https://devzone.nordicsemi.com/f/nordic-q-a/57337/how-to-isolate-a-broken-i2c-chip-and-keep-the-system-runningHi, I&#39;m using nrf52832 (SDK 15.3). And I have 24 chips on the same I2C bus. They all work fine right row. But if one of the chip is broken, the whole system will crash. Is there a good way to detect which chip is dead and isolate it, so that the systemen-USTelligent Community 13Tue, 04 Feb 2020 13:01:22 GMThttps://devzone.nordicsemi.com/thread/232564?ContentTypeID=1Tue, 04 Feb 2020 13:01:22 GMT137ad170-7792-4731-bb38-c0d22fbe4515:1109ddb9-3ee2-479e-8c70-4e79e4fb32edDmitry<p>A method suggested by Kenneth will help in a most lucky case when some chip just stopped to respond. But if a broken chip locks the bus, you have no way to isolate it.&nbsp;</p> <p>In one of my projects (there was an i2c bus with hot-plug components), in case of any bus error I&#39;m sending 16 CLK pulses then STOP condition, then scanning for topology changes. Though it doesn&#39;t help when attaching a broken device.</p><div style="clear:both;"></div>https://devzone.nordicsemi.com/thread/232536?ContentTypeID=1Tue, 04 Feb 2020 12:11:05 GMT137ad170-7792-4731-bb38-c0d22fbe4515:eccc4430-7f8d-46f0-9a1a-a895f00aeccaKenneth<p>Hi,</p> <p>I suggest to simply add/start a timeout timer to each twi transfer operation, then in case of timeout you can uninit() and init() the twi peripheral again. Then you can choose to try again a few times before you&nbsp;consider it as &quot;dead&quot;.</p> <p>Best regards,<br />Kenneth</p><div style="clear:both;"></div>