cleaning switching noise

Hi Board,

I have a tight spec, need to filter DC/DC for MCU and analog sections . should I slightly increase the O/P voltage of my converter (say 5.2V) and then CLC /RC filter it is this enough rejection for MCU or analog circuits . I couldnt use Linear regulator instead as it would dissipate good amount of heat.

Can you help me with How do clean powersupply when power is derived directly for DC converter

-D

An RC filter with a large C is likely all you need to filter out the DC/DC hash. Use a C with a low ESR (tantalum, ceramic, special low ESR electro.s) as this will be the main cause of propagated ripple. Of course the resistor will drop voltage proportional to current; but this is unlikely an issue.

dark said:

I have a tight spec, need to filter DC/DC for MCU and analog sections .

Click to expand...

So where's this spec? If you can state converter ripple, voltage, desired ripple (filtered), output current, output voltage tolerance (filtered), the actual values can be calculated.

dougy83 said:

An RC filter with a large C is likely all you need to filter out the DC/DC hash. Use a C with a low ESR (tantalum, ceramic, special low ESR electro.s) as this will be the main cause of propagated ripple. Of course the resistor will drop voltage proportional to current; but this is unlikely an issue.


So where's this spec? If you can state converter ripple, voltage, desired ripple (filtered), output current, output voltage tolerance (filtered), the actual values can be calculated.

Click to expand...

Hi,

The switching freq is 330KHz , accuracy is 2% , 150mV ripple , Output 5V@1Amp . Should the voltage at output be trimmed (+-15%)to higher side and then accomodate LC or RC drops .

Thanks

dark said:

Hi,

The switching freq is 330KHz , accuracy is 2% , 150mV ripple , Output 5V@1Amp . Should the voltage at output be trimmed (+-15%)to higher side and then accomodate LC or RC drops .

Thanks

Click to expand...

2% acc is the converter acc I take it. Well, you haven't specified an output ripple so what do you want? Anyway, the example below should give you an idea.

1 Amp (pretty high for an MCU) is going to be better filtered with an LC than an RC. For 5V/1A 150mV ripple in, L = 10uH + 0.05R, C = 47uF + 0.08R gives output of 4.95V with <1mV ripple.

An LC circuit is a cheap way of doing it. You can get linear regs with very low dropout (200mV at 1A), so this would keep the heat down and you also get a specified output noise and power supply rejection ratio, so you can determine what your output is going to look like.

I am not sure what your system looks like, but do the MCU and analogue sit on the same rail? Can't you provide the high current rail for your MCU from a dc/dc (as micros tend to be less sensitive to supply noise) and provide the sensitive analogue rail using a dc/dc + LDO or straight LDO?

Just a thought to consider...

Electron works hit the nail on the head here...an LDO.

I have been knocking up a lot of simple switch-mode power supplies for audio recently, both boost and buck. Radiated noise can be shielded, but filtering out the ripple is probably best done with an LDO linear regulator. I use them as post-switch mode filters all the time. They can be very efficient depending on current/drop out, more-so than an RC filter.