Sending signal by voltage...or by current? 2

[schnell]

Hello,

The basis of this thread is:

If you have a signal that you are sending across a PCB to a uC…..is it better to send the signal as a current , or as a voltage?

(incidentally, the PCB has three ~5W SMPS’s on it.)
- - - - ---- - - - - - - --- - - - - - - - - - - - - -- -- - - -
….is it true that sending the signal as a voltage is bad because it may get corrupted by noise?

Here is the circuit diagram…it’s the signal from a thermistor which gets buffered by an op amp then sent to a uC……(the uC’s ADC reads the voltage)

CIRCUIT DIAGRAM:

http://i56.tinypic.com/125qqs9.jpg

(Vcc = 5V)
(-the actual thermistor is off the board, and is connected by wires and a shrouded header on the PCB)

The thing is, the PCB track from U4A output to PTA4 on the uC, is about 10cms long, ……..this is a voltage signal, from the opamp.

Would it have been better to put the opamp nearer the uC….and then basically the output of U4A would not have to travel so far to the uC ?……………..

…….the actual thermistor signal (current flowing through the thermistor) would then have to travel further to get to the opamp, but that’s ok(?), because that is a current…..the actual thermistor signal does not get converted to a voltage until it goes into resistor R31 , which is next to the opamp.


So which is best for sending a signal, current or voltage?

-and also, why is such the case?

 

[zekeman]

Myu sense is that any signal from a thermister is inherently a fairly low frequency signal and it would make virtually no difference.
I would assume that the signal is probably too strong to be corrupted in any meaningful way.

 

[OperaHouse]

I'm not fond of sending board power out to a thermistor or anything. That is just waiting for a short to happen.

 

[IRstuff]

There are way too many variables to say without knowing way more details about your system. As a general rule, though, you do not want your low-amplitude analog signals and high-gain amplifiers near processors, since the clock switching noise can infiltrate into your analog signals that you want to keep quiet.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[2dye4]

Keep you opamp and its surrounding components as close to the thermistor as possible. Amplification is always best done near the source of the signal.

Make sure the ground for the components of the op-amp is at the same potential as the ground at the microcontroller. Keep this ground as separated as possible from the grounds carrying the switcher power currents.

As for the signal current or voltage the usual board level solution keeps the signal voltage if it originated that way. You may put an RC filter near the micro A/D converter input to cover noise coupled into the line after the opamp. For a thermistor signal which will not change rapidly no matter what use heavy filtering. 10k / 1uf ceramic.

 

[zappedagain]

Don't forget that your A/D typically wants to see a voltage, so you have at least a current to voltage converter at the end. Staying with voltage all the way through is simpler.

I typically use a circuit similar to yours but a bit simpler. I put the thermistor on the bottom and the bias resistor on top so I don't have to send the power supply directly to the cable. I also use a voltage reference to bias the thermistor instead of a power supply, otherwise there is no power supply rejection in your temperature circuit. I digitize this voltage divider directly and use a LUT in the micro to convert the voltage back to temperature.

John D