i.mx233 DDR Routing

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

i.mx233 DDR Routing

Jump to solution
932 Views
yanir
Contributor II

Hi, I am developing a board using the i.mx233 based on the EVK. I have implemented the DDR portion of the EVK schematic exactly and I have some routing questions. I have been reviewing the PCB layout guidelines app note provided by Freescale and the Micron app note and I see some differences. I hope someone can shed some light on this.

1) The Freescale app note stresses the importance of keeping the DQ,DQS DQM and Clock traces shorter so that 30% of the signal edge appears on the trace.

They use an equation:

Trace Length ≤ (0.3 × Rise/Fall Time × 15 cm/ns).

And the result is (based on FR4 and a rise/fall time of 5ns) 2.25CM or 984mil.

They then go on to say that routing multiple DDR devices slows down the rise and fall time so the trace length can be longer.

The Micron data sheet wants you keep traces <= 2inches for series termination. Since this is un-terminated can I use the same guide line? Why is this such a large difference compared to the 1 inch( 2.25cm) length restriction suggested by freescale.

I am implementing two DDR devices as in the EVK, how can I calculate my tracelength based on the formula above? I see in the gerbers of the EVK that they are:

Controller to Via 1265mil

Via DDR #1 291mil

Via to DDR #2 158mil

In the EVK they seem to have matched all of the first segment traces very closely to 1265, and there is more variability in the branch traces. The sum of all of these traces is 1714 mil with and the length of a full connection to a DDR module is 1423 to 1556 for this particular trace.

I can do the same, but I would like a rule to go by, like keep my traces 1500mil +/- 25mil.

I tried looking up rise and fall times in the micron data sheet but I could not find any spec for that. Does anyone know how to determine this value?

2) The EVK implements an un-terminated system. The micron data sheet provides a lot of routing guidelines but mostly in reference to series termination or parallel termination. I would prefer to stick with un-terminated as I cannot simulate to determine the resistance values. It obviously works for the EVK, should I be concerned about this?

Thanks in advance for your help.

0 Kudos
1 Solution
641 Views
RaymondChoi
Contributor IV

Hi Yanir,

I reviewed the MX23 AppNote (AN3989) that the equation (l<=(0.3*rise time*15cm/ns)) is correct.

For the <2-inch rule-of-thumb in Micron document, pls consult Micron representatives.

Best regards

Raymond

View solution in original post

0 Kudos
1 Reply
642 Views
RaymondChoi
Contributor IV

Hi Yanir,

I reviewed the MX23 AppNote (AN3989) that the equation (l<=(0.3*rise time*15cm/ns)) is correct.

For the <2-inch rule-of-thumb in Micron document, pls consult Micron representatives.

Best regards

Raymond

0 Kudos