Design Con 2015
Breaking News
View Comments: Threaded | Newest First | Oldest First
Max The Magnificent
User Rank
Blogger
Very Interesting
Max The Magnificent   12/3/2013 3:04:37 PM
NO RATINGS
I have to admit that (coming from the days of much slower clocks and signals) I'd never really thought about this -- thanks for sharing.

tom-ii
User Rank
Blogger
Re: Very Interesting
tom-ii   12/3/2013 3:12:37 PM
NO RATINGS
From a Motorola Semiconductor Design Guide:

 

A handy rule of thumb to determine if an interconnect trace should be considered a transmission line is if the interconnect delay is greater than 1/8th of the signal transition time, it should be considered a transmission line and afforded all of the attention required by a transmission line.

MS243
User Rank
Manager
Re: Very Interesting
MS243   12/3/2013 7:02:38 PM
NO RATINGS
This can be daunting on any type of vehicle, where many different systems signals can mix in a small space.   There are somespecial  techniques for managing this in one's reciever / baseband chain.

tom-ii
User Rank
Blogger
Re: Very Interesting
tom-ii   12/3/2013 7:27:45 PM
NO RATINGS
Often, for more complex systems, you can break it down some, and choose the "prime" frequencies - it will often be that much of the equipment uses similar frequencies, and you can throw these into your plan as well.  But you are correct - this can get onerous - and in those cases, you likely just have to use your best judgement and also lean on the experts that are working your grounding/shielding to tell you which frequencies will likely be the worst problems.

MeasurementBlues
User Rank
Blogger
Re: Very Interesting
MeasurementBlues   12/4/2013 8:19:32 AM
NO RATINGS
Openings in enclosures should also be less than ¼λ for EMI purposes. openings with lengths of that size or longer are more likely to let emissions out of the box that smaller openings.

tom-ii
User Rank
Blogger
Re: Very Interesting
tom-ii   12/4/2013 8:54:04 AM
NO RATINGS
Excellent point.  It might also be noted that those openeings can create really strange effects and "nodes," too.

MeasurementBlues
User Rank
Blogger
Re: Very Interesting
MeasurementBlues   12/4/2013 9:06:20 AM
NO RATINGS
Of course, there are cable grounding and shielding issues. The most significant is to make a 360° connection with a shield around a connector. No pigtails or you get emissions.

Understanding grounding, shielding, and guarding in high-impedance applications

Shields are your friend, except when... (Part 3), has links to Part 1 and Part 2

Grounding and shielding: No size fits all a piece I wrote in 2001, still relevalnt.

 

zeeglen
User Rank
Blogger
Re: Very Interesting
zeeglen   12/4/2013 9:42:20 AM
NO RATINGS
>Of course, there are cable grounding and shielding issues. The most significant is to make a 360° connection with a shield around a connector.

And make the shield connection to the chassis, not digital ground.  I once had to clean up a horrendous mess where the designer connected coaxial cable shields to his pcb digital ground plane and could not understand why the radiated emissions were so strong.


MeasurementBlues
User Rank
Blogger
Re: Very Interesting
MeasurementBlues   12/4/2013 8:25:42 AM
NO RATINGS
"coming from the days of much slower clocks and signals"

I won't comment on Max's slower days, but another issue is power supply clock frequencies, especially when high-speed digital links are involved. power supply or other clock frequencies can couple into digital clocks, creating periodic jitter. The higher the clock rate, then less jitter it takes to ruin signal integrity.\

Signal integrity engineer Daniel Chow explains periodic jitter in Jitter: Measurement References Matter. The animated graphics really bring it to life. The article has links to numerous other articles covering jitter.

MS243
User Rank
Manager
Buses and software loops
MS243   12/4/2013 8:53:48 AM
NO RATINGS
One often also has to factor in external buses and software loops -- software  loop repetition frequency can also add spurs and harmonics off of each fundamental, as do the SMPS switching spikes.

MeasurementBlues
User Rank
Blogger
Re: Buses and software loops
MeasurementBlues   12/4/2013 9:07:55 AM
NO RATINGS
MS243, thanks for the tip on software loops. Being more of a hardware type, I hadn't thought of software as a possible interference problem.

Max The Magnificent
User Rank
Blogger
Re: Buses and software loops
Max The Magnificent   12/4/2013 9:57:41 AM
NO RATINGS
@MeasurementBlues: Being more of a hardware type, I hadn't thought of software as a possible interference problem.

I agree -- this had never struck me -- very interesting

MeasurementBlues
User Rank
Blogger
Re: Buses and software loops
MeasurementBlues   12/4/2013 10:07:33 AM
NO RATINGS
When performing an emissions test, you usually want worst case and thus many systems have diagnostic software for the purpose of emissions testing. It would be like exercising more FPGA gates than are used in normal operation. Or you might drive a a whole lot of gates at once for a worst-case condition.

MeasurementBlues
User Rank
Blogger
Re: Buses and software loops
MeasurementBlues   12/4/2013 10:16:57 AM
NO RATINGS
Also with regards to FPGAs, another test is to drive as many gates as possible all at once for testing power integrity on a board. That causes the FPGA to draw maximum current so you can see what it does to the power rails due to inductance in the power delivery network. Too deep a dive in power can cause havoc and you probably need better or more bypass capacitors on the board.

zeeglen
User Rank
Blogger
Re: Buses and software loops
zeeglen   12/4/2013 9:35:17 AM
NO RATINGS
>software  loop repetition frequency can also add spurs and harmonics

At a radiated emissions test site a manager type once told me to "Keep moving the cables around until it passes".  I can see the new version now - "Just keep tweaking the code until it passes".


MS243
User Rank
Manager
Re: Buses and software loops
MS243   12/4/2013 9:47:15 AM
NO RATINGS
SW Loops --And you change to a different memory type, with different wait states then everything changes also.

MS243
User Rank
Manager
Re: Buses and software loops
MS243   12/4/2013 9:53:54 AM
NO RATINGS
Display's also radiate  -- CRT's were kind of bad with the magnetic deflection, and the new LCD, etc displays scan the array to update it -- this means that there are effectively long rows of horizontal and vertical ITO traces being switched on and off to make each pixel light or dark -- and ITO must be placed over the whole array to reduce these emmisions -- DO-160 is often hard to pass, for a display designed only to pass  FCC/CE

Flash Poll
Like Us on Facebook

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
EE Times on Twitter
EE Times Twitter Feed
Top Comments of the Week