Electronics > Projects, Designs, and Technical Stuff
EMC Pre-Compliance Visit: What to Expect and Bring?
<< < (4/6) > >>
techy101:
Well, ready or not here I go! Thank you again for the comments and opinions.




--- Quote from: T3sl4co1l on April 10, 2019, 10:13:19 pm ---Are your immunity results actually going to be representative?  Without the ~100pF bag of mostly water at the end of those cables, you can't expect the same RF currents as in practice.

Human analogs exist, which I imagine you could rent for a test; or if that's rather over-the-top, it might be okay to use a box of about the right size, with some resistors to it to model the ESR of a body.
--- End quote ---

This is going to end up being something I might make a separate post about. I'm not sure how exactly to test this properly since the skin impedance changes over the duration of a treatment. It's also dependent on the person, their skin, location on the body, etc...




--- Quote from: Berni on April 11, 2019, 09:23:28 am ---I can already tell you that the radiated emissions are going to be your new nightmare when i see it contains "four 5VDC->120VDC flyback converters"

--- End quote ---

Yup. This is exactly what has been making me so nervous about these tests.



--- Quote from: coppercone2 on April 11, 2019, 01:08:50 am ---king james bible

--- End quote ---
:-DD
coppercone2:
yes, this needs to meet class B emissions because its medical for iec 60601. it's considered tough. If you never did it before you will be suprized by how chaotic it is (i.e. you might think you can disconnect certain subassemblies to get a clear idea).. it probably won't end up being very clear.
mayor:
How'd it go?
techy101:

--- Quote from: mayor on April 13, 2019, 10:31:29 am ---How'd it go?

--- End quote ---

Sorry for the delay, I went on vacation right after my visit to Intertek and ignored work completely  :popcorn:

Overall the testing went well. It was nice to actually see the test setups and go through the process. The unknown made it much more stressful and now that that’s gone I’m a lot more confident moving forward. I went out to PCB West last fall and took a few sessions by Rick Hartley and Daniel Beeker which talked about techniques and tips for reducing emissions, etc.. I implemented some of these and would like to think they helped out, but I suppose you never really know.

Anyway, I passed radiated emissions by about 20dB, which was a huge surprise. I also passed conducted immunity and most of radiated immunity. One frequency, 870MHz, however caused a major failure. The output current shot up and hit one of our safety cutoffs (set at +- 0.1mA), triggering the device to log the fault and partially power-down, as it’s designed to do. The delay on the analog meter we had in the chamber, combined with some image vibration in the camera they use, makes it impossible to tell if the current actually increased or it caused some sort of fault in the ADC. My money, however, is on the current actually increasing. This only occurred when the antenna was in the horizontal orientation, not the vertical.

I haven’t had much time to dig into this yet but did notice an awfully big coincidence; the wire length inside the device, between the external connector and the board header, is darn close to the 870MHz wavelength of 344mm (see attached image). I’m hesitant to make any large jumps based on this alone because there is a lot more than just the wire itself involved, but I do find it interesting. Ferrite clamps on the cables didn't make a difference and the issue occurred whether or not out long applicator cables were attached to the device or a small dummy load, so I see that as pointing to something internal to the device itself.

The other failures I had were pretty bad on the ESD front. The good part was that it passed 15kV air discharge on the keypad. 15kV air discharge on the graphic LCD did cause some arcs which caused some pixels to go black, but they returned to normal within 15-20 seconds and there was no change in device behavior.

Now the bad. The connectors on the back of the device (the ones shown in the photo) failed by arcing to the outside of the connector at just 4kV. The bodies of the connectors aren’t tied to anything and are mounted to an ABS enclosure. When the system was not under a heavy load (stimulating) this caused the LCD backlight to drop out then come back on. When the device was under heavy load it caused the whole system to shut down. To me, this is indicative of a brown-out because the microcontroller is used to hold the system 3.3V power supply on or shut itself down. I think the ESD strike is causing a power blip that, when not stimulating, the onboard capacitance enough to prevent the microcontroller from losing power and shutting the system down. But, when stimulation is happening, the large power draw from the flybacks is enough to cause the microcontroller to lose power, shutting down the 3.3V system power supply.

At this point, I’ve not had time to dig into it, but we have tossed around a couple of thoughts. The device has four channels, three of which are isolated from the fourth channel which shares a ground with the rest of the system. However, there is no difference in behavior between ESD strikes on any of the four connectors. The isolated nature of this device has caused headaches all throughout the process and has me scratching my head at how to approach the problem. I’ll be digging into it over the next few days and we have placed a rental order for an ESD gun and IEC 61000-4-2 spec test table setup (I wanted to build one, but the boss went for rental instead). That should be here the end of next week and will be very interesting to play around with. On the one hand, I’m happy that the device failed “safe” by shutting down, but I also kind of wish it had failed catastrophically so I might be able to see the path it took.

Finally, there were failures at 15kV through the gap between the battery door and the body of the enclosure that, I think, went to the battery terminals. This will require further investigation with the device opened up.

Overall though, for my very first design, I’m quite pleased. I was worried everything would fail all over the place, but it passed many of the tests, especially the radiated emissions. With my lack of experience, I’ve spent a lot of time researching and focusing on design for EMC for this project and I’d like to think it paid off.

Now, on to problem solving!
Gyro:
It sounds like you did really well there. Yes, a trip out to the test house can make a fun outing (unless things go really badly). When CE was first being introduced, our test house used to run night slots due to the demand - the incentive was free pizza at 11pm!  20dB below the line on emissions is very respectable, especially in an ABS case.

For the susceptibility and ESD, it's worth checking that the requirements for your product are 'no effect', 'disruption but recovers without power cycling', or 'must recover after power cycle'. Floating the shells of connectors is always a bad one - you can get solder tags for those connectors to allow them to be connected together and to the right grounding point.

For the 870MHz susceptibility, very low value capacitors from I/O connections to ground can make a difference. I once had a case where a 10pF capacitor in the right place solved a problem in that frequency range. Larger values had too much self-inductance to work effectively! The ferrite beads failing to cure it probably indicates that the cables weren't the way it was getting in (case again?). Shunting with a capacitors will hopefully be more successful.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod