FEDEVEL
Platform forum

, 09-18-2024, 11:28 AM

Hello Everyone!
I'm currently working in an almost "Start-up company" specified for object detection on Sea (via AI, Image processing) and the demand increased for a Hardware as well which is a camera system on a gyroscope with Nvidia Jetson inside for Image processing. The device is operating with 24 VDC.
Unfortunately, we are not able to pass our EMC standard (acc. to IEC60945), since our radiation occours to be too high in the frequency range 30 MHz - 1 GHz.
The housing of the components are designed in an injection moulded part with a copper coating inside. Data communication is on Ethernet. During our last 'pre-compliance' testing we realised that the gyroscope what we are using is certified to industrial standard (US: FCC Class A / EU: EN-55032 Class A) which has higher limit than the residential (US: FCC Class A / EU: EN-55032 Class A) what we would more or less need in our requirements. This is one of the confirmed root causes (other devices shut down, ethernet off, only gyroscope powered and failed the test).
Do you guys have any idea if we could solve this issue in a normal way? Sadly, I'm currently out of ideas, that's why I approached this wonderful community.
Wish you a nice day !

QDrives , 09-18-2024, 11:35 AM

You mean class B for residential.
Is the gyroscope a 'big' mechanical one or inside a MEMS?
What is your noise spectrum?
Have you pinpoited it to the gyro?

, 09-18-2024, 11:52 AM

Thanks for the answer, actually the gyroscope is this one: https://www.flir.eu/products/ptu-5/?vertical=mcs&segment=oem ; attached a picture about the noise spectrum. This one with one colour is only about the gyroscope, the other one is our system powered up with / without Ethernet connection (blue: w/o, green: w/). The components are attached to the gyroscope and data is transferred through wires through the gyroscope. So we are using the interface of the gyroscope.

QDrives , 09-18-2024, 11:58 AM

Looks like you have 2 problems, as the green also goes over the limits.
Is the metal housing of the gyro connected to you gnd plane of the board?
What if you add a ferrite core (snap-on) to the ethernet cable for a test?

, 09-18-2024, 12:05 PM

Yes, the ferrite core on the Ethernet cables definitely helped. Managed to put them on limit , with QP it could pass that way. The metal housing of the gyro is connected to the gnd plane.

QDrives , 09-18-2024, 12:10 PM

How is the connection made?
With ferrite core working, it is common mode noise.

, 09-18-2024, 12:12 PM

The boards are attached to the metal frame and housing with screws and the ethernet is connected to grounded switch. We also tried to ground the housing as well, however did not bring any result.

QDrives , 09-18-2024, 12:38 PM

In an extreme 'solution' - what if you enclose the gyro in an EMC shield?

QDrives , 09-18-2024, 12:43 PM

https://www.digikey.com/en/products/filter/rf-shields/867

, 09-18-2024, 12:53 PM

Thanks for the tip! I will check with the colleague if we can fit this in, but unfortunately I think we don't have the space for it.

QDrives , 09-18-2024, 01:29 PM

First see if it would help.