I interface an image sensor, and the design schematic for power supply is: enter image description here

How do I choose one from the available 3uH inductors based on the specs? I figured the maximum current and dc resistance specs, but how important to consider are the following specs for this particular application?

  • self resonant frequency
  • Q minimum

The sensor clocking can range from 5-30MHz.


2 Answers 2


Inductors are commercially designed and supplied in three main types:

  1. RF inductors, intended to be used in tuned circuits, as part of a filter, oscillator or modulator/demodulator block.
    • The key specifications for these are self-resonant frequency and Q (or stray capacitance / equivalent parallel capacitance).
  2. "Energy storage" inductors, commercially known as "power inductors". Commonly found and used in Switched-Mode Power Supplies (SMPS). You may find this inductors in non-shielded (cheaper) and shielded versions.
    • The key specifications for power inductors are the saturation current, self resonant frequency and construction details (fully shielded, partially shielded, not shielded). This last parameter sets a hard, upper limit on the maximum usable switching frequency at which you store energy in the inductor.
  3. "Chokes", ferrite beads or blocking inductors, intended to block as much as possible all noise, clocks, switching noise and non-DC frequencies.
    • The key specifications for chokes is maximum saturation current (DC rated), equivalent series resistance and frequency response (or stray capacitance). Self resonant frequency and Q are not usually specified. Instead, a table or curve will indicate "how well" and "up to what" frequency the choke is doing its job: blocking RF currents. The cheapest chockes will not even specify the frequency response curve, which is not as bad as it may sound (depending on your application).

Answering more specifically your question:

  • You are mainly interested in inductors marketed as "chokes", "beads" or "blocking". Though you may also find that some manufacturers market these as filtering or noise supressor inductors. However, beware of the marketing lingo, any inductor fulfilling the requirements below will do.
  • In your circuit you are showing a typical use for a chocke: a low pass "L" filter. For a proper design, one would try to estimate the kind of noise to expect at V33 and the maximum affordable noise at AVDD. I assume you selected the 3.3uH value with this information at hand.
  • With the same input information you used to arrive the 3.3uH value, you now can choose a suitable inductor, usually applying the following criteria:
    • Equivalent Series Resistance (ESR) so that the voltage drop at the maximum current consumption is less than 5% of your DC voltage at the input. The 5% is fairly typical, however, it may be off for your actual application. For instance, let's say your circuit is powered at 3.3V and demands 100mA from AVCC under a worst case scenario. Then, you could do with any inductor with an ESR lower than about 33ohms.
    • The maximum (saturation) current for your inductor should be higher than the maximum (peak) expected current demand from AVCC. Following the example before, at least 100mA.
    • The frequency response curves for your selected inductor and ceramic capacitor (ey, the capacitor is also part of your "L" shaped filter, you should check also its frequency response) should effectively work up to your maximum expected noise source at V33.
    • Low priced, which can be sourced from multiple vendors, avaiable in a common/standard mounting package.
  • 1
    \$\begingroup\$ Great answer! The 3.3uH value was provided in the datasheet. However, based on what information about my circuit should I choose the inductance value for such application? \$\endgroup\$
    – Nazar
    Commented Sep 24, 2014 at 19:26
  • \$\begingroup\$ @Naz: I agree that it would be a good addition to the answer. \$\endgroup\$
    – Rev
    Commented Oct 14, 2014 at 12:47
  • \$\begingroup\$ @Naz The selection of the inductor value will have to take into account 2 main things: the amount and type of noise (actual or expected) in the input and the amount and type of noise which we can "afford" on the input to the analog section. Both topics are hugely broad... To narrow them a little bit I would need an specific scenario (which sort of input noise, which analog IC/circuit). \$\endgroup\$ Commented Oct 14, 2014 at 13:17
  • \$\begingroup\$ @jose.angel.jimenez Maybe some classical most common example? From your practice? \$\endgroup\$
    – Nazar
    Commented Oct 14, 2014 at 13:36

The inductor and capacitors form a 2nd order low pass filter and one thing you have to be careful about is the Q of that filter. For instance, if you have a repetitive noise on the 3V3 that matches the resonance of the filter you could end up amplifying that frequency (or a harmonic) and making the analogue supply even noisier. Worst case scenario for circuits that don't have "damping" (R damps that resonance), is that applying power rapidly can damage some devices on the analogue rail.

So, in short you don't want Q to be that great.

I have, in the past on some designs, put small series resistors with the inductor to avoid this when I'm unsure about how things will pan-out in a design and it's no great deal if the analogue current is about 10mA milliamps or less - maybe 10 ohms will "lose" 100mV and if you can live with this I'd recommend going down this route. The graph below shows an LCR 2nd order LPF - the less peaky frequency response is when Q is lower and this is when series R is greater.

enter image description here

Clearly as the analogue load current increases this also applies dampening to the possible resonance of the tuned circuit so, for greater loads the requirement for a series resistor also diminishes BUT, watch out for loads that are intermittant because this might also cause ringing on the analogue rail despite the 3V3 rail looking steady.

Other things to avoid are saturating the core of the inductor because it's inductance rapidly reduces and it becomes a less-effective filter as load current increases.

Ironically, having lowish properties of self resonant frequency can be a benefit to blocking a repetitive 3V3 noise so this sometimes is worth factoring in.

I would also consider using a third capacitor to reduce really high frequency noise - the two you have already might not be that great above several tens of MHz - check the capacitors SRF too. Above SRF they turn into an inductive reactance and may cause problems.

  • \$\begingroup\$ So, keep the Q as low as possible - got that. Speaking of res. frequency - not so sure. Could you expand on that? What things/components to watch out for that might interact with the inductor? The third capacitor would be a 10nF, 1nF, or some other? \$\endgroup\$
    – Nazar
    Commented Sep 24, 2014 at 18:50
  • \$\begingroup\$ I would also like some thoughts about how to determine L and C if no values were given for the OPs above example. \$\endgroup\$
    – Rev
    Commented Oct 14, 2014 at 12:51
  • \$\begingroup\$ @Rev1.0 I'm unsure how to simply answer that or, are you directing this at the OP? \$\endgroup\$
    – Andy aka
    Commented Oct 14, 2014 at 14:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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