M24128-BRMN6P stopped acknowledging commands from microcontroller.

BGome · ‎2020-07-21

Hi,

We have been using the M24128 EEPROM IC on our access control products to store user data.

I have attached the schematic of the chip below. There are 1k ohms pull ups to 3.3V on the SDA and SCL lines which are not shown in the image.

Recently we found that one of out devices was malfunctioning in the field after operating fine for around 2 years. On debugging the hardware we found that the micro-controller was unable to communicate with the EEPROM. We saw that the commands were not being acknowledged by the EEPROM on the oscilloscope. The 9th bit of the clock cycle remained High. Have attached the waveform image below.

We did continuity test, checked resistor values, searched for shorts but the the hardware seemed alright. We thought something internal to the chip must have gone wrong.

But then accidentally while probing I touched the E2 pin to ground and the EEPROM started to acknowledge and communicate again. Going further we noticed that if we directly ground the any of the address pin E2, E1 or E0 the chip would respond.

But this doesn't make sense since the they are pull down by the 10K ohms resistor as always and used to work fine for 2 years. To verify we saw the voltage on those pins as 0V. But still if the direct short to GND is removed from the address Pins (but 10k pull down is there on all 3 pins) the chip stops responding again.

Can you help me in understanding why this should happen and in what way we can prevent this from happening in our other devices.

Thanks for you help

Best,

Brosnan

Andreas Bolsch · ‎2020-07-26

First: Did you observe input current on E_i pins during access cycle? Unusually high? If yes, this would suggest ESD damage.

The datasheet explicitly says that E_i inputs should be tied directly to V_SS or V_CC (footnote 6 and 7 below 'DC characteristics' and section 2.3 "These inputs must be tied to V_CC or V_SS ..." ).

Input impendance for E_i pins for V_IN < 0.3 V_CC is specified as >= 50 kOhms, so with a 10kOhms pull-down one might expect to be on the safe side at around 0.16 times V_CC.

BUT: The footnote "E2, E1, E0 input impedance when the memory is selected (after a Start condition)." suggests that these inputs are somewhat special (maybe transmission gates only activated after start condition).

Together with even little ground bounce (supply current increases by a factor of 1000 when selected) 10kOhms is rather tight.

S.Ma · ‎2020-07-26

after power cycle still the same?

seems related to esd shock or toomany write cycles?

BGome · ‎2020-08-04

Thanks for your response Andreas.

I tried to measure the input current of those pins but as soon as a touch the pins with my probe, The chip starts to work fine and keeps working fine as long as the probe is connected. Also since the pins are pulled down (via 10K), am I looking for current in the outward direction from the pin?

Regarding the datasheet comments, should I change the design to directly connect the enable to ground without any resistor pull down?

.,

Yes it remains the same after power cycle (not working condition).

Since the unit was in the enclosure for around 2 years (without any human intervention) and suddenly stopped working, how can an esd shock cause this, are there other ways.

Thanks

Brosnan

Andreas Bolsch · ‎2020-08-04

Yes, I think a direct connection either to V_CC or GND is a good idea as this is suggested in the datasheet. The point is the input structure is not clear from the datasheet. It does suggest that the E_i are not simple digital inputs, but no further details.

If there is an ESD damage, one would expect a rather high leakage current. In order to measure this without disturbing functionality to much, I'd replace the 10k by a 1k (to GND) and 9k1 in series and measure across the 1k resistor.