DDR calibration issue on a custom iMX6Solo board

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

DDR calibration issue on a custom iMX6Solo board

Jump to solution
6,198 Views
michael_novotnu
Contributor II

#Mark Middleton 

I'm trying to make DDR calibration on my custom iMX6 board. DDR memory chip is IS43TR16128B-125KBL.

I used MX6DL_SabreSD_DDR3_register_programming_aid_v2.2.xlsx as a reference to create registers settings.

For calibration I tried DDR_Stress_Tester_V1.0.2, DDR_Stress_Tester_V1.0.3 and ddr_stress_tester_v3.00

version 1.0.2 and 1.0.3 got the same result as in my comment

version 3.0.0 result:

============================================
        DDR Stress Test (3.0.0)
        Build: Dec 14 2018, 14:12:28
        NXP Semiconductors.
============================================

============================================
        Chip ID
CHIP ID = i.MX6 Solo/DualLite (0x61)
Internal Revision = TO1.3
============================================

============================================
        Boot Configuration
SRC_SBMR1(0x020d8004) = 0x00000000
SRC_SBMR2(0x020d801c) = 0x20000001
============================================

ARM Clock set to 800MHz

============================================
        DDR configuration
BOOT_CFG3[5-4]: 0x00, Single DDR channel.
DDR type is DDR3
Data width: 32, bank num: 8
Row size: 14, col size: 10
Chip select CSD0 is used
Density per chip select: 512MB
============================================

Current Temperature: 44
============================================

DDR Freq: 396 MHz

ddr_mr1=0x00000004
Start write leveling calibration...
running Write level HW calibration
  MPWLHWERR register read out for factory diagnostics:
  MPWLHWERR PHY0 = 0x00000000

HW WL cal status: no suitable delay value found for byte 0

HW WL cal status: no suitable delay value found for byte 1

HW WL cal status: no suitable delay value found for byte 2

HW WL cal status: no suitable delay value found for byte 3
Write leveling calibration completed but failed, the following results were found:
    MMDC_MPWLDECTRL0 ch0 (0x021b080c) = 0x001F001F
    MMDC_MPWLDECTRL1 ch0 (0x021b0810) = 0x001F001F
Write DQS delay result:
   Write DQS0 delay: 31/256 CK
   Write DQS1 delay: 31/256 CK
   Write DQS2 delay: 31/256 CK
   Write DQS3 delay: 31/256 CK

Error: failed during write leveling calibration

It is my first experience with iMX6 processors and DDR calibration.

My .xlsx file and DDR schematic are attached.

Thanks

Labels (2)
1 Solution
5,418 Views
michael_novotnu
Contributor II

Found mistake in board assembling, incorrect value of  C89 and C91 (signals DRAM_SDQS0 and DRAM_SDQS1)

After changing to correct value, board works as expected.

View solution in original post

0 Kudos
8 Replies
5,429 Views
michael_novotnu
Contributor II

Hello,

I faced with the similar issue on our new board with imx6ULL and MT41K256M16TW-107.

 

One of our device with imx6ULL and MT41K256M16TW-107 works correctly.

But other one with the same imx6ULL and MT41K256M16TW-107 does not work.

Both devices DDR schematic are the same.

Here is voltage at power up:

+1V35_DRAM voltage is 1.35V

+VREF_DRAM voltage is 0.68V

R91 (SDCKE0) voltage is 0V

R92 (DRAM_RESET) voltage is 0V

In our design we have 32kHz and 24MHz crystals. I checked both signals.

Screenshot_2021-01-28_18-46-09.png

 

DDR Stress Test can not calibrate DDR, see log:

============================================

DDR Stress Test (3.0.0)

Build: Dec 14 2018, 14:13:23

NXP Semiconductors.

============================================

 

============================================

Chip ID

CHIP ID = i.MX6 UltraLiteLite(0x65)

Internal Revision = TO1.0

============================================

 

============================================

Boot Configuration

SRC_SBMR1(0x020d8004) = 0x00000000

SRC_SBMR2(0x020d801c) = 0x00000001

============================================

 

ARM Clock set to 528MHz

 

============================================

DDR configuration

DDR type is DDR3

Data width: 16, bank num: 8

Row size: 15, col size: 10

Chip select CSD0 is used

Density per chip select: 512MB

============================================

 

Current Temperature: 45

============================================

 

DDR Freq: 396 MHz

 

ddr_mr1=0x00000000

Start write leveling calibration...

running Write level HW calibration

MPWLHWERR register read out for factory diagnostics:

MPWLHWERR PHY0 = 0x00000000

 


HW WL cal status: no suitable delay value found for byte 0


HW WL cal status: no suitable delay value found for byte 1

Write leveling calibration completed but failed, the following results were found:

MMDC_MPWLDECTRL0 ch0 (0x021b080c) = 0x001F001F

Write DQS delay result:

Write DQS0 delay: 31/256 CK

Write DQS1 delay: 31/256 CK

 

Error: failed during write leveling calibration

 

The log output is similar that I got early with imx6DL.

Can someone advise me what to check?

0 Kudos
5,419 Views
michael_novotnu
Contributor II

Found mistake in board assembling, incorrect value of  C89 and C91 (signals DRAM_SDQS0 and DRAM_SDQS1)

After changing to correct value, board works as expected.

0 Kudos
5,814 Views
TheAdmiral
NXP Employee
NXP Employee

Hi Michael,

You have our latest stress test, that is good.

And what the test is telling you is that the board cannot even complete a Write Leveling routine, which requires the least amount of things to go wrong in order to complete.

There is something fundamentally wrong.

Have you confirmed that +1V5_DDR actually reads 1.5V?

Have you measured the +DDR_VREF voltage on C182? Does it read 0.75V?

What is the voltage on R81 (SDCKE0)?

What is the voltage on R82 (RESET_B)?

Is the layout in Allegro format? Would you be open in sharing it?

One last question: Does your design use a 32kHz crystal, or do you rely on the internal ring oscillator?

5,814 Views
michael_novotnu
Contributor II

Hi Mark,

Thanks for your help.

I can share DDR layout only next week.

+1V5_DDR voltage is 1.5V

+DDR_VREF voltage is 0.75V

R81 (SDCKE0) voltage is 0V

R82 (RESET_B) voltage is 0V

In our design we have 32kHz and 24MHz crystals.

pastedImage_1.png

I checked RESET_B signal on oscilloscope and found that is never goes to high so memory is always in reset state.

I have two assembled board and they have the same behavior.

0 Kudos
5,814 Views
michael_novotnu
Contributor II

Found that NVCC_LVDS2P5 is required for DDR controller on iMX6, in our design it is off.

I think it will fix our issue.

0 Kudos
5,814 Views
TheAdmiral
NXP Employee
NXP Employee

Hi Michael,

Yes, without power to the pre-drivers, the DDR pins don't work.

This is documented in the HWDG:

HWUG.png

If you have to redesign the board in order to fix this problem, I would highly recommend downloading the HWDG from the NXP website and go through the checklist to make sure you don't get hit up with any other gotcha's.

i.MX 6Solo Applications Processors | Single Arm® Cortex®-A9 @ 1GHz | NXP 

Cheers,

Mark

5,814 Views
michael_novotnu
Contributor II

Hi,

I fix power issues on my board and DDR calibration and tests passed successfully.

Thanks for help!

0 Kudos
5,814 Views
TheAdmiral
NXP Employee
NXP Employee

Hi Michael,

I'm happy you were able to fix your board so quickly.

Good luck with your project.

Cheers,

Mark

0 Kudos