diagnosing 9160 crash due to hardware error

[paul_tanner]

I have a working (hand-wired) prototype that connects to the internet without any problem.

Now I have made a PCB to replicate the same thing and loaded the same software.

Evidently there is a wiring error because the application crashes to the point where connection is made (lte_ready event).

Just before it crashes I get "E: Could not get GPIO Device Binding"

I am looking for the source of this error but cannot find it. (I test each GPIO binding before trying to activate LTE.)

Any suggestions as to where I might look?

[paul_tanner]

Thanks.

It's not from that board.c file as that would show the error with a module prefix.

I'm using NCS 1.5 and have carefully reviewed the schematic.  I suspect a short or something.

The devicetree is correct as it works perfectly on the earlier prototype.

Regards.

[Øyvind]

Thanks for clarifying. It does sound like you need to review the schematic, and debug the HW. 

Kind regards,
Øyvind

[paul_tanner]

Correction.  That (2 x 4.7uF) usually fixes it.  Any reason why this should be?

[Øyvind]

Can you please provide more information on what board you are using? Are you using the Actinius Icarus? 

From the Product Change Notification (PCN134) for nRF9160 rev 1 (B0) to rev 2 (B1) it states:
SiP external cap, DEC0, to be changed from 47uF to 4.7uF

[mike_actinius]

Hi,

This is not about the Actinius Icarus and of course all our products have the correct capacitors and are well-tested designs.

The customer has been in contact directly with us as well and this is about a custom board with 1 boost DCDC and 1 buck DCDC in series, powering both the VIN and VCC of a SoM, our solderable module with the nRF9160. We have advised the customer to add more capacitance to his design at the point of feeding VIN (initially there was 100nF which for most applications is fine for our SoM).

Kind regards,

Mike

[paul_tanner]

Thx. 

Yes.  Initial testing was on B0 Actinius SoMs.  That worked reliably.

Then I got the next batch of B1 SoMs which initially failed consistently (rebooted on lte_ready event).

I saw that PCN and had already changed to 4.7uF on both VIN and VCC.  However, these were connected together.

Actinius suggested larger capacitors which has not been a reliable solution.

Finally, I disconnected these 2 pins (annoying as this will mean another rev of the PCB) and took VIN straight from the LiPo cell.  This works reliably in my tests so far.

Any observations?  (The LiPo cell gets up to 4v when fully charged.)

Regards.

[Øyvind]

mike_actinius and paul_tanner, as mentioned earlier:

Øyvind said:

It does sound like you need to review the schematic, and debug the HW. 

As this ticket does not mention any details about the PCB other than nRF9160 and "based on working (hand-wired) prototype", it is very difficult for me to provide any good answers. From my side I will refer to the Product Specification and reference circuitry.

paul_tanner said:

I have tried increasing the capacitors on VIN and VCC (as the Actinius doc calls them) 4.7uF

The reason for asking about Actinius, was based on the comment above, as I am not aware of the products from Actinius, other than the Icarus.

Please issue a new ticket for a HW design review.

[paul_tanner]

Thx Øyvind,

If I decide to use the nrf9160 directly, the reference circuitry will be a great help, as would be a design review.

However, since I am using the Actinius SoM on the current design I had better continue this discussion with them.

Regards, Paul

PS.  They agreed that tying both power pins together might not be the best solution.  I will stick with powering VIN from the LiPo cell.

What makes this difficult is that the Icarus SoM is closed-source so I can't show the full circuit here.