2
\$\begingroup\$

With my firm we're trying to get the CE certificate for our device and one of the tests is the protection against power surges. For this, the test company is using a device that's connected to our device's 12V power input and gives +500V surge for 1.2 microseconds. For this test, we included a TVS diode parallel to ground to our schematic, and the final schematic looks like this:

D18 is an 18V TVS diode, SMBJ18A-HT. U12 is the voltage regulator,

D18 is an 18V TVS diode, SMBJ18A-HT. U12 is the voltage regulator,

In its datasheet, it's stated that SMBJ18A has a surge current of 20.5 Amperes, and a clamping voltage of 29.2 Volts. This looked satisfactory to our circuit, as our voltage regulator works between 5-36 Volts. In the actual test however, the TVS diode failed and shorted out, causing the power wires to burn and melt.

Is there any error on the choice of TVS diode for this purpose or should we use a different sort of protection for the 500V surge?

Regards, Alp.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ What specific requirement (directive?) are you testing against? 500 V surge sounds like the kind of thing you'd want for the mains input, not for a 12 V DC input, but I am not an expert on CE requirements. \$\endgroup\$
    – The Photon
    Nov 7, 2019 at 6:35
  • 3
    \$\begingroup\$ For example, this app note mentions 500 V as a requirement for electrical fast transient (EFT) and IEC61000-4-5 surge testing, both of which are applied to the AC power input, not to DC power after conversion from AC AFAIK. \$\endgroup\$
    – The Photon
    Nov 7, 2019 at 6:39
  • \$\begingroup\$ Yes the device's being tested in regards to IEC61000-4-5 surge testing. I'm not sure if it should be done, but the testing laboratory insisted that it's the right way for the certificate. The device is sold standalone without the 12V adapter, I think this is the reason it's being applied. \$\endgroup\$
    – bassumex
    Nov 7, 2019 at 6:49

3 Answers 3

2
\$\begingroup\$

Yes the device's being tested in regards to IEC61000-4-5 surge testing

In that case you need to know the surge voltage series impedance that is appropriate for your device. The appropriate series resistance may range from 2 ohms to over 40 ohms and you need to talk with your lab to find out what is the correct impedance for your device to meet the specification you are trying to achieve.

In the worst case scenario the impedance is 2 ohms and this means that the peak current (with a 500 volt surge) can be as high as 250 amps. But, dependent on the test and your own circuit and how it is intended to be isolated from earth, the actual surge current may only be a handful of amps.

In the actual test however, the TVS diode failed and shorted out, causing the power wires to burn and melt.

It seems to me that this means your circuit isn't isolated from earth and will "receive" the full 250 amps. I see there are nodes called "GND" but it's unclear that these nodes are "earth". So decide what is right and what is appropriate for the test setup and select the appropriate TVS.

Just in case you think there may not be a suitable TVS, rest assured that there will be. For instance, the SM30T is capable of hundreds if not thousands of amps of surge handling and, is specifically rated for the IEC 61000-4-5 surges: -

enter image description here

The above is one that I used on some equipment that needed to be rated for 2 kV IEC 61000-4-5 surges (peak of 1000 amps). However, the current surge was somewhat reduced by cable impedance so, do not neglect this because it will help.

\$\endgroup\$
2
  • \$\begingroup\$ Thank you for your valuable input, I will research the things you've said, but I wanted to clear the fact that board itself worked fine when we switched the TVS diode and connected power. Also, how does one figure out if GND nodes are actual Earth and not just a node? \$\endgroup\$
    – bassumex
    Nov 7, 2019 at 10:21
  • \$\begingroup\$ As the designer, you must know what these nodes mean and whether they connect to ground. I cannot tell you this. \$\endgroup\$
    – Andy aka
    Nov 7, 2019 at 10:27
0
\$\begingroup\$

You should be able to pass that test using TVS diodes. Talking of our experience, with diodes we were able to achieve class 2 (1KV with 2ohms) using SMCJ7.5CA. For higher tests we needed to add a TBU and a GTD, with that we were able to achieve up to class 4 (4KV)

But a very important point besides the chosen diode, is the layout. It is very important in this case to provide an easy way to the current to flow to the system earth.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you for your comment. Currently, upon such a surge, TVS diode is literally right next to the power input so that the surge current has the quickest way to the earth. I will check out the diode you've suggested. I think my problem is that I didn't take into consideration of the test equipment's impedence and therefore the chosen TVS can't handle such a current and fails catastrophically. \$\endgroup\$
    – bassumex
    Nov 7, 2019 at 11:10
-1
\$\begingroup\$

I am not an expert on CE requirements. But I dont think single TVS diode can handle that kinda Energy. Cuz TVS diode reduces its internal resistance to take all power on its hand. But not for 1.2us. I highly advise to try Switching Surge Stopper. like (https://www.analog.com/en/products/ltc4366.html) .I am sure It will handle so much better.

\$\endgroup\$
2
  • 2
    \$\begingroup\$ -1 for "cuz" and the surge is 1.2us not 1.2ms. Edit: also diodes are not resistive devices, they are semiconductors. \$\endgroup\$
    – MIL-SPEC
    Nov 7, 2019 at 9:38
  • \$\begingroup\$ Being semiconductor doesnt mean that they dont have "internal resistance". Diode has dynamic resistance which change after threshold voltage is applied. \$\endgroup\$
    – emre iris
    Nov 7, 2019 at 14:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.