ADP5585 going BAD after few days usage.ESD ISSUES?

Thanks RSchnell for your reply. Below replies from my side to your questions

1. Regarding - Whether the part operate correctly next time it is powered up - Yes , 2-3 times when the KEYPAD stopped working we powered OFF the board and checked for short between any of the ROW/COL lines to GND we did see about 120 ohm or 30-40 ohms to GND. After leaving the board OFF for same time when the board was powered ON again, KEYPAD continued to operate normally. However after still more continued usage of the board the ROW/COL lines developed a permanent short to Ground.

2. Regarding Voltage  from CPU - yes it 3.3V.

3. Regarding KP_12V_OFF - It is connected to gate of another FET which is again driven only from 3.3V level signals and not any higher. As well since there is a zero ohm in the path of that signal we have isolated that signal and we have seen the problem occur even under that situation as well

4. Regarding the capacitors - yes our plan is to keep them un mounted itself.

5. Regarding chasis_GND and IC ground - They become same once the system is mounted inside the metallic enclosure we have the chasis_GND gets shorted to IC ground , however yes there is a possibility that in our initial units when the testing was being done chasis_GND might be a floating node. Do you think if this Node was floating it could have caused the malfunction of IC as reported above.

6. Regarding voltage spikes getting coupled to ROW and COL pins - I do not see that possibility in the way in our system because the only other cable that is close proximity to the KEYPAD cable is LCD DISPLAY cable and a USB cable  which have voltages in the region of 5V

Further questions from my side

A. What is the transient voltage limit on the ADP5585 pins because the external ESD protectors we are using may have clamping voltages in the region of 10-15V for few ns. You can look at the datasheet of ESD protector from TI that we used - web link for same was provided above. We are also planning to change the ESD protector to that from NXP since it has even higher series resistor of 200 ohm and a higher capacitance of 15pF (compared to 100 ohm series resistor and 8.5pF capacitance in TI part). The selected NXP P/N is IP4253CZ16-8-TTL.

B. Do you recommend any other specific ESD protector to be used ?

B. Do you think if chasis_GND was floating then ,  it could have caused the malfunction of IC as reported above?

The ADP5585 family has very good ESD protection within the part, and we haven't seen ESD issues that I know of.

A 15 inch ribbon cable can sometimes come with a lot of capacitance, which could slow down the polling on the keypad key scans, but this would not cause the part to short to ground.

When the failure happens, does the part operate correctly the next time it is powered up, or is the pin permanently weakly shorted to ground?

Just to make sure, are the voltages from the CPU also 3.3 V? What is the KP_12V_OFF line connected to?

C2,C4,C6,C8, C67,C68,C69, and C74 should remain unpopulated as they aren't necessary, and can actually interfere with the key polling.

Is there any chance the chassis ground is different from the IC ground? Could there be a large voltage spike that could couple into the row and column pins somehow?

RSchnell

Hello,

A. The abs. max for the row and column pins on the ADP5585 is -0.3V and (VDD + 0.3 V).

Having excursions above this risk turning on internal body diodes, and repeated excursions can cause damage. These limits are very common as almost any CMOS design will have these body diodes in this configuration. If you scope very close to the part, the voltage excursions might be very hard to notice, as the body diodes will clamp the voltage to about these levels as they turn on. If voltage spikes are expected in a system, sometimes schottky diodes clamping input pins to VDD lines works to prevent this possible damage. The cathode of the schottky would be on VDD and the anode would be on the pin being protected. There can also be a schottky from the pin being protected to clamp to ground in case of a negative voltage.

This probably isn't feasible in your final design, but it adding schottky diodes on all the row/col pins might help find the root cause of the failures.

B. The ADP5585 was originally designed for very compact keypads with short distances between the keypad and the part. It was designed to not need external ESD devices in these cases. The ADP5585 works fine with longer cables, as long as the total line capacitance is low. If the capacitance is too high, the symptom would be multiple key presses being registered for single key press event, so it looks like your cables are fine capacitance-wise. The question is whether the long cables are somehow picking up voltage transients large enough to affect the part. The ESD structure you have right now does not have a clamping voltage low enough to prevent excursions above abs. max. One possible thing to try is putting 3.3V zener diodes on the row/col pins right near the part, anode on GND.

B(C). The hypothesis about the chassis floating was about possible mechanisms that would cause voltage transients getting coupled in to the long cable lines. It is possible if the chassis ground were dramatically different than the system ground, and then they were shorted that a large spike would capacitively couple, but that is not very likely.

These failures feel more like an over voltage situation than an ESD failure.

Is the ESD protection very close to the ADP5585, or is it on the other side of the cables? Closer to the part is the best location.

RSchnell

I should add a little clarification. By excursions above abs. max, I mean long sustained events. Short high voltage spikes like ESD events do not persist long enough to damage the body diodes. The body diode damaging events are much slower and persistent.