Question about power supply output parasitics (scope shots attached)

[Salvador]

Ok folks, I haven't got much answers lately here and almost none with my last smps ground wire thing but here is another problem I'm facing ,
on the outputs (DC) from my smps after the bridge rectifier diodes and a smoothing filter cap bank I still get about 1V or bit more parasitics which I am almost certain are due to the very switching itself and not inductively induced from the EMI around the smps , I tested this with a scope whenever I put the probe leads across output wires whether ground or +- I get the parasitic waves on the screen on top of my flat DC. when i disconnect and simply fool around with other wires I don't get them , even when very close to the switching parts I still get very small induced spikes much smaller than the ones in the output wires.
I have two transformers each having a secondary and a mur1560 diode bridge and then filtering caps (6x1000uF/160v) per transformer.Also I got additional 100nF poly caps for additional filtering across the outputs and 15kOhm resistors across each transformers rectified secondary to keep the max idle voltage stable etc.Apparently this is still not enough and some ripple or whatever that is gets through, I mean I simply put the scope ground on one side of a fuse and the probe at the other side so that i measure directly across a fuse on either the outcoming + or - and across those 2+cm of wire I still get those parasitics , is there any way to get rid of them ?
otherwise everything is fine but these little spikes get through into my amplifiers and I can hear them in the speakers and that is not nice.I somehow thought about maybe a linear regulator at the output that could drop only a few volts from the output and maybe somehow filter out the noise ? Give me some advice.

P.S. sorry for the bad scope shots it's just that the scope itself is not exactly NASA technology and it's shot by a webcam.

 

[sophiecentaur]

Firstly, what is the switching frequency and what is the frequency of that ringing? 1V certainly does sound bad.
Is you circuit layout sanitary? To reduce the problem of ringing, it is always best to use a ground plane construction with short connecting lands between components and good decoupling capacitors (small value ceramics in addition to the high value electrolytics).
There are so many reasons for circuits to play up like this if you don't follow the rules. Could you post circuits and photos of the layout?

 

[Averagesupernova]

Good luck. This could be a lot of things. High frequencies may be impeded by ground traces and such so this shows up as a voltage. Things we take for granted at low frequencies need to be considered at higher frequencies. I recall helping out on a scope design team and there was a nasty ripple on a high voltage switching supply that they were having trouble getting rid of. I poked around at it for a while and found that there were some things done in circuit layout that caused longer traces to some bypass capacitors and they were next to worthless due to the impedance between them and where they needed to be. Again, good luck.

 

[tech99]

I am unclear where you are measuring when you say you disconnect and still see the spikes.
Also, when you measure across the fuse, is it at ground potential one end or both ends hot? If the latter, remember one side of the CRO input is grounded.

 

[sophiecentaur]

We can't make progress without some more practical details about the circuit, layout and measurement method.

 

[jim hardy]

Is this the SMPS you're working on from your earlier thread?
https://www.physicsforums.com/attachments/igbt-nananana-png.94245/

I really dislike SMPS's around audio or precision analog gear for just this reason, the doggone things are 'noisy'..

Those big electrolytics on output should have small high frequency capacitors in parallel with them
do you have some small tantalums, on the order of a uf , handy ?

 

[Salvador]

Yes Jim that is the very schematic , when averagesupernova mentioned earlier twice the phrase good luck in one post I did kind of realized how hard it will be to eliminate the noise escaping through my filters into the DC output.

at the output in parallel with the big electrolytics stand 100nF poly WIMAS.well the mains transformers are not much better in terms of noise as they too inject quite some noise especially when they become big for high powers , I have had my fair share of trying to eliminate all kinds of hums in amplifiers.
I like the smps in this very amp in which I put it because the amp is for low frequency high power audio and it delivers strongly , I measured the voltage under heavy load and it doesn't sag and you can feel that stronger kick quite literally, for the mids and simply when listening mozart I use either tubes or something else.

Ok so I made a few pics again with the only instrument I have at hand that makes photos - webcam.the pics are bad quality but you will understand the very overall.
all high frequency traces from driver to switching IGBT's are as short as possible , the only wires that I left going from the pcb to various sockets are either the incoming mains wires and the filtered output DC wires coming out from the cap bank into the fuse plate on top of the smps.

Since my scope doesn't support very high voltages , I simply measured across the length of the outcoming DC wires.since this is high frequency noise it will show up basically anyway.So at first I did a simple test , I took a piece of wire , some 30cm or longer and attached it to my scope ground at one end and probe at the other.
then I switched the smps on and put the wire on various locations around the smps very very close to the parts that are the main emitters of HF EMI noise.
I got very very small to almost no parasitics on the scope screen , then i proceeded and put the scope ground on the gnd wire for example and the probe at the other end were the gnd wire enters the amplifier boards and there I have it , the 1-2volt peak to peak strange looking parasitic with spike tops and sinusoidal rise and fall tops after each peak.
and this very waveform is on every outcoming DC wire on each separate transformer and cap filter bank. I basically have two secondaries and two DC sources for each two channels (4 in total) so it's 90-0-90 from one transformer/rectifier/filter and the same from the other, so in total 6 wires , each one of the has the same exact thing on it , they literally don't differ even though each transformer core is a bit different.
that's all I can say because that's all I know for now.

P.S. in the pictures the primary side of the smps has traces soldered and shining the secondary has simply solder points with lacked copper traces.It wasn't meant t be this way but I simply got tired of soldering all the traces so i decided to stop at the primary :D
anyway the traces are thick so their current capability isn't a problem.
in the pics its quite hard to understand the structure of the smps as it had multiple layers at some places also the second transformer as it;s hiding underneath the aluminum heatsink enclosure that I made around the smps it takes heat away not just from diodes and switching transistors but also from both transformer cores as they do get hot under very big loads when I tested close to 2kw.i could make better pics at the weekend when i will have acess to a good camera.

I do understand diagnosing from far away is hard and probably I will have to solve this myself as with many other things but maybe there are some tips or things I could use to get rid of the noise somehow.But here I'm almost definitely certain that it somehow comes through the very power wires and is not due to the EM field created by the HF switching of current through both transformers and their attached wires.
there must be some way i can block the ac that's riding on top of my DC.

maybe i should use multiple smaller caps with low value resistors in series and put them across the rectified filtered output so that the ac noise would ride through those caps from both + and - to gnd wire and diminish itself through the resistor to levels that become inaudible in my speakers that would be a major breakthrough for me now.

 

[jim hardy]

I do understand diagnosing from far away is hard and probably I will have to solve this myself as with many other things but maybe there are some tips or things I could use to get rid of the noise somehow.

Identifying its source would help.

Magnetic coupling? Capacitive?

Your layout is intuitive to you now , but not to me.

Here's a mental crutch i use
trace out the path taken by your high frequency current then use right hand rule to estimate where the resulting flux goes, and what does it couple.

I have no idea what it'll look like on your board

Object being to minimize areas of the AC current loops
by short wires between transformer, rectifiers/switching transistors, and filter capacitors especially the hf ones.
If any are wires leaving the board use twisted pairs

Ampere and Faraday laws applied !

Then look for large surface areas with substantial AC voltage, envision them as capacitor plates radiating an E-field.

MIght help, i don't know.

old jim

looks like a fun project.

 

[tech99]

Still a bit uncertain about your exact probe arrangement.
I have found that the better way to search for noise is to use a current probe. I use a ferrite toroid cut in half, so it can be placed around the various leads coming from the power supply. Make the test with a dummy load on the power supply. The toroid has a winding on it connected to the CRO using coax. With your present set up, you have a ground loop caused by the CRO mains ground and the amplifier mains connection. Check particularly for noise on the mains lead coming out of the power supply. The cure for noise such as this may be capacitors of all sorts of values or it may be ferrites on all emerging leads. Another test is to hold an AM radio receiver near the various leads.

 

[Salvador]

well here's the fun fact , i recently rebuilt this smps so that it now looks like this , before it was basically an old pcb adapted to what was at the time just an experiment whether it works at all , so the previous version looked like a thousand times more terrible it had not only ac wires coming in and out of the pcb but also the HF wires from the switching transistors to the transformers going above the board via physical wire , for the first traffo the wire was shorter but to the second one i had to use some 10cm of wire simply going above the board like an antenna.
Now you might say it should have had way more noise , NO it had actually the same amount of parasitics as this smps has now , i measured that and now I am measuring this and to my surprise not much has changed , at least the sound through the speakers is the same.on the GND wire there are only these spikes seen in the pictures above but on either + or - there is also this constant small vibration of the DC line itself I can't zoom into that vibration as to see it's waveform because my scope isn't so precise and I can't focus into it but there is definitely something there.Maybe it's just pure ripple leftovers? Somehow the electrolytics even though I used plenty of them and bought them new from a respectable source aren't enough for the ripple.I have this feeling.

 

[jim hardy]

So at first I did a simple test , I took a piece of wire , some 30cm or longer and attached it to my scope ground at one end and probe at the other.
then I switched the smps on and put the wire on various locations around the smps very very close to the parts that are the main emitters of HF EMI noise.
I got very very small to almost no parasitics on the scope screen ,

Sounds like a good test. If i understand, you made a loop antenna and checked for induction ?
And found none?

then i proceeded and put the scope ground on the gnd wire for example and the probe at the other end were the gnd wire enters the amplifier boards and there I have it , the 1-2volt peak to peak strange looking parasitic with spike tops and sinusoidal rise and fall tops after each peak.

So the spikes appear when scope gnd is connected to power supply 'gnd' and probe is on amplifier gnd?
As if the voltage were produced across that piece of wire?
What do you see if you connect scope gnd and scope probe to same end of that piece of wire?
In other words, is switching frequency current flowing back to mains through the scope's gnd wire?

Is there a path for switching frequency current ?
Might there be a capacitive voltage divider ?
Is this plausible ? (Sorry, i think better in pictures than words)

How could we check for that ?
Will it run without that 2.2nf ? It does seem to bypass your line filter...

Just thinking, not asserting...

old jim

 

[Salvador]

well as for the EMI test yes Jim I connected simply a straight piece of wire between the scope gnd crocodile clip and its probe which represents most of the power wires both in length and shape coming from the psu to the amp boards.I found next to none induced EMI from such an antenna test.I mean if the field around the smps would be so high as to cause all these spikes I should have seen them on a 30cm wire since my frequency is 50khz.

Now just now I measured with scope again after inserting series LC snubbers across the secondary traces. No result whatsoever.This time I measured both across a single wire coming out of the psu and also across gnd and + and then gnd and - and basically everywhere is the same ripple parasitic I was talking about.

As for the grounds , no the secondary grounds that are with respect to the +- DC voltage are not connected to chassis nor to the primary side , the primary side is totally unconnected to anything in the secondary side. the aluminum box chassis is simply connected to the mains earthing plug in case lethal voltage would appear on it.but my wall socket as with almost every socket here in former USSR has no earthing wire so that trick is basically useless I just did it for fun.:dbut that doesn't concern me since I am very cautious about electrical stuff due to my various " dance routines" in the past when mains AC got me dancing.

at this point I'm almost as I said sure that what I measure and see is nothing else than ripple , twisted and bent but ripple.it's on both secondaries , its nowhere else but only in the DC filtered wires coming out of the psu and its in them no matter how you measure it literally rides on top of each wire no matter whether gnd or +-.

I also did the AM radio test and yes there is an audible interference especially right after the moment when t's plugged in then it tends to get lower.but putting the radio further than about 60-80cm seems to completely eliminate the interference.But radio I think is very sensitive to weak signals.
well either it's emi induced or it's ripple there isn't much else it can be I think but if it were emi it should have showed up in the wire I had around the smps a various places with scope attached because none of my dc power wires coming out of the smps are that long.

 

[Salvador]

P.S. I read on the net that people write about using linear regulator to drop a few volts and also reject the output ripple of various power supplies.maybe there is such a regulator that I could use at the output.I'm just not sure whether they are rated at high power levels such as mine. I have approximately 90-0-90 VDC output with some 5 to 10 amps of max output for the amp.

 

[jim hardy]

it literally rides on top of each wire no matter whether gnd or +-.

the primary side is totally unconnected to anything in the secondary side.

What about that 2.2 nf from primary side to gnd ? The one above red arrow in post 11 ?
You show same gnd symbol on secondary. That 2.2nf cap and the interwinding capacitance of hf transformer completes a ground loop for 'switching frequency current' . See red loop.
What experiment can you devise to rule out that hypothesis ? Will the supply run with that cap temporarily removed?

 

[Salvador]

oh sorry Jim that cap is non existent in my smps , i never bothered to install some stupid caps across strange places.
i will add the updated schematic to this post if that makes it any better.
i mean what is interesting is that before this setup with the previous mumbo jumbo layout i had like 4x470uF used electrolytics on each secondary bridge that i found lying around in house and in terms of ripple reduction and noise i see no difference whatsoever , and now I'm having 6x1000uF new nippon chemicons inside on each secondary for a total of 12 caps.
I assume the only thing I have gained from them is a higher reserve for sudden very demanding current transients that could arise if the amp is driven to it's limits on club music and that's about it.

here's another fun fact, the ripple is there on the DC outputs no matter whether i connect the longer wires that go into the amp or whether I disconnect them and leave just the small wires that come out prom the very pcb into the fuse plate.at first i thought that the noise maybe somehow comes from the amp boards due to the amp interfering with the smps EMI.but that's also not the case. as I've measured whether the amp is attached to the smps or not the parasitics don't change a single bit.
that means they come out of the very filter bank dc traces which means the electrolytics and smaller poly films and even smaller ceramics arent helping a bit in eliminating them or maybe they just can't
I really thought that large caps put together with smaller ones can almost eliminate such AC high frequency waves from a otherwise DC supply but it seems they can't.

P.S. when i compare what I see in the IGBT driving signal and the output noise ripple whatever you call it there is this one thing i find common it's the small spike on each square wave when it crosses zero or in other words each time the switch is turned on or closed.
I think what I see in the output is not simply a ripple it's the very spike just amplified through the transformer of the IGBT opening and closing , if I remember the spike was higher at opening and softer and lower when closing.I will try to catch it with a better camera at weekend.

 

[jim hardy]

i think it's common mode not normal mode.

Use a small capacitor to block DC to your scope or select AC coupling..

Are the spikes in phase on + and common wires? - and common ?

The schematic keeps changing. That makes it difficult to maintain a logical train of thought.
But removing that cap is what i wanted to try, just to see if something changed.
Installing a temporary one would answer that.

 

[Salvador]

Well sadly I have no way of telling whether the spikes are in phase or not , I mean each spike consists of multiple small and weirdly shaped sinusoidal like ups and downs I can barely read them out on the scope , a double probe input scope would be much better for such cases.
I'm pretty sure if the switching was done using a classic sinusoidal waveform nothing like this would show up in the output , the only problem is a mosfet type transistor would have huge amounts of heat produced in it's small body if it had to conduct all that current through itself with anything higher than a quarter ohm of resistance.
But I'm having IGBT's which means the main current path is not like that of a mosfet but rather a bipolar I wonder how much I could round the sharp edges of a square wave to get away with acceptable noise levels and heat combined.
But then again I assume the driver and oscillator IC's don't have a sinusoidal like waveform function just squarewave.

anyway I think how could I make the switching bit softer , I kind of think it's not the high frequency itself but rather the sharpness of it's waveform that makes this awful ripple parasitic thing , much like the cheap little smps that are now used in every bulb and small device which have this sawtooth waveform in their switchers which causes a lot of bad noise.

 

[jim hardy]

hmmm.

Inductance can cut sharp edges without producing heat

i wonder what would happen of we relocated your series inductance to primary side of transformer? They're 1::1 aren't they?

Can you patch that up without wrecking anything?

at this point we still are unsure of the source.

Troubleshooting is 'poking at things and observing response" - is that deductive?
Figuring out from basics what is a useful "poke" - is that inductive?

Moving that impedance from downstream of secondary to upstream of primary might cause a change that'd help us understand that transformer. Or it might not at all.

When nothing will show itself wrong one has to make everything show itself right.

end troubleshooting digression...

Hmmmmm. How much impedance is there between that transformer centertap and your power supply common terminal where you see the spikes? That ringing has the look of inductance - suggests back to trace the physical current loop and minimizing its area.

What does your intuition tell you is the source? How can you test that hypothesis ?

old jim

 

[jim hardy]

BTW I'm 6 hours behind GMT. What is your time zone? Seems to me you're possibly a night-owl?
I do better in mornings...

 

[Salvador]

the time zone i believe is GMT+2, and yes I usually go to sleep at either 12 at midnight or 1 in night.
Well the transformers are literally 1:1 the primary runs at about 155/160v since its half bridge and the secondary needs to be +-90 so that is about 180v rectified about 160v AC.So yeah primary is 12 turns and also the secondary is 12.
I had the chokes that you talked about in but they too gave no results whatsoever so for the valuable space I abandoned them.i can't put them on the primary side there would be no space but quite frankly I doubt that would do anything, unless they would have to be really big but then hey would influence the whole working of the transformer probably and limit my power output and heat themselves up.

I will have to search for a split probe I can put the scope across higher DC voltages while switching it to AC only detection and it's okay but since it has only about 50v max swing it can reproduce at the screen it would be hard for me to see the primary side current waveform although I need to see that.I will be closer to the cause if I will test down the current waveforms coming in and out of the traffo.

 

[jim hardy]

I had the chokes that you talked about in but they too gave no results whatsoever so for the valuable space I abandoned them.i

So they too are on the schematic but not in the circuit?

I doubt that would do anything, unless they would have to be really big but then hey would influence the whole working of the transformer probably and limit my power output and heat themselves up.

Hmmm. Think about that statement. When you had them in they were in series with your transformer , just there was a bridge in series too. Primary and secondary current are the same.

since it has only about 50v max swing it can reproduce at the screen it would be hard for me to see the primary side current waveform although I need to see that.I will be closer to the cause if I will test down the current waveforms coming in and out of the traffo.

A voltage divider, 1meg::10K would give you ~100::1 turndown but it'd round off the corners a bit due to scope capacitance.

With all your constraints it's pretty tough going.

my approach would be
1. Try to find and fix the source upstream of transformer
or, failing that
2. Block it downstream of transformer.

1 will involve changing things to observe changes.

Just out of curiosity
is this line filter also not in the circuit?

If not, does shorting it out change anything?

Tech99 suggested in post #9 an ingenious "Poor Man's Current Probe" .

Lastly -

Are these two caps necessary?

50 khz means ten microseconds per half cycle?
An amp into a microfarad is a volt per microsecond... cap accrues ten volts per half cycle, subtracting from supply voltage to transformer
but gets added back to supply voltage across transformer when transistors swap state
that sounds spikey to me.SMPS's are not in my zone of competence
but I've chased a lot of noise in my day.

Keep us posted , I'm really curious. I try to learn something every day in hope of someday knowing something !

 

[Salvador]

thanks Jim for sticking with me since there are only a few people here that do that. :D
I have a simple line filter , it consists of two caps in series across the AC input just before a choke.that's about it.

the capacitors that you've outlined and asked whether they are necessary make the half bridge topology if I'm correct, removing them means that the upper high side switch would put the rectified +325v DC right between the mains filter caps so at ground potential at one side of the cap and +325 at the other now the two switching transistors would switch 325 volts through the primary instead of 160 which is half the mains rectified because there is another cap in series.
I don't know how that would work but one thing it would do it would double the output voltage from my transformers. probably also double the power.
but some folks once told me that it doesn't work like that without that capacitor.

 

[jim hardy]

Okay , thanks, i learned something today , the caps are traditional . I'll fgure it out more deeply over next day or two, plodder that i am. As i said SMPS is not my world.
from http://www.irf.com/electronics/topology-fundamentals

Now - how should one choose the size of that righthand capacitor ? (your 1 uf's)
What happens to your spikes if you tack another 1uf across each to make them 2 uf? Anything? If so it's a clue, if not, well it's back to the drawing board... a quick enough experiment.

Also study your layout - how much Z is between those two 'ground' symols with your layout (they're your little triangles) ?

Further down that same IR appnote

i do not yet grasp quite how that works . Your young facile brain will pick it up more quickly than my old dinosaur one..

old jim

 

[Salvador]

the second schematic differs from my topology since the half bridges also come in various forms.
about the capacitor I actually have 2.2uF in there.

what do you mean by Z?

 

[jim hardy]

what do you mean by Z?

Impedance.

2.2 uf is 1.4 ohms at 50khz

What happens to your spikes if you tack another 1uf across each to make them 2 uf? Anything? If so it's a clue, if not, well it's back to the drawing board... a quick enough experiment.

Is amplitude of your spikes affected by changing the value of that capacitor?

 

[Salvador]

that's a good question definitely not one I can find out easily.
I will try to desolder the caps and put some other in.but I wonder can the spikes be transmitted further if they happen at the gate drive level.
I have seen these spikes at the turn on of the upper transistor moment and a smaller one at the turn off.
but that is gate drive , the rectified DC power is + so when the upper transistor comes on all it can do is to switch that rectified DC to the load which is the primary winding so I tend to think that wherever they begin they can't be associated with the gate drive also because of one reason that gate voltage of a given IGBT can only open the transistor either a little or fully , anyway the current direction is the same just the resistance through the device changes and the square wave rise is sharp and unaffected by the small spike at the gate, so I guess the main problem could be that the secondary output spikes are happening at the transformer somehow.

 

[jim hardy]

Rather than unsolder them, might you just tack in some more across the pads?
I have no phenomenon in mind, just looking for something the circuit is sensitive to...

so I guess the main problem could be that the secondary output spikes are happening at the transformer somehow.

Stray inductance and capacitance gives this sort of symptom
see around pages 10 to 15 here
http://ecee.colorado.edu/~ecen5797/course_material/layout.pdf

and this short one from TI about snubbers
http://www.ti.com/lit/an/slva255/slva255.pdf

seems to me you addressed snubbers long ago in a thread far away... may be snubber circuit needs a tweak?

I still think you have a common mode trouble because it's n all 3 of your output wires.
The ecee.colorado link discusses how that comes about, starting around pdf page24 of 38.
That capacitor you don't have in is one of the suggestions...
Remember this loop ?

 

[Salvador]

Ok reporting back , what I did was take out one of the two transformers , disconnected it electrically , the one I wasn't sure about whether it dosn't have an airgap in the middle.
Now the noise is much much lower.Before it was about from peak to peak about 2 volts in amplitude now the shape has changed and i's amplitude is 0.2v or about 150-200 mV.when I connected the speakers I couldn't hear any noise anymore just the typical " SHHSHSHSH" hissing type noise coming through an amp if it's input (pc soundcard this time) is turned all the way up with no signal.

also another interesting observation is that I have a rheostat potentiometer soldered in for both the frequency control and also for the manual adjustment of the permanent duty cycle and i turned the pot so that the duty cycle expands at the end of the pot i started hearing parasitic noise through the speakers again so i tunred it back until the noise stopped altogether.
I think what I need to do is either make one bigger transformer for this half bridge and just power all 4 channels from that or take that second one i disconnected out inspect it and replace it's core with hopefully the same ferrite that the other one has or very close otherwise they are a bit mismatched and that causes problems but I guess I will go for the one big thing and done solution.
I mean people talk about that if each channel has a separate supply it sounds better and etc but then again they say that tube amplifiers outperform transistor ones and I've listened to both and apart from softer and more likeable mids everything else is the same or worse so that is pretty much nonsense i believe.
it could only sound worse if the rails sagged badly with the one transformer , although maybe i will try swapping the core of this traffo as I already have made all the chassis so that it fits and I'm lazy to overdo the darn thing .

the transformer playing now has a core that I attached from 4 separate E core with glue and each E core ahs numbers F1 67 written on it when I google that I get no results.

 

[jim hardy]

no idea on f1-67

http://4sale.sbszoo.com/xformer-cores.htm

it could only sound worse if the rails sagged badly with the one transformer ,

At a power level high enough to sag THAT supply
i cannot imagine being able to hear a little electrical distortion added to what the speakers themselves will be doing..

 

[Salvador]

well these are ferrite cores so they don't have lamination etc.
you are right that it's hard to hear distortion at very high levels , maybe if the distortion is huge then.

 

[jim hardy]

what I did was take out one of the two transformers , disconnected it electrically , the one I wasn't sure about whether it dosn't have an airgap in the middle.
Now the noise is much much lower.Before it was about from peak to peak about 2 volts in amplitude now the shape has changed and i's amplitude is 0.2v or about 150-200 mV.when I connected the speakers I couldn't hear any noise anymore just the typical " SHHSHSHSH" hissing type noise coming through an amp if it's input (pc soundcard this time) is turned all the way up with no signal.

Aha there's something about that transformer ?
Could be
core - high reluctance draws more magnetizing current ? Does it rum warmer than other one?
windings - leakage inductance, interwinding capacitance
leakage flux, especially if core has air gaps.

or its physical location on the board...
Again - how does its primary current flow?

 

[Salvador]

i don't know exactly but I think the core differs in material from the other one.
I took the core apart and it seems like it's not ferrite but that other material they used back in day for TV smps which was running on lower power and most importantly frequency.
Metglass or something similar.i broke the core apart and it has that weird electronics smell I think ferrite didn't had that smell.

I basically hope to simply buy a matching or one that is up to spec with my needs ferrite core and put into the already wound bobbin from which i took the cube shaped core out.

what do you mean by?

Again - how does its primary current flow?

 

[jim hardy]

what do you mean by?

Trace the current path on the board

like in post 8
(though i have no idea how current flows on your board, that's just a 'how to' example)

apply right hand rule

there's just one turn on the circuit board of course,

does the magnetic flux B created by the current loop couple your output wires?

 

[the_emi_guy]

well as for the EMI test yes Jim I connected simply a straight piece of wire between the scope gnd crocodile clip and ...

Get rid of the crocodile clip that came with the scope probe and unscrew the plastic sleeve. Use a dental pick or equivalent to make low inductance ground connection between PCB ground and ground sleeve on probe tip. Next, determine how much of the noise is instrumentation by touching probe tip to the exact same place as ground connection.

 

[Salvador]

nop it's not the measuring wire fault I've tried many times attaching wires leaving the tips hanging in mid air etc etc and nothing produces such waveforms as can be seen attaching the leads o the smps output, he parasitic is definitely there.
ok here's the brightside I've attached 10k resistors across gate to source to tie down the gates as noted in some application notes by semiconductor makers.To my surprise when I plugged the thing it the IGBT's went bang and short.So now that's I'm changing the igbt's with my spare ones I also ordered a new etd59 transformer , previously I used some off the shelf ferrite cores let's see how a good solid ferrite core maybe changes this problem because it has atleast partially to do with the transformer as I already found out earlier , why the IGBT's blw with the gate resistors not sure , maybe the gates had dropped the driving signal too much and left the gates either floating or half opened at full current.

either way the transformer will be here today so I will see how this goes for me.
the thing I hate once again about semiconductors is that they are so demanding I can't even get to see what the scope shows that the gosh darn thing's already let it's magic smoke out.

(editted by mentor to change some words to protect sensitive eyes and ears)

 

[the_emi_guy]

Firstly, what is the switching frequency and what is the frequency of that ringing?

Salvador,
The first responder to your query asked about the frequency of the ringing. I don't think you provided this yet and its quite important info for debugging this.
There is no way to tell from your scope shots if we are dealing with KHz noise of MHz noise.

We can often use the ringing frequency as a way to figure out what is resonating. For example, if you suspect that the capacitance of the resonating tank circuit is a reversed biased diode, then you can put a small capacitor in parallel with the diode and see if it shifts the ringing frequency. This is a common method used to design snubbers.

 

[Salvador]

ok sorry for not telling that fully , it's just that I'm mostly doing everything myself including the thinking because I lost faith in this or any other internet forum for that mater long ago, also because nobodies responding etc.
the frequency fs the output ripple like pulses matches the turn on and off times of each IGBT, I saw this while measuring both the output and then the gate drive signal and it had this ringing , bigger at turn on less at turn off but it's always the moment when the square wave just begins to rise or is about to cut off.

I have a big ETD59 transformer which I am about to wind right now as i finish this and well let;'s attach that and see how this goes.

 

[jim hardy]

I lost faith in this or any other internet forum for that mater long ago, also because nobodies responding etc.

How to win friends and influence people...