Wish to test noise immunity of electronics products.

[Flyback]

Hello,

We wish to see if our offline Switch Mode LED lights are immune to noise.

Therefore, we will make a special “noise producer” as in the attached. It is simply an LLC converter with one of its switching nodes dragged out of the enclosure for 30 metres. This 30 metres is loosely twisted (it’s also insulated wire). We will drape this 30 metres of “noisy wire” over our products as they are working –to see if it makes them go wrong.

Do you think any of the products will go wrong when the “noisy wire” is draped over them? Might any operating electronics products be disturbed by being brought near to such a wire?

The waveform of the square wave of the wire is also shown. It is a square wave with 7ns rise time.

LTspice simulation of "noise producer" attached also.

By the way, the LLC converter ('noise producer') shown feeds off the output of a fully isolated full bridge converter, which itself feeds off the output of a 240VAC mains PFC stage.

 

[kubeek]

Sorry, but my crystal ball is in repair. Nevertheless, this has nothing to do with the actual noise immunity tests and either of the possible outcomes is meaningless.

 

[alec_t]

We wish to see if our offline Switch Mode LED lights are immune to noise.

If they are CE complient then they will be immune. If not, they shouldn't be on the market .

 

[Cicero]

If they are CE complient then they *should* be immune. If not, they shouldn't be on the market .

Haha, fixed it for you

The only way to properly test for immunity is to go to a certified lab.

 

[DerStrom8]

The only way to properly test for immunity is to go to a certified lab.

This is what I was going to say. Nobody is going to accept "CE certification" performed by the manufacturer who makes the parts being tested. You need to go to a proper testing facility. You can't do it yourself. No cutting corners here.

 

[Flyback]

Thanks, we are still interested in the possibility of a "dragged out" switching node exactly like shown causing other electronics to fail. -Or even making the power supply itself fail (the one thats had its swithcing node dragged out as shown)
Its a long story why...if you wish i can tell you?
This isnt anything to do with official regulatory noise immunity testing.

 

[unclejed613]

the wire is 30 meters long, and each side of it is 180 deg out of phase with the other, so the net inductance is close to 0. you might get some noise at 2.5Mhz because your "node" acts as a 1/4w shorted stub transmission line. it's not going to be much noise, and certainly not enough to influence the operation of LEDs.

 

[RadioRon]

Do you think any of the products will go wrong when the “noisy wire” is draped over them? Might any operating electronics products be disturbed by being brought near to such a wire?

I do not think that this test will cause your products to go wrong. There are some operating electronics that this test might disturb.

I see what you are after with this test setup, but it is a very specific circumstance whose results are difficult to extrapolate from. I mean, that, for example, let's say that your test causes your LED devices to malfunction. Does that mean that the LED design is deficient, or that the test is too severe? Does this mean that other LED devices that you haven't tested will malfunction? Does this mean that your LED devices will malfunction after being installed by customers? You see, there is no calibration of this test to recognized or calibrated standards or to other real-world circumstances. It is, as I might call, a shot in the dark, so to speak. If your test result pointed to the need for a design change that costs money, how will you justify this expense?

Your test attempts to impose a broad spectrum of interference onto a test subject. The choice of square wave with a fast rise and fall time does indeed generate considerable spectral components and because there will be strong current flowing in your test wire, the spectrum components will likely have significant strength, however the method used to couple that energy to the device under test is not good. One way that you can improve this test is to open up your 30 meter wire by separating the two conductors and form it into a coil around the test subject. Most of the coupling will be through magnetic field, kind of like a transformer, and your 30m wire makes a poor primary. It will couple considerably more magnetic field if it is formed into a large loop that surrounds the test subject. Another way that you can make your test setup much more effective is to allow the square wave switching frequency to be variable and controlled externally. With this feature, you can sweep the frequency up and down to look for sensitivities that may not be apparent when using only one square wave frequency.

I believe that others perform calibrated testing for electromagnetic compatibility by generating one spectrum component at a time (ie. a sinusoid) and then radiating them, with significant strength, onto the test subject. When dealing with one frequency at a time it is possible to calibrate and know the field strength being imposed onto the device under test and so then be able to compare results to others.

This link offers a quick explanation, for others reading this thread, of what EMC is about.

https://www.emcbayswater.com.au/blo...romagnetic-compatibility-emc-testing-methods/

Your test falls into the category of susceptibility/immunity. Have you given any thought to testing for excessive emissions from your LED product? This is of greater concern to the community.

 

[Cicero]

I just want to add, that testing houses are pretty reasonable. You can book half a day for a few £hundred, and do all the testing you require with proper results.

Think about how much time, cost, and effort it will take you to build your noise generator, considering that any results you achieve will be met with large amounts of scepticism making it all worthless. ie: If you see no change, you'll wonder if the noise gen is even working. If it breaks your work, you'll wonder if its generating too much noise. Its just far, far easier to simply book a few hours somewhere, and get it done properly.