Help to deal with ESD

[alex.calori]

Hi, Folks!

I'm a noob in designing electronic circuitry and I hope you guys can help me!

I'm facing an issue to deal with ESD!

My electronic circuit is a LED string powered by a switching integrated circuit (IC) that supports up 60V@1A and 500 V ESD.

I’m powering the circuit with 48 Vdc to feed 16 LEDs which sum about 46 Vdc at 700 mA. I’ve no problem to operate the system, LEDs work fine for a long time until I decide to put some stress.

For stress, I just manually unplug the power wire (48 Vdc) and plug it again – as you guys know, 48 Vdc generates a small electric arc (ESD) which is sufficient to breakdown my IC and ruin out the regulated current  .

So, in searching to find a solution, I found transient voltage suppressor (TVS) diodes as an option, but I do not know how to size (select) a proper device for my application. Also, I’m not sure if TVS will help me!

Please, help with!

Thanks a lot!   

[jeffjmr]

 

Interested in the answer myself.

[Benta]

I think the problem is somewhere else in your design.
But as the circuit is secret, I can say no more.

[mclute0]

my silly mind would be looking at something like (no this one doesn't meet your voltage spec)

https://www.ti.com/lit/ds/symlink/tps22965.pdf

since I am guessing it is not an ESD problem but an inrush problem...

[alex.calori]

Hi, Benta!

I think there are no problem with the design- I guess.

The LEDs run well and the problem happens when I put an ESD in the power line!

I plugged my oscilloscope probe in the input and I saw peak-to-peak voltages (e.g., 120V) far higher than the maximum voltage supported by the IC (65 V).

Afterward, I connected a TVS diode in parallel with the main line and no more transient voltages were noted!

I'm still running my test and ASAP I find the solution, I'll make you guys knows!

Thanks a lot!
 

[alex.calori]

Hi, mclute0!

An inrush would will be a problem too!   

But how can I know for sure?

In series with my circuit, I connected an analogic amperemeter and, when I power the line, visually I cannot see spikes higher than the threshold stated in datasheet.

Are there a better method to analyze it?

Thanks!

[m98]

While a TVS-Diode on the input is certainly a sensible protection measure, it doesn't solve the cause of your particular problem. Your voltage spikes are generated by the supply line and your low-ESR input capacitors forming an underdamped LC-Resonator. Add about 10x of your current input capacitance with a higher ESR (either electrolytic, or MLCC in series with a few parallel resistors to get about 1 ohm) to act as an RC-Snubber.

[Marco]

The problem is that your power supply is shit.

[mclute0]

Hi, mclute0!

An inrush would will be a problem too!   

But how can I know for sure?

In series with my circuit, I connected an analogic amperemeter and, when I power the line, visually I cannot see spikes higher than the threshold stated in datasheet.

Are there a better method to analyze it?

Thanks!

I would say you need an oscilloscope to see the problem clearly; the inrush happens fast and is over quickly.

Without much information on what you are doing, I am just guessing. like doing sudoku.

The spark you see is because the manual switching method is not fast enough. The current jumps the air gap between the contacts before the contacts touch creating a transient. This is called a spark gap and it is how the first radio transmitters operated, so there is EMI; but, I doubt at your stated voltage it is damaging the IC.

I would say the basic idea of a load switch is to use an inline capacitor to absorb the initial transient charge into a capacitor instead of your chip and a parallel reverse biased diode to reduce parasitic capacitances.

A decoupling capacitor is the same idea. If the input voltage drops, then a decoupling capacitor will be able to provide enough power to an IC to keep the voltage stable.
If the voltage increases, then a decoupling capacitor will be able to absorb the excess energy trying to flow through to the IC, which again keeps the voltage stable.

It all depends on what your doing and why... I like to use packaged solutions to do most things, but that's just my way of reducing work.

A more complicated solution would be using a mosfet to switch power on and off...

An inline MOV (Metal Oxide Varistor) is another possible way to reduce transients.

or NTC thermistor.

A TVS diode alone would work, but I have found they eventually fail without a capacitor somewhere to take the transient charge.

I trust someone will correct me if I misstated something...