EMI or ESD problem

Mina Anton said:

When other surrounding inductive loads (Fluorescent lamps with normal transformer and starter) switched off

Mina,

If I read it well these FL lamps are not connected to your board’s relays but via AC-mains and a switch at a distance of 2 meters from your board’s AC power supply. And the MCU reset’s when switching off the lamps.

These older type of FL lamp units can generate a lot of noise to your AC-mains, not only when switching off but also when switching on. If you would extend the distance between them it will disappear.

Over the FL switch you can place a capacitor up to 1uF 400VAC.

I guess you are using a AC mains adapter to supply your board. A good quality one should have an LC filter at AC input, if yours don’t have replace it by such one or add a separate AC filter.

Leo.

Hi Leo,

Thank you very much for your response; you’re 100% correct, I thought about changing the transformer and starter with new electronic starter, the problem disappeared.

Also I have to say that both (the lamp and my circuit’s adaptor) are supplied from the same AC line (from different points).

But the problem is I need to install this circuit in several areas, all with the same problem and I don’t have the option to change the transformer (or add a capacitor or increasing the distance); in other words I need to increase the circuit immunity without changing to high quality adaptor for my circuits or the transformer for the lamps.

Also I need a simple way to regenerate the problem in my lab to make sure that the final solution will work fine in the customer location.

Thank you.

Mina

Hi Leo,

Again thank you very much for your prompt response.

 Regarding the power supply, I’m using a single power supply to supply multiple boards in one building in one or two floor level(s) (several rooms), I’m using a 2 x 2.5mm wires to distribute the power in a bus topology, I’m using 24VDC output power supply to allow good margin of voltage drop and a DC/DC “TPS5430” to output 5VDC (to power the relays), then LDO “TLV1117-33” to output 3.3V (to power the microcontroller), so I believe that even with better power supply I will still suffer from this problem due to the long wires (about 100 meters) hidden in the wall in a plastic pipes. I don’t have the option to of adding separate power supply for every circuit, as I’ve to use this infrastructure.

The power supply I’m using is SMPS not the old ones with transformer, but you can’t grantee the quality of the filters in Chinese one from unknown brand (and this is the best available option in my local market).

Regarding earth:

Yes I know it helps a lot and it is available in most cases (90%) but unfortunately not available in my current test location that I got from my customer.

Regarding relays in my board:

Yes it caused huge problem when I tried to connect it with FL lamp as it was resetting the board during switching on, I added a snubber circuit to the relays, and it reduced the effect but didn’t eliminate it.

I don’t have much experience in snubber circuits but I did a simple search and found that the R & C values are by somehow variable depending on the load type and since I don’t know what kind of different loads may be used by the customer so this increased my problem.

Regarding regenerating the problem:

I’ve just remembered that I faced this problem once before with an old inductive door bell so if I can got one now it’ll regenerate the problem again but it’ll be a noisy test.

As a conclusion: I estimate that adding strong filters (and protection) on the DC and proper design of the reset pin circuit will eliminate my problems and this is exactly what I need your support in, even if I’ll add additional components and make several trials.

Thank you again.

Mina

Hi Leo,

Thank you very much for your response, it took me some time to import the required components since it’s not available in my local market.

By somehow the problem is solved… and this is the good news.

The bad news is I discovered that I miss an important design consideration point (the EMI protection).

It may be the biggest favor you could do to me and many others that may refer to this conversation is to guide me to a good tutorial (or lectures, books, etc.) that teaches how to make proper protection to input lines, power, RS485, USB, etc. so in my next design I don’t have either to wait till I meet problems and then search for a solution or to do a try-and-error technique till I get the best case which may not be the optimum, also to increase the reliability of my products.

However I’d like to thank you for powerful support that solved my problems.

Mina

Hello Mina,

Good to hear you’ve solved your problems.

There are several documents, here on the TI site as well on other sites about Board Design Guide lines and other design rules. But it all needs a good understanding about Magnetic fields and the several current flows. EMI is one of the most difficult cases, you might have a good understanding about magnetic currents but you never know where they really goes. It’s all about experience and experimental.

The first rule is to avoid EMI problems, to filter them out is in most cases very difficult, avoiding is much more easier as you might think.

First have a good look surround your board, how are connections made and with what type of cable (twisting, screening etc.). What are the current flows through the connections and – very important – the screen, in many cases it’s used for earthing the parts which makes the screen a noise transmitter instead of a noise protector.

The Power line Earth might look a solid base for a clean reference, but in many cases it’s more dirt as the power lines themselves, so very much care has to been taken about the use (routing) of it, and certainly don’t use it as (board) electronic reference (GND) point.

The onboard; Input, Power, RS232, USB protections must been more seen as a ‘mayday’ protection to keep your board alive in case of an external attack. But Ferrite in general is a good product to short noise.

Leo.