PCB routing & signal integrity issues at 50-100mhz?

Hey guys,

Just a little background, I’m an engineer doing mechanical design and software engineering. I’ve been working with AVR projects to help bridge the gap between the software and the machine. Now I want start laying out my own ARM based boards with LPC2000 series chips.

When you guys are laying out a board for something like a LPC2148 how much concern is there with respect to signal integrity issues around 50-100mhz? I understand the basics as far the concepts of reflection and cross talk. However, I don’t have much of a gut feeling for working with it at the design level. Trial and error design, simulation? Which tools help me get my hands dirty here?

All the high speed stuff is inside the chip, you don’t have to worry too much about it. Just decouple every supply pin and use good PCB design practice. I’d use a four layer PCB.

Leon

tybeede:
Hey guys,

Just a little background, I’m an engineer doing mechanical design and software engineering. I’ve been working with AVR projects to help bridge the gap between the software and the machine. Now I want start laying out my own ARM based boards with LPC2000 series chips.

When you guys are laying out a board for something like a LPC2148 how much concern is there with respect to signal integrity issues around 50-100mhz? I understand the basics as far the concepts of reflection and cross talk. However, I don’t have much of a gut feeling for working with it at the design level. Trial and error design, simulation? Which tools help me get my hands dirty here?

The internal clock, as said above, is a phase-locked-loop frequency multipler from a much lower frequency external crystal.

great, I was looking for the added level of confidence going into the 32 bit world. All of the signals on the board I fabricate will be low frequency relative to the signals contained inside the actual micro package. The clock signal would be the most likely to be sensitive but it runs at a multiple so even it is relatively safe. More layers is always the “better” way if your fabrication budget and skills can swing it, right? Smaller footprint?

tybeede:
great, I was looking for the added level of confidence going into the 32 bit world. All of the signals on the board I fabricate will be low frequency relative to the signals contained inside the actual micro package. The clock signal would be the most likely to be sensitive but it runs at a multiple so even it is relatively safe. More layers is always the “better” way if your fabrication budget and skills can swing it, right? Smaller footprint?

take a look at the many ARM7 and ARM9 small boards on the market. Olimex, FutureLec, ETT, many more. many are 2 layers.

great… only problem is now I’m being tempted by the STM32. The more I learn about the Cortex-M3 the more it really feels like the direction with momentum at the moment. Who knows, maybe it is marketing buzz but I like what I see. I also think the STMicro website is a little better than the NXP site.