Ferrite bead VS RC VS LC lowpass filter

Thread Starter

Baron

Joined Jun 15, 2009
31
What is the different between those three types of lowpass filters and what should I considering when I choose one.
I'm looking a low pass filter for an ADC VCC supply and reference voltage.
Sample rate of ADC is 1KS/s
Switching power supply frequency is about 800Khz.


Thanks
 

marshallf3

Joined Jul 26, 2010
2,358
Ferrite beads are out, you'd need a large ferrite choke or better yet one wound as a common mode transformer.

LC beats RC by a mile but you've got to design it tuned to that frequency for the maximum effect. Properly tuned can mean 100 times as efffective.
 

timrobbins

Joined Aug 29, 2009
318
I wouldn't discount beads too fast. Like caps, different inductor types have different useful frequency ranges. A switchmode supply may well have harmonics well into the MHz, as well as noise that is unrelated to switching frequency. Beads may well get you attenuation where a wound inductor will be going through SRF peaks and troughs. Essentially you want to brickwall filter the power supply from Hz to daylight.

You also want to pay just as much attention to locally bypassing the ADC induced digital noise from getting to the analog section - most ICs have split 0Vs, and the datasheets often have great recommendations on how to implement proper layout and component selection.

Ciao, Tim
 

SgtWookie

Joined Jul 17, 2007
22,230
Sometimes just winding a few turns of a conductor on a ferrite toroid is enough to block most of the noise out; adding a 10nF metal poly cap on the ADC input gives a low impedance source.

If you want to get technical about it, download "Elsie" which is available as freeware. Plug in your desired parameters for a lowpass LC filter, and view the results. You'll have to go through a number of iterations to get a really good filter. You also need to know your input impedance.
 

marshallf3

Joined Jul 26, 2010
2,358
When I think of beads I think of the ones that only allow a single wire to pass through. Great for VHF and above, almost useless at lower frequencies.

Toroidial coil forms are available as two different constuction types and of many different material compositions, each being better in a certain frequency range. There are some good (and quite a few poor) articles on the 'net about these.
 

Thread Starter

Baron

Joined Jun 15, 2009
31
Thanks for all replies.
The sampled signal is going to be a DC. So I don't care of phase shift issue.
The accuracy and low ripple noise of the measured DC signal is very important.
(Max noise and accuracy should be less than 3mv which is the 10 bit ADC self LSB noise)
In the ADC application note they are using LC filter between VCC and AVCC with the following values: L=10uH ,C=100nF. There is no explanation how these values were chosen and I don't understand why it's better than RC filter if I use values which gives much lower cutoff than the LC. For example R=100Ohm, C=10uF gives cutoff frequency at 160Hz.
 

timrobbins

Joined Aug 29, 2009
318
10 bit ADC is not a very critical application - in a relative sense! So you should be able to achieve noise less than LSB if you follow all the app note recommendations.

The L=10uH will provide higher impedance than 100R above 1.6MHz - if the inductor has a high SRF (eg. an smd type), and may have a lower DC voltage drop - which is likely why it is recommended.

You can probably get away with R=100 and C=10uF, but suggest you try and get a smaller cap, such as a smd type 10nF-100nF, on the IC terminals or very close for both Vcc and AVcc (to their respective 0V).

If you have a switchmode supply, then I suggest you look at it's app note to see what improved filtering/routing can be done.
 
Top