EMI problem with Motorola MPC8527 board

[BenKropp]

EMI problem

Hi,
I have a problem with my board. I can't pass the conformance check because of the high emissions at 250 and 450MHz. I'm using a Motorola MPC857@100MHz (50MHz busclock).
Has anyone experience with such problems or can tell me where to locate the problem or what to do to reduce it.
I want to avoid using a shielded housing but it seems it's the only chance.

 

[flatulent]

loop

You take a coax cable and put a small loop on the end. Put the other end on a spectrum analyzer. Move the loop over the board.

 

[wolf69]

after finding the source of the problem as indicated by flatulent
maybe you can attenuate the problem placing RC circuits on the high speed lines slowing down rise-fall signal time
.. not so nobile act
..but maybe it's right

Bye!

 

[VSWR]

Re: EMI problem

If you're lucky, adding ferrite cores on all attaching leads will cure the problem (or at least get you below the radiated emission specs). Use core material designed for the appropriate frequency range. There are very handy snap-on cores available today.

Careful PCB design is also important for high speed logical circuits. Do you use multilayer PCB, having a designated unbroken ground plane?

 

[mr_ghz]

Hi BenKropp,

Often such emission is caused by the current consumption of a device
using the specified clock. Because it seems to be an overtone of the bus-
clock and not of the CPU itself, i guess a SDRAM. Do you have one?
Are the decoupling-caps close enough? Is the tracing to/form the caps
correctly?

Bye

 

[BenKropp]

Hi,
thanks for all the replies. I have 2 planes (power and ground) and also like motorola adviced a seperate isle (using an inductor) for the PLL and a lot of decoupling capacitors. I have 2 SDRAMs.

@flatulent
I already did a measurement with our DSO (FFT capability) and an active probe with a short loop wire, formed to a ring, connected to it but I'm not sure if I do measure the radiated waves or only magnetic fields. I'm not very experieced with this, what do you think?

@wolf69
I've foreseen serial termniations for the most signals (and additionally parallel for the clocks) and did the pcb layer stackup accordingly (~50R impedancy), but which signal is the bad guy (-:

@GaAs FET
adding ferrites helps really to reduce the 50MHz emissions. I put it at the supply cable and everything was fine. But for all lines it's pretty expensive ... I'll do it if nothing else helps

@mr_ghz
good hint, I first thought it's the CPU/core current but it does not fit with the overtone. I'll concentrate on the SDRAMs.

 

[flatulent]

magnetic field

Radiation has magnetic and electric fields. You only have to sense one of them. The osciloscope is not sensitive enough to find the radiation. Since you know the frequencies, a radio receiver will also work.

The loop being small will allow you to identify which lead on the PC card is doing the radiation.

 

[BenKropp]

ok, thanks flatulent.
The reason why I asked for the fields was that I always got peaks in the area around inductors. Maybe I have to increase the distance to only measure electromagnetic waves ... what distance would you recommend?
Today I had a try with an antenna pre-amplifier, an oscilloscope (spectrum analyzer) and a wire antenna. Not too bad .... I saw the expected peaks and could influence them with changing the bus frequency.
Tomorrow I'll get a scanner antenna. Let's see what happens

 

[flatulent]

inductor problem

The reason for using a small loop is to localize the source of radiation. It looks like your inductors on the board are part of the problem.

Try putting shunt capacitors on the logic lines to slow the rise and fall times. This is a crude method, but will show you which lines are radiating.

Try using shielded inductors.

 

[BenKropp]

Hi,
thanks, I found 2 peaks in the area around the clock output of the CPU using the loop cable. Seems that the Motorola boosts out the busclock very powerful. There's a register to select "half strength busclock driver" and this reduces the 450MHz emissions very considerable. Let's see if I can now pass the test.
Besides, the scanner antenna and the rf pre-amplifier work very good. I can now see changes depending on what I try to reduce the emission.