0
\$\begingroup\$

As part of a university project I'm trying to produce a current through an inductor (610uH), that'll eventually be used as a solenoid. I'm producing this using a PWM voltage produced by a TI GaN inverter board (3PhGaNInv) powered from a bench-top power supply.

schematic

I'm having an issue that on the edges of the PWM, there is a huge spike in current through the inductor (measured using an in-line 8.9 Ohm current sense resistor). Note that I don't care about the current ripple at this point, just the spikes that occur when switching.

inductor current spikes

This seems to be caused by the bench-top power supply drooping, then overshooting back whenever the output of the inverter board is switched (see bottom graph). I'm confused as to why this is happening, as the schematic for the inverter board shows a 220uF and 3.3uF capacitor over the supply lines which I thought would prevent this.

ripple in power supply current

I've tested this with three power supplies, including a DC regulated one, and the voltage ripple is very similar in all of them. Any ideas what's causing this, or how to prevent it?

\$\endgroup\$
3
  • 2
    \$\begingroup\$ Please show the circuit schematic (actual components; more than a representative diagram) and layout, preferably with pictures showing arrangement of units (PSU, wiring, EUT, probe(s), scope). What are the inductor and resistor types/part numbers? \$\endgroup\$
    – Tim Williams
    Commented Apr 27 at 8:07
  • \$\begingroup\$ The resistor is a massive variable power resistor, I've asked around and no-one knows that actual part number for it. The inductor is hand-wound, air core. I can't attach a picture now as I'm away from the setup, but I will ASAP. \$\endgroup\$
    – Emerson SD
    Commented Apr 28 at 0:44
  • 1
    \$\begingroup\$ The problems you are seeing are being caused by parasitic components that are not shown in your schematic. Do you have any capacitors in the device under test (DUT) which are not shown in your schematic? Please post a photo of your apparatus, showing: 1. Over-all shot showing all electrical equipment, including power supplies, DUT, and measurement equipment. 2. Zoom in to show details of a) DUT, b) connections between DUT and power supply, c) connections between DUT and scope. \$\endgroup\$
    – Fabio Barone
    Commented Apr 28 at 3:03

1 Answer 1

Reset to default
1
\$\begingroup\$

These are not spikes in inductor current (you can't get those with 12V available -- consider V=L.di/dt).

Likely you are picking up stray signals across your 8.9 Ω (strange value) R. To confirm this, short the resistor and verify you pick up nothing.

Note that for these levels of signals and voltages you don't need GaN transistors.

\$\endgroup\$
5
  • \$\begingroup\$ An ideal inductance, yes. An inductor (real component), it's entirely possible. Also, shorting the load resistor in a converter under test, doesn't exactly sound like recommended operating conditions. \$\endgroup\$
    – Tim Williams
    Commented Apr 27 at 20:04
  • \$\begingroup\$ It's not likely to be the inductor; unless it has a lot of parasitic capacitance, or some very severe eddy current effects (making it effectively ˆnotˆ even close to 610 uH). \$\endgroup\$
    – jp314
    Commented Apr 27 at 21:41
  • \$\begingroup\$ Right, that makes sense. Thanks for the suggestion! Just to clarify, if I short the resistor and still pick up a signal across it, that means that the spikes are actually just stray signals and not actually occurring in the current? Also, resistor is a variable one, that's just the value I have it set to for doing the testing. It'll eventually be taken out of the system. \$\endgroup\$
    – Emerson SD
    Commented Apr 28 at 0:47
  • \$\begingroup\$ It's a not-very-widely-known effect, but ringing can indeed occur within inductors. See for example, this waveform I obtained in a QR PFC converter: Ch.2 is the drain waveform (and thus applied inductor AC), Ch.4 is an inductive pickup near the winding. There is a series resonant mode, more or less where the winding (effectively short-circuited at both ends) becomes a 1/2 wave stub. \$\endgroup\$
    – Tim Williams
    Commented Apr 28 at 8:09
  • \$\begingroup\$ "esistor is a variable one" ==> it probably has significant inductance and magnetic pickup and is not really well suited for current sensing. Use a simple 'carbon' resistor. \$\endgroup\$
    – jp314
    Commented Apr 28 at 17:46

Not the answer you're looking for? Browse other questions tagged or ask your own question.