NRF21540 Shield + NRF52833 DK - lower RSSI with shield than without

[lbe]

Good morning everyone,

I need to develop a firmware for a custom board that features an NRF21540 + NRF52833.

I started the development using the official evaluation kits: an NRF52833 DK + the NRF21540 shield.

I started from the example BLE peripheral "nrf_uart", which I compiled for the board nrf52833_nrf52833 and to which I added the -DSHIELD=nrf21540_ek

as indicated in the section "Working with FEM". I used two SMA 2.4GHz antenna with the nrf21540_ek and I used the nrf_connect smartphone app to

record the RSSI of the advertising.

It turned out that if I use the 21540 shield, I got a slightly worse RSSI than if not using the shield (I created a separate build configuration without using the -DSHIELD option).

I have about -55dBm without shield (about 1mt distance) and between -60 and -65dBm with the shield.

Clearly, I put the board and my smartphone in the same conditions and distance.

So, apparently, the 21540 is not used or worst, is badly affecting the signal....

Do you have any clue why this happens? Should I initialize it somewhere in the main.c source code of the uart app to make it work?

I have a fair experience with nrf52 using the old SDK, but I am rather new to the nrf_connect and zephyr environment, so it is possible that I am missing something really basic....

Thanks in advance for your help!

Lorenzo

[JONATHAN LL]

What version of the nRF Connect SDK(NCS) are you using? there is a bug in 2.00 -2.2.99.
This PR should fix
https://github.com/nrfconnect/sdk-nrf/pull/10120/files#diff-9a498e7310940c0b624cfbe919175b9cd29b76a33bf89e703376229ec8ac5c24 


You could also share the two hex files with us and we can test to see if we get the same result on the same setup.


What version of the nrf52833DK and 21540EK are you using?

Regards,
Jonathan

[lbe]

Hello,

sorry for the late reply but I have been busy on other things.

Yes, I am in fact using NCS 2.2.0 (latest stable at the time of this writing), so I believe I am affected.

I tried to modify the indicated files following the PR you linked in my local installation, but I still get the exact same behaviour: lower RSSI with the -DSHIELD option than without it.

I am using a shield version 1.1.0 (2021.33) and a NRF52833 DK version 1.1.0, 2022.6

I am linking the two hex files as you requested; as I said I did not modify at all the example application bluetooth_peripheral_Uart, but I just tried to compile it for the 52833 + 21540 shield.

Thanks for your feedback, I hope we can figure this out.....

Best,

Lorenzo

zephyr_Shield.hex

zephyr_NoShield.hex

[JONATHAN LL]

Hi Lorenzo, 

Did some testing and you need to add ```CONFIG_BT_CTLR_TX_PWR_ANTENNA=20 ``` to the prj.conf file as well as the fix from the PR. 

Did test with your files and did a clean build wit no fix, then added the fix and then added the CONFIG_BT_CTLR_TX_PWR_ANTENNA=20 and only when i add the antenna power that is when i am able to see any notable improvement in signal strength. So try that. 

Also, was comparing the RSSI values that is present when the device is connected  might not show a huge improvement but if you look at the values of the advertisements then there is a larger jump in signal strength. 

Red circle is the values i get with the bug or the fix but wit out the prj.conf change. Blue circle is the values after the config addition to the fix. 


Regards,
Jonathan

[lbe]

Hello Jonathan, 

thanks for your help. Actually, I don't see any improvement in the RSSI with or without the shield, even when adding the parameter you mentioned....but at least I don't see the signal getting worst anymore, which is already a good result for now.

Maybe I am doing something wrong in setting up the hardware? I put a generic 2.4GHz antenna in the SMA connector indicated as "ANT 1" in the 21540 EK, than I plugged it to the 52833 in the only possible direction.....Interestingly, if I try putting two antennas (ANT 1 and ANT2) I got a worse RSSI.

In the current situation, if I use a build configuration with the DSHIELD option, I get almost the same RSSI values as the configuration without the DSHIELD option. I tried both by leaving the shield on and removing it in case of the no shield option, I also tried to remove the TX_POWER option from the prj.conf when building for the NO_SHIELD version, but I still get almost identical RSSIs. 

I should be able to find another shield and another 52833DK....I will try with those, just in case I have a defective board.

If you have other ideas, they are welcome....

[JONATHAN LL]

lbe said:

.Interestingly, if I try putting two antennas (ANT 1 and ANT2) I got a worse RSSI.

I will test this out, with a shield that matches your and some different ones. 


Do you have a different nrf52 board that you can use with the RSSI viewer app ? Then you can try with the radio test sample and see if that works as it should or not. 

Regards,
Jonathan

[JONATHAN LL]

Trying to replicate your results but I am not able to get the same results. 


Testing the "adding an extra antenna to the EK worsens the result" situation but I do not see any change in the that I can prove to be due to the added antenna on the ANT2 port. 

And if you only use RSSI to check the strength of the connection it does seem like it does not improve with the code that has shield enabled. So there I do see the same behavior that you see, only comparing the code with the fix and shield and then unplugging vs plugging in the shield. But using the radio test sample or if I am using the RSSI Viewer app then I do see that signal strength is higher when the shield is connected. So just checking with the phone does not provide the best picture. 

I did test two different versions of the shield, both v1.1.0, but one made in 2021, week 33 and on made in 2022, week 8. They did preform a little different when measuring just RSSI. I have not tested the two version in the lab to see how different they perform as there is often some silicon variation from chip to chip\batch to batch. 

Regards,
Jonathan

[Muhammad Usman]

Hi Jonathan LL  and Lorenzo, 

Thank you for the nice discussion. I have the same working environment and I followed all your instructions as it is. the only difference is that I am using nrf21540 DK.

In the second, case I also have the custom board which is nrf52833+nrf21540. 

I am facing the same issue. Can you please help me in this regard?

Thanks 

Muhammad Usman

[JONATHAN LL]

Hi Muhammad Usman , can you create a new ticket and reference this ticket and if you want you can name me and I will take the ticket. 

You can feed in additional info from the following questions:

Are you getting bad results with the nRF21540 DK? or is that your reference when testing with the nRF52833DK+nRF21540 Shield, and that is what performs bad?

Have you also tested with the RSSI Viewer app from nRF Connect for Desktop?

Are you using the latest version of NCS or version 2.2.0 as Lorenzo or something else?

Regards,
Jonathan