filtering SMPS power supply for Audio

I would probably recommend using a SMPS to generate the +/- voltage rails and use an LDO follower on each side. You could use this approach to generate teh +/- voltage from a single IC.
www.ti.com/.../getliterature.tsp;fileType=pdf
Depending on your current requirements, you may want to choose a different IC from that family. Generate say +/- 10.3 V then regulate to 10 V using LDOs.

Hi John,

Thanks for the reply.
I was bit worried with the PSRR of the LDO , at high frequencies like the ones used in SMPS. Would using LDOs alone filter out the SMPS ripple and noise? Do we need to filter it with ferrite beads and capacitors before feeding SMPS to LDOs?
I need to power an audio preamplifier which expects very low noise power supply, and is generally used with linear power supply with transformer by others. I am thinking to reuse the available SMPS which is already available powering the main AMP, to make a compact system.

thanks and regards
S Sarath

I know that for high fidelity applications, the amplifier is typically powered from a linear supply. It is difficult to remove all of the switching artifacts from a SMPS, but fortunately much of the noise is way beyond the audible range. Years ago, SMPS was limited to maybe 50 kHz maximum switching frequency, but now most operate at 500 kHz or greater, with may options over 1 MHz. So now you do see switching supplies used in audio applications. Even the amplifiers themselves may use switching technology such as class D or tripath. I have not personally designed for your application, but I have used the technique above for powering sensitive RF amplifiers for extremely low noise research grade and military GPS applications. You may require some additional filtering. It will depend on what solution you choose and its operating frequency.

Hi John,

One doubt: is there a limit of capacitor value that we can put at the output of SMPS for filtering (if we use CLC filter between SMPS and LDO/LM317)? 

Is there a way to test various options for this, without a oscilloscope, to see if the SMPS (laptop power supply brick) output is filtered well?

Or do we need to use any simulators for this?

thanks and regards

S Sarath 

Usually there will just be an LC filter at the output of the SMPS.  Depending on the controller used, there may be a limit on the output capacitance, but it will vary by device.  In your case I would probably recommend an RC snubber from the switching node to GND to reduce overshoot and ringing.  after the output of the LDO you may want to use various sizes of ceramic filtering capacitors, such as 0.1uF, 0.01uf and 0.001uF to reduce high frequency noise further.

Depending on the devices you choose, there may be pspice models available.  In your case, you will probably need to model the parasitic circuit components as well.  You will probably be best served to build a prototype.  I think you will find it difficult to debug and fine tune your circuit without an oscilloscope though.

Hi John,

Instead of using again SMPS to generate positive and negative rails, for the laptop adapter (SMPS), can we use a rail splitter like TLE2426 (virtual ground) along with a opamp to get positive and negative rails. And then LDO ( and a LC filter if needed) on both the positive and negative rails?

As the laptop adapter gives around 19v and i need around +/- 9 to 10 volts.. so using rail splitter we can get around +/- 9.5v and then filter and regulate it further with LDO and LC (if needed)? Also required current would be max 60-80 mA only, as it would be used to power a preamplifier

how will be this approach, of not using again a SMPS for getting positive and negative rails?

thanks and regards

S Sarath

That approach might work but you could reliably only get +/- 10 mA from each leg. Now that I understand what you are trying to do, let me see if there is a better solution.

Here are some other suggestions...Let me know what you think.
www.eetimes.com/author.asp;doc_id=1280919