Is this a power supply filter? why inductor?

[wylee]

I think what I have here is a power supply filter (see the attached picture, extracted from MAXIM evaluation kit)

But what I don't understand is, why use inductor? and is there any reason for choosing these component values?

 

[flatulent]

The inductor reduces the ripple by forming a voltage divider at the ripple frequency between its series impedance and the shunt impedance of the capacitors. The capacitors are several values spaced by over a factor of 10x. This is because of the self resonant frequency of the capacitors. The larger values have more inductance. With the three of them in parallel, there is a small impedance in shunt with the line at a wide range of frequencies.

The circuit looks like a local power line filter for the close by circuitry and is fed from a power source farther away. This is a very common circuit for systems when the local circuitry is sensitive to power line ripple. It is more common on analog circuits than digital circuits.

 

[wylee]

From what I know, introducing a inductor instead or resistor to a LPF configuration will cause peaking in frequency response.

I ran some simulation an it proof this, if you were saying using inductor is better, why I don't see and ripple reduction in my simulation result?

Here is my netlist
**********************************
*power supply filter

.subckt sup_fil1 in out gnda
L1 in out 1.2u
C1 out gnda 33u
C2 out gnda 2.2u
C3 out gnda 0.1u
.ends

.subckt sup_fil2 in out gnda
R1 in out 100
C1 out gnda 33u
C2 out gnda 2.2u
C3 out gnda 0.1u.ends
.ends

x1 vdd18 vdd18a gnd sup_fil1

*power supply
vdd18 vdd18 v1 dc=1.8 ac=1
*10kHz sinusoidal ripple noise
v1 v1 gnd sin 0 50m 10k 0.0 0.0 0.0

.tran 50n 200u

.ac dec 200 1 100G
.graph par('vdb(vdd18a)-vdb(vdd18)')

.alter
x1 vdd18 vdd18a gnd sup_fil2

.end
**********************************

Attached is the result simulated from HSPICE
-light blue plot is my incoming rippled power supply 1.8
-pink plot is the output voltage with 1.2uH inductor LPF configuration
-darker blue is the output voltage with 100ohm resistive LPF configuration

From the frequency response (bottom graph), I can tell
1. at lower frequencies 100Hz-10MHz, inductive power supply filter will induce more ripple than resistive configuration
2. only at higher frequecies > 10MHz, inductive power supply filter will produce a more stable output voltage, since the attenuatuation is rolling off faster in magnitude.

So, can anyone help me one why people still use inductor as power supply filter? I still see no reason why inductor is better than resistor (I may be wrong, but please CORRECT my understanding)

 

[achilles.wang]

maybe, you should add a resistior as load and simulate again.

 

[barrybear]

The clue to the use of an inductor in series with the dc feed to the decoupling network is its value. Such a small inductor will only effect very high fregency pulse - oscilation. this is quite a comonn problem when feeding logic ic`s with dirty DC.

Barrybear

 

[wylee]

So is there any way/formulas/procedures to determine on how to choose an inductor and capacitors values in designing power supply filter?

 

[Fom]

This is second order LPF. So it will supress any riple much more efficient than RC based LPF (40 dB per decade instead of 20).
It can be used if you want to supress VDD ripple of DC-DC convertor. The cut-off frequency should be calculated from ripple frequency, it's amplitude and attenuation. If your capacitor is fixed and defined by load, the ripple attenuation will be defined by inductor value only.

 

[ME]

By using a inductor instead of a resistor, it becomes a second order low pass filter with 6dB attenuatuation per decade, instead of a first order low pass filter with 3dB attenuatuation per decade.


When you simulate the filter you should remember to add a load resistor.

 

[platonas]

The inductor in old power supplies was used to reduce the ripple. That was true when a large inductor was introduced in the circuit. In the ages of valve circuits, an inductor of some Henrys was very common. It forms a low pass filter which has large impedance at the ripple frequency.
In the attached circuit, I believe it is used as an RF chock. It stops any high frequencies to pass through. It is mostly used in supplies of transmitters in order to block any RF from entering the supply lines. Another use is to block RF interferance comming from the supply lines. Used in sensitive equipment or when best shielding for RFI is needed.
Usually these inductors are of low value (ranging from some uH to some nH) and wound on RF cores (ferite cores).

 

[Fom]

Sorry ME, I think you meant 6db per octave not decade.

 

[ME]

Sorry ME, I think you meant 6db per octave not decade.

Yes, sorry about that, I meant octave not decade.

 

[radiohead]

Please note that putting a resistor in the power line would result in a DC drop over the resistor, depending on current consumption. So depending on the operation mode of your chip, you might have 1.8V in one state and 1.5V in another. This is clearly not the way to go. That is why inductors are the preferred solution.

 

[jinboqiu]

use resistor may lead larger DC drop in high supply current, but it is stable in any circumstance.
use induce lead much smaller DC drop and have better atteuntion ,but in some condition ,may lead resonance in some frequence.this will generate greater ripple in power supply.

 

[sibeen]

As a new poster to these forums, but not new to electronics, I'd like to make a small observation about this thread. As soon as I looked at the circuit, it was obvious, to me, that this was a power rail filter, with the inductor helpfull for high frequency attenuation.

Now, my comment concerns "simulation". Is this the way that electronic education is going, that someone will look at a circuit and immediately begin a simulation of the circuit to work out its properties. Perhaps, the old days before spice et al became popular, students (or others) would have taken more time to actually examine the circuit.

Hopefully not too contreversial for this thread, delete if required