Oli Glaser wrote:
>
> It's part of a LC filter on the output of the DC to DC converter on my 
> isolation board. Output is 5V with an approx 100mV ripple (not sure what 
> waveform) at around 50 KHz, 400mA peak current. I was thinking a low 
> resistance (<0.5 ohm)100uH choke with a 10uF low ESR cap. The previous DC to 
> DC I was looking at (with a 90KHz switching frequency, 100mV ripple) 
> recommended a 47uH inductor with SRF >20MHz, and a 4.7uF cap, to reduce 
> ripple to approx 1mV.
> The same (or very similar) circuit (dual voltage, 90KHz switching) will be 
> used to supply the +/-5V opamps in my USB scope, so it has to be pretty 
> quiet. 
>
>
>   
You might even then isolate it completely and connect input + and 0V via 
a common mode choke and output + and 0V via a common mode choke.
The DC currents cancel as they are opposite and  equal.
The PSU should be completely  separate  ground plane at least  with its 
OV only connecting to rest of circuit via the decoupling capacitors at 
input and output.

Layout can be more significant than the 100uH, and negate series filter :-(
Put a series  coil on  input and output  too.

I tend to put 10uF or 20uF tantalum, 100nF or 220nF ceramic, and 1nF to 
10nF ceramic, all three. RF stuff I do would have multiple 100pF, 10nF, 
essentially at every point where something connects to power rail.

A good solid ground plane over whole bottom of PSU section and all other 
tracks on top, and PSU ground plane only connecting to other OVs at same 
point as SMD ceramic decoupler on  power rail  in & out.  Vin and Vout  
track  via  the  decoupling  caps  even if it makes track longer  
rather  than  single track to cap. Pretend the PSU part is a 100MHz 
amplifier. Easier then to get the noise from PSU very low.

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist