SDRAM Layout for newbies

[NivagSwerdna]

I am designing a simple expansion board for my FPGA dev kit which amongst other things has...

A 256Mb SDRAM device AS4C16M16SA-6TCN which is TSOP-54 II package.  I need to route these pins to a 2x20 pin connector.

Now I realize 166MHz/5ns might not be considered high for some but I would like to make a reasonable effort of this...

I am planning to...

Try and keep traces roughly the same length (hard given the device and connector are likely to end up side by side) and keep the clock traces around median length.

Is there anything else I should know?  Any opinions on routing under the device versus going around?

Any suggestions for reading on the subject appreciated.

(I'm using DipTrace but that is probably inconsequential)

Thanks in advance

[stmdude]

I'm by no means an expert, but I have designed a few (working) boards with 166MHz SDRAM and an MCU.

So, for my first board, the process went something like this:
1. Figure out the wavelength of 166MHz
2. Realize that it's close to six feet
3. Laughed at myself for being worried

Seriously, I just kept the SDRAM close to the MCU, and kept all the traces as short as I could without needing any vias. (I.e, I "went around" instead of under)

Everything is routed on one side of the PCB. I have a few vias, but they're for VDD and GND.

As for reading recommendations, while it doesn't mention SDRAM at all, I'd wholeheartedly recommend Toradex Layout design guide since you're starting to approach high-speed stuff.
It's free to download here: https://docs.toradex.com/102492-layout-design-guide.pdf

[Bassman59]

I'm by no means an expert, but I have designed a few (working) boards with 166MHz SDRAM and an MCU.

So, for my first board, the process went something like this:
1. Figure out the wavelength of 166MHz
2. Realize that it's close to six feet
3. Laughed at myself for being worried

It's not the frequency of the signal, it's its rise time that matters for signal integrity.

[stmdude]

It's not the frequency of the signal, it's its rise time that matters for signal integrity.

True, but I'm pretty sure the frequency matters when it comes to length-matching. But, as I said, I'm no expert.