When we power down our product VCC33RTC is drive to ground. We don't use a RTC backup battery.
VCC33RTC is directly connect to pin V3P3_RTC from DA6011 (DIALOG).
At now we observe the issues below:
- Some (10/3000) boards doesn't start proprely when we plug main batteries. In fact they start properly but the LCD backlight stay black.
- They start properly when we
- 1 / change tunnel creek ( E620) or
- 2 / add a 10K pull-up to 3.3V on signal LVD_BKLT_PWM ( pad P1) and LVD_BKLT_EN (pad N4) or
- 3/ start the board with the ON/OFF button ( PWRBTN_B pad J11 DA6011)
- the problem is reproducible
1/ need signals LVD_BKLT_PWM ( pad P1) and LVD_BKLT_EN (pad N4) a 10 K pull-up ?
2/ if yes what is the risk having this signal without pull-up ?
3/ If a product start properly today what appends later?
Please fill free to contact me if you need more details. This issues are priority to us.
Here at our company Eltan we have done customizations of Phoenix BIOS for a lot of TunnelCreek projects.
Maybe we are able to create special debug BIOS to detect difference of both GPIO_SUS signals and/or backlight behavior of the video BIOS between good and defective boards. Please contact us directly by email if you are interested.
According to the datasheet both GPIO_SUS signals do not have strap functionality, so you are free to add pull ups without any problems. In case adding the pull ups will solve your problem, I recommend to implement also the pull ups on the other working boards for future precautions.
I expect the root cause of your problem is that the RTC crystal requires 2 seconds to run properly after power up of VCC33RTC. You might confirm by measuring with low pF probe on RTC crystal signals. The initialization of the backlight by the video BIOS might be moved away from bad RTC clock time frame by adjusting the boot BIOS sequence.
Backlight is a "sticky" subject. If you are using panels with CCLF backlights *the ones that need the power inverters, it could be that the in-rush current for the power supply startup is too high for you power supply and by changing the timing a bit, will allow it to work or not. You could try changing the T1- T5 timing for the panel sequencing to see if giving the backlight more or less time changes anything. My guess is that putting pullups on those signals is manually changing the startup timing and that is what changes your sequence.
Many panels are VERY touchy about their startup timing (Chi Mei aka CMI panels require very strict timing and MUST be operated in spec- others are more lax on their power sequencing timing in my experience).
Alas, I am a graphics software person so I cannot speak to what the H/W ramifications of your signal pullups may be doing hardware-wise.
Hope this helps.