I2C problem

[amr nabil]

Hi all,

We are developing a tablet board which has android platform during studying your "Advanced PCB Layout Course" .

we face a very complex problem in the capacitive touch screen so we need your advice

our touch is connected with processor via I2C

the problem is (while charging only) capacitive touch detect fake touch in wrong positions which is caused by noise as illustrated in the following pictures

we need to know how to deal with this case? How to fix this problem and eliminate these fake points?

please see attachments

[kahlenberg]

It can depend on many many factors. It is almost not possible to make suggestions only by looking some osciloscope pictures. But let me gues it:
Charging circuit has a "big" inductor and it is close to I2C signals. Charging circuit drives the inductor although at low frequency such as 20KHz but it is powerful relative to I2C signals. Therfore I2C signals are getting noise from this inductor.
Either redesign charging circuit layout or place I2C signals far away from charging circuit.
Can you send a picture of the board?

[cometa]

what is the value for pull up resistor? for high speed i use 3.3k 3.9k 4.7k....

[robertferanec]

As kahlenberg pointed out the charging circuit - I would ask, is the charger step up regulator? These can make some noise. The noise can be picked up by the cable connected to touchsreen or what sometimes can do problems is also noise on GND and poor grounding of display + touchscreen.

Is the touchscreen working ok when tablet is dissasembled and display is placed away from the board?

​PS: Also cometa has a good point, some reference designs are using high pull up resistors.

[amr nabil]

I hope this picture of our PCB help

[Muhmmad Ahsan]

Hello Arm
Check you ceramic capacitor which is connected with LVDS connector .
Check you Cable which is connected with Board
and then check layout ,may be you routed I2C near to High Frequency signal as Mr,Feranec said ,

[amr nabil]

As kahlenberg pointed out the charging circuit - I would ask, is the charger step up regulator? These can make some noise. The noise can be picked up by the cable connected to touchsreen or what sometimes can do problems is also noise on GND and poor grounding of display + touchscreen.

Is the touchscreen working ok when tablet is dissasembled and display is placed away from the board?

​PS: Also cometa has a good point, some reference designs are using high pull up resistors.

the power supply IC is AXP209
also a question please, how to make a good grounding for display+touch screen???
during disassembly the problem still as it is..
we attached some photos of the Board please kindly check

thanks

[cometa]

i was having a similar problem with sd-card, and i solve follow this steps:
1-small coil for power sd card
2-change the pull resistor I used under 4k
3- for next pcb i never routed (for long distance) digital signal on top or bottom, always inside and like this is shielded by ground plane from top and bottom
I hope this can be useful for you,

[amr nabil]

what is the value for pull up resistor? for high speed i use 3.3k 3.9k 4.7k....

thanks a lot for your help, we use 2.2 K ohm

[amr nabil]

i was having a similar problem with sd-card, and i solve follow this steps:
1-small coil for power sd card
2-change the pull resistor I used under 4k
3- for next pcb i never routed (for long distance) digital signal on top or bottom, always inside and like this is shielded by ground plane from top and bottom
I hope this can be useful for you,

1- I am not sure how we can connect this coil
2- we use 2.2 Kohm
3- our PCB is 6 layers with this arrangement "1-Top(signal) ,2-GND ,3-Signal(POWER traces also) , 4-VCC , 5-GND ,6- Bottom (signal)"
we routed the I2C on the top and bottom layers and it is long really

[robertferanec]

amr nabil I apologize for late replay, I was traveling (I was speaking at CDNLive conference).

From your pictures, I noticed, that you are using a level translator on the touchscreen - what is it translating? Also, what is the circuit on the top of the translator, is that the boost converter?

[amr nabil]

dear Robert,
thanks a lot for your reply, the translator is for different voltage levels between CPU(3.3V) and the capacitive touch(1.8V).
The boost converter is for the LCD boosting from 5V to 24V for the backlight..

we actually found the most effective reason for the problem is the quality of the charger..

this link may help others who may face the same problem

http://www.righto.com/2014/05/a-look-inside-ipad-chargers-pricey.html#ref11

[cometa]

right now, I'm working on a "fast charger" project and this document will be very useful for me
thanks

[robertferanec]

"we actually found the most effective reason for the problem is the quality of the charger.."
- I heard about this kind of problems. Maybe try to add some filtering on the power input, that could also help.

"The boost converter is for the LCD boosting from 5V to 24V for the backlight.."
- Problem could be theoretically also noisy signals coming from the capacitive touchscreen, did you check this option? Especially if on the other side is the backlight converter. The board could be running on the edge and if you add the noisy power charger, then the problems combined together could become visible.

PS: thank you very much amr nabil for sharing the solution with charger