Reducing noise of unbalanced shielded twister pair to coaxial cable

[Plasmateur]

I'm working on a project where I am using Cat7 cable as feed through wire into a chamber to make voltage measurements.

I'm using one wire from each of the 4 pairs to make the voltage measurement, making the pair unbalanced.

I cannot use a differential amplifier, only passive components.

For now, the other wire which is not making a measurement is floating. I'm looking for any advice on how to best deal with this wire in order to increase the measurement performance of my circuit.

I've attach the schematic of my circuit.

Are there any best practices for doing something like this?

[DimitriP]

I'm not a best practices kind'of a guy but If it were me, I'd use all the odd wires for one conductor and all the even pairs for the other.
Less resistance, that depending on what kind of current is creating the voltage drop accross your resistor may or may not be as important.

I could be wrong but not many people here are too shy about making corrections !  

[Cerebus]

As always with this kind of question it helps enormously if you provide more context than you have - impedances, bandwidths, etc.

If you connect the other half of each twisted pair to ground, as you've illustrated, you're effectively connecting your signal to ground via the cable's distributed capacitance - this is probably not what you want and, depending on the length of cable, may seriously reduce your measurement bandwidth.

You can minimise the capacitance of the pairs by having the second wire in the pair driven to the same voltage as your signal. In an ideal world this would come from a low impedance source such as an instrument's guard terminal. You won't have that in your setup so you're better off connecting the pairs together as effectively one wire, or just leaving them floating. Grounding them is probably your worst option.

[Marco]

It will still be a transmission line any way. One which not correctly terminated with nothing to damp the reflections, little late to worry about bandwidth.

[Cerebus]

It will still be a transmission line any way. One which not correctly terminated with nothing to damp the reflections, little late to worry about bandwidth.

You just can't say that until you know the bandwidths and cable lengths. If it's 1kHz and 1m then we're a long, long way from worrying about anything to do with transmission lines.

[Plasmateur]

As always with this kind of question it helps enormously if you provide more context than you have - impedances, bandwidths, etc.

If you connect the other half of each twisted pair to ground, as you've illustrated, you're effectively connecting your signal to ground via the cable's distributed capacitance - this is probably not what you want and, depending on the length of cable, may seriously reduce your measurement bandwidth.

You can minimise the capacitance of the pairs by having the second wire in the pair driven to the same voltage as your signal. In an ideal world this would come from a low impedance source such as an instrument's guard terminal. You won't have that in your setup so you're better off connecting the pairs together as effectively one wire, or just leaving them floating. Grounding them is probably your worst option.

Thank you for the response. I like the idea of just connecting both of the wires, I will give that a try.

The bandwidth I need is around 200kHz, but I would really like to get that up to 1MHz.

The cable length of the CAT7 cable could be up to 3 meters but no more.

I did attempt to ground the twisted wire but it seemed detrimental to my signal. I wasn't sure if I should use something like a capacitor to ground for that wire. So leaving that wire floating seemed to give me a better signal.

I am curious though about the effects of the grounded foil shielding surrounding the twisted pair wires.

[eb4fbz]

Read about bob-smith terminations used in ethernet.

[Richard Crowley]

The bandwidth I need is around 200kHz, but I would really like to get that up to 1MHz.
The cable length of the CAT7 cable could be up to 3 meters but no more.
I did attempt to ground the twisted wire but it seemed detrimental to my signal. I wasn't sure if I should use something like a capacitor to ground for that wire. So leaving that wire floating seemed to give me a better signal.
I am curious though about the effects of the grounded foil shielding surrounding the twisted pair wires.

You did not respond to the question about impedance.  What is "R"?
A low impedance circuit won't be significantly affected by the conductor to shield capacitance. While a high impedance will find the capacitance rolling off your high frequency response.   
When you say your (unidentified) "O-scope" has "Z=50", are you talking about the cable impedance (not terribly significant down at 1MHz) or the input impedance of the test gear? (very significant)

[Plasmateur]

The bandwidth I need is around 200kHz, but I would really like to get that up to 1MHz.
The cable length of the CAT7 cable could be up to 3 meters but no more.
I did attempt to ground the twisted wire but it seemed detrimental to my signal. I wasn't sure if I should use something like a capacitor to ground for that wire. So leaving that wire floating seemed to give me a better signal.
I am curious though about the effects of the grounded foil shielding surrounding the twisted pair wires.

You did not respond to the question about impedance.  What is "R"?
A low impedance circuit won't be significantly affected by the conductor to shield capacitance. While a high impedance will find the capacitance rolling off your high frequency response.   
When you say your (unidentified) "O-scope" has "Z=50", are you talking about the cable impedance (not terribly significant down at 1MHz) or the input impedance of the test gear? (very significant)

I meant the input impedance of the test gear. I should probably use 50 Ohm terminators as the scope is set to 1M Ohm.
I know the impedance of the twisted wire is 100 ohm and the coaxial cable is 50 ohm. What is being measured at the end of the CAT7 wire has a dynamic impedance.

I hope I am answering these questions correctly.

[Richard Crowley]

Asking for the third time:  You have a resistor at the center of your schematic labeled "R".
What is the value of that resistor?  Without that, we have no idea what the impedance are here.
The impedance of the various wire and cable are of minimal importance here.
The impedance of your "dynamic load" is very important (and not disclosed). Yes, if it is "dynamic" then what range of values do you expect?  Surely you cannot handle anything from zero (dead short) through infinity (open circuit).
And the impedance of the measurement loop load (resistor "R" in your schematic) is very important (and also not disclosed).

You appear to be concentrating on things that are not important and ignoring the factors that ARE important.
Answering the questions in a straightforward manner will get you much farther than guessing what other things might be (or not) important.

It seems very odd that you are using a scope with a 50 ohm input impedance. Exactly what oscilloscope are you talking about?  Typically only high-frequency test gear uses 50 ohm input impedances.  And you aren't measuring anything here remotely approaching "high-frequency".

[Plasmateur]

Asking for the third time:  You have a resistor at the center of your schematic labeled "R".
What is the value of that resistor?  Without that, we have no idea what the impedance are here.
The impedance of the various wire and cable are of minimal importance here.
The impedance of your "dynamic load" is very important (and not disclosed). Yes, if it is "dynamic" then what range of values do you expect?  Surely you cannot handle anything from zero (dead short) through infinity (open circuit).
And the impedance of the measurement loop load (resistor "R" in your schematic) is very important (and also not disclosed).

You appear to be concentrating on things that are not important and ignoring the factors that ARE important.
Answering the questions in a straightforward manner will get you much farther than guessing what other things might be (or not) important.

It seems very odd that you are using a scope with a 50 ohm input impedance. Exactly what oscilloscope are you talking about?  Typically only high-frequency test gear uses 50 ohm input impedances.  And you aren't measuring anything here remotely approaching "high-frequency".

I apologize for not being clear and misunderstanding.

The resistor value for R be 500k Ohms.

The dynamic load will rise starting at 10k Ohms and ending at "infinity" over approximately 20 miliseconds. The rise of this dynamic load will likely be logarithmic.

Later on, this device will be used to measure a dynamic load which will periodically vary between 10k Ohms and approximately 50k Ohms at a frequency around 200kHz.

The scope is set at 1M input impedance with a 50ohm cable connected to it.
 

[Richard Crowley]

So you are talking about a rather high-impedance circuit.  Where the cable capacitance will start to be an issue with high-frequency signal loss (attenuation).

You could actually measure the capacitance of your Cat cable in all three configurations:
1) With both wires of the pair connected together, and the capacitance measured to the shield/ground.
2) With one wire of the pair used for the signal and the other wire of the pair grounded (connected to shield).
3) With one wire of the pair used for the signal and the other wire of the pair left floating.

Many (most?) cables actually publish the capacitance values for these various combinations in their data sheet.

And only YOU can measure the capacitance of YOUR particular setup.
Especially when you consider the special conditions of the chamber feed-through, etc.
Cat7 seems like an odd choice for this measurement. Not clear why you chose that?

[Marco]

Are there any best practices for doing something like this?

Do a silly dance before each measurement.

You want to take an differential measurement without galvanic isolation and pretend it's a floating measurement, while using an oscilloscope as a very poor differential amplifier, while using normal coax as very poor probe cables.

There is so much fundamentally wrong with all this that it's just par for the course

[Plasmateur]

So you are talking about a rather high-impedance circuit.  Where the cable capacitance will start to be an issue with high-frequency signal loss (attenuation).

You could actually measure the capacitance of your Cat cable in all three configurations:
1) With both wires of the pair connected together, and the capacitance measured to the shield/ground.
2) With one wire of the pair used for the signal and the other wire of the pair grounded (connected to shield).
3) With one wire of the pair used for the signal and the other wire of the pair left floating.

Many (most?) cables actually publish the capacitance values for these various combinations in their data sheet.

And only YOU can measure the capacitance of YOUR particular setup.
Especially when you consider the special conditions of the chamber feed-through, etc.
Cat7 seems like an odd choice for this measurement. Not clear why you chose that?

The data sheets showed significantly lower stay capacitance on the Cat7 than the coax. Also, jamming several coaxial cables into my chamber is a challenge of it's own. But I might just end up having to do that.

I've never made a capacitance measurement before, so I will have to look up how to do that.

[Richard Crowley]

Capacitance is measured with a capacitance meter.
Fortunately, here in 2017 we can get a quite decent meter that also measures resistance, inductance, transistors, etc. etc. for < $20   For example: http://www.ebay.com/itm/172557508334 is just one of literally dozens of similar items on Ebay.