PLC for Hamilton Visiwater DO Probe


Thread Starter


We have an old probe which we would like to add some additional functionality by connecting it to a PLC for process control. The signal output from the probe is stated at being the following from the manual and I am looking for ideas as to what PLCs will accept this signal. Direct quote from the manual:

"The VisiWater DO sensor generates the 4 to 20 mA signals by pulse with modulation (PWM) which is not compatible to all PCS systems. Also a galvanic separation between the power supply and the PCS is necessary for correct sensor functionality when used in 4 to 20 mA setups."

There is also further detail about the signal:

"Hamilton VisiFerm DO sensors use the method of pulse-width modulation (PWM) to adjust the DC currents of the 4 to 20 mA interfaces corresponding to the measured values. In principle, the pulse width (ti) of a rectangular signal with a constant frequency, the pulse duty factor (ti/T), is modulated and afterwards demodulated by a low-pass filter to generate continuous analog DC signals. The resulting value yi corresponds to the average of the PWM signal (see Figures 12 and 13). The PWM-loads of the Sensors have low-pass filters which are not able to eliminate all AC fractions of the used PWM frequency of 5 kHz due to technical impossibilities. Therefore, the current signals of the 4 to 20 mA interfaces are still overlaid by a certain AC which should be masked by lag smearing or input filters of the current input card of the process control system (PCS). Recommended PCS settings are a sampling rate below 3.5 kHz, an averaging over more than 1 s, and the use of galvanically separated inputs to avoid oscillations. It is also possible to use mathematical functions or isolating amplifiers for signal processing filtering if necessary."

Any help and hardware recommendations will be gratefully accepted.
Wow. It must be an really old DO probe because the honesty with which the manufacturer presents the issues with the analog output read like a legacy manual from yesteryear. I can almost feel the crinkle-painted chassis and see the backlit VU meters.

Anyhow, my exposure to dissolved oxygen has been on very slow processes - ppm ranges for biological processes (wastewater) and ppb range for boiler water. Either could be comfortably sampled at a rate of once per second or slower (particularly the wastewater) and it would be adequate.

So when the vendor warns that sampling has to be less than 3.5Khz, I go, whoops, why kind of oxygen detection is this? Is it anything needing samples every 300 uS? I seriously doubt it. But one has to ask.

Assuming your sampling needs are down around every 100mS or less, like every 1.0S or every 5 seconds, then what you have to take into account is the galvanic isolation and some averaging/damping/filtering.

To avoid a research or an experimental project on exactly what needs to be isolated from what else, I'd use a stand-alond 4-20mA 3-way isolator, typically a $200 - $300 DIN rail mounted electronic module. That way, the isolator's output which connects to the PLC analog input will be isolated from everything on the probe (source) side. That way you can use a less expensive analog input, and shop for one the just meets your resolution requirements (14 bit, 16 bit, whatever), instead of getting isolation at the analog input.

Most 4-20mA isolator modules do not have a spec for throughput or cut-off frequency. But, I found that Dataforth has its DSCT32 which has "Signal filtering . . . accomplished with a five-pole filter, which provides 80dB per decade of normal-mode rejection above 100Hz". The oxygen probe's PWM noise is normal mode noise so the DSCT32 filtering should be suitable.
check it out here:

With regard to filtering/damping/averaging in the controller, I work primarily with PID loop controllers and PAC (process automation controllers) which always have easy-to-configure filter/damping/averaging settings on analog inputs. But I see a large number of threads on the PLC forum sites asking for logic to do filtering/damping/averaging on analog signals so apparently filter/damping/averaging is not as common on the low and mid level PLCs as it is on PACs (filtering does seem to be available on the high end PLCs). Others here can comment on provisions for filtering on the PLC. But if you do filter in the isolator, you shouldn't need to filter in the PLC (again, assuming the is low speed data acquisition).