Fast Surge protection

Thread Starter

Ioannis66

Joined Nov 7, 2012
45
Hi Forum.
I need a very fast surge protection on a power line feeding a LED driver at 24 volts. The system was tested under IEC 61000-4-x specifications and failed on pulse disturbance induced on the power cable.

A TVS diode was added (1.5KE33), but maybe it was not fast enough. The test pulses induced are in ns level.

My next actions include a ferrite ring on both the power cables and a faster TVS after them.

Any other suggestions for faster TVS or other kind of protection please?
Thanks,
Ioannis
 

Deleted member 115935

Joined Dec 31, 1969
0
welcome to the world of pain.

Two ways to solve this
a) stop the pulse getting in
b) cope with the pulse.

(a) is difficult to retro fit,
You are on right track with a surge device, but, they have a limited number of hits before they stop, and they still let a large spike in before they attenuate.
even having a power line coming inside the case, the power line radiates the pules to any high impedance lines, such as resets.

The other help is to put large inductors and capacitors Pi filter on the power lines, to slow the edge of the pulse down.


(b) is the way to design at the start. basically , design th3e system not to magnify a glitch.
classic is in a large system is a global reset line, high impedance and picks up all the noise.
make the reset needing to be long before it does anything is the classic answer to this.

In the end,
there is no easy , "do this and it works", its a black art,
Ok, I'm biased as I get paid big bucks to sort this sort of stuff out,

http://learnemc.com/emc-resources
 

Thread Starter

Ioannis66

Joined Nov 7, 2012
45
Thanks for the replies.
Well, the circuit is proprietary and I am not allowed to publish. It is based on ARM processor.
There are no power lines or reset lines involved. Everything is internal to the circuit in earthed metal case-shield.
The device is powered by a DC source of 12-28 volts through a silicon cable. The light fixture is designed for fountains and is waterproof in a very tight shell.

Everything works ok and does comply with standards, except this specific pulse test that locks the driver chip. After power cycle it works OK again.

Ioannis
 

Thread Starter

Ioannis66

Joined Nov 7, 2012
45
I do understand Ian0.

My next step is to try ferrite bead on the power cable. I think TVS in not enough on its own, as the test shows...
Thank you.
 

Deleted member 115935

Joined Dec 31, 1969
0
ferrites wont stop the surge,

Suggest you look to your software,
the cause could be that the software can not cope with an interruption,
I don't know what you have, but I have seen that on USB and serial links, the software is expecting constant data , the surge causes a data glitch, and the cpu locks up.

also look with a scope at the PSU and reset circuit, see if they brown out or generate a reset on the spike.
 

Marc Sugrue

Joined Jan 19, 2018
222
Thanks for the replies.
Well, the circuit is proprietary and I am not allowed to publish. It is based on ARM processor.
There are no power lines or reset lines involved. Everything is internal to the circuit in earthed metal case-shield.
The device is powered by a DC source of 12-28 volts through a silicon cable. The light fixture is designed for fountains and is waterproof in a very tight shell.

Everything works ok and does comply with standards, except this specific pulse test that locks the driver chip. After power cycle it works OK again.

Ioannis
If the spike is being coupled in via the power lines it sounds like you need a line filter with some formal inductance to damp the spike. You could try a PI or LC filter on the line, a ferrite won't be sufficient in that case.
 

Marc Sugrue

Joined Jan 19, 2018
222
ferrites wont stop the surge,

Suggest you look to your software,
the cause could be that the software can not cope with an interruption,
I don't know what you have, but I have seen that on USB and serial links, the software is expecting constant data , the surge causes a data glitch, and the cpu locks up.

also look with a scope at the PSU and reset circuit, see if they brown out or generate a reset on the spike.
I've experienced this issue also.
 

Ian0

Joined Aug 7, 2020
9,668
Certainly, the CHEAPEST way to stop it is to see if you can get the microprocessor to be able to deal with it, via the watchdog timer or brownout detector.
Depending on what level of test it is, does it pass if the processor gets reset and starts the program again from the beginning? )(I'm guessing that it's not safety-critical!)
Or does the driver output latch, even though its input continues to work?
 

Thread Starter

Ioannis66

Joined Nov 7, 2012
45
Thank you for the replies. Yep, the idea of a heart beat LED to check if the MCU or the driver is failing might get me somewhere.
Ioannis
 
Top