TLK105 temperature problem (self heating after voltage transient burst)

Hello Mattias,

Does the TLK105 get damaged by the transient or does it survive the transient burst but require a reset or power cycle?

When the transient is applied, is it directly applied to the VCC pins of the TLK105 or does it come through system voltage regulators?

Best Regards,

Hi Rob.

The TLK105 does not get damaged as far as I can see, but it needs a Power cycle to function properly again. Maybe it could get damaged if we don´t manually turn the system off when temperature rises, but I have not tested that, since we have limited amount of hardware to test on.

The transient burst is applied to the 230V AC Power lines.

Then we have an AC/DC converter from 230VAC to +12VDC and then a Step Down converter from +12VDC to +3.3VDC.

So the TLK105 only receives the +3.3VDC voltage, but the transient seems to find it´s way from 230VAC all the way to the TLK105.

/Mattias

Hi Mattias,

If you have to bring down the power rail to recover the device, you are likely entering a latch-up scenario. This can happen to any IC if a transient exceeds the maximum voltage rating of the pin. You then sink a large amount of current which may lead to heating of the device.

I'd recommend looking at the transient on the 3.3V rail to determine what voltage level you are experiencing when the phenomenon occurs.

It can also be helpful to increase bypass capacitors on the affected rail when you have behavior like this.

The preferred solution would be better isolation against transients entering the device.

Best Regards,

Hi Rob!

I have spent the day doing additional testing and will try to explain what I have found out.

In the first test I modified the board with a lot of additional bypass capacitors, but I got the same result as earlier.

In the second test I disconnected the unshielded Ethernet Cable so that there was no Connection between my device under test and the external Ethernet switch. When there is no Connection to the Ethernet switch I am not able to recreate the problem. This makes me Think that there is no transient on the +3.3VDC, if there was I would have been able to recreate the problem with this setup as well.

In the third test I used a shielded Ethernet Cable between my device and the external Ethernet switch. When using the shielded Ethernet Cable I am not able to recreate the problem.

So what I am thinking is that when I apply the voltage transient burst to my device the disturbance is directed to the Chassis GND and lifts the entire system for the duration of the burst. When I have no Connection externally it is ok. When I use a shielded Ethernet Cable I get the Ethernet switch to have the same chassi potential as my device. When I use the UTP Cable the Ethernet switch will have a different chassi potential than my device for the duration of the burst and that would mean that the transient would find it´s way into the device through the differential Ethernet pairs, causing the latchup.

So this is what I am thinking, using the STP Cable would be an easy solution but unfortunately the customer requires UTP Cable so I must find a way to protect the PHY, do you have any further ideas that could help me?

Best Regards!

Mattias

Hi again.

Maybe the problem I have seen is a non-relevant problem.

We have to fulfil the standard Railway applications EN 50121-4:2015, which specifies in the immunity part of an AC supply that the transient must handle a burst of +/- 2kV.

However the test instrument we used can test 500V, 1kV, 2kV and 4kV and I realized that the temperature problem only seems to occur for burst of +/-4kV.

Just realized that testing at +/-4kV maybe is expected to fail since the rating of the Ethernet magnetics within the connector is 2.25kV/1.5kVrms. So starting to Think that performing theese tests is a bad idea....

I will continue the testing with burst set to max +/- 2kV in order to verify that I can´t recreate the problem at this level.

Best Regards!

Mattias

I have not been able to recreate the problem today with burst levels set to max +/- 2kV.