Kirk Humphries

EMI schielding

Discussion created by Kirk Humphries Employee on Jan 29, 2006

This message contains an entire topic ported from a separate forum. The original message and all replies are in this single message. We have seeded this new forum with selected information that we expect will be of value to you as you search for answers to your questions.

 

Posted: Thu Sep 29, 2005  8:30 pm

 

Hi,

 

My HC12 project needs protection against EMI. The problem is, I dont know which kind of metal will be the best for an enclosure shield. Aluminum and stainless steel are not magnetic - will it work the same as a iron steel enclosure to shield against EMI.

 

Thanks

 


 

Posted: Thu Sep 29, 2005  10:18 pm

 

A solid ground plane and proper connection of external cables/connectors will get your pcb most of the way there. Copper foil sheets are readily available to coat the inside of a plastic enclosure if aluminum or other metal enclosures aren't available. Production floors hate copper foil but its perfect for a prototype pcb. Steward (and others I'm sure) make little ferrites that you can place on top of the processor - never tried them myself.

 


 

Posted: Thu Sep 29, 2005  11:02 pm

 

It depends on how the EMI gets coupled into your project.

 

Aluminum or copper shielding is a good start.

 

Its likely coming in through the leads to the outside world, or the power supply. In this case, being careful about how the outside world conductors come into the shielded enclosure becomes important.

 

You really need to tell us more about the sources of EMI expected and how the project will be tested.

 

For some really demanding environments, double shielding with filters where the leads pass through each shield enclosure are needed.

 

For more ordinary requirements, careful board layout and a ground plane may be good enough.

 


 

Posted: Thu Sep 29, 2005  11:59 pm

 

Yes, in this instance i was thinking radiated (rf) EMI noise from outside the box - assume conductive radiation taken care of using filters etc.

 


 

Posted: Fri Sep 30, 2005  1:23 am

 

>Yes, in this instance i was thinking radiated (rf) EMI noise from outside

>the box - assume conductive radiation taken care of using filters etc.

 

A tight conductive box, connected to logic ground inside at ONE point only, should be fine.

 

The only reason to want an magnetic box is if there is a danger of magnetic coupling. This would only be likely if there were high power coils or transformers around with magnetic flux escaping.

 

There are special "high-permeability" alloys for shielding in these cases. They aren't cheap.

 

This can happen next to a magnetic deflection CRT. I had one experiment where we had to move the high speed analog board about 5 feet away from the computer monitor to get rid of analog noise from the CRT deflection system. As we were debugging the hardware, it wasn't in a box.

 

Luckily, we had a scope with FFT, so we looked at the noise spectrum, and saw a big spike in the 10s of kHz. It didn't take long to figure out that the monitor was the only thing around at that frequency.

 

Unless you're next to a transmitter, you shouldn't have much magnetic coupling problems for frequencies above 100 MHz.

 

Static induced EMI should be cured with a conductive box, but you should be aware that the peak power can be in the megawatts for a few nanoseconds.

 

You really need to be careful to keep the static discharge path outside the box, and you may need very effective filtering on the wires going through the box.

 

In the bad old days, when computers were larger than refrigerators, I had a few adventures with static discharges disrupting computer operations from 3 feet away. We ended up removing all the plastic chairs, and walking around in stocking feet.

 

A method that some radio hobbyists recommend for making shield boxes is to use copper clad PCB material. You saw out the box sides, and solder the box together.

 

A standard quick test is to use aluminum foil as a shield on a circuit board. If this helps, then you need a box. (You need to be very careful to keep the foil from shorting the pins and traces.)

 


 

Posted: Fri Sep 30, 2005  6:20 am

 

Hello,

 

The radiated noise (immunity) only becomes a problem if it is converted to a conducted noise that takes its current path through sensitive electronic circuits on your board, causing voltage potentials that are high enough to disturb normal operation. The most likely candidate for such a conversion to take place is in attached cables that work as antennas for the frequencys in question. Make sure that this current isn't induced into your electronic circuits by either bypassing it to a good groundplane or by blocking it with inductors (or both).

 

Usually, shielded cables will protect sensitive signals very well if the shield is connected correctly. Use EMI cable glands (if you have a conductive housing) or connect the shield directly to the groundplane without a "pigtail" (use some sort of clamp around the entire cable directly on the board). The shield should be connected the same way at both ends.

 

What are your immunity requirements? With a good groundplane, good decoupling and shielding, it isn't impossible to pass a test at 10V/m up to 1GHz even with a plastic enclosure.

 


 

Posted: Fri Sep 30, 2005  1:58 pm

 

<snip>

 

> Usually, shielded cables will protect sensitive signals very well if the shield

> is connected correctly. Use EMI cable glands (if you have a conductive housing)

> or connect the shield directly to the groundplane without a "pigtail" (use some

> sort of clamp around the entire cable directly on the board). The shield should

> be connected the same way at both ends.

 

Just be careful not to introduce ground loops when connecting the shield at both ends. They are subtle and really cause noise problems.

 


 

Posted: Sat Oct 1, 2005  6:34 am

 

> <snip>

>

 

> > Usually, shielded cables will protect sensitive signals very well if the shield

> > is connected correctly. Use EMI cable glands (if you have a conductive housing)

> > or connect the shield directly to the groundplane without a "pigtail" (use some

> > sort of clamp around the entire cable directly on the board). The shield should

> > be connected the same way at both ends.

>

 

> Just be careful not to introduce ground loops when connecting the

> shield at both ends. They are subtle and really cause noise problems.

>

 

> <snip>

 

Very true,

 

However, in my experience HF noise is a much bigger problem than noise (or other problems) caused by ground loops. Just make sure the ground current only flows in the shields and chassis. If this is a problem one end can be AC coupled with a capacitor but that mostly means a much worse HF shield.

 


 

Posted: Thu Oct 13, 2005  1:30 pm

 

That’s easy to explain: Most stainless steels are not magnetic.

Outcomes