Paralleling R for increased reliability in HV. Opinions welcome

[johm]

Hello! Came across this dilemma while designing an AC 240 V 50 Hz mains filter. When this circuit is implemented on a PCB with standard parameters, which option do you believe is most reliable in terms of:

• Parameter variability
• Thermal dissipation
• Routing complexity
• Breakdown voltage
• Failure modes of specific resistances (short circuit, open circuit, drift)

Resistors are  1W 2512 SMD.

It is basically a thought experiment as the difference is apparently minimal, but IMHO worth exploring.

Cheers

[moffy]

There is no effective difference between the two as you suspect.

[CatalinaWOW]

I disagree.  In failure modes there are significant differences.  For a simple example look at a fail open of R2.  In the left hand case this results in a resistance across the line of 150 ohms, with 2/3 of the line voltage across R1.  Power dissipation in R1 nearly doubles.  In the right hand case the resistance across the line is 200 ohms, with the voltage across R1 remaining the same at 1/2 the line voltage and no change in R1 power dissipation.  There are similar differences of other failures.

Whether this makes a difference depends on the components chosen and the goals for the circuit. In general I would choose the right hand configuration.

Power dissipation in the no faults case is only marginally changed, with the change all due to the actual layout and copper used for routing.  It may go up or down.

[Kleinstein]

In this application the current is mainly set by the capacitor, not the resistors. So for the current and power to the remaining resistor there is essentially no difference. Usually the resistors are way more reliable than the capacitor. One may want a fusible resistor and not an oversized resistor, though this would be required only with a non X rated cap.

The more failure prone part is the capacitor anyway.

[CatalinaWOW]

Kleinstein is right.  I was over hasty, only looking at the changes in your proposals, not the overall circuit.  In that light it seems that substituting four resistors for one is questionable since the voltage drop across the resistor(s) is modest and there isn't any other big change that occurs on failure of a single resistor.  Hence Kleinsteins suggestion to use a fuseable resistor to cover the case of the capacitor failing short.  The multi resistor approach can only help reliability if for some obscure reason the corner of your filter is very critical.