Components in parallel to creepage\clearance

rothlis

Involved In Discussions
The fifth paragraph of IEC 60601-1/A1:2012 clause 8.5.1.3 specifies the requirements for each means of protection (MOP). The third bullet point says "components that are connected in parallel with an insulation, with an AIR CLEARANCE or with a CREEPAGE DISTANCE comply with 4.8 and 8.10.1".

Compliance with clause 4.8 means that it is used within its rated levels and, if there are relevant standards for the component, then the component should show compliance, or if there are specific clauses in the standard for that component (e.g., transformers per 15.5.3) then those clauses need to be met. In all other cases the standard says that "any other applicable source (e.g. standards for other types of devices, national standards) could be used to demonstrate compliance with the RISK MANAGEMENT PROCESS".

For this latter case, I am interpreting the standard to allow components to participate in the MOP so long as they are acting as a MOP within their rated levels and are securely fixed.

For example, a transistor on a PCB can be part of the MOP for a 5V, 1A path if:
  • The pads and pins for the relevant nodes are spaced to meet the clearance requirement (e.g., 0.8mm, using air clearance per clause 8.9.1.3).
  • The transistor is rated to at least 5V and 1A.
  • The transistor is adequately fixed (e.g., soldered to PCB).
  • The resulting leakage current falls below the limits.

Is this a correct interpretation? If not, what is the proper response to the conjunction of the third bullet in 8.5.1.3 and clause 4.8 when there aren't any specific standards or clauses addressing a type of component?
 
Last edited:
M

MedMartin

Hello,

I think your interpretation is correct.

But the example is maybe not the best for several reasons:
- I would normally not consider a low voltage transistor as a MOP. This depends of course on the application, it may be critical in an ECG front end. But you start with citing clause 8.5.1.3 which is for MOOP (operator protection) thus the transistor is not for patient protection. Requirements for MOOP start from 42,4 V peak AC or 60 V DC.
- Furthermore the creepage and clearance distances are really for safety critical pathways, galvanic isolation distances, fuses and similar paths.
Todays components on the board like transistors, processors, resistors, ... are so small that they often do not comply to the creepage / clearance requirements of the 60601-1. Therefore for isolation distances on secondary side circuits one could use 60664-1 with much smaller distances.
- A part rated for 5V can in my opinion not be used to isolate a voltage of 5V. Just look at table 6 with test voltages. For MOOP the test voltage is 1 kV for a working voltage of 60 V DC. That's 60601 :)

This is just my opinion. Does this help?

Best regards,
Martin
 

Pads38

Moderator
I would not consider a transistor as being suitable for connection across a MOP.

Remember that failure of any one component is considered to be a 'Single Fault Condition'. For a transistor that failure mode could be short circuit or open circuit.

Your device has to remain safe in SFC and the values for leakage currents must remain below the limits of tables 3 and 4 with the transistor short circuited.
 

rothlis

Involved In Discussions
Thanks MedMartin. I would like to test the notion that the example may not apply. In particular, note that:
  1. Clause 8.5.1.2 (MOPP) refers to the tests of 8.5.1.3 (MOOP) for compliance. It seems pretty clear to me that the requirements under consideration are applicable to MOPP.
  2. This case is most likely only being relied upon as a MOPP for the patient auxiliary current limit. Note that the creepage / clearance Table 12 for MOPP starts with an entry for "up to 17Vdc". The distance there is readily met by even relatively compact FETs if the source and drain are on opposite sides of the chip. It seems perfectly applicable.
  3. Table 6 is clearly intended for solid insulation and is identified separately from components as a possible form of protection. If the dielectric test is meant to apply to the component itself, what is the purpose of that separate 3rd bullet point for components?

What do you think?
 

rothlis

Involved In Discussions
I would not consider a transistor as being suitable for connection across a MOP.

Remember that failure of any one component is considered to be a 'Single Fault Condition'. For a transistor that failure mode could be short circuit or open circuit.

Your device has to remain safe in SFC and the values for leakage currents must remain below the limits of tables 3 and 4 with the transistor short circuited.
This is not intended to void the 2 MOP requirement. We can consider the case where the example is expanded to have two independently controlled FETs in series.
 

Peter Selvey

Leader
Super Moderator
The problem here is that 5V is not really that dangerous for a number of reasons. Even the distance of 0.8mm (using 8.9.1.3) is overkill - these numbers have been derived based on an assumption of 330V transients that could exist between a secondary circuit and earth ... and even that could only occur under very special conditions.

Also, while the potential of harm from 5V exists, both the probability and severity is far lower than for mains voltages. There are a number events (probability factors) in the way. For example, you need to bypass skin impedance and also you need a fairly large surface area to get any significant current. There's no practical risk of fibrillation.

It is important to consider patient auxiliary currents where electrodes are used (ECG, simulators, some probes). There have been burns reported in the past from 5V rails being shorted to the patient with an ECG. But transients don't exist so the whole dielectric strength, spacings (creepage and clearance) makes little sense. I don't apply these limits and it would be impossible to do (every ECG on the planet would fail if followed). I do however perform a fault analysis and follow the 2 MOP principle. In this case, I consider any component being used within specification as 1 MOP (resistors, diodes, op-amps). Then I inspect the circuit (and test) to make sure the current stays below limits in fault condition (only one component fault at a time).

So in that context treating a transistor as 1 MOP is OK. But, as long as it is recognised as a workaround for low voltages, and not extended to the serious stuff such as mains or higher voltage situations.
 

rothlis

Involved In Discussions
Thanks Peter. I find everything you say here to be reasonable and in line with the interpretation I offered, though it seems you would even go a step further than I proposed. When you say "I don't apply these limits" (speaking of dielectric and spacing), are you saying that you consider that using a component within its rated values is sufficient on its own, so long as the current limits are still met during SFC? Given that the standard requires that each means of protection be either insulation, spacing or PE (whether components are in parallel or not), would you then document this as an alternate means of compliance under clause 4.5, citing the risk analysis associated those components?

By the way, this question was inspired by the experience during review of a stimulator that uses electrodes, so you're spot on regarding the applicable cases.
 

Peter Selvey

Leader
Super Moderator
For me, I would consider this a case of "blatantly obvious" mistake in the standard, just overrule and move on.

I used to think these cases needed to be written up and justified e.g under 4.5, but these days I use common sense and move on. Again, this is reserved for "blatantly obvious" cases, not nuanced or special cases. The older ECG standards, for example, contained so many outright mistakes that it just a bit overwhelming and the people upstream (reviewers, auditors, accreditation) don't always have the expertise to understand why it was "blatantly" obvious, so the less said the better.

And theoretically, you can't use 4.5 because it uses the requirement in the standard as a point of reference (the alternate solution has to be "comparable"), so if the requirement is illogical, you are back to square one.

In this particular case, it's clear that the principle of 8.1 a) 3rd and 4th dashes cannot be applied for auxiliary currents. You might have 15 components between a +12V volt rail and the patient, but none of them could withstand a 500Vdc test, so they all need to be shorted. Do that, and auxiliary current limits will definitely be exceeded. So it falls into the category of "blatantly obvious". I never had trouble to apply this interpretation and have used it many times.
 

rothlis

Involved In Discussions
theoretically, you can't use 4.5 because it uses the requirement in the standard as a point of reference (the alternate solution has to be "comparable"), so if the requirement is illogical, you are back to square one.
As I read it, clause 4.5 says only that the residual risk needs to be comparable. That seems easily achievable in this case since a design which strictly complies with the standard and one which simply uses the component as intended are going to have effectively identical risk profiles.
 

Peter Selvey

Leader
Super Moderator
Of course it's up to the individual engineer how you want to write it up.

My worry would be that a reviewer would see it this way:

- standard says 500Vdc dielectric is needed for every part (1 MOP)
- manufacturer claims a 25V transistor rating is equivalent (1 MOP)

Clearly, the probability of failure is vastly different between these two solutions, perhaps the 25V rating is 1000 times more likely to fail than a part that can handle 500V. They are not comparable.

So why is 25V rating acceptable? The answer comes from considering the overall probability and severity of harm. In that view, both solutions result in negligible risk. The 500V option is a orders of magnitude better, but both solutions have tiny risk that we can call it negligible, and hence comparable.

But it seem unfair to force manufacturers to do this kind of analysis every time there is an obvious mistake in the standard. Keep in mind 4.5 says "scientific data" and "comparative studies". It's not walk in the park. And there is a risk that upstream reviewers might not understand the negligible concept or the rationale why the standard is wrong.
 
Top Bottom