Slew rate failure while DDR3 read burst validation

[titusstalin]

Hi All,

Currently I'm involving DDR3 validation.

Here is the information about my setup.

Processor : MCIMX6U6AVM08AC

DDR3 Memory : MT41K128M16JT-125AITK

DDR3 Clock frequency : 400MHz

ARM Frequency : 800MHz

Chip Select used : CS0

No of chip’s DDR3 chips used : 4 no’s connected in fly by topology

Oscilloscope : 16 GHZ DPO

Probing Points on the Processor:

Clock (Channel C1) DDR_CLK_0 & DDR_CLK#_0

Data Strobe (Channel C2) DDR_SDQ0 & DDR_SDQ#0

Data 0 (Channel C3) DDR_DATA0

Observations:

We have done the below methods to make slew rate (SRQdiff & SRQse) pass in DDR Read Burst,

1. Modified the ODT Termination with different values like 30 ,40 , 60 Ohms etc
2. Modified the Drive strength of the DDR in MR1 register for DDR3
3. Modified the Drive strength of Processor DQS , DQ which we are probing.

PS: We have calibrated the board with DDR3 stress test tool.

We are getting failure DQS slew rate measurements.

Can any one please help us to solve this problem.

[Yuri]

Hello,

  You may try simulation a) on die and b) on pin.

And quite acceptable waveforms on die look not so good on pin,

just as yours. It is important to have proper signals on die.

Have a great day,
Yuri

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[hermann_ruckerb]

Hi,

according to the picture you do have a reflection on your measured signal. Most likly due to your testpoint. So this is a measurement artefact.

Are you using any interposer and do  any de-embedding ?

But I have to admit, that I do not understand the Report from the Tek scope.

standard slew rate on DDR2 is still 1v/ns. not sure what the 5V/ns and 10V/ns really stand for and I'm also not sure why some of them are pass if you do not reach the low limit ..

But I have to admit, that I usually work with the tools from other scope vendors.
I would not expect that Tek does have better tools, so expect also wrong measurements ..

Hermann

[titusstalin]

Hi Hermann,

Thanks for your reply.

I don'see any wrong pass/fail here, I think that is -ve values right so more negative value means low and less negative value means high right?

So it seems measurement are correct.

PS: We able to make write burst pass.

Any recommendations ?

Thanks for your support.