors out into two coaxial cables (equal length) with BNC adapters. I connect the coax cables to Channels 1 and 2, respectively. I then do (Ch.1 - Ch.2) to get the differential signal. I assume the capacitance would be similar to that of the cable. Do I need a special probe? I had read somewhere that I might need an active differential probe to remove common mode voltage, but I don't know how important that is.

 

[Svein]

The line is terminated with a 78 ohm resistor when I measure at the stubs along the line. When I measure at the end of the line, I remove the terminating resistor and plug it directly into the 50 ohm scope. Would this (goo.gl/YlikNS) 75-to-50 ohm connector be a good choice to interface to match the line/scope impedances?

Oops - if you attach stubs to a transmission line, you get all kinds of reflections. On a multidrop line, the "stubs" should preferably be less than 5mm.

 

[Baluncore]

When I measure at the end of the line, I remove the terminating resistor and plug it directly into the 50 ohm scope. Would this (goo.gl/YlikNS) 75-to-50 ohm connector be a good choice to interface to match the line/scope impedances?

You are using twinaxial cable, not coaxial cable. Coaxial adapters will not help you.

 

[berkeman]

I initially was using another piece of twinaxial cable to make a single-ended measurement. Upon realizing that I was grounding the negative conductor, I picked up a test cable that split the positive and negative conductors out into two coaxial cables (equal length) with BNC adapters. I connect the coax cables to Channels 1 and 2, respectively. I then do (Ch.1 - Ch.2) to get the differential signal. I assume the capacitance would be similar to that of the cable. Do I need a special probe? I had read somewhere that I might need an active differential probe to remove common mode voltage, but I don't know how important that is.

With respect, your whole measurement approach appears flawed. Can you post your statement of purpose for this experiment, and your whole approach? I don't think we can help you much if you post small bits of what you are trying to do.

A balanced diff probe is fundamental to any TL measurements, and if you are not understanding why, it's very hard to give you good quality help on your TL experimental investigations, IMO.