Ringing from switch-mode PSU poluting power rails

Hi,

I have a switch-mode powersupply unit that's injecting a high-frequency
ringing onto the input power rails (which is from a battery). Does anyone
have any ideas how to stop this? Would a simple diode and capacitor work?
Or an inducer? In which case, how do I chose the value of the inducer?

The DC-to-DC converter is:

http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1831

Thanks a lot,
Jack

 

Most likely the ringing comes from radiation. Too long powerlines that are
not properly decoupled or long and/or thin traces in the power circuit. The
coil itself also may be a radiation source. You may have to redesign the
layout. Keep the traces around chip and coil thick and short. The same for
the other components. You may need some shielding or a ground plane. Keep
the input lines away from the powersupply.

I see no use for a diode in the power supply input. A decoupling capacitor
is absolutely necessary as shown in the example circuit. If that capacitor
is good there is no use for a coil. A good decoupling requires a low ESR
elco of some tens uF (or larger, depending on the current) parallel to a
ceramic capacitor of let's say 10nF.

Main rule: Prevent the disturbing "signal" to escape from the circuit.

petrus bitbyter

 

Hi,

Thanks loads for your ideas. Yes, I originally thought it was radiation.
But then I cut the PCB in half, separating the powersupply and the mic
pre-amp. This didn't solve the problem. Then I noticed that if I powered
the switchmode PSU from a separate battery then the intereference went away.
I've also done some tests with an oscilloscope and the PSU is DEFINITELY
injecting noise into the battery.

Here's a quick box diagram showing the circuits:


Battery---+---------+
| |
PSU Amp
|
Camera

 

You might be getting ground/supply bounce on common return/suppy lines.
Use a star point technique for both ground and supplies to minimise this.
Make sure the prime supply ac impedance is low too. Put big C's at the
SMPS ,Mic and star point. Use ground plane at the SMPS.

If that doesn't work try one of these

http://www.animationlibrary.com/Animation11/Creatures_and_Cartoons/Fairy_Tal
e/Magic_wand.gif

Brian

--
Brian Howie
BAE SYSTEMS Avionics Limited
Sensor Systems Division
Crewe Toll Phase II, 1st Floor,
Edinburgh EH5 2XS
Phone +44 (0)131 343 8769
FAX +44 (0)131 343 8941
Email

 

This circuit makes noise when the internal switch disconnects the
LXP,LXN terminal from ground. This allows the inductor to snap
positive, forward bias the output diode and dump the inductor current
into the output capacitor. So the ground terminal of the output
capacitor has fast rising current coming out of it. If this capacitor
is not connected very closely to the bottom of the input capacitor and
to the ground terminal of the chip (where the internal switch suddenly
turns off the current), this current path switch produces voltage drop
between these three points on the ground line. This voltage tends to
leak out into the rest of the system.

If you already have these three points closely connected, then you may
have to add another small inductance in series with the input supply
line. I would first try a bead on lead or other small inductance with
a low DC resistance. It has to be rated to not saturate at the
maximum current that will pass through it. Increasing the value or
paralleling a couple caps at the input side may help. also.

 

So the battery and its lines are prime suspect. You have to decouple the
battery and the amp the same way you have to decouple the PSU. Then you have
to use separate power lines for PSU and Amp directly from the decoupling
capacitors of the battery so the powerline parts used by both simultaniously
are as short as possible. That should be enough. But, as you know, "should
be" is not always enough. You can try a coil in the powerline to the PSU, as
close to it as possible. You can also try ferrite rings in the power lines.

To rule out radiation further you can experiment packing camera, PSU and
the powerlines to it in alufoil. (Pack it in a plastic bag first to prevent
shorts.)

(As an aside: I'm not interested in discussions about top posting. IMHO it's
a free choice for everyone. I just brought down your responce for my own
conveniance.)

petrus bitbyter

 

I don't have tons of experience with this kind of problem, but I'll share
my views anyway.

For this configuration, it seems to me that adding another inductor
between the battery and the power supply might not be necessary, since the
current is already being drawn through an inductor.

So, try a bigger (or lower ESR) capacitor to ground. Put the capacitor
closer to the output inductor, and keep the ground lead short and fat. The
chip itself undoubtedly produces current spikes when it turns its
internal FET's on, so try putting a high frequency cap as close as
possible to the chip (again, keep any leads as short as possible).

If all else fails, put an inductor in series with the power supply line.
Experiment with different values. You could always start by putting in
another one of the main inductors, since you probably have them lying
around. If that works, and if you want, you can try lower values until you
find the smallest value that will work.

Good luck.

--Mac

 

Thanks a lot, everyone!

All these ideas have helped... but I'm still getting noise reaching the amp.
Urg! So now I'm using two separate batteries! Not the most elegant
solution but it works.

Jack