Issue
- What is an acceptable induced voltage on to I/Net signal terminals
Environment
- I/net
- SCU
Cause
If you have 10V which is induced on to an input cable. And The installer ran
24Vac alongside signal wires in a multicore cable with a common screen. And have a plan to eliminate the induced voltage by reducing the number
of pairs carrying voltage and using other cables
Resolution
Of the input signal types, the Thermistor type is considered the most vulnerable to the pickup of stray voltage from sources in the local area. This is because the AI input is a high input impedance (as it is with all voltage measurement modes of operation), AND the signal source is also very high impedance (the 10K thermistor itself). With one side of the circuit grounded, this presents the 10K ohm impedance of the signal wire as a prime candidate for inducing AC voltage noise from and local adjacent AC voltage sources AND especially any such run in the same cable.
There is no “acceptable limit” definition on such AC pickup.
Acceptability is directly related to the system application and/or customer tolerance for fluctuations of temperature readings.
In the room temperature range, a thermistor signal changes by only ~ 54mv for each degree C.
Without filtering, this 54mv/C sensitivity would produce constant variation form small induced noise levels.
We provide both hardware and software filtering on the analog signals that allows rejection of AC noise seen on the input terminal, BUT you still want to minimize the level of the noise pickup as seen on the analog inputs.
In the I/NET controllers the hardware filter provides a 94/95 % reduction of the 50/60 hz AC noise level presented to the controller. The controller firmware follows with implementation of deglitching and averaging filters that further reduce the effect of noise on the AI point value. We also have a input filter technique specifically targeted at reducing the impact from AC line induced noise. That filter works on the principle of making the AI readings at a scan rate of 1.5 times the AC line frequency. This is the purpose of making the AC line frequency a configuration item (50 or 60 hz) in the CLAN controllers. On the SubLAN, this is scan rate is fixed at 25ms (1.5 times 60hz) which make the line frequency filter not as effective for 50hz, but still very helpful. As the analog inputs are scanned, we capture a reading on the negative or positive cycle of the induced noise and then we average that with a reading taken from the opposing cycle of the induced AC noise as seen 25ms (or 20ms) away from the previous sample. The average of the positive and negative cycle of the induced noise creates a near complete cancelation of the induced error – for a sinusoidal wave noise signal. This technique does a great job of producing a notch filter centered on the AC line frequency and reducing/filtering the impact from such line frequency noise. This combination of several filters allows you to see AC line induced noise on the analog input terminal, and still maintain a stable temperature reading in many cases. Note -- If the AC noise is non-sinusoidal, due to anomalies in the noise transfer or originating from an AC source other than the sine wave AC power line, then the affectivity of the noise reduction will be reduced and signal reading impact will be larger.
The 10V of induced voltage you describe is very large (huge) and far beyond anything I have seen, but then I have never seen 24VAC wired in the same cable with thermistor signals. Separation of the 24VAC as you describe below is needed.
Your described past tolerance for 5V is also above what I would expect to accept.
However, it is possible that the AC line frequency filter has been making such high noise level appear acceptable on the temperature reading display.
For possible work-around use, the high impedance of the thermistor source makes the addition of a capacitor on the AI terminal to ground very effective in reducing the applied voltage from AC noise pickup. Here a capacitor value in the range from 2 to possibly 100 micro-farads can sometimes be useful to reduce or work-around an installation wiring problem. The capacitor should be of the “Low Leakage” type. Leakage in the capacitor would be exhibited as a small reduction in the DC voltage and associated increase in the temperature reading. The higher the micro-farad capacitance value, the more reduction in noise you will achieve, but it is easier to obtain low leakage capacitors in the smaller capacitance values. If polarized electrolytic capacitors are used, be sure to position the negative terminal toward ground.
