Smps-current-filter

You can add an LC or CLC lowpass filter, and maybe also a three-terminal linear regulator circuit, to the SMPS output. You should be able to get the noise+ripple to be extremely low, with that; possibly in the tens-of-microvolts range.

What output voltage(s) do you have available, from the SMPS?

Do you know what frequency the SMPS runs at? Or, have you looked at the output with a scope, and know what the magnitudes and frequencies are, for the two main ripple/hum components?

For the LC or CLC filter, you can probably just use a small, high-current toroidal inductor, such as the Bourns/J.W. Miller models (2100, 2200, or 2300 series, e.g. Mouser.com # 2301-V-RC, 10 uH, 20 Amps, $2.78 qty 1). Put the L in series with the output and put a large electrolytic to ground after it (and also before it, for a CLC filter).

For my 60 kHz SMPS, using 2200 uF 50V Nichicon UHE-series caps worked well, with a 10uH inductor. e.g. Mouser.com # 647-UHE1H222MHD6, $1.81 qty 1.

You can probably increase the L and C values, if 10uH and 2200uF don't give enough attenuation. A CLC filter, alone, should easily be able to get your ripple/noise down to less than 100 mV p-p, and probably to less than 10 mV p-p, and maybe to less than 1 mV p-p.

At any rate, you will probably want to think about varying the L and C values, for use with your SMPS. If you know the approximate frequencies and amplitudes of the SMPS's output ripple/noise components, you could easily simulate the filter, and the SMPS's output noise/ripple (with voltage sources), with LTspice (free download from linear.com), and quickly find some L and C values that would do a good-enough job. (Remember to add the DC resistance to the simulated inductor, and add the ESR to the capacitors, in your simulation.)

In my case, having only one positive SMPS output voltage, I followed the CLC lowpass filter with an LD1084 (an LT1084 knock-off) 5-Amp adjustable regulator circuit, with the adjust pin bypassed to ground with a small-ish electrolytic (say <= 47 uF), and 150uF to ground from the reg's output (plus the other recommended components, etc, from the regulator's datasheet). Simulating it, with the SMPS, the output noise+ripple, into a resistive load, was around 15 uV.
 

Paswa

Member
2007-08-10 9:24 am
thanks friend gootee, your infos were very very useul.

I dont have a oscilloscope. But i borrowed my friends osc and find it is same as yours. it is 60Hz.

I'm trying to make the LC filter. Thank u.

I've a question. look,

A processor fan, and side fan and a smps fan running inside the cabinet. is the fans create some hum?

A computer power supply should not produce any humm (low frequency ripple). If you are experiencing humm, the source is probably anywhere else.

you are wrong, friend. This is a computer SMPS where i drew the power, It is a 800W SMPS from NavTech
 
You're welcome.

Maybe you could try disconnecting each fan, only for a few seconds, while watching the scope display of the output voltage.

Also note that many SMPS types need some minimum load, to function correctly.

P.S. In my experience, Eva is rarely, if ever, wrong about any SMPS information. So you should probably assume that she is likely to be correct that any rectified-mains-frequency "hum" is actually coming from somewhere else, if your SMPS was well-designed and has no circuit faults.