The Current Sense Transformer

In past blogs we've discussed the current measurement features of the ADALM1000 (SMU). The ALM1000's SMU channels can measure DC currents from -200mA to +200mA. Because of the 100 KSPS sampling rate it can actually measure AC currents as well. But the current to be measured has to be flowing into or out of the SMU channel. This limits the range of voltages that the current must be "referenced" to, 0 to +5V. To measure current over a wider range of voltages a current shunt monitor IC like the AD8210 from the ADALP2000 Analog Parts kit can be used. The ALM1000 SMU channels use this same chip to measure the current. The operating input common-mode voltage range of the AD8210 is −2 V to +65 V with respect to the ground pin of the IC. A larger voltage range but still not enough to safely measure the current of a household appliance or lighting fixture operating from 120 V AC. So what can we use to do that? Enter the Current Sense Transformer.

The current sense transformer, is a transformer that is optimized or designed to produce an alternating current in the secondary winding which is proportional to the current being "sensed" or measured in the primary winding. Like any transformer, current transformers isolate the measuring of currents in high voltage circuits to a much lower voltage and provide a convenient way of safely monitoring the actual electrical current flowing in a high voltage AC power line. In our case this could be with the SMU of the ALM1000.

The principal of operation of a basic current transformer is slightly different from that of an ordinary voltage transformer. Unlike a power transformer used to step up or down voltages, the current transformer often consists of only one or a few turns as the primary winding. This primary winding can be of either a single flat turn, a coil of heavy duty wire wrapped around the core or just a wire inserted through a central hole like that shown in the photo of a clamp-on current probe transformer, (model LCTC-0250) figure 1. With this current "probe" the jaws open so that it can be clamped around the conductor carrying the current to be measured without having to disconnect the conductor. Current probes such as this are designed for use in 50/60 Hz AC power applications. The LCTC-0250 probe has a current measuring range up to 100 Amps and a built-in current to voltage (burden) resistor so the output voltage is specified as 15 mV/A.

Figure 1, Clamp-on Current Transformer, model LCTC-0250

Many manufacturers make a range of current sense inductors that are a toroidal coil with a hole in the center that the user passes a wire (or loops of wire) through to sense the AC current, figure 2. Depending on the particular model and the specification these types of current transformers are designed for use in switch mode power supply control systems and can operate over a frequency range from 20 kHz to 200 kHz. The PE-51718 center tapped 100:1 20 mH version is shown. The size, excluding the leads, is 20mm tall x 11mm wide x 10mm deep which is small enough to fit on a solder-less breadboard.

The nice thing about using coil like this as a sense transformer is that you can choose any number of turns you want for the primary winding. Up to when the center hole is filled based on the gauge of the wire used.

Figure 2, Pulse Engineering Center tapped 100:1 20 mH example

A current transformer with a built-in primary winding from CoilCraft is shown in figure 3. Since it is totally encapsulated, we can't tell how it is constructed. The design frequency range for this 200:1, 80 mH example is from 1 KHz to 1 MHz.

Figure 3, CoilCraft 200:1 80 mH example CS4200V-01

In figure 4 we have a surface mount current sense transformer from Würth Elektronik. In this example the primary winding is simply the wide metal strap that goes up and around the central secondary winding. The manufacturer specifies the inductance for this 200:1, 20 mH example at 10 KHz so it probably is also not designed for lower frequency power line applications.

Figure 4, Würth Elektronik 200:1 20 mH example from MID-SNS Family

To test these current transformers I used an ALM1000 and the test circuit shown in figure 5. A 4 V peak-peak sine wave signal is generated by the channel A AWG. The signal is then AC coupled through a large capacitor to the 10 ohm load resistor which converts the voltage into a 400 mA peak to peak current. The current is sensed by the primary winding which is connected to ground. On the secondary side the 100 ohm burden resistor is connected across the coil winding and the resulting voltage is measured by channel B in Hi-Z mode. The other end of the coil is referenced to the fixed 2.5 V rail to center it in channel B's input range.

For the clamp-on probe and the PE coil one of the longer wire jumpers from the Analog Parts kit is inserted through the center hole to use as the primary.

Figure 5, Frequency Bandwidth test circuit using M1K

The input frequency is swept from 20 Hz to 1 KHz in all the following tests. The first bode plot is for the LCTC clamp probe with a single wire through the clamp. Remember that the clamp has a built-in burden so the 100 ohm external resistor was not included for this test case. The magnitude response is very flat, within a dB down to 20 Hz.

Figure 6, LCTC current probe bode plot

Next up is the PE-51718. As we see the response below 1 KHz is not at all flat which is to be expected given the 20 KHz minimum frequency specification. The lighter set of curves are for one wire as primary and the darker set is for 4 turns as the primary.

Figure 7, PE-51718 bode plot

Next we have the CoilCraft CS4200V-01 and Würth Elektronik 750316796 examples. Both are 200:1 turns ratio. The darker curves are for the 80 mH CoilCraft device and the lighter curves are for the 20 mH Würth device. As expected the higher inductance of the CoilCraft device gives the better low frequency response. The CoilCraft device meets and exceeds its 1 KHz minimum frequency specification and the Würth device is probably only flat above a few KHz using this value of burden resistor.

Figure 8, CoilCraft CS4200V-01 and Würth Elektronik 750316796 bode plot

Electronic Burden, I to V converter

One way to improve the frequency response of any current transformer is to replace the resistive burden with an electronic solution, i.e. an op-amp I to V conversion circuit. The single supply rail-rail AD8541 CMOS op-amp is used as an I to V converter as shown in figure 9. The virtual ground at the summing junction, pin 2, presents a very low impedance load on the secondary. The 1K feedback resistor converts the current into a voltage that is measured by channel B at pin 6.

Figure 9, Op-amp I to V converter circuit

To test the frequency response using the op-amp the CoilCraft CS4200V-01 (dark trace) and Würth 750316796 (light trace) are again compared in figure 10. Note the vertical scale is now 3 dB/div. There is a huge improvement in the flatness of the response compared to figure 8 with less than a dB of roll off at 60 Hz. The CoilCraft response is now about as flat as the LCTC current probe in figure 6.

Figure 10, CoilCraft CS4200V-01 and Würth 750316796 op-amp I to V bode plot

Making real world measurements

As a real world example let's use the model LCTC-0250 clamp-on current sensor we saw in the first figure and the M1k to measure the current waveform of an LED Holiday light string. The LCTC-0250 probe has a built-in current to voltage (burden) resistor so the output voltage is specified as 15mV/A. The string consists of 35 white LEDs in series. About a foot of the wire was untwisted and one leg was wrapped 5 times around the clamp. The sensitivity should be about 75 mV/A ( 5 * 15 mV/A ).

The probe is connected directly to the input of an M1k without any additional amplification or filtering. As you can see in figure 11 the current is simple 1/2 wave rectification and the peak current is between 35 and 45 mA. Hard to see exactly with the noise and the signal is too small to trigger on properly and apply trace averaging.

Figure 11, Current waveform without any signal processing

By applying some mathematical wizardry we are able to clean up the noise and make the "signal" big enough (by 10X) to trigger on and use trace averaging. A simple 20 tap box car digital filter with an overall gain of 10 is applied to the captured waveform trace and trace averaging is used (set to average 8). The waveform is nice and clean now and the p-p current is 42 mA.

Figure 12, filtered current waveform

DIY current sense transformer

I'm all about re-purposing stuff for other uses. So I decided to see if I could make my own current transformer. To start we need a high inductance coil to use as the secondary. We also need to be able to wrap a few turns of wire around the secondary coil for the primary. I happen to have a few different 12 V relays in my junk bin that I have no real use for. These relays have a lot of windings of fine wire. The Inductance can be fairly high. So after some destructive deconstruction on three examples (before and after modification) this is what I've come up with:

Figure 13, Making a CST from a 12 V relay

Starting with the first attempt on the left, this relay, most likely for automotive applications, took probably the most deconstruction. The black plastic cover just pops off but I had to cut away most of the other plastic holding the spade connectors and added jumper wires for the primary and secondary coils. For this one I added just 4 turns for the primary but there is room for more. A little hot glue was added to hold the wire in place. The coil inductance was about 350 mH with a DC resistance of 82 ohms. It worked well enough at higher frequencies but the relatively low inductance limited the response. The one in the middle needed the least deconstruction. After popping off the cover I just wrapped 6 turns around the coil and connected the leads to the contact pins (I cut the actual contacts from the pins first). It plugs nicely into a breadboard. The coil inductance was the highest for this example at about 1.5 H with a DC resistance of 350 ohms. This higher inductance works down to the lowest frequency. The third one on the right had the cover glued on so it took more cutting with a rotary tool to end up with just the coil. Again, 6 turns were added for the primary. The coil inductance was the about 1.0 H with a DC resistance of 260 ohms. A 100 ohm burden resistor was used for all the frequency testing as was done for the other CSTs. The ESR of these coils is much larger than the commercially manufactured examples. The ESR for the CoilCraft example is only around 4 ohms. This will have a material difference to the frequency response of the DIY coils.

These DIY current transformers were tested alongside the commercial CoilCraft product as a baseline to compare the frequency response. First plot is for relay example 1 (left side of photo). The darker curves are always for the CoilCraft product. The magnitude of the response is significantly lower but slightly flatter above 100 Hz.

Figure 14, DIY coil number 1 vs CoilCraft.

Next we compare the second DIY coil. The magnitude of the response is also lower and is even flatter above 100 Hz than DIY coil 1.

Figure 15, DIY coil number 2 vs CoilCraft.

Next we compare the third DIY coil. The magnitude of the response is similarly lower and is even flatter above 100 Hz than DIY coil 2.

Figure 16, DIY coil number 3 vs CoilCraft.

For completeness the final bode plot compares DIY coils 2 and 3 which are very close to each other in response.

Figure 17, DIY coil number 2 vs coil number 3.

These kinds of coils are not going to have real tight coupling between primary and secondary. There is not the same amount of magnetic field closure from pole to pole like a toroid coil or E shape coil. In fact the inductance changed from 1 H to 1.5 H for the middle example when the moving contact part of the relay was pressed against the core.

I have a fourth example coil from a 12 volt reed relay. I removed the reed switch but the coil needs a magnetic core of some material inserted where the switch was to have any kind of inductance.

But still these home-made transformers worked well enough and it was interesting to take the relays apart and see how they were constructed and then remake them into something else.


For measuring AC power line current the current sense transformer can safely isolate the high line voltages from the measurement circuitry. This is a very important safety consideration. The wide dynamic range of the 16 bit ADC in the ALM1000 allows the use of high current (100 Amps) probes like the LCTC to measure currents a low as a few 10's of mA directly without any signal processing amplification or filtering.

Leave your comments and suggestions below as usual.


Below is attached a great treatment of transformers written by Ken Connor.

  • The purpose of the transformer analysis I did was to show where the basic frequency dependence comes from. If the current transformer is to actually work like one, its inductive impedance ωL must be significantly larger than the combined resistance of the inductor plus the burden (load). It is how we use terms like 'larger' and 'smaller' that this all gets really interesting for engineers. 

    I really enjoy teaching students about transformers because they provide a very clear lesson on how to think like an engineer. They have three uses or, more precisely, can provide three types of transformation: Voltage, Current and Impedance. Of course, they are related, but, they have different frequency dependence. 

    First, what do words like ‘small’ and ‘large’ mean in the specific context of current transformers or transformers in general? If  ωL is to be much larger than R, there has to be some kind of a condition to be satisfied. In this case, the ratio of the currents must be the turns ratio to within some percentage error. An engineer must decide how much error can be tolerated, which will depend on the purpose of the transformer. The transformer has a job to do and a spec to meet. For current transformers, we can see that there will be a minimum frequency for satisfactory operation. 

    Second, as implied by the first comment, specs matter. I remind my students that if someone asks them to build a widget that does some function they have to ask ‘under what circumstances’ or they will likely lose their job. One has to design the minimum cost device to satisfy the given spec. Current transformers work in a particular frequency range (above a minimum frequency) so the cheapest one to design will likely be the one that operates near that minimum frequency. Also, any design one comes up with must be compared with off-the-shelf options. I suspect that current transformers are also a good example of something that should be bought off the shelf because something out there and cheap because so many have been made will do the required job.

    Both of these comments also relate to a common learning issue with students. Once they start doing experiments, they conclude, too often with the concurrence of their instructors, that theory and experiment do not agree. They have to be taught a different point-of-view. That is, devices can be made to behave like theory suggests, within reasonable bounds.