0
\$\begingroup\$

I have custom PCB with nRF52832 MCU and two I2C devices sharing same bus, IAM20380 on 0x68 and IAM20381 on 0x69.

All works fine, but after a few hours of testing I2C interface freezes when I2C command RX or TX is requested. This strange behavior lasts event after power-cycling and MCU reset. I have also tried discharge potential parasitic capacitance with connecting SDA and SCL to GND.

This behavior can be repaired by reheating the PCB again even with temperature which is lower than melt-point of used solder-paste. After reheat all works fine. I2C interface is running on 400kHz and SDA and SCL lines have about 1.5cm in length and both lines have 10kΩ internal pull-ups. PCB is running on approximately 3V. brdsch

\$\endgroup\$
6
  • 3
    \$\begingroup\$ A schematic would be nice \$\endgroup\$
    – Voltage Spike
    Sep 5 at 18:31
  • 2
    \$\begingroup\$ You didn't ask a question in your question. \$\endgroup\$
    – TypeIA
    Sep 5 at 18:35
  • \$\begingroup\$ Sounds like you didnt clean the flux and it misbehaves. Can we have a picture of the physical board? \$\endgroup\$ Sep 5 at 20:48
  • 3
    \$\begingroup\$ 10k pull-ups with 400 kHz? Sounds way too high resistance for that speed. Are you sure it is the reheating that fixes the issue, or just the time it takes to reheat the PCB, which would allow for all kinds of power supply capacitors to discharge enough? Post the schematics to see what else is suspicious, besides the 10k pull-ups. \$\endgroup\$
    – Justme
    Sep 5 at 21:20
  • \$\begingroup\$ I have added pictures of both board and schematic. \$\endgroup\$
    – Pepam
    Sep 6 at 6:23
1
\$\begingroup\$

This strange behavior lasts event after power-cycling and MCU reset.
This behavior can be repaired by reheating the PCB again even with temperature which is lower than melt-point of used solder-paste.

Reheating, cooling to freeze, twisting the PCB makes the circuitry work momentarily, but breaks eventually. That is a popular problem due to marginal PCB design and assembly.

Since your circuitry works when heated to the melting-point, it is likely a mechanical failure as described. The IC pins to PCB contacts are insecure, or the PCB routing clearance is not enough for the assembly quality. That, often, is the case with fine pitch boll-grid devices in the circuitry.

You can ask the assembly house to examine the PCA under X-ray.

\$\endgroup\$
4
  • \$\begingroup\$ The OP states "... even with temperature which is lower than melt-point ...". \$\endgroup\$
    – brhans
    Sep 5 at 21:57
  • \$\begingroup\$ @brhans , Yah I know. I was exaggerating about heat. :-) If solder balls get melt under the devices, there is a lot more chance to mess it up than re-setting it. \$\endgroup\$
    – jay
    Sep 5 at 23:15
  • \$\begingroup\$ @jay Can you please take a look at schematic and board attached to the post? \$\endgroup\$
    – Pepam
    Sep 6 at 6:33
  • 1
    \$\begingroup\$ @Pepam It looks like VDDIO (of both devices) is floating. It has to be tied to the microprocessor VCC (IO). BTW, is it 4mil clearance? Looks closer than 6mil. \$\endgroup\$
    – jay
    Sep 6 at 14:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.