Creepage requirement Wearable ECG

CuriousMonkey

Registered
Good morning, All!
I'm designing an internal battery powered (3.7V DC Max 5V DC) wearable device.
Which has ECG as Patient connection and nasal canula connector (pressure sensor) as an applied part.
I have following queries which needs your expert scrutiny;

Referring to attached image;
Creepage requirement Wearable ECG


  1. Does the protection required by R9 and R10 combined = 1MOPP considering Internal battery (type BF) provides another 1MOPP?
  2. Creepage and clearance required between Point B and C = 1.7mm and 0.8mm respectively (as per table 12 for 1MOPP)?
  3. wrt point 2. the minimum footprint required by R10 to be 1206 (3216M) to give 1.7mm creepage?
  4. Does it required to have 2 separate Resistors (R9 and R10) to make the design single fault safe or declare R10 as critical component and use only R10 but not R9.
  5. if R10 is declared as critical component what kind of certification or specification it should have?
  6. Same Creepage/clearance is requirement between point A and D as B and C?

    Warm regards,
    CM
 
Last edited:

Loekje

Involved In Discussions
Hi CM,

As you cannot detect a failure of either resistor R9 or R10 you must assume they are failing. Failing of the 10M lead-off detection resistors you can detect (BTW: they must have proper creepage and clearance also).
We had a long discussion with a test house: "but these resistors never fail short circuit as they only have the OpAmp Bias/Offset current and lead-off detection current to carry" Answer: "Yes, but what IF they fail?" ... Sigh... FWIW: we found a Panasonic document in which 10^14th resistors where evaluated after their service life (within current and temperature limits) and literally zero failed. But that was not accepted as scientific proof.

So it is not the point if you would declare a resistor as a critical component, you must have resistors of the type (60601-1 3.17) "COMPONENT WITH HIGH-INTEGRITY CHARACTERISTICS", as these are assumed never to fail.
You cannot just buy these off the shelf, but if you buy a single batch of simple resistors and have the batch tested at a test house under your operating conditions and then resistors of this batch may be declared as such components. Buy another batch and you must have this batch tested again.
In this case you can do with a single current limiting resistor, but then you have the problem to implement 2MOPPs creepage/clearance. You end up with parts that have 3.4mm creepage, like through hole parts.

Another option the test house approved is to use 3 resistors in series: one is failing without detection so this one does not count, a second one is failing which counts as a single fault condition and the third one is to restrict the current in single fault condition to applicable limits.
Depending on the willingness of your test house you may argue that to have the resistors function as current limiters for your supply voltage the input amplifier must fail first which might be detectable, in that case you can do with two resistors. For your DRL you definitely need 3.

Hope this helps you analyse your design,
Loek
 

CuriousMonkey

Registered
Hi CM,

As you cannot detect a failure of either resistor R9 or R10 you must assume they are failing. Failing of the 10M lead-off detection resistors you can detect (BTW: they must have proper creepage and clearance also).
We had a long discussion with a test house: "but these resistors never fail short circuit as they only have the OpAmp Bias/Offset current and lead-off detection current to carry" Answer: "Yes, but what IF they fail?" ... Sigh... FWIW: we found a Panasonic document in which 10^14th resistors where evaluated after their service life (within current and temperature limits) and literally zero failed. But that was not accepted as scientific proof.

So it is not the point if you would declare a resistor as a critical component, you must have resistors of the type (60601-1 3.17) "COMPONENT WITH HIGH-INTEGRITY CHARACTERISTICS", as these are assumed never to fail.
You cannot just buy these off the shelf, but if you buy a single batch of simple resistors and have the batch tested at a test house under your operating conditions and then resistors of this batch may be declared as such components. Buy another batch and you must have this batch tested again.
In this case you can do with a single current limiting resistor, but then you have the problem to implement 2MOPPs creepage/clearance. You end up with parts that have 3.4mm creepage, like through hole parts.

Another option the test house approved is to use 3 resistors in series: one is failing without detection so this one does not count, a second one is failing which counts as a single fault condition and the third one is to restrict the current in single fault condition to applicable limits.
Depending on the willingness of your test house you may argue that to have the resistors function as current limiters for your supply voltage the input amplifier must fail first which might be detectable, in that case you can do with two resistors. For your DRL you definitely need 3.

Hope this helps you analyse your design,
Loek

Hi Loek,
First of all, thank you so much for taking the time to reply to this post.
Your reply clears many doubts, at the same time made me curious about few more points.

  1. Just to clarify these 3 resistors in series need to have at least 1MOPP (with Creepage and clearance of 1.7mm and 0.8mm respectively)?
  2. These resistors are used as "protective impedance" as well as current limiting. Is there any other components can be used? e.g Analog switch (CMOS Bilateral Switch) for "protective impedance" and a resistor in series for current limit?
  3. Just thinking about this because none of the wearable device (Alivecore, Apple Watch..etc) with FDA approved ECG has resistors with 1MOPP or 2MOPP creepage but they do have CMOS Bilateral Switch for some reason! Is there something obvious that I'm missing?
  4. Would "conformal coating of assembled PCB" reduce the creepage requirement = clearance? which may allow me to use smaller footprint of resistors?
 

Loekje

Involved In Discussions
Hi CM,

1: yes. If you assume one fails short circuit that MOPP is also gone.
2: Problem with any active component used for protective impedance is that unless you have very thorough scientific proof that provides otherwise you must assume it fails short circuit just like the resistor but then also directly to power supply (and then you need a way to detect the correct functioning, probably with an active component that again you can detect is is correctly functioning?).
3: I have never seen a wearable to date that is up to the standards. That said, at our firm the bottom line is: "Are you willing to apply it to your daughter (.. or prefer to use it on your mother-in-law)". Usually these device ARE safe, only need to have luck with a test house that applies the standards to the spirit instead of to the letter. The 60601-1 sucks especially on table 12 and the utterly useless notion of "resistors bridging a MOP".
In theory a manufacturer of wearables may have something implemented like a nifty single fault safe leackage current detection circuit that switches off patient connections before tripping the limits over the MEASURING DEVICE. But I can bet some good bottles of anything that in practice they are essentially non-compliant.
4: Yes! But beware: you have to proof that your coating process never fails. Just using a pot of coating and a brush usually is not up to standards, but Parylene coating like the firm SCS does is a certified process. We have tried embedded resistors within PCB layers a few years ago but then that process was not (yet) certified by the PCB manufacturer.

Good luck with your design efforts.
Loek
 

CuriousMonkey

Registered
Hi CM,

1: yes. If you assume one fails short circuit that MOPP is also gone.
2: Problem with any active component used for protective impedance is that unless you have very thorough scientific proof that provides otherwise you must assume it fails short circuit just like the resistor but then also directly to power supply (and then you need a way to detect the correct functioning, probably with an active component that again you can detect is is correctly functioning?).
3: I have never seen a wearable to date that is up to the standards. That said, at our firm the bottom line is: "Are you willing to apply it to your daughter (.. or prefer to use it on your mother-in-law)". Usually these device ARE safe, only need to have luck with a test house that applies the standards to the spirit instead of to the letter. The 60601-1 sucks especially on table 12 and the utterly useless notion of "resistors bridging a MOP".
In theory a manufacturer of wearables may have something implemented like a nifty single fault safe leackage current detection circuit that switches off patient connections before tripping the limits over the MEASURING DEVICE. But I can bet some good bottles of anything that in practice they are essentially non-compliant.
4: Yes! But beware: you have to proof that your coating process never fails. Just using a pot of coating and a brush usually is not up to standards, but Parylene coating like the firm SCS does is a certified process. We have tried embedded resistors within PCB layers a few years ago but then that process was not (yet) certified by the PCB manufacturer.

Good luck with your design efforts.
Loek

Hi Loek,
Once again, thank you for detailed response!
I like the analogy of Daughter vs mother-in-law :)

Last query;
  • What are the required creepage/clearance between two ECG leads (part of same patient connection)? In the image its point A and D.
  • Does the TVS diode falls under creepage/clearance requirement as well as Single fault?
 

Loekje

Involved In Discussions
Last answer ;-)
  • Patient connections of the same applied part do not have separation requirements, and all your ECG connections are the same applied part. If your ECG patient leads have connectors on the patient side, like to snap electrodes, then beware to comply to clause 8.5.2.3.
  • Your TVS diodes are to ground, in general there is no need to have creepage or clearance to ground unless the rest of your schematic shows that there may run an unacceptable current from patient connections to that ground. What happens in SFC when a TVS fails short-circuit is that it is detectable (signal gone) with no patient hazards, but when a TVS fails open circuit it is not detectable and your input resistors must be able to handle ESD pulses (at least up to a voltage where the TVS will arc). Usually resistors are not specified to these high voltage pulses, and then you might get into problems.
    A good engineer can solve all these possible problems, but each design has its own preferred solutions.
If your wearable goes commercial, please let me know :)
Regards,
Loek
 

CuriousMonkey

Registered
Last answer ;-)
  • Patient connections of the same applied part do not have separation requirements, and all your ECG connections are the same applied part. If your ECG patient leads have connectors on the patient side, like to snap electrodes, then beware to comply to clause 8.5.2.3.
  • Your TVS diodes are to ground, in general there is no need to have creepage or clearance to ground unless the rest of your schematic shows that there may run an unacceptable current from patient connections to that ground. What happens in SFC when a TVS fails short-circuit is that it is detectable (signal gone) with no patient hazards, but when a TVS fails open circuit it is not detectable and your input resistors must be able to handle ESD pulses (at least up to a voltage where the TVS will arc). Usually resistors are not specified to these high voltage pulses, and then you might get into problems.
    A good engineer can solve all these possible problems, but each design has its own preferred solutions.
If your wearable goes commercial, please let me know :)
Regards,
Loek
Thanks Loek, you are a star!
Will keep you posted!

Warm regards
CM
 
Top Bottom