DS90UB927Q-Q1: The block screen issue

Hi Zhang,

Thank you for providing the schematics and registers.

Zhang Liang said:

We read the register of  DS90UB927Q-Q1, the 0xC4 register is 0x78, but the register is 0x68 when the black screen is happen,

Register 0xC4 is the HDCP_STS register. This register is not public in the UB927 datasheet because HDCP is not present on the UB927. Register 0xC4 is public on the UH927 datasheet.

If you are curious, the change from 0x78 to 0x68 is because of "Downstream Hot Plug Detect"; however, because HDCP is not on this device, this register is not valid.

Zhang Liang said:

there are the random black screen when the system running,

Was the lock status on the 928 monitored when the screen went black? For the register dump you provided, was this taken after the screen has went black?

Some further questions on black screen

• Is this present on all systems?
• Is this easily reproducible?

Best,

Jack

Hi Jack,

Thanks for your professional comment.

Jack Scherlag said:

If you are curious, the change from 0x78 to 0x68 is because of "Downstream Hot Plug Detect"; however, because HDCP is not on this device, this register is not valid.

The 0xC4 is 0x68 because we read the register, and if this is not valid, please kindly review the register print information and advise if there are any error.

Jack Scherlag said:

Was the lock status on the 928 monitored when the screen went black? For the register dump you provided, was this taken after the screen has went black?

The print information is got when the screen went black;

Jack Scherlag said:

• Is this present on all systems?
• Is this easily reproducible?

The present on all systems and this is random, we have no sure when the block screen happen, thanks.

Hi Jack,

please refer to below issue description:

1.When the screen is went black,the control channel such as I2C,GPIO works normally which is reflected in the screen touch and backlight work normally.
2.The PIN27 LOCK status, PIN28 PASS status,PIN44 CMLOUTP and PIN45 CMLOUTN of 928 always output high levels regardless of whether the screen is black or normal display.
3.It’s no information printed when the screen is went black.
4.This problem is present on all systems.
5.This problem is easily reproducible.
6.When the screen is went black,we try to reset the 927 and 928 bay the register,but no improvement.
7.When the screen is went black,we power down the 927,reboot it and re-initialize in ,but no improvement.
8.When the screen is went black,we power down the 928,reboot it and re-initialize in ,it return to normal display.

9.we think the reason for this problem is that 928 works abnormally.
10.We set the adaptive equalizer to a long Cable Mode(LCBL) by the register 0x35,but no improvement.

Hi Jack,

Please review above and provide your comment that how to solve this issue, thanks.

Hi Jack,

We still has no idea to solve this issue, and we modify the 0x04 (Mode select) = 0x90 that there are no improve the screen, thanks.

Hi Zhang,

Thank you for confirming. In order to figure out where to go next, we will need to use the 927 PATGEN to identify if this issue is related to the display output or the CPU input. If you let me know the display timings, I can help generate a script. How did you confirm that the CPU signal was normal for the system with black screen?

Also, are you able to provide register dump for good and bad 928?

Best,

Jack

Hi Jack,

Sorry, we write the 0x04 (Mode select) = 0x90 and the black screen issue is solved for the 1024*600 screen, and we will test the 1280*800 screen,

many thanks for your kindly support;

BTW, can you explain whey we need to the set the 0x90 that we can solve this issue, that we can avoid such issue next time, thanks.

Hi Jack,

This is very useful for our design, many thanks for your support.

BTW, we have the other project that it is also has the black screen(12 inch screen) and please kindly review below discription for this issue:

1. Our tablet will have a black screen in the course of use, and the black screen cannot be automatically restored.
2. When the screen was black, we used oscilloscope to measure LVDS and I2C signals of 927 and 928, and the test results were normal.
3. When the screen was black, The PWM backlight control signal has no output on the 928 but its input on 927 is normal.
4. It is precisely because the 928 does not output PWM backlight signal, resulting in the screen backlight is not bright. At this time, after the backlight control signal is provided to the screen externally, the screen can return to normal.This experiment shows that the LVDS channel is normal.
5. Why the control channel is unnormal?

Hi Zhang,

Can the black screen issue be easily recreated? Does it always happen at the same time?

Zhang Liang said:

It is precisely because the 928 does not output PWM backlight signal, resulting in the screen backlight is not bright. At this time, after the backlight control signal is provided to the screen externally, the screen can return to normal.This experiment shows that the LVDS channel is normal.

My theory is that there is an unintentional reset of either the 927 or 928 which removes the GPIO mapping. What GPIOs are used for the PWM backlight signal? What's the frequency of the signal?

One way to prove whether or not there is an unintentional reset is to dump the registers before and after the black screen issue. The LVDS channel requires no additional programming so it is not surprising that it easily returns to normal.

Best,

Jack

Hi Jack,

We are use the GPIO1 and our PWM frequency is 40KHz, and plesae advise if it is error? thanks.

Hi Zhang,

Can you see if there are register differences in the 927 and 928 when the screen is normal and when the screen is black? The PWM frequency is supported. I do not have any concerns on the frequency.

If the LVDS output of the 928 is normal and you can read the registers of the 928 remotely, this indicates that there is not an issue with the channel but the GPIO configuration.

Best,

Jack

Hi Jack,

When the screen blackening, the registers 0x0a, 0x0c of 927 and the registers 0x02, 0x0d, 0x0e of 928 are changed, please refer to below:

i2cset -f -y 5 0x0c 0x04 0xb0     //Clear Counter

i2cset -f -y 5 0x2c 0x1d 0x29

i2cset -f -y 5 0x2c 0x1e 0x95  

and then the screen is on, and after In a few seconds, the screen is block and the register is back to the error.

Please refer to the pricture and then provide your comment, thanks.

Hi Zhang,

Thank you for pointing out the registers that are different before and after black screen. If we look at the registers in the 928 that are different, the values we read after black screen are the reset values. For example, 928 register 0x1D is set back to 0x20.  This indicates that the 928 is reset sometime during operation.

The following can cause device reset

• PDB pin brought low
• I2C command to 928 register 0x1
• 928 power dropping below minimum

Here are some next steps we can take

• Probe the PDB pin and see if it is steady during operation
  • The minimum high level input voltage is 2.0v
• Probe the LOCK pin and see if it is steady during operation

Are there any device resets issued by the SoC/MCU during operation? Are the power up requirements met by the hardware?

Best,

Jack

Hi Jack,

When the screen is black ,the register 0x0a of 927 is error;

What mean of the values in register 0x0a of 927 , and does each value have a corresponding error type 2;

The PDB PIN and the power supply is steady during operation;

Hello Zhang,

The US team is currently out of office due to the US public holiday and will return on 5/28. Thank you for your patience at this time 

Best Regards,

Casey

Hi Zhang,

Zhang Liang said:

What mean of the values in register 0x0a of 927

927 register 0x0A (CRC Errors) indicates that there were bit errors on the back channel. A brief device reset will interrupt back-channel operation and cause the 927 to count this as bit errors.

Zhang Liang said:

The PDB PIN and the power supply is steady during operation;

Can you please attach scope captures so I can confirm?

Jack Scherlag said:

• Probe the LOCK pin and see if it is steady during operation

Are there any device resets issued by the SoC/MCU during operation?

Have both of these questions been investigated?

For this system, is this done on a lab bench or in the vehicle? Are there any other systems operating nearby the 927-928? Let's try to rule out any external influence on this system.

Best,

Jack

Hi Jack,

We Compared the black screen and before black screen register, the 0x1d, 0x1e, 0x35 & 0x3b of DS90UB928Q-Q1 is charged, 

and we check the register of 0x35 & 0x3b,please advise if this is the AEQ issue?

We also test the power rail design:

1. We test the DS90UB928Q-Q1 waveform of PDB and Lock pin, the PDB has much noise interference(refer to the waveform yellow curve), and the lock 

signal will be dropped about 10+ms when the black screen block is happen;

2. The black screen  is probability of occurrence to rise when the operator temperature is rise;

3. We push up the PDB by oursalves, the register is same to the phenomenon of automatic black screen;

4. Is there any softwave solution to solve the issue for the power crosstalk to causing black screen