cancel
Showing results for 
Search instead for 
Did you mean: 

STKNX circuit, high power loss at single fault condition: CFH (capacitor) shorted.

SBart.5
Associate II

We based our circuit on the "STKNX Reference Design" (V1.0, 14. Nov. 2019)).

In the case of the single fault condition of a shorted CFH capacitor during operation almost the whole KNX bus voltage (21-30V) is permanently over the 68R resistor (RTX) and generates a big power loss of over 5-12 W (depends on the current KNX voltage).

In the case of  a single fault condition at 30 V KNX bus voltage, the overload of RTX leads to an broken RTX with high impedance, so the overload ends automatically --> no danger.

However in the case of a single fault condition at 21 V KNX bus voltage (where the power loss is lower) the lower overload do not lead to an high impedance RTX (maybe the impedance increase a bit, but not enough). So that a power dissipation of 4-5 W are permanentely available, wich heats our case illegally.

For RTX we are using the following type: YAGEO RC2512JK-0768RL

 

Can you help us to fix this problem?

 

5 REPLIES 5
AScha.3
Chief II

>single fault condition of a shorted CFH capacitor

1. This should never happen ! So choose a CFH cap with min. 50V rating, it will get no short. Problem away.

2. if you want the RTX be "safe" , choose a PTC (or R + PTC) for this; the PTC has to "switch" (get hi resistamce) in case of permanent too hi current here. But no."1" is better solution.

If you feel a post has answered your question, please click "Accept as Solution".

Hi AScha.3,

thank you for your quick response.

To 1):

We are using an MLCC, 50 V type. But for these kind of capacitors, as far as i know, a short circuit is a common failure mode ("Low Insulation resistance (IR) or short, along with low or no capacitance are the two most common failure modes for MLCCs.", Source: https://www.kemet.com/en/us/technical-resources/failure-analysis-of-capacitors-and-inductors.html).

So in my mind a shorted CFH cap. is a possible single fault which i have to consider.

 

To 2):

To put some PTC in this reference design circuit for KNX communication may lead to other problems i don't know, because of the different behavoirs of the resistance on different temperatures and communication load conditions.

 

My suggestion

For the shorted CFH, my suggestion is to put 2 capacitors in series with a total capacity of CFH according to data sheet. So if one capacitor breaks (shorted), the other capacitor is still there and prevents the problem described above.

 

Further error case:

The problem of permanentely overload of RTX exists also if i consider a single fault at the STKNX (for example a short circuit between pin R68  and ground).
(My opinion regarding to singel fault condition consideration and ICs (integrated circuits): In the single fault consideration theoretically i have to short every pin with every pin. And Pin R68 shorted to ground is problematically in my mind). 

 

Please correct my opinion if am wrong.

 

 

 

1. If you so scared about a mlcc /short , (happens almost only on mechanical stress, bending the board) then dont use it: take film cap , through hole, this cannot short by itself.

MKS2-63 47N MKS2 PET-Kondensator, 47 nF, 5 %, 63 VDC, RM 5

 

2. What are the allowed tolerances here ? (To calculate /select a matching PTC , we have to know.)

If you feel a post has answered your question, please click "Accept as Solution".
SBart.5
Associate II

to 2): What are the allowed tolerances here? (To calculate /select a matching PTC , we have to know.)

We use the given circuits according to the "STKNX Reference Design" (V1.0, 14. Nov. 2019)).

With RTX = 68 Ohm  (>= 1 W power dissipation, 5% tolerance).

Our max. environment temperature in the case is +85 °C

Our minimal environemnt temperature is -20 °C

In the case of a single fault its ok, if the environment temperature increase to +115 °C.

 

I don't know the max tolerated deviation of the RTX for proper functionality (there is no information in the STKNX datasheet or STKNX Reference Design).

 

 

 

 

>2> I don't know the max tolerated deviation of the RTX for proper functionality.

Me too. 🙂   So we cannot design a proper replacement for RTX . It is 68 r.

Maybe you could use a "fuse" resistor...(i dont know, whether you want+ can get this kind of resistor.)

660-TPR1CTA680J (mouser)  1 W type :

AScha3_0-1720447694144.png

 

 

If you feel a post has answered your question, please click "Accept as Solution".