Hey,

I have just been out performing EMC/EMI tests on our systems and found out that one PCA (printed circuits assembly, that's what it is called? a mounted PCB?) in the system is very susceptible to radiation and also radiates a lot. In the previous version of this circuit, the ground plane seems to have been spread on the top and bottom layer if I understand correct. However in the new version, our supplier seems to have dropped this idea and have not covered the first or last layer in a ground plane. Only a small part of the top layer seems to be connected to ground.

My boss claims this could be the reason for the bad EMC/EMI results. I have absolutely no experience whatsoever from layout of PCBs and don't really know how that ground plane can affect the results and why you spread the gnd connection that much (except to decrease resistivity and thus make all nodes have "the same" ground potential).
Does anybody care to enlighten me?

I have just taken my master degree in computer engineering and is currently working with electronics, which I am very interested in. I just feel I lack the knowledge

 

Without a schematic and the actual PCB it hard to say anything conclusive. But a good book regarding this matter is "Electromagnetic Compatibility Engineering By Henry Ott" It is on Google books but somewhat crippled. But you can have a look and see.
http://books.google.com/books?id=qe...ntcover&dq=henry+ott&cd=1#v=onepage&q&f=false

 

Having the proper ground plane will have a drastic effect on EMC.

It's partly down to reducing the voltage difference between ground points.
That in turn makes decoupling more efficient and reduces power supply noise.

The other great effect is having the grounded copper immediately against all the general signal tracks reduces the 'antenna effect' that you otherwise get.

 

If you can show in public or in private schematics and the pcb layout I'm sure we can come up with some solutions.

 

HI all,
EMI reduction is a field in of itself. I designed and used to run my switching power supply's EMI Lab and was the EMI engineer in charge of making the smps power supplies meet FCC/VDE levels A or B. One of the key things that needs to be understood in this field is that whenever you have a high-frequency waveform on a single conductor that is not close to it's return (ground), then you have basically a radio transmitter and the single wire is the antenna and it will radiate EMI. Ground planes on PWBs with high frequency switching must almost always be at least 4-layer so that one or two of the inner layers can be used as ground planes. What this does is change the trace from an antenna to a transmission line, which greatly reduces the radiated and conducted EMI. Now, this can be done without multi-layer PWBs, but then you must have the return (ground) trace be parallel to the signal trace and not further away than 3*W, where W is the trace width of the signal trace. But there are lots of useful tricks to reducing EMI, like slowing the rise and fall times on a fast switching signal line, using terminating resistors to keep a constant impedance on long signal lines, and there are many others.

Also, this is not something that can be easily simulated with the usual tools. It can be done with Finite Element Analysis simulators (FEA), but these require the user to input the physical configuration of the PWB and components to perform the simulation. You can't just put in a schematic like in SPICE and run it. It needs much more detailed data. Shielding is another way to reduce radiated EMI by containing it. Conducted EMI can be reduced by using LC or LRC filters to bandwidth-limit the signals to below the required EMI specifications. The measurement equipment, spectrum analyzers and specialized radio receivers are very expensive and specialty RF amplifiers and a variety of antennas (not cheap) are required as well. Finally, a screen room that is big enough for the measurements being done is required, or else, they can be done in a naturally-shielded valley in the mountains, if the user can find a spot to measure that's quiet enough in terms of minimal levels of broadcast radio and TV, commercial 2-way radio signals, etc. so that the electronic equipment's EMIis properly measured and not confused with a taxi cab's radio or similar.

Those are just a few of the issues involved. At a minimum, just to have a set of equipment to do the tests, if bought 2nd hand will run you at least $ 100,000 easily and if new equipment, much, much more.

Regards,
Kamran Kazem

 

I agree with your boss.

Reject the PCA, return it to your supplier, and tell them to fix it.

Regards,
Ifixit

 

All excellent replies to the OP. And that Henry Ott book is worth every penny, he's the world leader in the field.

If it helps to understand it a bit better, think in terms of rise times rather than frequency. A high dV/dt will create interference by radiating electric fields. This may be contained somewhat by electrically conductive shielding, such as a ground plane. This works by turning the electrical energy into heat in the shield.
A high dI/dt creates magnetic fields, which will pass quite nicely through most conductors, and so here magnetic shielding (such as mu-metal) is required. This works by warping the magnetic flux around the item being shielded.

Radiated emissions and susceptibility are often two sides of the same coin - the same principles are valid for both.