TCA8418: Ghosting under electromagnetic radiation

Hi max,

We are planning to do the test early next week. Will let you know the results. Below are the configurations planned for the upcoming test.

1)Mounted 10k pull up to VCC(1.8V) on all Rows.

2)Mounted 1uF capacitor (Instead of previous 22pF) on every row & column.

3)Mounted 1uF capacitor on every column & 10K pull up to VCC(1.8v) along with 1uF capacitor on every row.

Note:we will be mounting ESD Diode(ESD9X3.3ST5G) on every row and column in all above configurations.

Please let us know your thought on the above plan. Also, let us know if we need to try out anything else?

Thanks & Regards
Gireesh Nair

Hi Max,

As discussed earlier, we had 3 configuration ready for our testing. But we have found that, on the board in which capacitors (1uF) were mounted, the functionality of TCA8418 was effected badly.The observations are as follows,

1)Multiple keys were getting pressed when a single key is physically pressed.(Looks like the keys of same column or row. Need to cross check on it.)

2)Few keys doesn't get pressed at all(This was not observed before mounting the capacitor).

We are assuming that this is because of the huge capacitance on the key and column lines. Let us know your thought on the same.

So, finally we could use only the pull up mounted board for the EMI/EMC testing . Below are the observation when the board with 10K pull up on every rows along with ESD diode(ESD9X3.3ST5G) on every row & column was subjected to radiated susceptibility test.

1)Very few ghost events were registered(Looks like these is in our acceptable range).

But, Once the test was completed,We tried pressing the keys, we observed that ,the key need to be pressed twice to get one interrupt (Example: when we press "1" twice we get one interrupt for "1" & if we press"2" we get the previous interrupt of "1" and so on).

Note:If the board was rebooted, there is no delay observed. Key pressing works fine. until the board is again subjected to radiation.

We were also getting multiple key press issues on few other keys. The multiple key pressed were consistent(Same combination of keys were getting press every time). This issue was observed even after the reboot. But, when we pressed the key after few hours of radiation testing the keys worked fine. 

Hope i din't mess up with the explanation.Please feel free to ask if any clarification required. 

Plan for next round testing:

1)Board with only 10K pull up mounted on every row.(ESD diode will not be mounted)

2)Board with different resistor values(eg: 20K,49.9K etc)  mounted as pull up on each row.

Please let us know your thought / suggestion on the above issue & next round test plan.

Also, we would like to know the maximum line capacitance allowed on TCA8418 IC(would be better if you could suggest us a ESD diode which can be used on the lines).

Regards

Gireesh 

Gireesh,

Thank you for the update. I agree the large capacitance is likely slowing down the active key scanning by such a degree that it is no longer functional - sorry for not realizing this would be the case sooner. I'm glad to hear that the pull-up resistances seemed to improve the ghosting during EMI testing, but the behavior shown after the testing is puzzling to me.

Prior to the reboot (i.e., when it seems there is a one-key-press delay), are the key events being recorded in the event table properly and only the interrupt indication is delayed?

Have you noticed any configuration registers changing values during or after the EMI test? (I assume the parts are configured prior to the test and kept in a static configuration.)

The multiple-press issue after the reboot is a bit strange as well, especially since it seems to go away after several hours. In that time, are any configurations modified? Or are there any other changes to the way the device is operated (i.e., is there a different VCC level, is the duration of each button press different at this point in the test, etc.)? Note also that there are some limitations on the VCC ramp rates in order for the device to properly initialize itself (see Section 10 in the datasheet) - it would be good to double-check that these are being followed and that any pulses on /RESET are above the minimum-required duration.

Regarding the diode selection, I'm not sure what else could be done to improve over ESD9X3.3ST5G. The working voltage could not reduce further since otherwise it would start to interfere with normal operation, and the peak clamping voltage seems reasonably low. The diode is not getting damaged so it does not need a higher power rating, and the parasitic capacitance seems to be low enough to not interfere with key scanning.

Regards,
Max

Hi Max,

*We had chance to check the event table during the delay. we observed that, there was no key event registered in the event table.

*We had configured the IC prior to the test and kept in a static configuration.(I believe TCA8418 will be working in Idle Key Scan Mode prior to test)

*W.r.t Multiple- press issue, No Modification in configuration or device operation is done(Board was just kept shutdown, until we rechecked it).

*We will try out testing without ESD Diode & let you know the result.

We would like to know, what are the possibility or reason for TCA8418 misbehave as in above scenario(In what & all scenarios TCA8418 can miss behave??) .

We are left with no clues. It would be great to have your suggestions based on your past experience.

Regards,

Gireesh

Hi Max,

We did one more round of radiated susceptibility testing, Below are the results.

As discussed earlier we had tested multiple boards with different configurations.

1) Board with 110K pull-up on all rows & ESD(ESD9X3.3ST5G) on all row and column and another board with only 110K pull-up on all rows of TCA8418.

Outcome: we found that the ghost interrupts events registered with 110K pull-up & ESD combination was comparatively lesser than that with only resistor board. But we were able to find a delay of 4 key press once after the radiation test is complete.

Therefore we tried out placing an EMI shield(Was connected to board ground) on TCA8418 IC. We found that the delay was not observed(we tried 2 times for verifying the consistency). But still few ghost interrupts were registered in all the above scenario.

PS: We had used TCA8418EYFPR in our earlier revision of the boards. we had changed to TCA8418RTW in order to avoid the BGA part. we believed that both TCA8418RTW & TCA8418EYFPR are similar except for the ESD protection capability. (We were not finding the issues during the compliance testing with TCA8418EYFPR boards).

Can you figure out why would the change of TCA8418EYFPR to TCA8418RTW should cause such an issue?

As we are in a critical stage to debug this issue. we would like you to recommend different keypad controller IC with similar specifications.

please let us know your thought on the above observation.

Regards,

Gireesh

Hi Max,

Thanks for the suggestion.

But, LM8330 is a 0.4mm pitch BGA part. As informed you earlier, we have a constraint on using BGA part. 

Please suggest a Non-BGA part if any. I will parallelly check for some other alternative too.

Regards,

Gireesh

Hi Max,

Sorry for the delayed response,

We have not found any clue on this issue yet. We have currently paused our debugging this particular issue.
We are planning to restart the debugging a couple of weeks later.

Please do post if you are able to find any clue regarding this issue.

Regards,
Gireesh