Maker Pro
Maker Pro

Filtering output of switching regulator

V

Valentin Tihomirov

Jan 1, 1970
0
Hi,
I have very sharp spikes about .3 volts at the output of a switching
regulator. I am unhappy because the ouptut does not become any smoother
neither when adding of a big electolitic capasitor (470uF) nor adding of
non-electrolitic 1uF capasitor (that should be enaught to filter these thin
spikes).
 
A

Active8

Jan 1, 1970
0
Hi,
I have very sharp spikes about .3 volts at the output of a switching
regulator. I am unhappy because the ouptut does not become any smoother
neither when adding of a big electolitic capasitor (470uF) nor adding of
non-electrolitic 1uF capasitor (that should be enaught to filter these thin
spikes).
a low ESR cap. tantalum or a small ceramic job in parallel with the
electrolytic.

mike
 
S

Spehro Pefhany

Jan 1, 1970
0
a low ESR cap. tantalum or a small ceramic job in parallel with the
electrolytic.

Maybe with an L (or R if you can stand a bit more drop) in series.
But maybe the spikes are not really there. If the scope leads are
shorted together and connected to the top end of the cap, is there any
spike still there?

Best regards,
Spehro Pefhany
 
T

Tilmann Reh

Jan 1, 1970
0
Valentin said:
I have very sharp spikes about .3 volts at the output of a switching
regulator. I am unhappy because the ouptut does not become any smoother
neither when adding of a big electolitic capasitor (470uF) nor adding of
non-electrolitic 1uF capasitor (that should be enaught to filter these thin
spikes).

Output spikes and their amplitude heavily depend on the PCB layout.
There are numerous application notes of the IC makers about the
PCB design rules for switching power supplies.

Besides that, low ESR capacitors are useful (as already mentioned),
and in some cases an additional LC filter will be needed. Depends
on your application.

Finally, sometimes the .3V spikes you are measuring are not really
there, because your probe simply picks them up...

--
Dipl.-Ing. Tilmann Reh
Autometer GmbH Siegen - Elektronik nach Maß.
http://www.autometer.de

==================================================================
In a world without walls and fences, who needs Windows and Gates ?
(Sun Microsystems)
 
A

Active8

Jan 1, 1970
0
Maybe with an L (or R if you can stand a bit more drop) in series.
But maybe the spikes are not really there. If the scope leads are
shorted together and connected to the top end of the cap, is there any
spike still there?

good test. might reveal a leaky probe cable.

i thought about L, but that's as obvious as controlling AC with a triac
;-)

mike
 
S

Steve

Jan 1, 1970
0
Since this is in the topic what is a good way to reduce RF generated from a
switching supply.
Lay out helps reduce emi from allowing the current to null its self out
but I found the schottky diode in one case, generating RF noise from 75-200
MHz . Adding a 1 nF cap across it dropped it allot. Any more ideas.
Switching frequency was 200KHz, and a FET switch.
 
K

Ken Smith

Jan 1, 1970
0
good test. might reveal a leaky probe cable.

It doesn't need a leaky cable. With most scopes, probe ground is the
chassis ground which is the third pin on the power cord.

If there are large current spikes in the circuit, a part of the current
will end up flowing in the braid of the scope cable. This is usually a
bad thing.
 
K

Ken Smith

Jan 1, 1970
0
Since this is in the topic what is a good way to reduce RF generated from a
switching supply.
Lay out helps reduce emi from allowing the current to null its self out
but I found the schottky diode in one case, generating RF noise from 75-200
MHz . Adding a 1 nF cap across it dropped it allot. Any more ideas.
Switching frequency was 200KHz, and a FET switch.

Use shielded inductors.

Make it so that all connections to the DC-DC pass through an inductor of
some type. This includes the "ground" connections. You may have to use
"remote sensing" to regulate the voltage on the far side of the inductors.

Use 0.1u ceramic capacitors across the big filter capacitors.
 
N

Nico Coesel

Jan 1, 1970
0
Valentin Tihomirov said:
Hi,
I have very sharp spikes about .3 volts at the output of a switching
regulator. I am unhappy because the ouptut does not become any smoother
neither when adding of a big electolitic capasitor (470uF) nor adding of
non-electrolitic 1uF capasitor (that should be enaught to filter these thin
spikes).

A common mode filter transformer and a capacitor do wonders in these
cases. But beware of the resistance of the filter's coil, it might
cause too much drop accross the filter.


+ ----^^^^^---+-- +
===== |
| === output
===== |
- ----^^^^^---+-- -
 
S

Steve

Jan 1, 1970
0
Thanks ken
I did have .47uF ceramic caps across the filter caps and one right after the
diode. I was using tolorid coils also . With the spectrum analyzer pick up
near the supply the RF was showing up, also on a local tv.
This will be put in a box, and I might need to use ferrite beads on the in
leads also.
Maby I should have gone with bipolar transistors low ft, and a 20KHz
frequency.
 
J

John Popelish

Jan 1, 1970
0
Steve said:
Thanks ken
I did have .47uF ceramic caps across the filter caps and one right after the
diode. I was using tolorid coils also . With the spectrum analyzer pick up
near the supply the RF was showing up, also on a local tv.
This will be put in a box, and I might need to use ferrite beads on the in
leads also.
Maby I should have gone with bipolar transistors low ft, and a 20KHz
frequency.

I think you would get a few more useful suggestions if you post a
schematic and a few close up pictures of your construction on a web
page, somewhere or on alt.binaries.schematics.electronic.
 
K

Ken Freeman

Jan 1, 1970
0
I would look for stray inductance in the path of the circulating switching
currents. You have voltage SPIKES, not a level shift, so ESR is probably
not the biggest problem. Given E=Ldi/dt, a fast current transition through
an inductance should give you a sharp spike. You should also be able to
estimate your current transitions based on the design, and hence estimate
how much inductance is causing this spike. That will enable you to perform
a targeted PWB layout inspection, hunting for stray inductance.

Email or dialog in the group if you have questions.

Ken
 
B

Ban

Jan 1, 1970
0
Ken said:
The scope just to the right of me here, has its ground lug cut off
and a heavy wire off the front panel to ground. This was done to
reduce the noise from its own power supplies. It works but it
generates a question now and then.

Gracious God,
throw that thing away.
For testing the power supply you need a small tektronix receptacle soldered
directly onto the output trace *and* ground-plane. Plug your 10:1 probe into
that receptacle. This is the only reliable test config.

ciao Ban
 
K

Ken Smith

Jan 1, 1970
0
Gracious God,
throw that thing away.
For testing the power supply you need a small tektronix receptacle soldered
directly onto the output trace *and* ground-plane. Plug your 10:1 probe into
that receptacle. This is the only reliable test config.

Note: I am not the one with spikes on the power supply.

The test configuration you suggest does not work on circuits that have
large currents in the ground plane and a connection to earth ground. You
still get current flowing in the probes braid and hence still get the
non-real spikes showing up on the scope.

The Tek test socket removes the loop of wire formed by the ground lead of
the probe and hence is very good when that loops impedance is the problem.
I believe that the OP's problem will turn out to be the non-zero impedance
of his ground plane.
 
A

Active8

Jan 1, 1970
0
We seem to be awfully informal here. The correct address is "Mr. Dammit"
dear mr. damnit:

i respectfully apologize for my informality and would like to point out
the misspelling of your last name, damnit.

hope you don't mind, god damnit.

respectfully yours and have a good god damned day,

Active8
 
Top