LNK306GN - Problem with regulation and output voltage drop

Could anyone help me ?

Note : I have updated pictures in my previous post.

Thanks

for the selection of your free wheel diode (DFW in your schematic)  ,  are you following the step 6 from AN-37?   the trr of the diode should be below 35 ns .  

Thank you for your attention.

The free wheel diode and feddback diode are the same, with trr(max) = 25ns

Hi,

Here is the inductor current scope shot, just in case. I used a 0.8 ohm resistor in series with the inductor. Referring to der231, page 16, it seems to be normal waveform.

Do you have any other suggestion on how I can improve my circuit ?

Your schematic look good with except  the inductor. I s quite big for this kind of current.   Can you please try around 500 uH?

 Could you check  your feedback branch to make sure there is not any cold solder joint ?   

Here is  a comparison between behaviour with 1mH or 680mH.

We noticed that the output ripple is smaller than previously but we still don't have a 5V output. We'll try to set an even lower value for L.

The feedback joint solder seems ok. I'll make some measurement and keep you inform.

We have new observations to share with you.

We watched the feedback capacitor voltage :

- when there is no load connected, this voltage is the copy of the output voltage (and we have 5V)

- but when we connect a load (still 45ohm), we notice that the output falls down around 4,3V (with variations, as mentionned before) but the capacitor voltage is still at 5Vdc voltage. That could explain why the lnk doesn't regulate the output voltage.

So there is problem with the capacitor voltage since it doesn't track the output voltage when we connect a load.

Do you think that the feedback capacitor value could be wrong and cause that, knowing that FW and FB diodes are the same ?

Waveforms are attached.

Thanks

I am having the same problem.

I changed Dfb to a 400v ultra fast and it only exacerbated the problem.

Based on the symptoms, Cfb is not getting charged via Dfb.

As the load (current draw) increases the FET stays on longer.  This allows more energy to leak into Cfb via the divider which in turn shuts off the fet before the output reaches regulation.  This would explain why the output voltage moves inversely to current.

My question:  Has anyone been able to get a 300mA power converter to work based on the AN-37 layout?

As the FB pin only needs 47uA, I assume that I can increase the values of Rfb and Rbias.  This will not cure the problem but it might reduce its magnitude.

As a test; I pulled up the resistor divider and put down an optocoupler with an a zener.  I then ran the converter at varying loads from 25mA though 300mA.  The LNK306 held regulation.  

I made my board look mostly like the optocoupler picture on the top of page 3 of AN-37.  I added a resistor between optocoupler and the FB pin just to keep from dumping an unthrottled 5.8 volts into the fb pin.   I am not sure that this was needed, but I did not want to have to replace the 306 for not being prudent.

The cost of the optocoupler is about the same as the 600v diode so in the long run I call it a push.

You need to adjust Rbias to have the desired output voltage when the supply is loaded.  Reducing RBIAS increases your output voltage.  If you need regulation at no load,  then you may need to increase your preload.

Could you finally get it work?

I casually designed the  same circuit but  have the exactly the same problem.

Any help will be apreciated

Ysaac