Failing Surge EFT test at 1kV.

[morzh]

Hi.

I have a 85-270V In / 14V out Flyback SMPS.
We are failing EFT (Fast transient) test with 1000V applied. We pass 500V one.
The surge is failed in the way that the equipment connected to it (an optical modem) sustains Bit Errors.

I am not sure yet how they apply the transient - we do not have the Protection GND connected: it is a 2-wire in (Line/Neutral) supply, so when they say the transient applied to each line, and then to both lines, I have yet to see it done (will see it tomorrow).

The input is protected by these in following order:

620V MOV (I think about 50 Joules absorption), then 5 Ohms NTC, then 0.68uF cap, 15mH CM choke, and tehn all connects to the rec. bridge.

Extra details:

The Y-cap is 2.2nF and connected between output GND and negative of the rect. bridge Cap (56uF, 400V electrolytic).
The transformer has no internal EMI shielding.

Any ideas? I am going to try different MOV tomorrow, including one for 470V I have handy.

Regards.

 

[Dmitron]

1) the surge is applyed between the input AC lines.
2) 4kV surge discharge can be applied between FC-line and the output of device as an isolation test.
in this case may be 2 mistakes: a) the optocoupler is opens or closes during surge- so device have no power supply. b) you have bad grounding - so the current through 2.2 nF cap. results fall triggering of logic.

 

[morzh]

OK, thanks, Dmitron.

I found out what the problem was. The design is fine. What happened, when we assembled boards, we were missing CM chokes, and for testing supply itself we shorted them out.

What I did not know was that one of those supplies ended up in the testing for EFT, which it was not supposed to do.

Once I opened the box I saw this right away.

I put 10mH choke I had available in there and we passed every single test under 1kV.

Thanks again.

Whew.....I wish every problem was that easy.