Author Topic: Reduce ripple in power supply  (Read 2265 times)

0 Members and 1 Guest are viewing this topic.

Offline rakeshm55Topic starter

  • Regular Contributor
  • *
  • Posts: 209
Reduce ripple in power supply
« on: May 04, 2019, 11:15:14 am »
I was trying to simulate the spectrum for output of a 5.5V DC to DC converter (LMT4613) . 500kHz switching frequency for switching regulator.

I expect a max current of 6A on Load .

The FFT of output voltage ripple which seems to peak at switching frequency 500kHz with a -50db.

How to reduce the ripple further??
Do i need to use an LC filter?? I guess ferrite beads cant suppress this ripple.

 



 

Offline MrAl

  • Super Contributor
  • ***
  • Posts: 1666
Re: Reduce ripple in power supply
« Reply #1 on: May 04, 2019, 05:38:36 pm »
Hi,

Yes a post LC filter is the typical help here.

Since the frequency is so high you can often get away with a hand wound inductor with the right wire size.
Even 2uH can help a lot with say a 100uf low ESR capacitor.
We can look at the numbers next too.  500kHz gets cut by about 66 db.

Just remember to keep the feedback the way it is now, dont connect it to the output of the LC filter.
« Last Edit: May 04, 2019, 05:43:28 pm by MrAl »
 

Online TimNJ

  • Super Contributor
  • ***
  • Posts: 1720
  • Country: us
Re: Reduce ripple in power supply
« Reply #2 on: May 05, 2019, 02:50:17 am »
LC will do the trick. Use the lowest ESR cap you can get. (Lower ESR means you can get away with less inductance for the same attenuation.)

Simple way to calculate the attenuation is to treat the capacitor as a resistor, based on its ESR @ 100KHz. The ESR won't change very much between 100KHz and 500KHz. Polymer cap ESR starts to go up somewhere between 1-5MHz (typically) so don't expect amazing incredible attenuation at 10-100MHz.

 

Offline radiolistener

  • Super Contributor
  • ***
  • Posts: 4135
  • Country: 00
Re: Reduce ripple in power supply
« Reply #3 on: May 05, 2019, 02:55:49 am »
there are a lot of thing that you cannot simulate. Such as HF leakage through ground, near field coupling and other.
If you want to get clean DC it's not easy, and for really quiet DC, you're need to use linear power supply with RF shielding.

Regarding to solutions, you can use 1 MHz switching power supply and ferrite beads with high resistance on HF and high quality low pass filter with high suppression. It can help to reduce HF leakage. Higher frequency for switching power supply helps to make more wide gaps between spurs, it helps to filter them more easily, because your filter may use more flatter slopes. But if your equipment needs very clean DC, it also cannot help. The only way is to eliminate all switching power supplies and use linear regulators.
« Last Edit: May 05, 2019, 03:08:55 am by radiolistener »
 

Offline MrAl

  • Super Contributor
  • ***
  • Posts: 1666
Re: Reduce ripple in power supply
« Reply #4 on: May 05, 2019, 02:19:36 pm »
LC will do the trick. Use the lowest ESR cap you can get. (Lower ESR means you can get away with less inductance for the same attenuation.)

Simple way to calculate the attenuation is to treat the capacitor as a resistor, based on its ESR @ 100KHz. The ESR won't change very much between 100KHz and 500KHz. Polymer cap ESR starts to go up somewhere between 1-5MHz (typically) so don't expect amazing incredible attenuation at 10-100MHz.

Hi,

It's only good to use a lone inductor when the load can not be disconnected.  Otherwise use an LC filter.
 

Online TimNJ

  • Super Contributor
  • ***
  • Posts: 1720
  • Country: us
Re: Reduce ripple in power supply
« Reply #5 on: May 05, 2019, 03:45:58 pm »
LC will do the trick. Use the lowest ESR cap you can get. (Lower ESR means you can get away with less inductance for the same attenuation.)

Simple way to calculate the attenuation is to treat the capacitor as a resistor, based on its ESR @ 100KHz. The ESR won't change very much between 100KHz and 500KHz. Polymer cap ESR starts to go up somewhere between 1-5MHz (typically) so don't expect amazing incredible attenuation at 10-100MHz.

Hi,

It's only good to use a lone inductor when the load can not be disconnected.  Otherwise use an LC filter.

When did I say anything about using a “lone inductor”?
 

Offline MrAl

  • Super Contributor
  • ***
  • Posts: 1666
Re: Reduce ripple in power supply
« Reply #6 on: May 06, 2019, 12:17:54 pm »
LC will do the trick. Use the lowest ESR cap you can get. (Lower ESR means you can get away with less inductance for the same attenuation.)

Simple way to calculate the attenuation is to treat the capacitor as a resistor, based on its ESR @ 100KHz. The ESR won't change very much between 100KHz and 500KHz. Polymer cap ESR starts to go up somewhere between 1-5MHz (typically) so don't expect amazing incredible attenuation at 10-100MHz.

Hi,

It's only good to use a lone inductor when the load can not be disconnected.  Otherwise use an LC filter.

When did I say anything about using a “lone inductor”?

Hi,

Oh maybe your drawing was a little deceiving you were looking at the effect of ESR i think.
 

Offline Simon

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Reduce ripple in power supply
« Reply #7 on: May 06, 2019, 12:31:18 pm »
LTSpice will let you add ESR to the component, or use a decrete resistor.
 

Online TimNJ

  • Super Contributor
  • ***
  • Posts: 1720
  • Country: us
Re: Reduce ripple in power supply
« Reply #8 on: May 06, 2019, 06:42:57 pm »
On that note, I think the ESR-only approach will tend to be optimistic because it implies that the capacitive reactance is 0 (at whatever frequency we're talking about.) It would probably make more sense to find Xc too.

Then, the magnitude of the impedance at that frequency would be Sqrt(Xc^2 + ESR^2).

So, yeah, the more accurate approach would  be to simulate it with a capacitor and add the ESR parameter. That said, ESR tends to dominate so the ESR-only approach won't be too far off, minus a resonance peak.
« Last Edit: May 06, 2019, 06:49:03 pm by TimNJ »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf