EE's - Laptops/Printer Change and Emissions Testing

MThomas

Involved In Discussions
A laptop and printer are shipped with our system....they are on the BOM. They are therefore a part of the ME. There are also many other custom components which produce emissions in the system

Let's leave safety and applied disturbances out of the discussion.

The laptops/printers are certified for US/CAN/EU to class B home use. The system must be US/CAN/EU class A, but, again there are other noisy emitters in the system besides the laptop and printer.

Question: If we change the BOM to use a different laptop or printer, do we need to repeat RF and conducted emissions

Our lab does not want to answer the question in a straightforward manner. They never say no. They have offered to rerun the tests and give us the test results but not a complete report.

There are many opinions within the engineering department

Opinion 1- Retesting is not necessary because you are substituting class B with another class B device-no need to worry that devices have different emission spectra
Opinion 2- You must re-test because the part numbers on the test report will not match the part numbers on the BOM
Opinion 3- Write a justification to eliminate re-testing...this is just as good as objective evidence

The justification would be a formal analysis that treats the new printer or laptop as point sources/currents sources with maximum class B emissions at all frequencies. It goes the math to add the additional energy to the existing test result in the form of addition dBuV/m. Finally, it is shown that the results are below class A limits.

Any thoughts?
 

Peter Selvey

Leader
Super Moderator
Shipping the laptop and printer with the medical device does not make it a medical device or medical electrical equipment (MEE). However it can be part of a ME SYSTEM.

IEC 60601-1 has many requirements for ME SYSTEMS. IEC 60601-1-2:2014 also has clause 4.2 for ME SYSTEMS which suggests the system may also need to be tested if the non-medical equipment could affect basic safety/essential performance (I don't have copy of the latest EMC standard, so not sure if this is still valid) .

Even though ME SYSTEMS is written into these standards, it is legally wrong.

The ultimate target of any modern regulatory system is the "declaration of conformity". This declaration covers not the samples tested, but the actual objects placed on the market. To make it clear: the declaration is NOT that a sample was tested and found to be in compliance, but that actual devices coming off the production line are in compliance. A test on a single sample is not sufficient evidence for this (a single sample is statistically meaningless), it must be combined with design features and production controls to ensure compliance in regular production. The person or organisation signing the declaration declares that they have the design/production controls in place to ensure that any test results on a single sample will be representative of broader production.

This is all specified in ISO 17050-1 which covers the declaration of conformity. This standard is mandatory in Europe and used by the FDA, and would be well used throughout the world, although I suspect not well read as most people making the declaration would just copy a pre-existing format, and probably think that a positive test result on a single sample is enough to sign the declaration.

Anyway, such a declaration cannot be made for an ME SYSTEM that contains devices with different manufacturers, unless there is a contractual agreement between the manufacturers. Without this agreement, one manufacturer cannot sign to say they have design/production controls for the other parts of the system. Such agreements are not practical and rarely exist in the real world.

In principle, the manufacturer of the MEE should consider the risks of an ME SYSTEM, including EMC. However, the outcome of this analysis should be actions taken for the MEE, not the system. For example, if there are concerns that the laptop or printer could affect sensitive diagnostic measurements, then actions can be taken such as the use of optical isolation, physical separation, shielding of sensitive parts. There may be formal and informal experiments to verify the effectiveness of these solutions. However, ultimately the compliance should be based on criteria for the medical device, and should not be dependent on the particular laptop or printer used in any system tests.

It should also be noted that EMC testing is a black art. The errors in test reports are typically in the order of 3 to 6dB which is huge. (3dB = ±30%, 6dB = ±50%). Plus the tests put more emphasis on repeatability rather than being representative. For example, the layout of cables is done in special way to try any make the tests repeatable, rather than being representative of the real world. EMC tests are necessary but when taking into account the need to retest or make decisions about whether changes to a system could affect results, the large uncertainty of the EMC tests themselves should be a factor.
 

MThomas

Involved In Discussions
I am not sure how your answer is related to my question. Can you assist me?
Shipping the laptop and printer with the medical device does not make it a medical device or medical electrical equipment (MEE). However it can be part of a ME SYSTEM.

IEC 60601-1 has many requirements for ME SYSTEMS. IEC 60601-1-2:2014 also has clause 4.2 for ME SYSTEMS which suggests the system may also need to be tested if the non-medical equipment could affect basic safety/essential performance (I don't have copy of the latest EMC standard, so not sure if this is still valid) .

Even though ME SYSTEMS is written into these standards, it is legally wrong.

The ultimate target of any modern regulatory system is the "declaration of conformity". This declaration covers not the samples tested, but the actual objects placed on the market. To make it clear: the declaration is NOT that a sample was tested and found to be in compliance, but that actual devices coming off the production line are in compliance. A test on a single sample is not sufficient evidence for this (a single sample is statistically meaningless), it must be combined with design features and production controls to ensure compliance in regular production. The person or organisation signing the declaration declares that they have the design/production controls in place to ensure that any test results on a single sample will be representative of broader production.

This is all specified in ISO 17050-1 which covers the declaration of conformity. This standard is mandatory in Europe and used by the FDA, and would be well used throughout the world, although I suspect not well read as most people making the declaration would just copy a pre-existing format, and probably think that a positive test result on a single sample is enough to sign the declaration.

Anyway, such a declaration cannot be made for an ME SYSTEM that contains devices with different manufacturers, unless there is a contractual agreement between the manufacturers. Without this agreement, one manufacturer cannot sign to say they have design/production controls for the other parts of the system. Such agreements are not practical and rarely exist in the real world.

In principle, the manufacturer of the MEE should consider the risks of an ME SYSTEM, including EMC. However, the outcome of this analysis should be actions taken for the MEE, not the system. For example, if there are concerns that the laptop or printer could affect sensitive diagnostic measurements, then actions can be taken such as the use of optical isolation, physical separation, shielding of sensitive parts. There may be formal and informal experiments to verify the effectiveness of these solutions. However, ultimately the compliance should be based on criteria for the medical device, and should not be dependent on the particular laptop or printer used in any system tests.

It should also be noted that EMC testing is a black art. The errors in test reports are typically in the order of 3 to 6dB which is huge. (3dB = ±30%, 6dB = ±50%). Plus the tests put more emphasis on repeatability rather than being representative. For example, the layout of cables is done in special way to try any make the tests repeatable, rather than being representative of the real world. EMC tests are necessary but when taking into account the need to retest or make decisions about whether changes to a system could affect results, the large uncertainty of the EMC tests themselves should be a factor.


The issue of consideration is direct harm to the electromagnetic environment of intended use; other ME from other manufacturers are in the environment.

There must be objective evidence or a justifiable analysis that the collection of components do not exceed the accepted limit between harm and no harm.The collection of components...laptop, printer, parameter monitor, etc on a shelf with line isolation. It has a single connection to the mains..we take this as a definition of ME not medical system.

EMC testing is a type test of a representative sample, performed once.

The question is whether this type test has to be repeated if the printer part number or laptop part number is changed. Let us go on to say that the part number change does not create any new hazards.


Does that make it clearer? Which option is this best choice?

As an aside, each device is tested for basic safety with regard to electrical hazards before shipping...this is outside the scope of emi testing.

The laptop hardware and software gather raw data, make a diagnosis, and this is printed.
 

Peter Selvey

Leader
Super Moderator
The point of the previous post is to highlight that as much as anybody wants to test a "system" because of perceived or actual risks (such as the laptop/printer interfering with other medical devices), there's no legal foundation to do this because your company is not the manufacturer of the of the other devices in the system (e.g. laptop or printer). Even if you do the testing, you can't use the report in a "declaration of conformity" which the legal basis on which access to markets is made. Often design teams are not aware of this document because the QA dept signs, but it will be there somewhere in the regulatory filings.

You mentioned "type test of a representative sample, performed once". This is correct, but the key point is the word "representative". In order to know if a test is representative, there has to be design and production controls. How can manufacturer A know if the EMI test results for a device made by manufacturer B are representative? They can't. So you can't have a "type test" for a system with multiple manufacturers. By definition.

Separately, the issue raised is not as serious as it might seem. The limits for immunity are orders of magnitude higher than the limits for emissions. For example radiated immunity uses V/m while radiated emissions use µV/m. The reason is that limits for emissions is about interfering with radio and TV transmissions. These TV and radio signals can be very weak by the time they reach an antenna. The Class B limits for household use are even stricter as the house could be far away from the transmitter (e.g. in the mountains or out of line of sight), hence the noise generated by equipment should be pretty low to avoid interfering with raido/TV reception.

In contrast, immunity testing has more to do with the device being able to handle much larger sources of noise such as being close to a TV tower, wi-fi router, mobile phone and the like, typically large transmitters but there are other sources.

Anyway, testing for CISPR 11 whether Class A or Class B has nothing to do with the potential for harm except perhaps the stress of the person in the room trying to tune into to the football on thier transistor radio.

Actually, the far more critical point is the special tests for bluetooth and wi-fi modules in the PC and laptop, since these are transmitters. In order to legally market the laptop and printer, they must already comply with regulatory limits, under the responsibility of the labelling manufacturer. Testing the output of these transmitters are not covered by medical device regulations, however the immunity testing does take into account the potential for such transmitters to be in the environment.

It is not reasonable for your company to re-test BT/Wifi standards just because there is a worry that the devices could interfere with other medical devices. It just does work for whole bunch of reasons. For example, the laptop manufacturer might source several different Bluetooth modules and switch them around depending on supply constraints. Your company won't even know which BT module is the particular laptop. Only the labelling manufacturer can deal with that kind of stuff.

Again, risk controls that could be feasible for your company are to turn off the BT/Wifi, if there really is a worry.
 

MThomas

Involved In Discussions
It is our judgment that laptops of the same part number but different serial number are equivalent. A laptop with a different part number is not representative...it uses different circuitry, has different sip/sops etc.



As an aside specific absorption rates of RF energy comply with FCC/CAN standards for the internal radios.

The laptop and printer are connected to custom designed medical equipment with the approprate insulation which produces a significant amount of emissions.

I do not wish to bring immunity to applied disturbances into the discussion because there are needless complexities.

The laptop and printer are part of the ME for safety testing. They are certified to 60950 or its successor. We do not rely on them for any insulation.


Are you suggesting to reconfigure the test so that the laptop and printer are ancillary equipment out side the chamber for emissions and applied disturbances?

No matter what the regulatory requirements are, the conducted and radiated emissions of some part of the device that we sell must be measured.
 
Last edited:

Peter Selvey

Leader
Super Moderator
Yes, in principle EMC is the same safety, it works by each labelling manufacturer being responsible for their device. If you are putting together a system, the entity placing the system on the market should always check the "other" system components (those not manufacturer by the entity) by asking for certificates, declaration of conformity etc for key standards/regulations, and periodically check for validity, but that should be the limit. The system entity should not be retesting anything.

There should be ways to keep the laptop and printer outside the chamber or shielded in such a way that they do not influence the test results. Of course, one option is to test the target device with the laptop/printer, and then if the system passes you can also judge that the target device on its own will also pass (this is a common method). But that method should not imply that the test also covers the peripherals, it's more of a short cut to avoid having to do the shielding etc.

Now ... the Part number ... no that's not the way it works. There's more to it.

Condersider for example you test your device with laptop made by Q and model X123 (in IEC standards we say "model number" rather than "part number").

So, you do the EMC test on the system of your device with X123. Result is pass. Great.

Then Q decides to change some parts, but they keep the same model number. This happens a lot especially in the first 1-2 years, it could be cost down, parts supply issues, addressing complaints from market, design mistakes, change of factory etc etc etc. Despite a lot of internal changes, the model number stays X123, the outside looks the same, which means marketing, sales, promotion can all stay unchanged. Very, very common situation.

Of course, with every change Q has to think about EMC and safety, and any other requirements. They can retest, or write a justification. They have their own team of engineers, QA department working to sort all that out. keep the certificates and regulatory side valid. However, they do not need to change the model number. Also they do not need to notify any users (such as yourself) about the changes. Internally, they can use serial numbers to provide traceability to the changes and the different versions while keeping the same model number.

This is one of the many reasons why a "type test" on "system" made up of devices from different manufacturers is meaningless or even impossible in a regulatory context. Only the labelling manufacturer has the ability to keep everything under control.
 

MThomas

Involved In Discussions
ok thank you very much. We are on the same page. We buy such a large number of laptops/printers that the contract is written to give us a product change notice if there are any changes to the internal design, processes, or plant of manufacture. Or so I am told by the head of quality. But technology changes quickly and after lifetime buys, there must be a change.
 
Top Bottom