Inductive Current Sensor


Current sensors may be used for many applications. Specifically they are suited for track clear detection, but with the condition of all wheel-sets of all cars being equipped with resistors. Furthermore they may be used for local contacts, where only locomotives or cab cars (with headlight) need to be detected. In these applications no resistor wheel-sets are needed.

While with analogue control an auxiliary voltage is needed to detect standing locomotives, there is no such requirement with digital systems due to the permanent track voltage. Furthermore due to the permanently changing polarity it is sufficient to test one polarity only. However the direction can't be detected that easy.

Digital systems like DCC also allow to use inductive sensors as with any other "AC" supply. The AC is set in quotes as the digital signal is not sinus shaped as AC usually is, but a square wave shaped signal. This results in the possible problem, that at the edges parasitic capacitors result in current peaks, leading to a false occupancy signal.

I have been inspired by the pages from Rob Paisley. There is an inductive sensor presented together with other sensors. The VT-5 coils used were from Coilcraft but are out of production. They were distributed by Jameco with the part number J9199-A, the only similar device at Jameco is the 164718PS as used by detector 5 at the Teton Short Line.

Advantages of Inductive Sensors

The most common method for current detection is the usage of the voltage drop along a diode. Diodes are well suited due to their almost constant voltage drop over a wide current range. But this voltage drop causes disturbance especially if not all trackage but only small portions of it is equipped with sensors. Especially if only a small section within (D)Streckenmodule are sensed, a compensation for the voltage drop is quite difficult. Schottky diodes are of little help here, as their lower voltage drop is only valid for small currents. Concerning the voltage drop with high currents they are not as good as standard silicon diodes.

Furthermore the diode method does not automatically provide galvanic insulation between track and the controlled circuit. This insulation is required for operation at FREMO, to avoid problems due to track currents entering the signalling circuit or undesired feedback of the sensors to the track. Especially ground loops may lead to major disturbances. With diode based detectors the galvanic insulation is implemented usually with opto-couplers. Either there are two diodes in series are used to supply the LED in the opto-coupler directly from the current drop, allowing the even higher voltage drop, or there is an amplifier supplied by the track voltage needed.

Inductive sensors offer galvanic insulation combined with a not measurable change of the track signal. Even with a locomotive across the gap between sensed and not sensed track, getting current via both districts, the current is shared and the locomotive is detected.


The coil used with the inductive sensor operates like a transformer. The wire passed through the coil acts as primary winding and the wire may be passed trough multiple times to increase the primary winding count. But it does not transform voltages but currents. The high track current is reduced to a smaller number. At the same time more voltage is available allowing to control a transistor. Using a coil with 50 turns and the track power wire passed through the coil once the current is reduced by a factor of 50. This explains, why a smaller number of turns at the coil makes the sensor more sensitive and a more turns makes it less sensitive. On the other hand a track wire feed through the coil more then once, i.e. increasing the primary turns, increases the sensitivity.

With a winding ratio of 1:50 the 1.5 mA drawn by a wheel-set with a 10 kOhm resistor and 15 V at the track are transformed into 30 μA; just enough to control a transistor. With a short circuit current of 5 A, 100 mA are delivered to the base of the transistor. But a standard transistor tolerates up to 500 mA base current.


Circuit with potentiometer (2K)    The circuit is quite easy: The transformer is in the diagram only a coil, as the primary winding is just the wire passed through the coil. The transistor is controlled via a resistor. With the potentiometer the sensitivity may be adjusted. A capacitor to stretch the pulse and a free wheeling diode completes the circuit.
Figure 1: Circuit of the sensor  

The resistor dampens any possible oscillations; the diode prevents high voltages across the base emitter section of the transistor with negative voltages from the coil. The capacitor is needed as the transistor opens with small loads only at the edges and a sampling micro controller may miss these short pulses. Also a turn off delay for applications without micro controller may be implemented by a larger capacitor. The transistor may be of any type. However for a high sensitivity of the sensor it needs to discharge the capacitor fast. Therefore I used the C type of the BC547. A BC337-40 reacts slower with tiny loads, but can draw more current. Therefore it leads to a higher sensitivity when a larger capacitor is used.

The key component is the coil. After a long discussion with Helmut Schäfer, who I would like to thank at this point, I decided for a 50 turn type, to get relatively sensitive. There is a commercial sensor from NCE using a type with 200 turn, designed for currents up to 20 A. As within FREMO there are no boosters used which deliver more then 5 A and even with 10 A the transistor would not be damaged, I regard 50 turns as a good value. Coils with less then 50 turns are also not very common.

     Circuit with capacitor (2K)
   Figure 2: Circuit of the track power switch sensor

With my application – the track power switch – I don't want to trim the sensor each time it is used, but the sensor doesn't need to be very sensitive. Therefore I tried to get rid of the potentiometer. But a test on a real layout showed problems with some modules, showing occupancy even without anything on the track. The capacitance of paired wires resulted in enough current at least at the edges of the DCC signal. The software delay made the sensor even more sensitive to this effekt. A capacitor at the base of the transistor solved the problem. There was no false detection by any of the 28 sensors used at the meeting in Alsfeld. even with the capacitor the sensor is sensitive enough for this application.

Apart from this simple basic circuit there are of cause many extensions and variations. Helmut Schäfer developed some circuits including one with an NE555 to generate a pulse with defined and length and to directly drive a small relay.

Circuit (4K) Figure 3: Circuit of the sensor with NE555 © 2005 Helmut Schäfer

Input Circuit

     Input circuit of the Track power switch (2K)
   Figure 4: Input circuit of the track power switch

With the track power switch I used the same complex input circuit as with the absolute block system (see figure 4). Especially when pulling wires along other peoples modules or let other pull the cable, you are never sure what the cable will be connected to. Using the sensor within your own station in the most simple case the sensor may be directly connected to the micro controller port. An external pull-up is needed, if the large tolerance of the internal pull-up is not acceptable. I you are afraid of external voltages applied a series resistor and a Zehner diode should be added. With the capacitor at the sensor and the debouncing software within the Atmel, the capacitor in the input circuit is not really required.

Where to Get the Coil

Coils for current measurement are produced by several manufacturers, but are not available at most electronic dealers serving private needs. Therefore you have to find a distributor for the corresponding manufacturer, who is prepared to deliver these coils in small quantities.

Manufacturer Coil with 50 Turns Distributor
Nuvotem Talema AS-100 Rutronik / RS (400-8575)
Rhombus L-11001 CompoTEK
Pulse PE-51686 / FIS-101 Spoerle
Coilcraft D1869 / CS1050 Avnet Memec

I bought the AS-100 from Nuvotem-Talema at Rutronik. A total of about 1000 units were distributed within FREMO.


   Sensor top (41K)
   Sensor bottom (43K)
   Figure 5: Top and bottom of the first prototype

Initially I was going to order small PCBs and populate them with SMD parts. But then I noticed that it is as easy to build it on a piece of prototype PCB. The connections of the coil were soldered flat on the board and the transistor was also placed flat on the board, to allow all being covered by shrinking tube. The connecting wire is feed through holes for stress relief.

The wire at the sensor is quite short and got a Cinch connector. Cables with application dependent length may be connected here. I fed a short cable through the coil and added plug and socket for easy insertion into the inter module wiring without any open connector pins.


This section with grow by time. Up to now there is only the first experience from the track power switch test in Braunlage.

Braunlage 2006:

Before the meeting I disabled the internal pull-ups of the micro controller and used only the external 100 kOhm pull-up. Even with a current of 3 mA (two times 10 kOhm as load) no occupancy was detected with the internal pull-ups enabled. The transistor was not able to discharge the capacitor fast enough. Without the internal pull-up 3 mA were detected reliably. To detect a single 10 kOhm load the track feed wire would have to be passed twice through the coil. With the track power switch application this is not necessary.

During set-up two of the four modules equipped with sensors showed occupancy even with nothing on the track. This can be explained by capacitive load. As initially I wanted to be more sensitive, I did not include the potentiometer for sensitivity adjustment. After enabling the internal pull-ups of the Atmel the false detection disappeared. But this shows that you may not simply add a sensitive sensor to any module. To detect 10 kOhm resistor wheel-sets the sensor needs to be calibrated to the module. I will do some testing, whether it is possible to distinguish between capacitive and resistive loads to avoid this calibration.


I developed the sensor primary for the track power switch, using any plain track modules with added sensors to detect trains. There is no modification of the modules needed. Also the sensor may be placed in any of the two feed wires. If the track profiles touch each other on one side, making any electrical break impossible on this side, the sensor may be placed in the other feed wire.

Sensor placed under module (269K) Figure 6: Sensor placed under a module at the test in Braunlage 2006

For the track power switch there are 4 sensors needed. At each of the two gaps two sensors, to detect the approach of a train from either side (from inside and outside of the switched section). In this case the modules used for detection are not connected at the other end. If a module within a booster district is used for detection, modules behind the sensed module need to be supplied by a separate wire to prevent their current to flow through the sensor.

Besides the track power switch there are many other possible applications:

  • Track contact for the absolute block.
  • Turn on contacts for lights and barriers at level crossings.
  • Track clear detection within stations and yards, if we succeed with in FREMO to equip all wheel-sets with resistors.

Valid HTML 4.0!