Hello all,

On an IOT device which I developed, I am using the RAC03-24SK to convert the 230AC to 24VDC.

Everything was working fine until I sent the device for ESD testing.

During the first tests for air discharge on 8kV, the AC/DC module fails, the input becomes short circuit and blows the fuse. The damage is permanent and replacing the fuse makes no sense.

How is this possible when the module's datasheet says that it is certified for those ESD voltage levels?



Its not the first time I develop a device that is going to be tested for ESD but its the first time that a top-brand module fails.

 

Everything was working fine until I sent the device for ESD testing.

This is where I would be sensing the failed parts back to the manufacturer with a copy of the testing lab reports. The testing lab should be able to provide the test method, procedure used and exactly when failure happened.

How is this possible when the module's datasheet says that it is certified for those ESD voltage levels?

Only the manufacturer can explain why the part failed.

Ron

 

I had similar on something else
when it turned out the test lab had done a contact test at the air gap test voltage ,
as @Reloadron says,
check with test house and talk to the manufacturer direct

 

The amount of energy is usually specified one way or anonther.

A Metal Oxide Varistor in the right may solve the problem. The classic problem is whether you can finish the testing before you exhaust all available modules or patience.

 

@DickCappels I have both TVS and Varistor on the input. Also there's extra protection on the PSU's output

 

It might turn out that you need a varistor or two on the output. ESD can be sneaky.

 

View attachment 293345

@DickCappels I have both TVS and Varistor on the input. Also there's extra protection on the PSU's output

the 250 V TVS on the input, is that correctly rated for UK mains ?
Its still unusual that the recom fails , they are IMHO quiet robust normaly .
do you need F2 ,is that not covered by F1 ?

Any chance of a picture of the input circuit of the board,
as @DickCappels says, ESD is a tricky one at jumping where you dont expect it to .

Id also double check the test house did the required test, not one it thought was good,
I have also had test house put the earth / gnd onto the -ve input, not onto the boards ground, which then floated.
which caused fun,

Have a chat with Recom where you are, in my experience in the UK, the representative is very interested in any failures to find the cause

 

@drjohsmith The TVS is rated for Greece, where we have 230VAC. The device's 24VDC is floating and its not connected to the Earth, I will contact the lab about their connections during the test.

The board:


The board fails when they apply air discharge near the screw terminals. The signals there are also protected with TVS.

 

@drjohsmith The TVS is rated for Greece, where we have 230VAC. The device's 24VDC is floating and its not connected to the Earth, I will contact the lab about their connections during the test.

The board:
View attachment 293461

The board fails when they apply air discharge near the screw terminals. The signals there are also protected with TVS.

Im guessing those terminals are on the output of the DCDC.

DCDC, the ratings are for the inputs,
the assumption is that the DCDC is inside a box so its outputs don't need protection .

these terminals are going to need there own full protection
if the dcdc output is not referenced to a ground, where is any protection you have going to discharge to ?

If I understand your circuit correct, this will apply 8Kv across pin 2 / 3 of the DCDC output, which its not rated for.

 

In the above picture, the enclousure is open, the upper part is missing.

Terminals:

1. The left connector, is the AC's input
2. The other connectors are inputs to the digital logic, there are no outputs at all.

Nothing else than the Recom module is damaged. The board has also a 24DC->5VDC DC/DC converter circuit and one LDO, nothing is damaged at all.

Do you are mean that the strike jumps to the circuits directly powered from module's output?

If I understand your circuit correct, this will apply 8Kv across pin 2 / 3 of the DCDC output, which its not rated for.

DCDC, the ratings are for the inputs,
the assumption is that the DCDC is inside a box so its outputs don't need protection .

This is an interesting case, I guess only Recom can answer it. I thought that both input/output of the module are rated for the ESD levels that are specified on the datasheet.

 

In the above picture, the enclousure is open, the upper part is missing.

The left connector, is the AC's input. The other connectors are inputs to the digital logic, there are no outputs at all.

Nothing else than the RECOM module is damaged. The board has also a 24DC->5VDC DC/DC converter circuit and one LDO, nothing is damaged at all.

Do you are mean that the strike jumps to the circuits directly powered from module's output?

can you share the circuit that connects to the connectors that caused the problem please.

If these are un protected digitla inputs, and they had 8Kv zapped on them,
you have probably blown the digital inputs.

also , the esd, will then have gone up the inherent diodes in "all" digital inputs, onto the internal power supply
and jumped who knows where !

have a good read through these sites
https://learnemc.com/emc-design-guidelines

https://www.ti.com/lit/sszb130

Bottom line,
8Kv, is one heck of a spike.
it can jump cross many mm of gap,
it will take the easiest path to ground,
if it can't make ground, then your design is a capacitor, and will charge up to 8Kv,till you touch it.
DCDC outputs are not designed to have 8Kv across them
and LDO will probably survive, as its reverse circuit internaly is a farliy big diode, so if you raise its output to 8Kv,its input will also rise to 8Kv
If the recom was plugged in, then its input is 230 v, its output would now be 8Kv, its not rated for 8kv on its output

 

The inputs and everything going on terminal is protected with TVS diodes. The TVS are placed right next to each terminal's pad and directly to the GND (GND = Recoms #2 terminal).

I got your point but from datasheet's point of view, I was sure that output was also protected :/

 

The inputs and everything going on terminal is protected with TVS diodes. The TVS are placed right next to each terminal's pad and directly to the GND (GND = Recoms #2 terminal)

ESD protection works , by attenuating with resistance / inductors in series , and with "shorts" to ground .
the ESD pulse has to have somewhere to go

I'd like to see your output circuit, but from what you have written Im guessing the output of the recom is floating, and the TVS go between the signals and the recom pin 2 output,

draw a circuit of the route the esd spike must take to disipate.

Id say, when the 8Kv esd pulse hits your digital input, the TVS "fire" and pump the 8kv direct onto the output pin 2 of the DCDC.
QED, your outside the spec of the recom

You need to have a safety earth , connected to the units earth stud / input.
You then connect as a minimum your ESD protection between the input pin and the safety ground, at the connector,
then when esd hits, the 8kv is shorted to the real ground and dissipated.

depending upon the circuit and specification your meeting, you might need more protection, such as serial resistance / inductance, or even optical inputs,

One other thing, ESD / EMC IMHO,does not work if retrofitted to a design, so it could be hard to meet the 8kv requirment,
e..g. if you add a separate safety earth wire and extra tvs to this earth on the inputs, the inductance of this earth wire could be such that its a open circuit to the very high speed edge of the ESD pulse.

if you could share a drawing of your digital input circuit we might be able to advise

 

I'd like to see your output circuit, but from what you have written Im guessing the output of the recom is floating, and the TVS go between the signals and the recom pin 2 output,

Exactly.

Its the first I have to design a device with floating GND and its a bit tricky.

 

Exactly.

Its the first I have to design a device with floating GND and its a bit tricky.

thats why you have your "gnd" and the safety ground,

the two are different,
safety ground is where all the ESD gets dumped.

if your medical, then you need very low leakage tranzorbs or such like
but if you can share your input circuit , we might be able to help.

 

@drjohsmith Do you think by connecting (tie) Gnd to the Earth might proof if that's the case? Just for some first tests.

 

@drjohsmith Do you think by connecting (tie) Gnd to the Earth might proof if that's the case? Just for some first tests.

re my above comment,
ESD and EMC are very hard to retro fit,
if you tie the pin 2 out of the recom to the input ground,
your losing your isolation,

sorry ,back to initial,
tell us what your digital inputs are,
and we can try to help.
else were guessing and probably giving you duff advise.

what is the specification for the digital inputs,
do they have to pass what specification ?
are you medical , only in Uk , only Europe, world ?
do you want CE marks
are you after VDE ,

if any of these are strange to you, then I'd advise that you employ a specialist,
its a good few thousand euros for each test house visit, it soon adds up

 

Its only for EU and want the CE mark. The device is nothing special than an IoT device for indoor use.

About the digital inputs, for example, the right connector (1x3) is being used to connect a dallas temperature device.

The pins are GND, 1-Wire, VDD (where VDD = 3V3 from the LDO). Same concept for the other terminals too.

I have seen several other similar devices with also floating GND, with CE mark. I guess they also passed the same tests.

During the design, I followed some rules like it is an battery powered device (floating).