Kevin KesslerMeaning of Saturation Current
The current handling capability of these inductors is not dictated by the max current the coils of wire can take before getting too hot, but by the max magnetic field the core of the inductor can support. For ferrite core inductors (the most common core type), as current through the inductor increases, the magnetic field increases, and the magnetic domains in the core start to align with the magnetic field of the coil, increasing the energy the inductor can store. Once the current through the inductor increases to the point where all the domains in the core are aligned, the “enhancement” in energy storage caused by the core stops, and the effective inductance of the inductor drops. This can be seen at the elbow of the curve, and operating circuits like switch mode power supplies, or LED drivers above the saturation current will result in lower efficiency and more heating.
PCB Design
After Googling around a bit to get some guidance on measuring inductor saturation current, I ran across this thread, https://www.eevblog.com/forum/projects/inductor-saturation-tester-alternative-route-to-dump-the-excess-energy/ Subsidiary Threads: https://www.eevblog.com/forum/beginners/help-with-testing-inductor-saturation-current/, describing a circuit to do just that. I breadboarded it up, and it worked fairly well, but I decided to get a PCB made to reduce the parasitics of the breadboard. While I breadboarded up the circuit using discrete transistors, along with the Digikey order for my well characterized inductors, I also ordered some DGD0215 MOSFET gate drivers. Since it takes about 2 weeks for the PCBs to get here from China, I drew up the PCB with both the discrete transistor MOSFET driver circuit and the DGD0215 drivers. In the case that I didn’t use the DGD0215 drivers correctly in the circuit, I could fall back to the discrete transistors without having to spin up a new board and wait for the slow boat from China for a new PCB. I also added footprints for extra capacitance to supply the turn-on current surge that I didn’t need, as well as a 12 volt regulator to supply the drivers from a higher voltage for the device under test. In the end I just used 2 bench supplies; one at 12 volts for the MOSFET drivers, and one at 12-20V to test the inductors. I also put in many test points spaced for an oscilloscope probe with a ground spring, to make debugging and measuring easier. Since adding all the extra pads is free, but getting a new board takes 2 weeks, I often try to think of what could go wrong and create alternatives for that situation as well as copious test points into the PCB.
If I were to create a second version of the PCB, I would remove a lot of these extra pads, as well as shortening the current loop which drives the current through the test inductor. The high current loop in this design runs around the outside of the PCB, increasing the parasitic capacitance and inductance, and hence the ringing seen in the traces, as well as increasing the EMI radiated from this device. This PCB was focused on functionality, but it works well enough that I will probably not bother to create a better one.
Circuit Function
The full schematic of the board is in the project gallery, but above is the functional schematic of the populated parts. The test power is connected to J1 and 12V to drive the gates is connected to J2. The Function Generator is fed into J12, and the current is measured by the Oscilloscope probe at J16. The various Test Points are scattered about, but the second channel (green) of the scope is connected to TP1. The set of grounds at the lower right are just the ground holes for the Test Points. The inductor under test is connected to J3 DUT.
The 5V pulse from the function generator drives the DGD0215. The DGD0215 inverts the signal, so that, when the pulse is high, the output is 0V, and the P-Channel MOSFET turns on. When the pulse is low, the output is 12V and the P-Channel MOSFET is off. On the N-Channel MOSFET side, the pulse is fed into the inverting input, so when the pulse is high, the output is 12V, and a low pulse output 0V. This double inverting of the signal is a bit confusing, and the DGD0216 doesn't invert the signal, but I bought the DGD0215, so I used them.
The end result of this is, when the pulse is high, both transistors turn on. VCC immediately appears on TP1, but the current flowing through the shunt resistor, R13, slowly builds up, because that is how inductors work. The voltage drop across the shunt is measured by the oscilloscope probe, and that reading is how we determine the inductance and saturation current of the inductor under test.
Using the tool
The PCB is used by putting an inductor to test in the fixture, and supplying 12V to the jumper to power the MOSFET drivers (or using the voltage regulator if it was populated). The power for the device under test is set to between 12 and 20V since higher inductances work better with higher voltages, because the current slope is larger with a larger voltage. A pulse from my function generator is delivered through the BNC connector on the left, and the oscilloscope probe is connected to the BNC connector on the right, to measure the voltage across the .01 Ohm current shunt. The pulse width from the signal generator that turns on the MOSFETs is varied between 1uS and 50uS, and I adjusted this pulse width and test voltage to get an oscilloscope display that shows the inductor going past the saturation current.
I also put a probe on the high side of the inductor, and used that to trigger the oscilloscope (green trace) when the MOSFETs turns on.
The output from the oscilloscope (yellow trace) shows both the saturation current of the inductor as well as its inductance. To find the inductance you start with the equation for inductance L=V/(di/dt). For my oscilloscope, there is a 10.7mV offset, so that it reads -10.7mV with 0 current. So a 10mV reading on the oscilloscope actually means a 20.7 mV reading above 0, and I=V/R, so I=.0207/.01 or 2,07 Amps. So calculating the inductance is a matter of taking 2 points on the linear part of the curve, calculating the current difference (ΔV/.01), and dividing that by the Δt between the 2 points, and then taking the voltage of the device under test and dividing that by value from above. In the example above:When the calculated inductance matches what the inductor says it is, I can be pretty comfortable that I have a good measurement.
Measuring the saturation current is just a matter of finding the point before the elbow. I tended to be conservative in this measurement, because I don’t want to push the specs in my designs.
so this inductor would be good as long as it stayed under 1.3A.
As you can see, under higher currents, the green voltage line sags. This is because of the resistance of the P-Channel MOSFET when fully (RDSON). While this doesn’t really hurt the measurements, because I’m not being that precise, if I was doing this again, I would find a better P-Channel MOSFET with a lower RDSON, or eliminate that fet all together.
Ferrite vs. Metal Alloy Core
The behavior of inductors with metal alloy cores is significantly different from ferrite core inductors. While the ferrite core hits a saturation point and the inductance changes abruptly at that elbow point, the inductance of metal alloy cored inductors, like the Bourn inductors I got for my LED driver, lowers slowly as current increases. This was a new piece of knowledge to me, and it is explained here: https://www.mag-inc.com/Design/Design-Guides/Inductor-Cores-Material-and-Shape-Choices.
