Flyback Airgap Doubts?

I recently tried to make an smps flyback
as to start.. i went to TI webbench for the design
I started one design - 12v , 3amp, @ 100khz
IC - UC3845

Transformer specs E25/13/7 . Pri= 38 turns sec-11, aux =11 turns ( I have few EPCOS ones)
Lprimary = 226 uh. I didnt see any airgap measurement

So i fabricated the PCB, then wound the transformer
So when i put together the transformer the inductance was over 2 milli henry(i think).
so i filed down the centre leg with a dremel (not even) . After getting the inductance reading around 250uh I taped the transformer.

After all assembling done I powered the unit thru a 40w incandescent Lamp . The out put was around 11.85 volts.. after I added some load of 200 ohm (100x2 nos - 2 watt resistors). The output was around 12.05.. Then i increased it to 22ohm, it showed around .49 amp.

After that if I add another 100ohm resistor , it collapses . volt READING around 6-7.0v.. current goes down.. and... the lamp lights up..if i remove the last 100ohm resistor - things go back to normal..
So my question is what could be wrong here. Did the core saturate??
If this is related to air-gap ?? the air gap became big or small? by decreasing or increasing the air-gap can i fix this issue?

In case you want someone to check your transformer design, you should give complete information, e.g. which input voltage was used during the test. Also what's the designed primary peak current.

if there is a current limit on the primary - this will limit max power out ...

The design is as in the ti diagram , i will attach the PDF.View attachment TI_12x3_100k _UC3845.pdf
...
In my design ... I first used a MUR460 as the output diode.
The RCD diode in the circuit is an SMD 3amp one. All i Had was a FR107.. one the circuit was powered this diode (FR107) got very hot.
(I tried to get FR207,307 HER207 etc) nothing available...

make Rsns smaller for more power - if the snubber diode is getting hot the transformer has too much leakage inductance ... also the Fr07 is a bit too slow ( trr)

- - - Updated - - -

For 3 amps out, in DCM, the peak current in the o/p diode(s) is 12A ...

- - - Updated - - -

with your Tx you only have 45 volts of flyback voltage - it should be closer to 150V - else it will take ages for the energy stored in Tx to discharge into the sec side.

this is part of the reason you are not getting enough power ...

- - - Updated - - -

the design is for 36 watts only - your 40W light bulb is then the full load - if you add more load the current limit on the pri side will limit the Vout - as you have seen ...

do you want more than the designed 36W ... ?

- - - Updated - - -

If you use the 38:11 turns ration - the output diode will see 325 x 11 / 38, + 12V = 106 volts at 230Vac in

this is why the app note says use MBRB40250TG as the o/p diodes, they are 250V, 4A each ...

The MUR460 will work, might get a wee bit hotter - use it in the snubber diode ...

Hi
I did your flyback in LTSPICE. It is attached for you….This is the free sim from analog.com…..you can run it by simply converting it to .asc instead of .txt

I gave you a vac input of 180-240vac so apologies if this is not what you have?

Your transformer seems ok, as long as you have it clamped together properly.

I did it for you with 38 turns of swg22 in two “sandwich layers” for the primary…for the sec I gave you your 11 turns made up of two strands of TEX-ELZ 7/0.15.
I used your efd25/13/7…the flyback and transformer design calcs are attached.

As to your not getting max power……well, you don’t need slope compensation, so take that off...…….then youll get more power through on the same sense resistor.

Also, I haven’t been through your startup circuit yet……also, I haven’t done it with opto isolation as this was just a basic checker.

I also changed your RCD clamp as yours looks too dissipative…see the sim for what I gave you there. (again this is based on my assumption of 180-240vac input)

You are obviously having trouble……so in the first instance…connect pri and secondary grounds and get rid of your opto…just set it up with a straight resistive divider feedback.……….like I do in the sim……then when you have that working…go back and sub in the opto etc
Be careful with live non-isolated mains….and use a diff probe just in case…do you have an isolated AC source?

BTW what is your exact vac input?....i assumed 180-240vac for you..

EFD25/13/9 TDK EPCOS.
https://www.tdk-electronics.tdk.com/inf/80/db/fer/efd_25_13_9.pdf

- - - Updated - - -

Actually I re-did the excel calc for your transformer turns ratio (as attached) ……..you are in CCM even at 339vdc on the HV bus…….to be honest, your transformer turns ratio is too high…ie NS/NP is too high….
CCM is bad for offline low power flybacks due to the reverse recovery of the secondary diode.
In the words of the Chief Engineer of a huge global electric drives company in the North East of England…”All Offline flybacks should be in DCM (or BCM)”

- - - Updated - - -

Oh heck sorry, I am having a bad day…I accidentally did the sim at 135khz…..and then left the incorrect 66khz annotation on there……here it is corrected and re-attached for 100khz…woops!

- - - Updated - - -

Also, have you got an isolated DC supply?...if so...just in the first place use that to power the uc3843 with...dont use the aux or the rc startup circuit.

Get it down to the bare bones and get that working..then start to put the rest in.

Did you use multi strands for the sec?...did you interleave wind?

Give me a shout if you want me to re-do your transformer for DCM over all VAC input?
Please also tell what is your VAC input range?

Great .... My mains voltage here is 230v
Thanks .. ""Easy peasy"" and ""treez"" for your detailed replies...Thanks for your analysis and I am going through each comments- one by one.

Give me a shout if you want me to re-do your transformer for DCM over all VAC input?
Please also tell what is your VAC input range?

Click to expand...

Great.. You are very welcome to do so ... mine is 230v

**************
Mostly with the fly-backs designs, I have seen calculations for Airgap . Here I don't think I saw any..(or I didn't see it)
I wound the transformer first then measured the Inductance with an LCR meter.. then started grinding the center leg until i reached the specified inductance(2.3uh). Is this method wrong ? Sometimes I have seen in designs that use air gap 0.5mm and ready made cores with such measurements are available.
and I use this core in my design if the inductance measurement does not match , still i can proceed with my design(ignoring the inductance reading)?

The most importance question i want to ask.. suppose I design a 12v 3 amp circuit... The air-gap... I didn't make one ... so the core saturates....
At what point will the core saturate.. 1) The circuit powered without any load . 2) any load is connected say..100ma.. 3)or a specified load is reached..say 1.25amp
**************
I made a mistake now ... with the FR107 heating up (RCD) , I replaced it with the MUR460 from the output stage(I had only one MUR460 then).
I replaced the output diode with sb560 (OR SR560 whichever with specs - 5AMP 60V ) . one terminal of this Sx360 was not properly connected ... so it went open circuit i think.
The unit stopped working.. I found the UC3845 VCC and GND shorted.. replaced it with a new one ... now the O/P reading fluctuates between 9.5-14.55
I replace OPTO and TL431 -- still the same... seems like i have to redo the whole thing

60V is too low for the o/p diode with your existing Tx, needs to be 150V at least - see posts above ...

No... I get it .. When i connected MUR460 in RCD , it didnt get hot... I didnt connect the O/P diode( the replaced SB560 - i checked it was not SR560) properly so there was no output (SB560 was never in a working circuit and as you said it would have been a mistake to use this 60v diode). As a result with not feedback from opto . so thereby maybe the aux supply went over 30+ volts (no zener clamp there) and burned the 3845? I now have a MUR460 at the output(I have MBR20100 but i think will be an overkill for this). Now i have to figure out what else got burned.

if you dont have airgap in flyback txfmr then it will saturate immediately because when you power up it is on full load charging up the output capacitor.
Can you get some gapped cores?...i feel sorry for you doing it with a file?......i tried that years ago and it was horrendous to do.

I will redo your flyback transformer for you....back soon.

So now you have 9-14.5v at 3A?..... what is the frequency of the oscillation?...is the output voltage "motor boating"?

So now you have 9-14.5v at 3A?

Click to expand...

oh.. no.. THe max i got was before and it was 12v at 500ma... any attempt beyond that collapsed it.
After i replaced the 3845 voltage around 9-14 and output now down to 200ma..

I attach here the new transformer in the ltspice schem for DCM at 230VAC...in fact, its DCM down to 270vdc input on the HV bus, so you have some margin.
The new NS/NP is 0.245
I kept your LPRI=226uH.......is that still what you have?

In this docement is all the math you need for flyback transformer
https://massey276.wixsite.com/maths

To find the saturation current Isat..... it equals B.A.N/L

where B=0.3 teslas
A = Amin (minimum core cross sectional area)
L = = inductance
N= turns number

I assume you know how to wind the transformer and do it all from here?

And it sounds like you have a feedback loop problem perhaps..........can you post latest schem and then it can be seen what compensation values you have used

Here also is guidance for course
https://massey276.wixsite.com/electronicsdegree

Here also is an SMPS course ..free download
https://drive.google.com/open?id=0B7aRNbu3Fes4TU92Mkw3YlA3ams

- - - Updated - - -

To calculate the airgap size.....use the reluctance to get the gap size dimension...but in truth, nobody does this.....its too inaccurate due to fringing... for a particular core, people just have look up tables....or do it trial and error.....or just put up with what you have if its satisfactory.

- - - Updated - - -

oh.. no.. THe max i got was before and it was 12v at 500ma... any attempt beyond that collapsed it.
After i replaced the 3845 voltage around 9-14 and output now down to 200ma..

Click to expand...

OK, do you have a latest schem?

@tree -- thanks for the info.. I will go thru the docs one by one.. may be I have a feed back problem... I willl reassemble the whole unit and come back in a day or 2.

I have a very silly question to ask... with the inductance given 226uh... what is the effect of reducing it to say.. 200uh... or by increasing it to 260uh..
ie,, by adjusting the airgap( even mine is not exactly 226 - it is around 250)..... is the 226uh given by TI is the maximum that should use..

Well...i made it just on the border of DCM/CCM at 270VDC with 226uH = LPRI and FSW=100khz......which is a good situation.

If you increase LPRI, then you will no longer be on the border of dcm/ccm at 270vdc input.

So if you increase the LPRI, then you will just need to adjust the switching frequency so that you get the pri and sec current profiles that you want.

..to be honest, if you play about with the sim then you can see what happens, and you can even "mould" the design into shape just by hacking on the simulator.
I will just change your LPRI to 300uH just for the sake of argument and resubmit it in a moment.

----------------------------
Remember your overall desire for flyback with hi vin and low vout and higher iout....

There is more current in sec than pri...so make the sec conduction time longer than the pri conduction time.
But dont make the pri conduction time so small that the current mode control trips off on your "turn on spike".
Also, remeber the leading edge blanking time.....you will always be on for that long (although uc3843 doesnt have LEB, just the delay in the RC sense filter)
Make it in DCM....but not so far in dcm that you get really high sec current peaks.
DCM is best becuase it avoids sec diode reverse recovery.
(ok if you just tip toe into slight ccm at low vin then thats ok)
if you make LSEC and LPRI TOO low...then you will have to really increase fsw in order to keep the peak currents down.

sunil21 said:

I have a very silly question to ask... with the inductance given 226uh... what is the effect of reducing it to say.. 200uh... or by increasing it to 260uh..
ie,, by adjusting the airgap( even mine is not exactly 226 - it is around 250)..... is the 226uh given by TI is the maximum that should use..

Click to expand...

not sure what changing the inductance will actually do, however, while the inductance is specified at 226 uH,
there is tolerance on that of at least 10%, possibly more. (that essentially includes 200 and 250 uH)
It comes from the tolerances of manufacturing the core material,
the dimensions of the core, and variations in the windings and etc.

you can set the air gap relatively easily and relatively repeatably by inserting 1 or more layers of paper.
For prototype purposes, plain printer paper will do, or parchment paper. it usually works best to gap all
the legs equally. If you're using an E type core, you can put the paper in the two outer legs.

For production purposes, you can buy craft paper at specific thicknesses, 1 mil, 2 mil, etc. Sorry, I don't
know where to get it.

here i attach ltspice with LPRI = 1mH...just for sake of argument...and to keep it in DCM......i had to reduce the fsw to 16khz......and adjust LSEC accordingly....
You can also play about with it......1mh woudl mean a much larger transformer than efd25

Or - you can increase the pri turns to 60 and double the gap - and hey presto - DCM ... and a decent flyback voltage ( 136V ) and a 100V o/p diode will do ...

- - - Updated - - -

I found this excellent article on linkedin:

www.linkedin.com/pulse/design-flyba...n-j-tuck/?trackingId=kGwlgunIRFKW/Jhgyvmudw==

That's a great link...see Dr Slobodan Cuk of Tesla is talking air gaps there.

talking crap unfortunately... - the gap length is not independent of gap area - where he gets this off the cuff nonsense is a mystery - perhaps he has never designed a flyback ...?