Modem library initialization failed, error: -5

[egood]

Hello,

We are currently bringing up a new hardware design that uses the nrf9160. As part of this bringup we load the out of the box at_client sample so that we can write certificates to the device. However on boot we are getting "Modem library initialization failed, error: -5"

I think this translates to: NRF_EIO Modem firmware update failure. Modem must be reprogrammed. (from nrf_modem.h)

The modem also doesn't respond to AT commands. 

We have attempted to re-program the modem with both v1.3.6 and v1.3.5 to see if it would have an effect, but both have the same results.  

Can you all help provide some additional insight into what this error means? And possibly some further debug steps?

We are using the following:

nrfConnect SDK v2.5.0

NRF9160-SICA  Rev B1

Thanks!

[Håkon Alseth]

Hi,

 

There was some changes from revision 1 to revision 2, as outlined in this document:

https://infocenter.nordicsemi.com/pdf/pcn_134_v1.0.1.pdf?cp=2_1_2_6

 

One crucial change is the capacitance on DEC0, which needs to be 4.7 uF now. Can you verify that this component is 4.7 uF on your design?

 

Kind regards,

Håkon

[egood]

Yes DEC0 is bypassed with a 4.7uF capacitor. 

[Håkon Alseth]

Thank you for confirming.

 

Q1: How many devices are behaving like this?

Q2: Have you tried flashing for instance mfw v1.2.x version to see if changes the behavior?

Q3: If there is only a few devices showing this behavior, have you tried reworking the nRF9160 (warming it up) to see if this could be caused by a bad solder joint?

 

Kind regards,

Håkon

[egood]

1. We have 5 total in the prototype batch, 4 of the 5 are having this issue. 

2. I loaded modem FW v1.2.2 and got the same result.

3. We have tried reworking one unit via hot air. However, the SiP is a very difficult package to rework. Our first attempt did not yield any difference. However, we are still planning to have some of these units X-rayd for solder quality. 

Although this may be a solder quality issue, I still want to understand the areas we should be looking at. It sounds like DEC0 is possibly one of the pins to take a look at. I have checked the voltage on the 4.7uF DEC0 capacitor under operation and it seemed to be in the same range as other boards we have that are functional. 

What other hardware related items could we check while we wait on the X-ray? 

Thanks

[Håkon Alseth]

Hi,

egood said:

Although this may be a solder quality issue, I still want to understand the areas we should be looking at. It sounds like DEC0 is possibly one of the pins to take a look at. I have checked the voltage on the 4.7uF DEC0 capacitor under operation and it seemed to be in the same range as other boards we have that are functional. 

What other hardware related items could we check while we wait on the X-ray? 

That is correct. DEC0 is critical for the startup of the nRF9160. As for other signals, it is hard to pin-point exactly what has gone wrong in this case.

Have you checked your voltages and current consumption of your board as a whole compared to the working sample? That can give an indication.

 

Kind regards,

Håkon

[egood]

We have checked the voltage and current consumption and it seems normal. Voltages are all nominal ~3.3V DEC0 is ~2.2V, current consumption was showing <5mA which is normal given our idle state.  

[egood]

We have received X-ray images back from our supplier and I would appreciate your opinion on if the soldering looks acceptable for the nrf9160. per IPC-A-610 we should have <25% voiding. To me a lot of these joints look questionable, but I'm not an expert on x-ray inspection images.

 

[Håkon Alseth]

Hi,

 

The voids in the middle are not too alarming, in terms of functionality of the device.

The lower left corner seems to have two points, which looks like pin 69 (GND_SHIELD) and pin 70 (RESERVED) have little solder. These two pins should not have such a great impact on the boot-up of the nRF9160.

 

There looks like some of the reserved pins are shorted?

It is a bit hard to tell based on the picture, but that can be a problem if they are.

 

Kind regards,

Håkon

[egood]

I see what you are saying, that for sure looks like they are shorted. Would this for sure cause the modem not to boot? For reference that would be pins 120,121,122,123. 

We will address this with the supplier regardless, but I would like to understand if this is the smoking gun /root cause for the boot issue.

Thanks!

Eric