LC Filter that seems not to work

[ruscoff]

Hi all,
   i'm trying to filter the output of charge pumps (TPS60150, step up from 3.7V to 5V) with LC pi filter. The load use about 1mA.
L = 100nH  (http://www.farnell.com/datasheets/1910214.pdf)
C = 10uF X5R 16V 0805

The cut frequency are around 160Khz.

Into the PCB the components are very near (1,75mm between them) and ground return paths are clear.

As you can see from the attachments there aren't any attenuations, and filter add some noise.

So, why the filter doesn't work correctly? i'm expected -40db around 1,5/2MHz

[Zero999]

The Q is too low. The impedance of the inductor is only 100m Ohm at 160kHz and the DC resistance is 260 mOhm plus the capacitor's ESR.

[ruscoff]

I suspect something similar..but i can't understand where is the problem, i didn't find any guide on google.
I think, If is only a problem of dc resistance the filter cut frequency should move up but i think i should view anyway the -40db slope.
Could you explaing why it won't work?
Thanks

[T3sl4co1l]

Can't really tell what the horizontal axis is... MHz??

dBu is a rather strange unit to pick for the vertical axis.  dBuV or dBm is more common.  What load impedance is this at?

Datasheet says 1.5MHz switching frequency, so there should be a peak at 1.5.  But it also has a pulse skipping mode, so your noise output will depend very much on load.  A no-load test is inappropriate here.

What does your layout and measuring setup look like?  These are critical for this measurement.

Tim

[ruscoff]

X axis is in MHz, it is write but maybe to small.

Yes, you have right it is in a skip mode because the load is large, it should sink 10uA with a peak 100mA when works but for very short time < 1ms. Oscilloscope has 1Mohm of input impedance. By the way, the output waveform is a sawtooth with superimposed high frequency ripple, i think the peak at 2MHz. The filter have to attenuate that ripple because the load is pratically ideal.

I try to reduce the probe ground wire lenght put it around the probe. The same peak are measurable on the air with the probe like an antenna but if i connect it to circuit peaks increase of 20/40db.

I don't know why the software use dbuV.

[Zero999]

Try using L = 100µH and C = 10nF