LNK574 2W clampless design doesn't work above 130VAC input

Are your windings in the correct polarity?

Have you gone through with scope probes to see what different components are doing and if they're behaving the way they should?  

Do you have your PIXLS design files and PCB layout information that you can attach?

-The Traveler

It was my first assumption but unfortunately winding polarity is correct. My next clue was the order of windings (the first samples had feedback windings outside). But the next version, with feedback winding inside, behaves in the same way.

Please find attached documentation.

Thanks

Witold

One thing I'm noticing is that it appears you have almost no SRC pin PCB copper area and that all your feedback and bypass pin components share the same return paths as your drain current.  This is typically a huge no-no is any power supply design.  Can you verify this?  It's hard to tell from your PDF layouts.  Figure 5 has a decent LNK584 PCB layout example and Fig 5 of the LinkSwitch-CV datasheet (http://www.powerint.com/sites/default/files/product-docs/linkcv_family_datasheet.pdf) has a nice PCB layout example showing proper return paths to the input bulk caps for both the SRC traces and bias winding connections.

-The Traveler

I can't add much more copper to the SRC signal because I'm surrounded by low volatge (safety) traces. But for sure I can improve things a bit. I can't copy Figure 5 because D3 (DB on the fig.5 ) and C7 (CB on the fig. 5) are considered as a hot devices and there is no enough room for hot devices under EE13 core.

I can try to jumper SRC with relatively thick wire

I have noticed one more strange behavior - my design works with 30VAC input voltage. I can't expect 1:9 input voltage ratio from so simple design. All standard designs are specified for 1:3 (85...265).

Witold

PCB layout can really make or break a power supply design.  It is especially important how you route and place all the feedback components as well as how you "steer" the different currents in different sections of the design.

Depending on your timeline, I would suggest making a small power supply proto design on a board by itself.  Get the bugs worked and make sure it works in an ideal layout/configuration.  Then maybe see about getting it to fit into your mechanical constraints.  This will really help you get an idea of what problems are design related and which ones are PCB/Layout related. 

-The Traveler

That last image you sent of your PCB layout was pretty helpful.  Could you send me something a bit more detailed though (not the individual layers from the big zip file) but that also includes the input section (bulk caps, rectifiers) as well as the rest of the output section.  It'd be easier if I can see what is going on in your design.  Feel free to sketch/draw out on the layout and leave me some notes. 

-The Traveler

Hi!

The first problem with my SMSP has been solved. It was not related to the layout. The output rectifier was the weak point. It must be really serious diode with breakdown voltage  well above 100V. Mine was good up to 60V.

Now I'm working on V(out) versus I(load) characteristics. Right now my power supply can't deliver much more than 1 W and voltage drops are huge.

Witold

Hello!

Why is my board running at 60KHz only? According to specs it should be close to 100kHz. Please look at the attached photo I took in the following conditions:

Vin = 89 VAC (111 VDC at the input)

Vout = 14V ( should be 24V, very poor output characteristics - see attached pdf file)

I load = 57mA (PIXI calculations were performed with 80mA)

The nominal oscillator frequency is 100kHz but the controller adjusts this frequency depending on load and line conditions. 

-The Traveler

You don't appear to have a snubber on your output rectifer and the drain waveforms in your previous posts appear to contain quite a bit of ringing at turn-off.  You definitely have some PCB layout issues from what I can see and I'd probably quess that there might be some transformer construction issues as well. 

-The Traveler