3

I have a mezzanine card, 60 by 40 mm, mounted near one end of a main PCB, about 160 by 40 mm. This card is a known source of EMI problems and, in fact, has a metal shield over it. But, I suspect that it is injecting EMI into the main PCB and it is radiating from there.

There is a single power/signal connector from the mezzanine card to the main PCB with two ground pins. These are connected to the main card ground plane via thermal reliefs. The card also has four mounting holes, plated through and continuous with its ground plane. These are used to mount the metal shield using metal hardware, but they are at present electrically isolated from the main PCB ground plane.

My question is: is it good that the main PCB ground plane connects to the mezzanine card only via the two pins in the connector or would it be better to mount the card using metal pillars and have the mounting holes on the main PCB plated through and continuous with the main PCB's ground plane?

There is a second connector between the boards, but it is just to route a signal from the mezzanine card off the board and there are no electrical connections from that connector to ground.

Troublesome emissions are in the range 25 to 300 MHz, some higher.

Graham Davies
  • 514
  • 4
  • 7

1 Answers1

3

Given the information provided it is very difficult to give anything but general advice.

  1. Understand what the source is. Try to reduce the amplitude at the source.
  2. Understand what the antenna is. Try to make the antenna less efficient.
  3. Shield the whole thing.

My experience with getting multi-board systems through EMI testing is that it is so much hassle than I always opt for a metal enclosure around any multi-board system. Use the metal enclosure as a Faraday cage and make sure everything passing though the enclosure is sufficiently filtered to the enclosure. And avoid having the enclosure itself be the (return) path for any high frequency signals.

My experience with single-board systems is that they can usually go through EMI testing with some basic care. Especially around power supply (PDN design - I use PDNTool.com to make a low impedance bypass across the whole frequency range) and filtering for connectors. Consider placing all connections along one edge.

Maybe these answers can help you as well:

(sorry for the non-exhaustive list, but I answered on these so they were quick to find for me and there is at least one coherent answer in those that follows my thinking :-)

In short: It sounds like you need a metal enclosure for the 2-board system you are building. But start by understanding. If it is too hard to understand what is going on - seek help. Two heads are better than one. Peer review is an underrated but very useful activity.

Community
  • 1
Rolf Ostergaard
  • 4,448
  • 18
  • 24
  • +1 I was not aware of PDNTool.com, sounds like a nice tool – Blup1980 Apr 24 '14 at 08:57
  • Thanks for this answer, I think it is technically excellent and I will read the reference topics as well. Unfortunately, I have tried most of what is suggested. – Graham Davies Apr 24 '14 at 11:08
  • 1 - As far as I can tell, nothing that I designed is contributing EMI; it's all the module, the use of which is fundamental to the product. – Graham Davies Apr 24 '14 at 11:08
  • 2 - If I'm right about 1, the antenna is the whole main board and the goal of my question is exactly what you say. – Graham Davies Apr 24 '14 at 11:09
  • 3 - From the outset, I advised my client to metalize the inside of the enclosure as much as possible (there is a Bluetooth transceiver in there) but he refused for cost reasons. I have also asked the client to hire an EMC expert to consult on this, first during the original product design and again for this EMI remediation exercise. Again, the client declines. – Graham Davies Apr 24 '14 at 11:12
  • @GrahamDavies: Thanks. Sounds like a pay me now - or pay me later proposition... If what you are talking about is your board + an off the shelf BT module, you should be able to make this work. The step you need to understand is 1b: How the source couples to the antenna :-) – Rolf Ostergaard Apr 24 '14 at 18:16
  • @GrahamDavies: Re 3... "metalize as much as possible" is no good. Either you have a Faraday cage or you don't. Re cost... do the math: Is production volume high enough to justify a lot of additional development time? (hint: if it is, you would not be using a BT module, but should have the BT chip on your own board and save even more). – Rolf Ostergaard Apr 24 '14 at 18:19
  • The Bluetooth module is not the source of the EMI; the mezzanine card is. The Bluetooth module, which has an on-board antenna, is the reason the case can't be completely metalized. You seem to be saying that a shield is not effective unless it is a complete Faraday cage. This is at odds with my understanding. If there is an aperture in the shield large enough for Bluetooth radio signals to pass through (2.4 GHz), it could still be quite effective at the problem frequencies I'm seeing. I am not privy to business decisions on this project. – Graham Davies Apr 25 '14 at 16:20
  • In that case your best bet may be a Faraday cage around the two boards and the BT module - and with a coax out to the BT antenna somewhere outside the cage. That - or prepare spend a lot of time :-) Good luck. Engineering is always about compromises between technology and business. – Rolf Ostergaard Apr 25 '14 at 20:07