Measure DC current with a PLC analog input

What the voltage divider leaves out

The divider drops array voltage into the input range and the controller scales that number back to volts. It senses one quantity. The small drop along wiring and busbars depends on the load and connection resistance too much to serve as a current measurement, so there is no current value in the controller to plot, log or scale. Adding the current channel means adding a device that converts conductor current into a signal. It does the same job the divider does for voltage.

Why an AC current transformer reads zero on DC

A conventional current transformer develops a secondary signal only while the magnetic flux in its core is changing. Steady DC makes constant flux. So a clamp or transformer rated for AC alone sits at zero on a photovoltaic string however many amperes flow. The sensor for this job is built for DC. A Hall-effect transducer measures the magnetic field in its window and outputs a signal proportional to the window current, with the controller fully insulated from the conductor. A low-value shunt in the conductor is the other option. It suits larger currents, as the FAQ below explains.

Match the sensor output to the analog input

Transducers in this class carry jumper-selectable input ranges and a choice of output: 0-5 V, 0-10 V, ±10 V or 4-20 mA. Common window ranges go from 150 A and 200 A full scale upward. Order the output variant the spare channel accepts directly. A current-loop output on a voltage channel needs a sense resistor across the input: 250 ohms turns the 20 mA full-scale loop into 5 V, and the 4 mA floor into 1 V. That 1 V floor is the live zero: a reading below it flags a broken wire or a dead transmitter, a fault a voltage output leaves invisible. The farther the sensor sits from the controller, the stronger the case for the current loop. The article on voltage vs current analog signals for sensor runs works through distance and noise. Check the sensor's documented supply voltage against your control power before ordering.

Multiply a thin conductor through the window

A window sensor responds to ampere-turns: every pass of the conductor through the window adds the conductor current once to what the sensor sees. Winding N passes multiplies the sensed value by N, so the controller divides the scaled window amperes by N to recover conductor amperes. Two limits hold. N × the maximum conductor current must stay inside the sensor's window range. Each counted pass must go through the window, because a loop that misses it adds nothing. For a conductor under about 1 A the working count is in the tens. The worked example below runs one case end to end.

What you need

• A DC-rated Hall-effect current transducer whose window range covers N × the conductor maximum and whose output matches a spare analog input
• A window that fits the conductor; 12 mm and 15 mm holes are common in this class
• The sensor datasheet: transfer function, supply voltage, direction marking
• A clamp meter to read the true conductor current at two operating points
• A 250 ohm sense resistor if a current-loop output lands on a voltage input
• Sensor supply wiring, with the sensor's output common tied to the input channel common
• The controller's analog input documentation: channel range, raw counts, scaling method

Wire, verify, then scale

1. Read the true conductor current with a clamp meter at the highest and lowest operating points the measurement must cover; the highest sets the range and the pass count. A PV string is live whenever light strikes it, so clamp around the insulated conductor, or isolate with a rated switch before breaking the circuit for a series meter.
2. Pick the pass count N and set any input-range jumper to the smallest range that holds N × the measured maximum. For a sub-amp conductor, 25 passes is the working count on the smallest commonly stocked window.
3. Wind N passes through the window with the conductor insulated, following the marked direction arrow, and count passes by the conductor's entries through the window.
4. Wire per the datasheet: output to the channel's positive input, sensor common to the input common, sensor supply as documented. For a current loop onto a voltage channel, wire the 250 ohm resistor across the input terminals so the loop current flows through it.
5. Verify the raw signal before touching the scaling. Meter across the input at both operating points and compare with the datasheet transfer function (window current × volts per ampere). If the meter disagrees with the datasheet, fix the sensor, supply or wiring first; scaling arithmetic cannot repair a wrong signal.
6. Scale the channel: counts to volts, volts to window amperes by the datasheet, then divide by N. Keep the voltage-divider scaling out of this channel. Every controller moves the raw counts into memory in its own way; one walkthrough is reading an analog input into V-memory with the ANLGIN I-Box.
7. Compare the controller's amperes against the clamp meter at both operating points and confirm the value rises and falls with the load. If meter and controller still disagree after step 5, the scaling arithmetic is the suspect; why the scale value differs from the PLC analog input walks the usual causes.

Worked example: a sub-amp photovoltaic string on a voltage input

Inputs: one PV string, 0.8 A at its measured maximum (clamp meter at midday; the case stands for any conductor under about 1 A); a Hall-effect DC transducer with a voltage output for a 0-150 A window, the smallest range this class commonly stocks; a 0-5 V analog channel assumed 12-bit, 4095 counts across the span (substitute your channel's documented resolution); and N = 25 passes.

Rule in words: window current = conductor current × N. Sensor output = window current × full-scale volts ÷ window amperes. Counts = output ÷ 1.22 mV, one count being 5000 mV ÷ 4095. Controller amperes = counts × window range ÷ 4095, then ÷ N.

Arithmetic:

• Window current: 0.8 A × 25 = 20 A
• Sensor output: 20 A × 5 V ÷ 150 A = 0.667 V
• Counts: 667 mV ÷ 1.22 mV per count = 546 counts
• Back through the scaling: 546 counts × (window range ÷ 4095) = 20.0 A at the window, and 20.0 ÷ 25 = 0.80 A at the conductor

One pass would give 26.7 mV, about 22 counts, with channel resolution of 36.6 mA per count. A few counts of noise would be a large slice of the reading. At 25 passes the resolution improves by the same factor: 36.6 ÷ 25 = 1.5 mA per count.

Result: the string reads 546 of 4095 counts, 13% of the channel span, and each count is worth 1.5 mA of conductor current.

Decision: keep 25 passes while the conductor maximum stays under 6 A (window range ÷ 25); above 6 A, cut the count so N × the maximum stays inside the window range (10 passes covers to 15 A, one pass covers the full window). Choose the count so one count of the channel is small next to the smallest change you need to see: at 1.5 mA per count, the string's swing from 0.2 to 0.8 A resolves into about 400 steps.

How to check it worked

• Two-point comparison: clamp meter and controller value at both operating points, moving together in direction and roughly in proportion.
• Live-zero pull test on a current loop: lift the signal lead at the sensor and the reading must fall below the 1 V floor, about 819 counts on the 12-bit channel (1 V ÷ 1.22 mV). A value parked mid-scale means the loop or the resistor is wired wrong.
• Pass-count audit: count the physical passes again. One missed pass in a 25-pass winding reads 24 ÷ 25 = 0.96 of true current, a 4% low error that can pass for sensor tolerance at a single operating point.
• Once the scaled amperes match the meter, plotting and spreadsheet export are the next job, on the controller's own logging tools or a trend display.

The decision

Recommendation: fit a Hall-effect DC current transducer whose output variant matches a spare analog input and whose window range covers the conductor maximum; wind conductors under about 1 A through the window 25 times, which carries up to 6 A of conductor current and puts the example's reading at 546 counts; verify the raw signal with a meter at two operating points before any scaling; and divide by the pass count as the last step in the controller. Take the 4-20 mA output when the sensor sits far from the controller or when you want the live zero to flag a broken loop; keep a voltage output for short runs inside one enclosure. A transformer-type clamp is the wrong tool here: steady DC leaves it at zero.

Questions