Advertising problems after change of chip revison (QFAAB0 -> QFAAE0)

[OlaHallstrom]

We have tested the same code on two boards with different revision of chips

nRF52832 QFAAB0  : advertising ok

nRF52832 QFAAE0 : advertising does not work

We are using SDK version 15.0.0 and softdevice S132. The hw schematic is identical when it comes to radio parts but differs in other aspects.

Are there any known compatibility issues between chip/SDK version? Tried to read erratas but its a bit confusing since chip revision is documented as numbers (1, 2, ...) when using letters on the chip package.

BR

/Ola

[Einar Thorsrud]

Hi Ola,

There are some errata that are fixed in revision 2 (QFAAE0), but none that should require any changes in your code, so I would not be surprised if your hardware changes are related to this. What are the most significant changes? Have you tried to debug to see what is happening on the device which is not advertising? What state is it in? I san error check hit, or is there a function not returning, or something else?

Einar

[OlaHallstrom]

Ok thanks. No I don't get any error codes returned (unless forgetting to pick them up). Advertisment start returns ok and looks exactly as when running on the older revision. I will involve our hw enginneeer in the loop as well. We will get back...

[Einar Thorsrud]

Hi Ola,

Sounds good.

Einar

[plamenbalk]

Hi Einar,

Thanks for your assistance.

The RF output is just the same as it has been during our latest 5 designs.

I did try to remove all impedance matching components, but still - we get no connection to the nRF connect app on the phone. than we have only a 1 cm PCB track with a 4 cm wire - at 5 cm distance on the "working" board I get -50 dBm, so the chip itself with this poor antenna shall work.

at the same time I measure current consumption on the board that do not advertise.

It looks good to me.

I use the power profiler board for the measurement. 

Similar current consumption is measured on the working board.

so - I am really confused.

The energy goes somewhere, but the messages are not read. Or - not read correctly.

Unfortunately my oscilloscope is not fast enough. And I do not have a spectrum analyser.

Just to eliminate the batch issue, I will ask the manufacturer to solder an IC from a previous batch.

Do you have any other suggestions?

What sniffer would you recommend to use as an option to catch any data?

Thanks.

Cheers / Plamen

[Einar Thorsrud]

Hi Plamen,

I see from the power profiler measurements that the nRF is clearly advertising. Since you are not able to receive the advertisements we must assume there is a HW issue.

I do not think randomly changing matching components is the way to go. Instead, since the first thing you should check is the 32 MHz crystal. Has that been changed? Which part number do you use now, and what is the value of the load capacitors?

Einar