SEPIC Converter oscillation

[QQ]

Hi,

I am trying to simulate a SEPIC converter in HSPICE. I find that there
is an oscillation at the switch node (drain of the mosfet) as well as
the second node (other end of the flying capacitor) but the output is
stable and at the correct value. The sinusoidal noise rides on the
switching waveform. Any idea what is the cause of this noise?

I suspect the noise may be because of some kind of resonance between
the inductors and capacitors but I am not sure what exactly or how to
eliminate it. Any help in understanding this is much appreciated.

Thanks
QQ

 

[MooseFET]

Hi,

I am trying to simulate a SEPIC converter in HSPICE. I find that there
is an oscillation at the switch node (drain of the mosfet) as well as
the second node (other end of the flying capacitor) but the output is
stable and at the correct value. The sinusoidal noise rides on the
switching waveform. Any idea what is the cause of this noise?

I suspect the noise may be because of some kind of resonance between
the inductors and capacitors but I am not sure what exactly or how to
eliminate it. Any help in understanding this is much appreciated.

I assume this is a high frequency thing on the drain of the MOSFET.
It is likely the capacitance of the MOSFET and the inductor
interacting.

Is it really a problem? It may just be the normal sort of ugly wave
form you see in switchers. If this is the case, don't try to get rid
of it. Anything you do to get rid of it makes a new and real problem.

 

[Joerg]

I assume this is a high frequency thing on the drain of the MOSFET.
It is likely the capacitance of the MOSFET and the inductor
interacting.

Is it really a problem? It may just be the normal sort of ugly wave
form you see in switchers. If this is the case, don't try to get rid
of it. Anything you do to get rid of it makes a new and real problem.

If it's not a ringout with a fast decay but a stable oscillation this
might fly into the OP's face at the EMC lab. When they give the thumbs
down for EMC cert.

 

[MooseFET]

If it's not a ringout with a fast decay but a stable oscillation this
might fly into the OP's face at the EMC lab. When they give the thumbs
down for EMC cert.

I don't think this is what the OP was talking about. It is fairly
hard to make a switcher have a sustained oscillation above the
switching rate. It it was below the switching rate, he would likely
have said something about the pulses being modulated or the like.

 

[Joerg]

I don't think this is what the OP was talking about. It is fairly
hard to make a switcher have a sustained oscillation above the
switching rate. It it was below the switching rate, he would likely
have said something about the pulses being modulated or the like.

If it's way below the switching rate some of it usually show up in the
output. Until a couple years ago I thought the same, that it is hard to
sustain oscillation on a switcher. Until one fine day where a client
asked me to see what can be done to improve the EMI of a system so it
would pass at the EMC lab. Found a nice stand of "trees" around 260MHz.
Before disassembling much I asked their engineers about clock rates and
such. The consensus was that there wasn't anything that could generate
this. Turns out it was the switch mode supply. Singing like a bird.
Forgot which manufacturer but it was one of the big famous ones.

 

[MooseFET]

On May 3, 9:49 am, Joerg <[email protected]> [....]
I don't think this is what the OP was talking about. It is fairly
hard to make a switcher have a sustained oscillation above the
switching rate. It it was below the switching rate, he would likely
have said something about the pulses being modulated or the like.

If it's way below the switching rate some of it usually show up in the
output. Until a couple years ago I thought the same, that it is hard to
sustain oscillation on a switcher. Until one fine day where a client
asked me to see what can be done to improve the EMI of a system so it
would pass at the EMC lab. Found a nice stand of "trees" around 260MHz.

The "stand of trees" effect can be a burst oscillation on the
switching edge. It can also be just the selection from the existing
spectrum by something resonant. I don't think the OP was talking of
either of these. He was looking with a scope. This sort of thing is
hard to see.

[....]

this. Turns out it was the switch mode supply. Singing like a bird.
Forgot which manufacturer but it was one of the big famous ones.

Power10?

 

[Joerg]

On May 3, 9:49 am, Joerg <[email protected]>

[....]

I don't think this is what the OP was talking about. It is fairly
hard to make a switcher have a sustained oscillation above the
switching rate. It it was below the switching rate, he would likely
have said something about the pulses being modulated or the like.

If it's way below the switching rate some of it usually show up in the
output. Until a couple years ago I thought the same, that it is hard to
sustain oscillation on a switcher. Until one fine day where a client
asked me to see what can be done to improve the EMI of a system so it
would pass at the EMC lab. Found a nice stand of "trees" around 260MHz.

The "stand of trees" effect can be a burst oscillation on the
switching edge. It can also be just the selection from the existing
spectrum by something resonant. I don't think the OP was talking of
either of these. He was looking with a scope. This sort of thing is
hard to see.

I also chased it with a scope, an EMCO near field kit and a wideband amp
in front of the scope. It was oscillation. Didn't bother the
functionality of the loop though but I guess they had designed a bit "on
the edge" there.

[....]

this. Turns out it was the switch mode supply. Singing like a bird.
Forgot which manufacturer but it was one of the big famous ones.

Power10?

I really don't remember, this was just one of the layers to peel in that
case. Like usual it ends up with a laundry list of changes. That power
supply was outta there afterwards. We didn't want to fix the problems of
an OEM supplier, just informed them about it. Of course, I had offered
to redesign it for them later but haven't heard back ;-)

 

[MooseFET]

MooseFET wrote: [....]

The "stand of trees" effect can be a burst oscillation on the
switching edge. It can also be just the selection from the existing
spectrum by something resonant. I don't think the OP was talking of
either of these. He was looking with a scope. This sort of thing is
hard to see.

I also chased it with a scope, an EMCO near field kit and a wideband amp
in front of the scope. It was oscillation. Didn't bother the
functionality of the loop though but I guess they had designed a bit "on
the edge" there.

Yes that is a scope, in the same sort of way a formula 1 car is still
a car. I have never seen a problem like what you had but can imagine
how it could happen if there was enough wrong with the design and
layout.

I once had a DC-DC converter of my own design act very strange. It
turned out that when the layout was done a large plane area was
connected to the drain of the MOSFET instead of ground. You could see
harmonics of the 150KHz switching frequency on a scope probe held more
than a foot away from the PCB. It made so much EMI it messed with
other supply circuits on the same PCB.

[....]

this. Turns out it was the switch mode supply. Singing like a bird.
Forgot which manufacturer but it was one of the big famous ones.

Power10?

I really don't remember, this was just one of the layers to peel in that
case. Like usual it ends up with a laundry list of changes. That power
supply was outta there afterwards. We didn't want to fix the problems of
an OEM supplier, just informed them about it. Of course, I had offered
to redesign it for them later but haven't heard back ;-)

 

[Joerg]

MooseFET wrote:
[....]

would pass at the EMC lab. Found a nice stand of "trees" around 260MHz.

The "stand of trees" effect can be a burst oscillation on the
switching edge. It can also be just the selection from the existing
spectrum by something resonant. I don't think the OP was talking of
either of these. He was looking with a scope. This sort of thing is
hard to see.

I also chased it with a scope, an EMCO near field kit and a wideband amp
in front of the scope. It was oscillation. Didn't bother the
functionality of the loop though but I guess they had designed a bit "on
the edge" there.

Yes that is a scope, in the same sort of way a formula 1 car is still
a car. I have never seen a problem like what you had but can imagine
how it could happen if there was enough wrong with the design and
layout.

Yep, but the fix was easy: Switched manufacturers, problem gone.

I once had a DC-DC converter of my own design act very strange. It
turned out that when the layout was done a large plane area was
connected to the drain of the MOSFET instead of ground. You could see
harmonics of the 150KHz switching frequency on a scope probe held more
than a foot away from the PCB. It made so much EMI it messed with
other supply circuits on the same PCB.

The most embarrassing scenario is when you turn on the new super-duper
design and the lab radio goes weeeeeooouuu ... phshhhhhhht....

 

[MooseFET]

On May 5, 11:39 am, Joerg <[email protected]>
wrote:
[.... about switchers ....]

The most embarrassing scenario is when you turn on the new super-duper
design and the lab radio goes weeeeeooouuu ... phshhhhhhht....

You must have lived a charmed life.

KABLAM FIZZZKKKKKKA-POOF and throwing the casings of the capacitors
clear across the lab is by far more embarrassing.

 

[Joerg]

On May 5, 11:39 am, Joerg <[email protected]>
wrote:
[.... about switchers ....]

The most embarrassing scenario is when you turn on the new super-duper
design and the lab radio goes weeeeeooouuu ... phshhhhhhht....

You must have lived a charmed life.

KABLAM FIZZZKKKKKKA-POOF and throwing the casings of the capacitors
clear across the lab is by far more embarrassing.

Oh, I had my share of those. The first one was as a teenager when I was
still blissfully unaware that even large electrolytics have their limits
on the ripple current you can burden them with. Transformer-less tripler
from 230VAC since a transformer just wasn't in the budget. Cranked out a
stiff 910VDC, sweet. Just don't ever plug in the power cord reversed. In
Europe those weren't polarised so a Sharpie dot had to suffice. Turned
it on, fired up the amp. 500W out, great, 700W, 900W, 1200W, yeehaw!
Then the fluorescents in my room dimmed in a weird fashion. Even more
weird because I knew for sure there was no dimmer. KAFOOMP!

One of the caps went straight into the ceiling, taking out a big chunk
of plaster, then returned from orbit still hissing. Or rather, its
wrinkled can returned because the rest had come out of it and was
raining down as fluffy snow. I guess it had celebrated its last seconds
by mutating into rocket fuel. Luckily I wasn't with my head over the
amp, else I'd be disfigured now. Had to mix plaster, get up on a ladder,
then re-paint the whole ceiling. Oh, and my shirt had lots of burn holes
in it and so did my bed and the carpet and a chair and... Mom was not
enthused about my hobbyist activities that day.

 

[DJ Delorie]

I guess it had celebrated its last seconds by mutating into rocket
fuel.

"I'm pretty sure we shouldn't be breathing the yellow smoke."

Been there

 

[Joerg]

"I'm pretty sure we shouldn't be breathing the yellow smoke."

Been there

The real stench came from tantalums. They let of a greenish cloud. Cured
me pretty much from ever using them.

 

[krw]

The real stench came from tantalums. They let of a greenish cloud. Cured
me pretty much from ever using them.

I've used kabillions of tantalums. The only time I ever had a magic
stinky (and violent) smoke release was when the dummy populating the
cards plugged them in backwards. We had some pretty strict quality
standards for 'em though.

If you want stink, try frying a selenium rectifier. ...eewwwww!