EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: jahonen on July 27, 2010, 04:42:01 pm
-
Hello all,
I finally realized one of my ideas, an inductor tester which can be used to measure the large-signal inductance of an inductor and saturation current with aid of an oscilloscope. If you wound your inductors yourself for SMPS, that information is invaluable. Small-signal inductance measured by ordinary inductance meters is generally not very interesting for a SMPS design, since it matters how the inductance depends on the current, how the inductor will saturate and how steep the saturation characteristic is.
Basically, whole thing is quite simple, just adjustable pulse generator, MOSFET power stage feeding the inductor from a large capacitor bank and current measurement. Here are the schematics (http://koti.mbnet.fi/jahonen/Electronics/Inductor%20Tester/InductorTester-01_Schematics.pdf) and layout quick-look (http://koti.mbnet.fi/jahonen/Electronics/Inductor%20Tester/Inductortester_LayoutPreview.png).
(http://koti.mbnet.fi/jahonen/Electronics/Inductor%20Tester/Inductortester.jpg)
On the right there are current measurement output and trigger pulse output (both are designed so that they are terminated to 50? at the scope end). On left there is external pulse generator input, also terminated to 50?. I managed to get the pin order of the LM317 wrong so there are some patches around it. I got the PCB done on the "hobby PCB" order at my company (they actually support that), so I got the 4-layer PCB for free (or, not free, I have to work for them :P).
The use of the MSP430F2272 MCU seems a bit overkill for simple pulse as this but at least the pulse length is stable and you get nice 7-segment display for indicating the pulse length (in µs). Software still needs work, but basic things are there, pulse width can be adjusted and it fires the pulse when button is pressed. Additional features might be burst mode test with adjustable pulse interval, so you can get N pulse averages with ease on the scope.
Pulse testing is nice since it enables to drive high currents without anything going up in smoke, and of course the pulse width can be gradually increased. Inductor is at low-side and high-side MOSFET feeds the pulse, so that no gate drive currents will go through the current sense resistor.
And finally, here are some sample measurements taken using this device and an oscilloscope. Channel 1 is the inductor current and channel 2 is trigger pulse, which is same than high-side FET control signal. Traces are averaged few times to reduce noise on the di/dt math function.
First, an ETD-core wound 100µH inductor with some air gap:
(http://koti.mbnet.fi/jahonen/Electronics/Inductor%20Tester/InductorTester-003.png)
Scope math function is set to show the di/dt of channel 1 current, and since the capacitor supply voltage is known, we can calculate the inductance at any current. Capacitor bank voltage was 15 volts in these measurements, but capacitors are rated at 50 V, so higher supply voltages are also possible. For values shown in that screenshot, the inductance is approximately 15 V/133.59 kA/s = 112.3 µH. Saturation point is also clearly visible where the current starts to rise steeply, around 6 A.
And then, toroidal iron powder core wound 100 µH inductor:
(http://koti.mbnet.fi/jahonen/Electronics/Inductor%20Tester/InductorTester-006.png)
This one has very shallow saturation characteristic, and no clear saturation point is visible, although gradual increase of di/dt means that there are some drop of inductance. Inductance around 8 A is therefore around 130 µH. Saturation current is higher than this thing can measure (it seems to saturate around 45 A), but I think for most of my use, it will be enough. That certainly confirms that this inductor is perfectly good where I eventually intend to use it.
Regards,
Janne
-
I use a signal generator and an oscilloscope myself.
My signal generator has two outputs: 50 Ohm and 600 Ohm. I set the output voltage to 1Vp-p, then connect the inductor, set the function to sine wave and alter the frequency until the voltage is 0.71V which is the -3dB point and make calculating the inductance easy: it's just an RL high pass filter circuit, with the internal impedance of the signal generator being R.
The saturation current can be measured by wrapping another turn of wire round the core and passing a steady DC current through it, although this might not be practical.
-
Neat! :) Very nice board. Mine looks scarier so you might want to cover your eyes.
I made a similar inductor tester a while back but a little more low-tech. :P It uses a quad comparator, a couple of BJTs, diodes and a heap of resistors to control the MOSFET. The energy stored in the inductor is directed to an amplified zener diode when the MOSFET turns off so that I am able to use a high enough repetition frequency to get a good display on an analouge scope.
The two timing capacitors can be replaced for independent period and on-time range selection. The schematic isn't 100% accurate but shows the general idea. The upper probe in the picture is for triggering and the lower one to watch Vds of the MOSFET. The BNC cable is for the current measurement which is done with a 0.01 ohm resistor.
-
How do you keep your copper-clad boards so shiny?
-
That's just glare from the desktop lamp.
It's not shiny at all in reality. ;)
-
I've just seen this topic because of your post in https://www.eevblog.com/forum/index.php?topic=2850.msg38027#msg38027 (https://www.eevblog.com/forum/index.php?topic=2850.msg38027#msg38027).
If there was no filtering, the waveforms are really clean! Even the derivative is so smooth...
-
I've just seen this topic because of your post in https://www.eevblog.com/forum/index.php?topic=2850.msg38027#msg38027 (https://www.eevblog.com/forum/index.php?topic=2850.msg38027#msg38027).
If there was no filtering, the waveforms are really clean! Even the derivative is so smooth...
Actually, the traces are averaged, no hope of getting such traces without averaging. In addition, one of the switch-mode regulators seem to be a bit noisy, something to sort out in future revision.
Regards,
Janne
-
... <lots of interesting stuff> ...
Thanks for that, you've given me some excellent info for doing some inductor measuring for a project i want to do.
-
I've just seen this topic because of your post in https://www.eevblog.com/forum/index.php?topic=2850.msg38027#msg38027 (https://www.eevblog.com/forum/index.php?topic=2850.msg38027#msg38027).
If there was no filtering, the waveforms are really clean! Even the derivative is so smooth...
Actually, the traces are averaged, no hope of getting such traces without averaging. In addition, one of the switch-mode regulators seem to be a bit noisy, something to sort out in future revision.
Regards,
Janne
Excuse my ingenuousness, I think I was fantasizing about a perfect electronics world, where all traces are smooth :-)
Since I wasn't able to find a marking on the screen capture, I guessed there wasn't any averaging. But there should be at least the scope's front end noise!
In the afternoon (if I dont' forget to) I'll check if on the Tek default screen the averaging acquisition mode is reported, since I think it's an important info to display, together with band limitations and scales...
However, the "instrument" is very useful for those working on power electronics!