RS485 Deterioration due to Overheating, self-healing

Thread Starter

Shaft123

Joined May 13, 2016
4
Hello!

I am making Conducted Coupling tests on an RS485 communication circuit by applying 10 V over a frequency sweep of 1 to 230 MHz. After some time of running the tests, I see that the communication fails, and I cannot get it to work at all, even without the added disturbances. After a couple of hours of waiting, I can communicate with the circuit again. Can it be due to overheating of the transceivers, and that they self-heal, i.e. cool down, after a couple of hours, or what is the phenomena behind this?

If this is the case, how can I protect my transceivers from this phenomena?
 

shteii01

Joined Feb 19, 2010
4,644
They are definitely not healing.
Like you noted, assuming overheating is the issue, they are simply cooling to their designed operating temperature. Note my choice of words: DESIGNED OPERATING TEMPERATURE. Again, assuming that you are correct and they are overheating, then the procedure that you are employing pushes the devices outside their designed operating conditions which means that they will eventually break. You might say: Well, everything breaks eventually. And you be right. The difference here is that your devices will break much sooner than normal. How soon? I have no idea. I will bring one example that is simple, your devices are likely soldered to pcb, the cycle of high temperature and then cooling may deteriorate solder joints so eventually your device will not have good connection to the traces on the pcb and simply stop receiving power or stop being able to pass the signals along those traces.
 

Thread Starter

Shaft123

Joined May 13, 2016
4
They are definitely not healing.
Like you noted, assuming overheating is the issue, they are simply cooling to their designed operating temperature. Note my choice of words: DESIGNED OPERATING TEMPERATURE. Again, assuming that you are correct and they are overheating, then the procedure that you are employing pushes the devices outside their designed operating conditions which means that they will eventually break. You might say: Well, everything breaks eventually. And you be right. The difference here is that your devices will break much sooner than normal. How soon? I have no idea. I will bring one example that is simple, your devices are likely soldered to pcb, the cycle of high temperature and then cooling may deteriorate solder joints so eventually your device will not have good connection to the traces on the pcb and simply stop receiving power or stop being able to pass the signals along those traces.
Thank you for your reply.

But to clarify, is it possible that the Conducted Coupling Disturbance is the reason for overheating?
 

shteii01

Joined Feb 19, 2010
4,644
Thank you for your reply.

But to clarify, is it possible that the Conducted Coupling Disturbance is the reason for overheating?
Lets start with basics. RS-485 uses 5 volt signals. You are sending 10 volt signals. Right there you are sending more energy though the wires than they designed to handle, so where is that energy going to? Heat.
 

Thread Starter

Shaft123

Joined May 13, 2016
4
Can you give more details about the test? Block diagram, voltages and currents, anything?

ak
I test according to the IEC 61000-4-6 standard. I have a Modbus Master in my computer communicating to my slave circuit with an RS485 interface.

The schematics are as shown in figure 5k following this link: http://www.compliance-club.com/archive/old_archive/011021.htm

What I do is that I apply 10 V on a frequency from 100 kHz to 230 MHz on the communication cables. The communication cables go through the CDN and the voltage is capacitively coupled on to the communication cables.

The RS485-circuit under test is the SN65176B.

Why is it that the circuit stops working and then after a couple of hours of resting starts working again?
 

AnalogKid

Joined Aug 1, 2013
10,987
Why is it that the circuit stops working and then after a couple of hours of resting starts working again?
Almost certainly thermal. Freeze mist might be a bit harsh, but if you do something to cool down the chip quickly and the recovery time shortens, you have confirmation.

ak
 

jsallas

Joined May 17, 2016
9
Hello,

My recommendations:
-Try to lower your voltage from 10V to 5V.
-If you need to perform 10V testing, you should use a device with integrated Failsafe (to ensure proper voltage levels) or add TVS diode to protect the bus from transients. At that testing frecuencies the wire impedance might be inducing voltage transients (v=L(di/dt)) in your bus resistors.

As an extra recommendation, if your network needs to be robust, you should consider to use galvanic isolated RS485 transceivers, like the ADM2587E.
 

Thread Starter

Shaft123

Joined May 13, 2016
4
Hi!

Thank you all for your kind responses. What I will conclude then is that it is possible to deteriorate the transceivers by conducting CDN (Coupling Decoupling Network) tests due to the induced RF power.

Thank you for the recommendations, I tried with TVS diodes and a small filter, the disturbing frequencies changed to other values but were in total less compared without protection.
 
Top