Is shielded cable necessary for a DC signal? 4

[robd2002]

My company makes a level sensing device that consists of a microcontroller board and a pressure transducer. The pressure transducer output ranges from 0.5 to 4.5VDC. The micro samples this line and sends the answer (10 bit number between 100 and 923) and from this and the density we figure out the height and ultimately the volume. The cable between the micro and the transducer has 3 conductors (Ground, 5VDC supply, and the output). Does this application call for shielded cable? A client has used unshielded cable in many installs with no problems but some erratic behavior from one unit is making me wonder.

Thank you,

Rob D.

 

[MacGyverS2000]

If there is any AC signal riding on the DC line (next to a power cable, radio station nearby, etc.), your signal could be sampled as anything. Shielding is one way, but the smartest solution is to take a running average of the samples so it can run without shielding (in most cases).


Dan - Owner

http://www.Hi-TecDesigns.com

 

[itsmoked]

You really haven't supplied enough info to say in your application or this case.

Generally it would be kind of bad form to not spend a few pennies more and run proper shielding. But, depending on a whole slew of things it might be reasonably acceptable to not run shielding in your particular application.

I would certainly expect a system that works fine without shielding at your facility and several others to eventually be completely hammered down the road somewhere. That is why the prudent engineer would probably run shielding.

With regards to various sampling methods.
You can sample till you're blue in the face with no improvement at all, depending on what exactly the noise source is, and its manifestation in your system.

Sampling and filtering are best used when all shielding and passive filtering options have already been included, and to try to bullet proof the system.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

"My company makes a level sensing device that consists of a microcontroller board and a pressure transducer. ... we figure out the height and ultimately the volume."

Where were you when we had The World's Longest Thread, six trillion posts, on the best way to measure the depth of water in a well. ;-)

As the others have already stated, shielding is probably a good idea.

The circuit could be designed (relatively easily) to be extremely immune to incoming noise on the PS input, and the output could be designed to use a form of signaling that is extremely noise immune. But if it isn't so designed, then it's a matter of the environment.

And if the PS is susceptible, then the micro may go insane.

 

[zappedagain]

You may have a DC signal, but your channel bandwidth is defined by the anti-aliasing filter in front of the A/D. The wider it is, the faster you can make a decision, but the more noise it will let in.

If you don't have a filter than you are very susceptible to noise, only limited by the analog bandwidth of the A/D, and that is possible much higher than the maximum sampling bandwidth (this is used in undersampling).

John D

 

[robd2002]

Thank you Macgyver, itsmoked, VE1BLL, and zapped. I was able to visit the site and saw the difference between shielded and unshielded cable on an oscilloscope. Suffice it to say the difference was dramatic. I cannot say from where the noise originated, however. It sure seemed that it would be easy to "digitally" filter the spikes (is an average or a median considered a digital filter?).

Alas the find may have been in vain. Boss manager won't allow software to be altered to print out the a/d values being operated upon, so I may never really know what effect the noise had on the calculation, and he went ahead and bought unshielded cable for the next installation!

Thank you again,

Rob

 

[itsmoked]

Good luck Rob.
You need it. LOL

Keith Cress
kcress -

http://www.flaminsystems.com

 

[itsmoked]

Oops! Certainly averaging IS filtering. It can often be done with just some capacitors of the right value.

Keith Cress
kcress -

http://www.flaminsystems.com