# Capacitor Across Resistor

Discussion in 'General Electronics Chat' started by mlkcampion, Dec 11, 2006.

1. ### mlkcampion Thread Starter Active Member

Nov 16, 2006
59
0
Hey Forum
I have often come across sensor circuits, working
on the 4-20mA scale where a campacitor (470uF) is usually
put in parallel to a 100ohm resistor, what is the reason for
this? What is the effect of putting another capacitor in parallel,

Thanks
Michael

2. ### JoeJester AAC Fanatic!

Apr 26, 2005
3,387
1,211
Could you put a sample of the schematic for us to view?

3. ### Papabravo Expert

Feb 24, 2006
10,338
1,850
It provides a low impedance path for high frequency noise around the current sensing resistor. I think it is questionable how much good this does especially if the thing which follows the current sensor is an A/D converter which stores charge on a capacitor to do the conversion. Either way you want a low pass characteristic for process variables.

Just as an exercise tell what the Reactance of 470 uF is at 1 Hz., 10 Hz., 100 Hz., 1 kHz,. and 10 kHz. This frequency range pretty much covers what you find in terms of process variables.

Hint: Xc = 1 / (2*pi*f*C)

4. ### mlkcampion Thread Starter Active Member

Nov 16, 2006
59
0
Hey
Cheers for the reply, it would be an analogue
to digital converter sensing the voltage across
the 100ohm resistor. Would i be right in saying that
the 470uf would really only start becoming usefull
around 10Khz.

The reason i was told for this capacitor was to slow down
the response of a pH sensor. Its quite jumpy and i think that
is due to the process as oppose to electronic interference.
I personally dont think a simple capacitor is going to have the
effect of averaging the incoming signal????

Thanks
Michael

5. ### Papabravo Expert

Feb 24, 2006
10,338
1,850
The point of the 470 uf capacitor is not to average, but to "short circuit", in an AC sense, high frequency noise, by providing a path with an impedance very much less than 100 ohms. If you want additional filtering you should place it after the initial opamp stage which detects the current and outputs a voltage. By placing the corner frequency at less than 10 kHz. you will in essence be doing the averaging that you seem to be looking for.

6. ### Dave Retired Moderator

Nov 17, 2003
6,960
145
To emphasise the point Papabravo is making:

At 1Hz:

Xc = 338.62 Ohms
R = 100 Ohms

Xc > R therefore the majority of the sensed signal is through the sensing resistor.

At 10kHz:

Xc = 0.03 Ohms
R = 100 Ohms

At 10kHz, the capacitor is effectively a short circuit bypassing the resistor, very little of the sensed signal would pass through the sensing resistor. For 10kHz noise impressed on a 1Hz signal, it should be clear that this arrangement acts as a simple (if rather crude) high-frequency filter.

Dave

Feb 24, 2006
10,338
1,850
Hiyo Silver

8. ### mlkcampion Thread Starter Active Member

Nov 16, 2006
59
0
Hey
Thanks for the response, basically there really
ain't much i can do with a simple capacitor
to stop the signal from varying so much because the 4-20mA signal which
is converted to voltage 0.4-2Vdc (100 * 4-20mA) is input to
How does some averaging program inside the PLC sound?

Michael

9. ### Papabravo Expert

Feb 24, 2006
10,338
1,850
The problem with digital signal processing is that high frequency signals may alias for lower frequency ones. Without a good analog anti-aliasing filter you're just plain SOL. My advice is to build a suitable front end for the 4-20mA input. If that is impractical then learn to live with it.

10. ### Dave Retired Moderator

Nov 17, 2003
6,960
145
Absolutely. A point to make about anti-aliasing filters is that they still can suffer from the effects of signal aliasing because no filter can provide a transition-zone of zero width. Say for example you wished to filter all noise beyond 5kHz, then sampling at 10kHz would in theory prove a filter with cut-off at 5kHz (the Nyquist point). However due to the transition-zone width constraint it is likely that you would experience with frequencies up to 6kHz, which would be aliased down to 4kHz, even with a high order filter.

The remedies; 2 options:

1) Sample at a higher rate, which has the effect of increasing the Nyquist point and would have the effect of removing the undesired frequencies which are being aliased around 5kHz. However this wastes bandwidth that you may not have at your disposal and high signal processing requirements.

2) Or increase the order of the filter, which reduces the transition-zone width, however you increase the risk of instability and the requirement for high precision components becomes unattainable.

In reality a compromise is the best option with a slight increase in the sampling rate and a logical decision to increase the filter order to managable proportions. In all reality you would probably be able to get a DSP chip of the shelf for a relatively decent price that will implement your requirements.