RE Suppression method

Question

I am working on a Single phase Welding machine with a input voltage of 230V and Current of 20A and output ( 100 A and 24 V). Now we are supposed to pass it CE certification, as per standard EN 60974-10. We conducted all the test like RE, RS, CE, CS, ESD, EFT, Surge, Voltage Dip, Harmonics and Flicker. All most all our test passed but it gets failed in the RE and CE. In CE it fails in the 25 MHz to 30MHz by some 10 dB and In RE it fails in 30MHz to 50MHzby some 10dB. Basically it fails in both the test in the higher frequency only that after 25MHz to 50MHz. The switching frequency are as follows IPS is 50KHz to 200 KHz (Topology Quassi Resonant Converter) PFC 70KHz ( Topology CCM Boost Converter) Main Inverter 47KHz (Topology Full Bridge Converter)

Kindly give some suggestion for the RE to pass.

Answer 1

At 230 V, 20 A, and above the limit by 10 dB, you don't just slap on a cap somewhere. You'll need a serious line filter at least, and that's just a bandaid instead of a cure at best.

The real problem is lack of attention to emissions thru the whole design process. You don't just build something, then test it, then poke around sticking filters on until it passes. It would take a large, heavy, and expensive filter to fix the problem after the fact like this. The fact that this is all you're asking about shows that lack of proper design up front is the real problem.

So the right answer is go back and design it right, this time actually thinking about proper grounding, minimizing high frequency loop currents, etc. There is no good way out of the mess you're in. Good design costs real money, but now you can see how bad design costs even more.

Answer 2

10 dB is a severe over-the-limit fail and you should be aiming for at least 5dB below the limit on this type of product (industrial welding etc.) when type testing at an EMC test house.

You have this product that has incoming AC wires and outgoing welding wires. Basically a lot of cable and there's no test house on earth that can distinguish (in the "standard" test) whether the radiated emission is from the "box" itself or the cables or both. So you have to take steps to understand where the emissions are coming from. (Halving the problem and dealing with each in turn).

The results you have so far indicate to me that there is a reasonable chance that a good measure of the radiated emissions are coming from the cables i.e. it is conducted emissions radiating from the cable. It happens quite often.

So, my advice is try and rule this out by doing a test with the shortest possible cable(s). Bear in mind that the radiation may also be emitting from the high current cables attached to the box that are normally used for welding. You can force cables to be in a different orientation so that the radiation impact on the test house's receiver is also minimized.

In short do everything to minimize conducted emissions turning into radiated emissions. Then you move on with the radiated emissions from the box and a possible culprit is the switch mode transformer - try shielding it. I had an emission problem from a transformer once and wrapped some copper tape around it (carefully) to prove that it was that. Then I engineered a better solution the next day.

Other than that I could suggest you use a "sniffer" aka near-field probe to try and narrow down the culprit(s) on your circuit board(s). PCB interconnection wires are also a primary example of conducted currents also emitting.

Answer 3

Sounds like your problem is pretty severe! Try slowing your switching MOSFETs down. It will generate more heat, but if you can tolerate that it might reduce some of the harmonics.