Twisted pair to BNC noise issue

[John_Doe_]

Hi,

I am in the process of compiling a sensor system for an offshore asset using accelerometers. The differential voltage signal is transferred in a twisted pair cabled and is connected to data acquisition (NI-9239) via BNC. In my testing it seems that a lot of noise is introduced in the connector between the cable and the data acquisition equipment. Do you have any recommendations on how to connect a twisted pair cable to a NI-9239 via BNC? Or commonly used connectors from twisted pair to BNC?

[Henrik_Volkers]

The accelerometers already have signal condinioners? (IEPE with 10 to 20 mA is a common long cable field application, however the 9239 has no IEPE support... and the usual 2-4mA IEPE supply can work but sometimes the extra low impedance help 😉 )

Output impedance?

How long is your twisted pair?

Can you  try a lower the input impedance (1k-10k) ?

[John_Doe_]

Thanks for the response Henrik.

I see that my original post could use a bit more information.

I have nine accelerometers where the only difference is the cabling. For four of them I use the following cable:

For the rest I use the following pigtail in combination a terminal to BNC adaptor:

With this setup I get the following RMS-values at idle:

It is pretty evident that something goes wrong for sensor 5-9. As the cables are from the same manufacturer I assume that it is the connector that is to blame. I have also tried splicing the pigtail to a BNC cable, but I am not able to achieve better results using that setup.

[Henrik_Volkers]

What are the types and sensitivities (cal-data) of your sensors?

The RMS voltage noise is about 3.5 times higher,  a 8315A2D0 4000mV/g to 8330B... 1200mV/g matches 3.3 times the sensitivity 😉

How does it look like if you compare scaled RMS values?

(assuming all sensors see about the same input :D)

[John_Doe_]

All sensors are 8316A2D0D0TA00:

https://www.kistler.com/?type=669&fid=78520&model=document

The sensitivity is very close to 4000 mV/g.

[Henrik_Volkers]

Have you checked (pin to pin) both cable connections to be equal? 

You want pin 1 & 2 to for +- power and 4&3 for +-signal..

The 1786D cable seems to be wrong according to this doc:

https://www.kistler.com/?type=669&fid=79055&callee=frontend

[John_Doe_]

It seems that I have used a wrong picture of the cable. The correct name is 1786C10 and it looks like this:

[Henrik_Volkers · Accepted solution]

And how is it wired?

 IF the power -  and signal - are swaped the sensor migth still work (due to protection diodes) but you get only half the amplitude... 

Just check the connections to be pin to pin equal and correct 😉

[John_Doe_]

I have just checked the cable and I am quite surprised. The "good" cable is single ended and not for differential voltage as it should be.

I am very grateful for your help Henrik. Thanks!