I am building a circuit in which an analog signal ( sine wave) of around 25 mV leaves the main pcb board to a second module pcb through an HDMI cable. Inside the HDMI cable, I am using one of the twisted pair for my signal. One wire of the twisted pair carries the signal on its outgoing path from main pcb to module pcb, and the other wire of the twisted pair carries the signal on its incoming path from module pcb to main pcb. On the module I have an unknown Impedance which I want to measure. The 25mV signal of a known frequency is generated by an IC on the main board. It then passes through the unknown impedance on the module pcb and returns back to the IC on the main board which then calculates the Impedance. I want to measure the impedance for a range of frequency from 100 Hz to 100 KHz.

1) I am using a double shielded HDMI cable and the shield I have connected to ground. My 25mV analog signal has a DC bias of 2 V. The shield will protect the signal from external ac noise signals but will there be a capacitance between my shield and signal which could be a source of error in my system? Should I connect my shield to 2 V?

2) Error in my results in the range of frequencies from 25 KHz to 100 KHz increases with increased error for a higher frequency. I am not able to understand whether this is due to ineffective shielding of my signal?

I have attached a simple drawing indicating the path of signal. I am looking for some guidance regarding my questions. Thank you.

 

There will be stray capacitances between the two wires and the shield and the other wires in HDMI cable. Twisted pairs are pretty good at rejecting external interference so your results may be better if the shield and all the other wires in the HDMI cable are left disconnected. None of those other wires have signals on them do they, for that would be another source of error.

 

There will be stray capacitances between the two wires and the shield and the other wires in HDMI cable. Twisted pairs are pretty good at rejecting external interference so your results may be better if the shield and all the other wires in the HDMI cable are left disconnected. None of those other wires have signals on them do they, for that would be another source of error.

Thankyou for your reply. Inside the HDMI cable, 3 other twisted pair are used to carry the same signal but not at the same time. I use a switch on main board to send signal to only one pair at one time. But there is DC signal on five other wires, which I use as control signal for switches on my module board, which is either 4V or 0V. These DC signals dont switch very fast and change with a minimum delay of 5 miliseconds. Also, I have one wire carrying 4 V DC as power signal for module board and one wire carrying ground signal.

If I leave the shield disconnected, will it still protect from external noise sources?

 

The fact that it is a twisted pair provides good protection. The other wires, whether connected to 4V or 0V will impose extra capacitance on the pair.

 

The fact that it is a twisted pair provides good protection. The other wires, whether connected to 4V or 0V will impose extra capacitance on the pair.

okay that I understand. But what about the shield? If I leave the shield disconnected, how will it still protect from external noise sources?

 

The twists cancel it out. Have a read here: https://en.wikipedia.org/wiki/Twisted_pair#Unshielded_twisted_pair_.28UTP.29

 

The twists cancel it out. Have a read here: https://en.wikipedia.org/wiki/Twisted_pair#Unshielded_twisted_pair_.28UTP.29

I was trying to find out how UTP works and read this somewhere that in the twisted pair, interference is induced to both the wires of the twisted pair which can be cancelled out when signal is measured in differential mode at the terminating equipment. If this is how UTP works, for my circuit I am not cancelling this common mode noise anywhere. The signal which passes through the twisted pair is just one signal and how will the interference cancel here automatically?

http://www.cablinginstall.com/artic...shooting/how-utp-deals-with-interference.html

 

Twists reduce EMI when differentially driven
Shielding reduces RFI
Shield at one end only - usually the source

You will need to do an ope/short calibration to cancel the effect of the cable.
Impedance measuring usually uses a force/sense scheme.

 

If this is how UTP works, for my circuit I am not cancelling this common mode noise anywhere.

Then that will be a problem.

Impedance measuring usually uses a force/sense scheme

And this may be the solution.

 

Shielding reduces RFI
Shield at one end only - usually the source

You will need to do an open/short calibration to cancel the effect of the cable.
Impedance measuring usually uses a force/sense scheme.

Shielding reduces RFI so I should connect shield to ground even though my shield and and my signal reference are at different potentials?

At the moment I calibrate system on few known impedances which are also on the second module pcb and then measure an unknown impedance which is also on the module pcb. This I thought would cancel the effect of cable but it only works till a particular frequency. At frequencies greater than 30KHz, the system behavior saturates in a way that all values higher than 500 KOhm are measured as around 500 Kohm. It seems there is some leakage noise current into the system but I dont exactly know what is happening.
could you please tell a bit about how would open/short calibration help?