Usage of OTS computer during IEC 60601-1-2 emission tests

[FelipeSchneider]

Hello folks,

I've been struggling to find a definitive answer on how to select an Off-The-Shelf (OTS) computer for IEC 60601-1-2 emissions testing, both in the standard and through general research. I'd greatly appreciate your advice.

I am finalizing the test plan for a Class II medical device with a Type BF applied part. The device is powered and communicates via USB.

Since an OTS computer (certified to IEC 62368-1 / 60950-1) only provides 2 Means of Operator Protection (MOOP), I have designed adequate insulation within my medical device to ensure it meets the leakage current and insulation requirements of IEC 60601-1 on its own.

My question is about the EMC testing setup for emissions (IEC 60601-1-2):

• I need to select a laptop to power the device and receive its data during testing.
• I understand that the computer will significantly influence the EMC performance of the overall system.
• What is the best strategy for selecting this computer? Using an overly "good" or overly "bad" laptop doesn't seem to represent a realistic worst-case scenario and could lead to non-representative results.
• In any case, any computer selection seems impossible to be reproduced by the end-user and I can't afford the option to do multiple tests (in any case, I have never heard of anyone doing multiple EMI tests to get statistical data out of it...).
• Another point is about the computer power supply, should it be connected to the mains or not during irradiated emission tests since it might also plays a relevant role in emissions?

Any tips or shared experiences on how you've approached this in your own testing would be immensely helpful.
Thank you in advance for your insights!

 

[Peter Selvey]

Although you are likely to get advice to the contrary, when you have a system that involves devices from other manufacturers (PCs, accessories, power supplies etc.) they cannot logistically be covered by a "TYPE TEST" such as for EMC. A basic tenet of "type" testing is that there must be a single identifiable entity that is responsible for design and production and ensuring any test results are "representative".

Let's say I manufacturer marbles that have to be 20±1mm in diameter. I send a single marble to a third party lab and get a result of 19.9mm. OK, the report looks nice but what does this mean? Statistically nothing, a test on a single sample alone is meaningless. However, if I have designed a production process that consistently produces 20±0.3mm marbles (i.e. 3:1 margin) and follow it with my own production testing with ±0.5mm criteria, and then send a representative sample to a third party lab for independent testing using the ±1mm criteria, then yes, the whole package together plausibly forms a "type test": as the responsible manufacturer I can genuinely declare that the third party test lab result, despite being only one sample, is "representative" and I have design and production controls to ensure this is the case.

Obviously this breaks down if the marbles are off the shelf and I have no idea what design and production controls are in place. A test on a single sample goes back to being meaningless. I'd have to test lots of samples, and continue to test to be sure.

Now, IEC 60601-1 and IEC 60601-1-2 are actually written to cover systems. But this is a clear case of overreach, an instinctive response that testing the whole system is necessary to be sure it is safe, without thinking about the logistics this requires and what a type test actually means. It is fundamentally wrong. There is no doubt this will be fixed at some point in the future.

ISO 14971:2007 had the same overreach, when it tried to include "systems" in the definition of "manufacturer". The committee surely got rapped on the knuckles about this, and the 2017 edition completely redefines a manufacturer being just the entity responsible for the individual "medical device" and makes no mention of systems.

So, how should an EMC test be done?

There are a few legally correct options:
1) keep the PC outside of the EMC test environment, using ferrites and other methods to prevent the PC from influencing the test result. The EMC labs should be able to help with this, it should be a common situation for both medical and non-medical testing
2) use a good quality PC that in principle should not affect the results
3) use a dummy set up, something that simulates the PC
4) a mix of these.

The point is you want to test your medical device, without the other manufacturer's devices interfering with the results.

 

[Hendi]

2) use a good quality PC that in principle should not affect the results

If you go this route sucessfully, you can list the PC model and say that it passed EMC with this in your documentation. Depending on your market and things you might want to add that the system integrator/distributor/customer should consider testing if he deviates from that setup.

 

[FelipeSchneider]

Thank you Peter and Hendi.

I agree with you, Peter, regarding the system and Type Test intention and that the standard does not make a good job explicitly explaining the ME System approach.

Additionally, the irradiance limits are so low that I assume that the committee when writing the standard must have considered that the interaction of the medical device with they system would likely not create safety issues.

What I can say is that we are going for option 2 as to reduce the risks of internal and external questions regarding our risk management approach.

 

[Peter Selvey]

The emissions and immunity testing are orders of magnitude apart (µV/m vs V/m). The emission limits are designed around not interfering with the communications signals received by devices (wifi, radio, TV, bluetooth, phones) which quickly drop off to very weak signals away from the source, hence the need for extremely low limits from devices in the same frequencies. In contrast, a device's immunity has to withstand larger signals assuming it could be located near a transmission source. With that in mind, keeping your medical device 1m from any transmitting source (including the PC's bluetooth, wifi) is a good rule of thumb. Of course, you could disable wifi/BT this but I think users would find it easier to keep the gap.

 

[Al_Z1]

If you declare your device as a system, where PC is part of the system, than you can go by another one approach. If your PC is out of patient environment, than according to p.16.1 of IEC 60601-1 you have no need to comply with medical device standards requirements.

Also, to exclude variety of different PCs, you can use several specific models and add it to your medical device, prohibiting the use of other PC models in IFU. AFAIK, ban of using other PCs, than from device shipment, is widely used.