TOP269 EN-55014 fails
Hi all,
I have a power supply based on the TOP269 with the output 24V @ 5A and the input range from 190 ~ 260Vac.
This power supply was developed and designed according to the technical data (and datasheet) of the TOP269.. for now it works OK.. but it fails in the certification process, in this case in the norm: EN-55014
In attachments I send the scheme details and also the PCB drawing, and the result..any idea for this bad EMC performance? Is there anything that we have made incorrectly.
Thanks in advance.
回應
Hi,
The Line and Neutral is placed along your EMI filter. With this arrangement noise may bypass the EMI filter stage and couples to the input line directly. Placing input connector right next to MOV might improve the EMI performance in high frequency region.
If that is not possible, then you may have to change the layout to arrange the EMI filter stage so that noisy nodes are separated from input lines.
Moving the Y caps close to the lines as also is some thing that you can try.
I saw that it is failing in low frequency region, so adding differential mode inductor or increasing Xcap (C93) value might improve the margin in the low frequency region.
Regards,
PI-NANO
Hi all,
Somehow I miss these replies.
Thanks for your feedback of both..
In the ancient pcb design I had made the changes.. put the capacitors of 2n2 in the input.. but the error/noise is the same.
I had also tried to put differential mode inductors, the error is same..increased the Xcap (C93), same... so I ran out of options..
At this moment, I'm trying to a new redesign of the pcb taking special attention to all the points referred here.. and also points from another thread.. by the way.. if you have more ideas, please let me know.
Thanks in advance
The results are surprisingly good for that layout, the only fail point is 152kHz, this is likely to be DM, so increase C93, C94 & C91...
I will try this in the next days.. I will keep in touch..thanks for the feedback
Your neutral and phase line traverse the whole pcb, they will pick up noise, perhaps shift the two 2n2 Y caps to the input terminals, this may well help ...