SI and EMI training

When you feed a mismatched line with a constant frequency, you are going to have standing waves along the line with a period of half a wavelength. Thus, there are local voltage as well as current minima and maxima along the line. A larger current will generate a larger external magnetic field strength and a larger voltage will create a larger external electric field strength.

Apparently the common misconception that only mismatched lines radiate, is that it is easier to measure high values at voltage or current maxima, but in the same time forget that there is a corresponding minimum between two maxima. To get a realistic picture of the total transmission line radiation, it should be measured at least over a half wavelength (in that transmission medium).

Paul

Reply to
Paul Keinanen
Loading thread data ...

If the wavelength is much longer then the height of the line above the ground plane, then the transmitting efficiency is proportional to the square of the height. The length of the line does matter also; it could be enough to fail CISPR 25 if the transitions on the line are faster then ~5ns.

Bad termination -> standing wave -> resonance. It could significantly worsen the situation.

Congratulations, You have learned the basics :)

Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

formatting link

Reply to
Vladimir Vassilevsky

I probably should have said a series terminated line that is close to the ground plane does not radiate much. Even without series termination, a standing wave requires a close match of trace length to the wavelength and periodic stimulation. Does that happen very often? A 200 MHz clock is 150 cm long. Even at a quarter wavelength this would be 15 inches. When was the last time you tried pushing 200 MHz onto 15 inches of trace on a board? Even at 200 MHz you would be looking at terminating or the signal would likely not work.

That is what the Ritchey course is all about, the basics are pretty much all you need to solve any of these problems.

Reply to
rickman

At 200MHz, the trace behaves more like a lumped system then like a distributed system. The wavelength consideration can not be applied directly. You should consider the stray inductance and capacitance of the trace separately as well as all of the pads, vias, inputs and outputs connected. It can very well resonate at 200MHz and/or its harmonics.

Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

formatting link

Reply to
Vladimir Vassilevsky

Just nitpicking, but you should also consider the transmission medium velocity factor, so for ordinary PCB materials, such as G10, the full wavelength at 200 MHz would be about 100 cm.

Paul

Reply to
Paul Keinanen

What would I expect to see on a clock line if it were being clocked at its resonant frequency and I put a scope on the various locations along it?

Reply to
rickman

job security , Luddite ! make it sound difficult ! You Wizard ( Vladimir Vassilevsky) .

-------------------------------------------- I am an EE , so its simple to see xmission lines everywhere . Learn transmission line theory and termination and you are half way there ! Transmission lines do not radiate EMI . If you do a good job and , then terminate and it can't radiate even at thousands of watts . It cancels ! The opposite of a Xmission line is an antenna ( im Ham ,KC7CC) Antennas radiate by coupling 2 conductors . Ground planes are Fiction .

------- Notice coils on all mainboards ? bad idea , but they can't design xmission lines , they arent engineers . I see lots of errors on computers. Like CAT5E cable , take all but one pair out and it will work better !

ha ha ha ha .

Have a serious cable problem ? Switch to CATV ! It is much higher bandwidth than CAT5E !

EMI ElectroMagneticInterference .... Permalloy 80 will waste and attenuate magnetic flds . But copper screen will not see mag flds ! It has not ferromagnetic ability . Its fortunate that RF is high enough that the holes in copper screen are smaller than 1/4 wave !! This is what stops the RF . Most trouble is from conducted , not radiated EMI . Hand held scope with a differenciating cap on input . The ONLY thing getting thru will be the higher freqs and no grnd loops ! Skin effect allows it only one mil surface while the same wire may have 60 hrz , 250 amps current ( uses .3" (?) ) , so you'll not see any of the line noise .

You will learn better engineering as a Ham Radio operator than

4 years in school ( EE ) !

Vladimir Vassilevsky wrote:

Reply to
werty

Not same caps .

0 .1 MF dont oscillate like a 10 MF .

About 4 ohms series to all large filter caps .

I use a MFJ209 antenna tuner to "Dip" circuits . And text SRF in caps . A coil and cap form a circuit at the end of the 209 that makes a 3:1 SWR on my meter , anything that absorbs RF from the 209 , will read a dip on meter . As the Cap oscillates from the 209's excitation , it draws down the bridge . Most oscillate below 10 mhz . But with a 4 ohm , your ckt sees nothing . I want to put a microProcessor on it and make it test everything electronic . Because after all , it is hard to understand what is happening on a RF bridge , the MP announces it to you ! [ Change in Z at 17 feet , possible kink in the cable . R U sure termination is 75 ohms ] Thanks for the reminder , i used the wrong terminator !

______________________________________________

If PCB traces are to high inductance ( too narrow ) , you'll need lots of storage ( 0.1MF ) . But sometimes , a delay in the Supply line helps . At Microwave freqs , where loss is tremendous , inductors are everywhere on supply lines . Sometimes they even use 1/4 wave shorted stubs to stop RF from taking an unwanted path but it "tunes" the circuit to only one freq . These stubs are so small you overlook them sometimes !

__________________________________________________

rickman wrote:

Reply to
werty

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.