Issue link: https://iconnect007.uberflip.com/i/1369942
MAY 2021 I DESIGN007 MAGAZINE 15 Olney: I guess I'm one of those crazy people because I'm a PCB designer, and the way I see it is everything has to be done properly. Whether it's a really fast DDR4 or a slow 200 MHz DDR, I route them all the same. I route them all to 10 picoseconds delay and it doesn't take any extra time to do that. People say, "Why should I have it so accurate?" Well, if you just do it out of habit every time then your design works perfectly. If it is 125 out, it might work on the test bench, but if you put it in the field and temperature cycle it, it will fall over. To make it reliable and performing efficiently, you need to have it spot on, and it takes little extra effort to do it that way. Hartley: at is exactly what we did on the result. We intentionally misrouted a few of them in the prototype because I just wanted to see how it would function. We straightened it up in the final design, but we didn't go to these crazy lengths to make things ridicu- lously tight the way a lot of people do; we just routed them. We put serpentining in a cou- ple of them to get them all within a reason- able skew margin and they worked perfectly. ey worked perfectly in the field—and this is avionics, where it had to go through tremen- dous temperature cycling. ere are many of these nav systems flying around in the world today, and the last I heard they were all work- ing perfectly. Barry Matties: Did you do any simulation on those, Rick? Hartley: Aer the timing analysis we did sim- ulation to verify that the timing analysis was right, and it said that it was right also. at was kind of a double blessing, and as Barry pointed out, DDR3 is even more forgiving, because it balances the lines. With DDR2, you had to do some amount of branch routing to get things to be timed correctly. With DD3, you don't have to do that. You simply route things in daisy- chain fashion, and it sends out a test pulse to figure out what the line links are, and then calibrates everything to be at the right time. ere's a more sophisticated way to say that, but you get the point. Olney: Also a lot of people think, Rick, that you can't use fly-by routing with DDR2. at's not correct, I have done it before, it just needs you to hard code the clock delays to each chip from the strobe. So, the data is captured at exactly the right time as the clock passes on the fly-by. If you hard code them manually you'll be doing the same thing as your auto-leveling would do in DDR3c. Hartley: Yes, DDR3 is automatic. at's inter- esting. I did know that was possible with DDR2. Olney: Fly-by routing is a lot easier to route than T-topology because you have a lot more room to route around the outside of the chips instead of going straight through them point to point. Rick Hartley