False trigger for gpio7.IO[13]

[yanissy]

Hi,

We have a circuit which is used to detect system undervoltage (scroll down for schematic screenshot).

The undervoltage signal (UV_WARNING) is being fed into the gpio7.IO[13] of iMX7 chipset.

Our software is using interrupt function to detect this UV_WARNING.

However, we notice this pin always give false interrupt whereby the pin voltage is not in high state even though we receive interrupt from this pin.

We measured the UV_WARNING signal using oscilloscope and found that it has noise level that less than 100mV which should not be triggering the interrupt.

We managed to get rid of this false interrupt by 

1. Removing R150 + adding 1kohm pull down in parallel with C129 + add another series resistor of 10ohm after C129, before going into iMX7. 

OR

2. Removing R150 + adding 1kohm pull down in parallel with C129 + reducing C129 to 220pF.

However, we still could not understand why did this happen.

Appreciate if someone can explain on thits.

Thank you.

Regards,

SY

[igorpadykov]

Hi Foo

issue may be caused by U101 slow transition times (rise/fall), also one can try to

set bit 3 HYS - Hysteresis Enabled in IOMUXC_SW_PAD_CTL_PAD_*, description

can be found in IOMUX Chapter of  i.MX 7Dual Applications Processor Reference Manual

Best regards
igor

[yanissy]

Hi @igorpadykov 

We did try to enable the hysteresis for this GPIO but the false interrupt still happened.

The noise level that we measured is less than 100mV at zero DC level, which is far way below the high state.

Therefore we could not understand why would this false interrupt happened.

Regards,

SY

[igorpadykov]

one can also check ground connections on board, so if there are analog and digital

parts their grounds should be properly connected, preferably in one point..

Best regards
igor

[yanissy]

Hi @igorpadykov 

Understand that it might pick up noise from ground but still couldn't understand why a 100mV noise would cause the trigger..it's almost zero voltage.

Thanks.

Regards,

SY

[igorpadykov]

this may be caused by poor soldering, please try to reproduce issue on NXP i.MX7D Sabre SD board.

Best regards
igor

[yanissy]

Hi @igorpadykov 

It should not be due to soldering issue as it happened to all of our PCBs.

The false interrupt disappeared when we removed C129 and it returned when we put back C129.

Anyway we managed to resolve the false interrupt just that it is still a myth to us why a mV noise would cause the trigger.

Thanks.

Regards,

SY

[igorpadykov]

>The false interrupt disappeared when we removed C129 and it returned when we put back C129.

>Anyway we managed to resolve the false interrupt just that it is still a myth to us why a mV noise would >cause the trigger.

removing C129 make rise/fall times faster, while internal hysteresis (HYS) seems not sufficient

to compensate it. Just for test one can try to check power supplies and add additiona filtering.

Best regards
igor

Best regards
igor