60601-2-10 clause 201.12.4.103 Output Indicator (itself in SFC?)

[Loekje]

Hi Forum,

The clause states that the device must indicate with a yellow lamp when it can produce more than 10mA, 10V or 10mJ over 1k, in normal and single fault condition.

Does this mean the designer has to either use a LED with high integrity characteristics (plus the rest of the steering electronics) or double the components that can have the light itself fail?
Or is it enough to have the circuit go though a Power On Self Test in which the operator must verify that this LED indicates during the appropriate self test?

As I never encountered this indicator doubled my guess is either the latter or this clause is not taken too strict?

Kind regards,
Loek

 

[Peter Selvey]

My guess is that the authors of the standard were thinking of faults resulting in unexpected output rather than faults in the indicator. However, strictly speaking faults in the indicator should also be considered.

In general an indicator would be unstressed modern electronics which has a very low failure rate, the fault in the indicator would have to be combined with either a user error (turned the output on unintentionally) or a second electrical fault, and the unplanned output would then need to flow through dangerous path (e.g. through the heart) and even then there is a low probability of a serious outcome (e.g. heart actually stops). So it may be that overall the risk is acceptable without any risk control measures. At the same time, fairly low cost solutions exist (e.g. have the MPU monitor the current through the LED, use two LEDs) so it's not crazy to use a risk control either. It's a grey zone item!

 

[Loekje]

Thanks for your reply Peter,

Gray zones are the place to go for asking agreement up front with your test house: "Do you agree with our solution on how to implement this clause?".
(and of course you start off with the implementation that suits you best)

 

[Peter Selvey]

Ignoring the test house angle, this is also probably a good place to test the "risk minimum theory".

This theory places less emphasis on the absolute value of risk (which is near impossible to measure or establish criteria) and more on the impact of resources, and in particular if the resources is small then just do it, don't argue the point.

So for example, instead of asking if a 13 meter high tower is acceptable risk for a backup generator for the Fukushima nuclear power plant, we should first ask the question "at what height does the risk become negligible?". If, for example, 15m (yes, just an extra 2m) makes it a once in a million year event, then why waste time on endless discussions about whether 13m tower estimated to be ok for everything except 1 in a 1000 year event is OK, whether the benefit to society justifies the risk, blah blah blah. Just build a 15m tower and be done with it.

So, coming back down to earth and looking at the original case study, the cost (resources) of the electronics to monitor the LED current is so small, it is really not worth the discussion. Just do it.

Unfortunately, our preponderance towards estimating risk and comparing against acceptable criteria (ala ISO 14971) encourages avoidance of low cost solutions that virtually eliminates risk. It will almost certainly be a case study in the future for how human beings can be silly. For decades.