SMPS power up problems.

[nigelwright7557]

post the current waveforms used to detect oc.

I cant, I dont have an isolation transformer. I can only look at the secondary safely.

I have got a bit further.
If I decrease the pulse width to the primary I find I can then decrease the gap between the pulses to almost meeting up.

I might be fooling myself a bit coz i am using a lamp in series with my setup to limit current.

 

[whizgeek]

Almost certain you are saturating the transformer.

thats a dual scope you have there.
Put a resistor in the return leg of the switching element, use probe each side of the resistor, differential mode on the scope

 

[nigelwright7557]

Only one channel works, I got the scope very cheap.

Yes the transformer is definitely saturating hence the over current problems.
I am using an etd34 with 2mm gapped core.
If I output just a short pulse to the mosfet it works fine.
The output looks OK now after I snubbered the primary.
I have tested the output upto 500mA, I am waiting for high power resistors to arrive so I can test higher currents.

 

[nigelwright7557]

I am trying to get around 5 amps at 40 volts out.
Input is 240VAC mains. I am running the PIC at 50KHz with 50% duty cycle.
I am using an ETD54 with n97 core, 50 turns primary and 13 turns secondary.

I am now running the PIC mosfet driver at 500KHz so there is a 2uS pulse to the mosfet. Anything longer causes an overcurrent trip.
I am now using an ETD34 with a 2mm gapped core.
I am managing to get 2amps at 30 volts out at the moment, but I would like to get more.
The transformer is getting quite warm which is a bit of a worry but I have read SMPS transformers often do.

 

[nigelwright7557]

I found a fault in the datasheet circuit for the L6565.
It shows 100nf with a 220k in series in the tL431 circuit.
This causes problems with the feedback signal.
If I make the 220k 0r then the feedback circuit works fine.

 

[nigelwright7557]

I added another small winding to the transformer to detect zero voltage crossing on the output so I know when to apply the next pulse to primary side of the transformer.
Seems to work very well.

 

[nigelwright7557]

Got a bit further tonight.

I found I was getting an over current event because I was firing another primary pulse before the previous secondary discharge had finished.

I can now get full secondary voltage that I wanted at a decent current level.

 

[whizgeek]

Good to hear you are getting there.
Re the datasheet; your application is substantially different, so values dont really apply.

What PIC are you using ?

 

[nigelwright7557]

I am trying to emulate the L6565 with a PIC.
The PIC now does output pulses, zero voltage crossing detect and primary over current detect. I also detect voltage feedback from the secondary.

I am just using a PIC16f506 running at 20MHz.
There is very little code to do.

 

[5th element]

Not that I know much about SMPS design (something i'd like to change at some point), but it seems like you've managed to solve the original problems. And even though there haven't been that many contributions you've still kept us informed with the process of how you figured it all out, which is rather helpful. What voltage/current will this thing now work at/provide?

 

[nigelwright7557]

What voltage/current will this thing now work at/provide?

It currently runs at 50 volts and 3 amps.

 

[5th element]

Now that you've solved the apparent problems, would you say that the transformer is running cooler then it was before, or does it just run hot? And have you measured the efficiency of the thing?

 

[nigelwright7557]

Now that you've solved the apparent problems, would you say that the transformer is running cooler then it was before, or does it just run hot? And have you measured the efficiency of the thing?

It is still running hot. I have also found the transformer to be very noisy on full power.
Its not quite there yet, I still have a bit of work to do.

 

[nigelwright7557]

I have returned to the ETD54 core and added a 1mm gap and the transformer whistles just slightly, should be ok in a enclosure.
I am now getting 40 volts at 3.2 amps which is passable.

 

[nigelwright7557]

Latest update.

I tidied up the design and ordered some new pcb's.
They arrived and I built one up.
I had problems previously with the transformer whistling so I wrapped the turns tighter this time then applied lacquer to hold the turns together.
Fired up the new pcb and it still whistles like mad !
I had another look at my software and it looks at the voltage feedback and if a pulse is required it gives a full power pulse. The gap between these was a bout 5KHz hence the whistling.
I tried adding a very short pulse to the transformer if a large pulse wasn't needed and the transformer suddenly went very quiet.
I am getting 50 volts and 3 amps which was my target so happy now.

I should have added I went back to an ETD54 transformer. I couldn't source any gapped cores so I added 1mm spacers to the core.

 

[nigelwright7557]

I did a full test of my SMPS with different loads tonight and the whistling is there again.

So I have altered the software to measure how long between feedback pulses and output a pulse proportional to the current required.
While this hasn't cured the whistling it has spread the whistling spectrum a bit so it isn't so loud.

The L6565 outputs pulses depending on the current taken by the transformer primary and so doesn't have this problem.

 

[nigelwright7557]

I found I got an improvement in current output by giving the transformer a larger gap.
This reduces the inductance a little and allows a longer drive pulse to the primary of the transformer.