New Products Flash Sale
Thread Tools
Old Feb 05, 2015, 07:20 PM
rocky79 is offline
Find More Posts by rocky79
Registered User
rocky79's Avatar
Los Angeles
Joined Aug 2004
388 Posts
Discussion
Reducing Switching noise and Conducted interferance

I have a switching regulator that operates at switching frequency of 260Khz that steps down a 28vdc to 3.3Vdc. The switching regulator output will supply current as high as 1 amp for the initial 5 minutes on power up then it falls off to several mAs.

Can you please suggest an input and output filter that dramatically reduces the input and output ripple as well as the high frequency noise?
HF noise is measured to be in the 200~300Mhz range. I have attached the schecmatics as well as the input and output waveform as captured by the scope
rocky79 is offline Find More Posts by rocky79
Reply With Quote
Sign up now
to remove ads between posts
Old Feb 06, 2015, 03:31 AM
jakestew is offline
Find More Posts by jakestew
Registered User
United States, ID
Joined Sep 2011
874 Posts
There are several online power supply designers that will calculate this for you. I suggest using TI's webbench.

Looks like you only have 50mV sag on the input. That's not significant and any extra capacitance will probably fix that.

Doesn't look like any significant output ripple either. The HF noise is probably an artifact, check your scope connections and play around with them a bit. I sometimes get noise like this from random sources. Could likely be from computer or USB noise.

In any case, that tiny bit of ripple is easy to fix. You can't really go wrong no matter what you do. A ferrite bead might even be enough.
jakestew is offline Find More Posts by jakestew
Reply With Quote
Old Feb 06, 2015, 08:09 AM
Al Offt is offline
Find More Posts by Al Offt
Registered User
Boston, MA subburb
Joined Sep 2004
1,202 Posts
Rocky,

Try placing a 100nf (0.1uf) ceramic capacitor across C2 (the 150uf electrolytic) at the output of the switching regulator. Electrolytic caps do not filter high frequency (>1MHz) components very well but the combination of the two might do the trick.

Alan

Just read the note at the right hand side of the schematic. They suggest placing a small resistor (10 to 100 ohms) and ceramic capacitor (470pf -2.2nf ) in series across the output switching pin and ground. This combination would also filter high frequency components from the output.
Al Offt is offline Find More Posts by Al Offt
Last edited by Al Offt; Feb 06, 2015 at 08:24 AM.
Reply With Quote
Old Feb 06, 2015, 08:14 AM
srnet is offline
Find More Posts by srnet
Stuart
srnet's Avatar
UK, Cardiff
Joined Dec 2008
3,459 Posts
Are you using low ESR electrolytics ?
srnet is offline Find More Posts by srnet
Reply With Quote
Old Feb 06, 2015, 11:32 AM
rocky79 is offline
Find More Posts by rocky79
Registered User
rocky79's Avatar
Los Angeles
Joined Aug 2004
388 Posts
Thank you for your replies.
At the output I am using a150uF 35v low ESR cap. Part#
AFK157M50G24T-F

and a 150uF, 50v at the input.
AFK157M50G24T-F

I already have an RC snubber across DS2 R=45Ohms, C=220pF.

I will add a ceramic 0.01uF across out and input and see if that gets rids of the HF ringing.
Attached is the waveform before and after a snubber was added across DS2
rocky79 is offline Find More Posts by rocky79
Last edited by rocky79; Feb 06, 2015 at 12:45 PM.
Reply With Quote
Old Feb 06, 2015, 06:44 PM
rocky79 is offline
Find More Posts by rocky79
Registered User
rocky79's Avatar
Los Angeles
Joined Aug 2004
388 Posts
I have added the 0.01uF and 3x1uF caps at the input. The input noise improved however I still would like to reduce the 80mVpp 260khz (ripple wave from the switcher) amplitude
Will a Pi filter works? if so, suggest values will be appreciated.
rocky79 is offline Find More Posts by rocky79
Reply With Quote
Old Feb 07, 2015, 08:56 AM
Al Offt is offline
Find More Posts by Al Offt
Registered User
Boston, MA subburb
Joined Sep 2004
1,202 Posts
Rocky,
A couple of questions:
1. In post 5, you say: I will add a ceramic 0.01uF across out and input and see if that gets rids of the HF ringing.
By HF ringing, do you mean the mean the 200-300MHz you mention in post 1 or do you mean the 260KHz switching frequency at the output?
2. In post 6, you said that you added the 0.01uF and 3x1uF caps at the input?
Did adding the 0.1uf capacitor to the output help attenuate the 200-300MHz component?
3. How much current do the connected devices require? If not much, you might consider an active LP filter (no inductor) using an Op Amp.
4. What maximum peak to peak level can you tolerate in the 260KHz switching frequency?

Alan

P.S. In general, adding filters to the input of a device is intended to give a more stable voltage to the device. Most of the time, filtering a device's input does not significantly affect its output.

Adding filters to the output should be more effective in reducing variations in the output voltage especially since output voltage appears to be fed back to the regulator (the FB pin).
Al Offt is offline Find More Posts by Al Offt
Last edited by Al Offt; Feb 07, 2015 at 10:35 AM.
Reply With Quote
Old Feb 07, 2015, 09:35 AM
Al Offt is offline
Find More Posts by Al Offt
Registered User
Boston, MA subburb
Joined Sep 2004
1,202 Posts
You have not provided any information about your application. Do the devices you want to connect to the supply require very low ripple?

At 3.3 volts, I would guess you are dealing with digital circuits. There may be ways of dealing with the 260KHz ripple at the device end. For example, some devices do not require a very quiet supply voltage since their threshold between '1' and '0' is wide. Or you can gain some filtering with decoupling capacitors at their pins.

Alan

Did some further investigation. The 80mv ripple (2% of output voltage) at the switching frequency appears to be typical of a switching regulator. Check out http://www.dimensionengineering.com/products/de-sw033

The cut off frequency of the existing L1/C1 filter is about 1.9KHz. That means you have roughly 60dB of attenuation at 260KHz across the existing filter. If you have the components available, try adding another series inductor of 47uH and shunt capacitor of 150uF after the C1 capacitor. At 40 dB/decade, it should give you about 120dB of attenuation overall. Don't change the feedback location, just add the parts after C1.
Al Offt is offline Find More Posts by Al Offt
Last edited by Al Offt; Feb 07, 2015 at 10:40 AM.
Reply With Quote
Old Feb 07, 2015, 07:45 PM
jakestew is offline
Find More Posts by jakestew
Registered User
United States, ID
Joined Sep 2011
874 Posts
Quote:
Originally Posted by Al Offt View Post
Did some further investigation. The 80mv ripple (2% of output voltage) at the switching frequency appears to be typical of a switching regulator.
I think somebody mentioned this 6 posts ago.

Without checking the scope setup there's no way to know if this is even actually present in the circuit. The HF noise is almost certainly USB noise. One big hint is the fact that it looks like he must be using a USB connected scope.
jakestew is offline Find More Posts by jakestew
Reply With Quote
Old Feb 07, 2015, 08:08 PM
vollrathd is online now
Find More Posts by vollrathd
Registered User
vollrathd's Avatar
Joined Jan 2007
991 Posts
Ferrite Toroids

Quote:
Originally Posted by rocky79 View Post
I have a switching regulator that operates at switching frequency of 260Khz that steps down a 28vdc to 3.3Vdc. The switching regulator output will supply current as high as 1 amp for the initial 5 minutes on power up then it falls off to several mAs.

Can you please suggest an input and output filter that dramatically reduces the input and output ripple as well as the high frequency noise?
HF noise is measured to be in the 200~300Mhz range. I have attached the schecmatics as well as the input and output waveform as captured by the scope
Have you considered running the switching regulators input and output wires through a ferrite toroid?

Back in the late 1990's I had an Astroflight 90 brush type motor, with 38 Nicad cells to power it. Radio interference on 72 Mhz was so severe, I only got about 20 feet of radio range. With the antenna fully extended.

The two motor power leads were run through two Toroids, then to the ESC. That pretty much took care of the interference problem.

Places such as Digikey.com sell toroids of all sizes and shapes.
vollrathd is online now Find More Posts by vollrathd
Reply With Quote
Old Feb 09, 2015, 12:29 PM
rocky79 is offline
Find More Posts by rocky79
Registered User
rocky79's Avatar
Los Angeles
Joined Aug 2004
388 Posts
Quote:
Originally Posted by Al Offt View Post
Rocky,
A couple of questions:
1. In post 5, you say: I will add a ceramic 0.01uF across out and input and see if that gets rids of the HF ringing.
By HF ringing, do you mean the mean the 200-300MHz you mention in post 1 or do you mean the 260KHz switching frequency at the output?
2. In post 6, you said that you added the 0.01uF and 3x1uF caps at the input?
Did adding the 0.1uf capacitor to the output help attenuate the 200-300MHz component?
3. How much current do the connected devices require? If not much, you might consider an active LP filter (no inductor) using an Op Amp.
4. What maximum peak to peak level can you tolerate in the 260KHz switching frequency?

Alan

P.S. In general, adding filters to the input of a device is intended to give a more stable voltage to the device. Most of the time, filtering a device's input does not significantly affect its output.

Adding filters to the output should be more effective in reducing variations in the output voltage especially since output voltage appears to be fed back to the regulator (the FB pin).
Hi Alan

Yes, I added the 0.01uF, 2.2uF and that got rid of the (200-300Mhz)HF frequency noise at the input.
I added a 150uF Low ESR electrolytic cap at the output and now I have about 20mvp-p ripples.
The max current the regulator supply is about 1A at the output.
rocky79 is offline Find More Posts by rocky79
Reply With Quote
Old Feb 09, 2015, 12:30 PM
rocky79 is offline
Find More Posts by rocky79
Registered User
rocky79's Avatar
Los Angeles
Joined Aug 2004
388 Posts
At this point I got the signal cleaned up from HF noise and large ripples. Thank you all for your input.
rocky79 is offline Find More Posts by rocky79
Reply With Quote
Old Feb 09, 2015, 03:24 PM
vollrathd is online now
Find More Posts by vollrathd
Registered User
vollrathd's Avatar
Joined Jan 2007
991 Posts
Hysteresis?

Quote:
Originally Posted by rocky79 View Post
Hi Alan

Yes, I added the 0.01uF, 2.2uF and that got rid of the (200-300Mhz)HF frequency noise at the input.
I added a 150uF Low ESR electrolytic cap at the output and now I have about 20mvp-p ripples.
The max current the regulator supply is about 1A at the output.
As previously mentioned, that output voltage "Ripple" is a characteristic of the Switching Regulator power supplies. These regulators are either "Switched On" or "Switched Off" at very high frequencies. The switching duty cycle or on/off ratio is controlled by feed back from the DC output signal.

The narrower the hysteresis allowed by the regulator chip, the lower that ripple voltage, but at some point, the circuit might become unstable, and go into oscillations.
vollrathd is online now Find More Posts by vollrathd
Reply With Quote


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Sold MPI Miracle Switch MKII *Reduced!* gravityking Aircraft - General - Radio Equipment (FS/W) 1 Mar 30, 2015 10:42 AM
Discussion HK la noi acasa renatoa Romania 2 Jun 17, 2011 07:45 AM
Discussion Noi more Red baron squardon(sp) gouda Life, The Universe, and Politics 8 Dec 05, 2007 02:17 PM
Discussion Belt Static Electricity- to conduct or not conduct? Axesea Micro Helis 4 Sep 29, 2007 01:44 PM