noise problem with differential analogue input

[johnsmith]

hi

I have noise problem with my differential analogue input circuit.

Input Part #1: http://john111smith.googlepages.com/DIOP0.JPG
Input Part #2: http://john111smith.googlepages.com/DIOP1.JPG

All inputs wires are coaxial ones(but very long;about 15 meters)
Case/Chassis has connected to Earth of system.
Also, there is an averaging function for 50Hz noise at microcontroller.

I will be appreciated with your comments in this

regard.

Thanks in advance
J.Smith

 

[Russlk]

What kind of noise? Why is it a problem? What is the signal level?

 

[RadioRon]

Possible problems:

The differential input impedance of your circuit is unreasonably high, at around 1 Mohm (R3 to R38). It would be better if you could lower the impedance to something in the range of a few hundred ohms, but I don't know what your source impedance is, so cannot say for sure what is the best value. If you are sending high bandwidth signals through the coax, you may also need to pay attention to impedance matching, which may force your load impedance down to 75 or 50 ohms.

Construction is important in cases like this. We need to know how you connected the shield of each coax cable, as this is not shown on your schematic. Please also explain how you have connected C103 and C107 since these are your common mode input filters and need to be connected to a common point.

Definite problems:
Your schematic is not drawn well. You should put all components on one page and directly connect the G and S nodes so that the signal flow is more clear. You should also use standard symbols if you want others to understand your circuit. In this case I mean that the method you use to show "No-connects" as if they were the legs of a spider on your op amp symbols is non standard. You should take greater care in allowing enough room so that your capacitor symbols don't touch each other and so that your component labels don't overlap any other lines or symbols.

 

[Roff]

And what the heck are those series back-to-back diodes about?

 

[mneary]

Diodes don't come up as a valid part number. Did you manufacture them yourself?

Is J1 your input? Where is the ground and/or common?

Unreadable schematic.

Please clarify.

 

[RadioRon]

The diodes appear to be for overvoltage protection. Tranzorbs?

 

[johnsmith]

Tanx for your posts.

The main problem was fluctuation of common lines that solved with sharing all commons with some resistors to GND of circuit. but it till suffer from some (about +/- 2%) reading fluctuation especially at one of inputs. Because of some field problems.unfortunately, I have to record scope trace of signal fluctuations in few next days.
Signals are DC levels and the circuit read them at 50 reads/sec (each read is average of about 194 [10bits ADC] samples at 20 milliseconds)
J1 is input of circuit. regarding cable sending & receiving; it is both ends. also, it is audio coax.

 

[Russlk]

With such high impedance the coax will be microphonic. Also, you should ground only one end of the coax otherwise there will be a ground loop.