%XRFTEST: conducted test for tx power levels - not getting expected results

[amsIoTdeveloper]

Setup:
I am doing a conducted test with the nrf9160 connected directly to a Signal Hound SA44B spectrum analyzer.
I have a dc block and 30dB attenuators installed to protect the spectrum analyzer. 

Pin 61 is using the same components as the nrf9160 Feather design

Test:

Command I'm using to test max power is:

AT%XRFTEST=1,1,12,7000,23,1,1,6,0,0,3,3,0

Command I'm using to test a lower power level is:

AT%XRFTEST=1,1,12,7000,10,1,1,6,0,0,3,3,0

For the max power test the nrf9160 is returning 276/16 = 17.25dBm 

Here is my spectrum analyzer screenshot of the test measuring -19.39dBm

Questions:

1. Why isn't nrf9160 returning a transmit value for the max power 23dBm as expected by the command? 

2. I have done several different tests with/without attenuators to try figure out why my spectrum analyzer is reading about 6dBm lower than the nrf9160 is telling me on B12.  Why is my spectrum analyzer reading 6dBm lower than the nrf9160 response value?  On some other bands it is even 10 to 20dBm difference.

2. I see the nrf9160DK is using different components and topology on pin 61 for the antenna so which reference design is the best one to follow? 

3. We are sending a lot of systems to the fringe which is why I need max power so what is a good baseline test and ideal results so I can conclude this PCB is as good as the nrf9160 can be for fringe operation? 

I would really appreciate some help to try better understand how I can maximize the performance of the nrf9160 for operation on the fringe. 
Thank you.

[mathiaso]

Hi, Gary!

Before we begin a proper investigation on the issue, could you try to remove the matching components between the ANT pin and the connector (remove L9 and C34, short C35)? As both are supposed to already be matched to 50 ohms.

Best regards,
Mathias

[amsIoTdeveloper]

Mathias, thank you for a quick response.  I have done the test and the results are very encouraging!! 

Same test setup as OP 
L9 and C34 removed and C35 shorted
so now it's just a 4 Layer PCB trace designed with 50Ohm impedance 

AT%XRFTEST=1,1,12,7000,10,1,1,6,0,0,3,3,0

nrf9160 response: 154/16 = 9.625dBm
Spectrum analyzer: 9.5dBm 

AT%XRFTEST=1,1,12,7000,23,1,1,6,0,0,3,3,0

nrf9160 response: 328/16 = 20.5dBm 
Spectrum analyzer: 20.41dBm

So I'm very happy to see my PCB impedance design is basically perfect. 

I also tested B5 and B2 with same results

AT%XRFTEST=1,1,5,8400,23,1,1,6,0,0,3,3,0

nrf9160 response: 323/16 = 20.19Bm
Spectrum analyzer: 20.1dBm 

AT%XRFTEST=1,1,2,18500,23,1,1,6,0,0,3,3,0
nrf9160 response: 329/16 = 20.56Bm
Spectrum analyzer: 20.43dBm 

I'm satisfied that resolved question 2.

So my biggest question now is this all I need for production and will this pass compliance when we get to that stage or do I still need some matching components on pin 61?  What does Nordic recommend as best practices for a reference design because I'm seeing huge variations from one reference design to the next. 

Sorry I see my OP question numbers had a duplicate question 2, but I'm hoping you can help me with the other 3 questions please? 

[mathiaso]

Hi again, Gary!

I'm happy to hear that it helped you!

amsIoTdeveloper said:

So my biggest question now is this all I need for production and will this pass compliance when we get to that stage or do I still need some matching components on pin 61?  What does Nordic recommend as best practices for a reference design because I'm seeing huge variations from one reference design to the next.

Ideally, you shouldn't need any matching components, as both parts already are supposed to be 50 ohms. However, theory and practice might deviate from each other and thus, making the inclusion of matching components favorable. I will have to double-check with some of my colleagues to hear if it would be wise of you, based on prior experience, to spend time on finding and creating a matching network and if this would be able to bring you closer to an output power of 23 dBm.

2. I see the nrf9160DK is using different components and topology on pin 61 for the antenna so which reference design is the best one to follow?

This matching network is for the antenna only and will vary between antennas, layouts, casings, etc.

3. We are sending a lot of systems to the fringe which is why I need max power so what is a good baseline test and ideal results so I can conclude this PCB is as good as the nrf9160 can be for fringe operation?

I guess that performing tests where you see that the conducted output power is as high as possible, in addition to checking that your antenna has a match that is as close as possible to 50 ohms, would be good indications. Will let you know if I get some more input on this.

Best regards,
Mathias

[amsIoTdeveloper]

Mathias:
Thank you for clarifying these items for me, it is greatly appreciated. 

So if I'm understanding correctly, the nrf9160 is already impedance matched for a 50Ohm antenna and connectors so as long as I use good quality 50Ohm antennas with low SWR then no other matching components are needed on pin 61?  

I understand there is already an internal TVS on pin 61 however an additional external one is also beneficial for additional protection. 

That seems to  be the solution for the plan to move forward.  Please confirm.

Thank you,
Gary 

[mathiaso]

Hi, Gary!

Apologies for my delayed response. I've been out of office for a few days this week.

amsIoTdeveloper said:

So if I'm understanding correctly, the nrf9160 is already impedance matched for a 50Ohm antenna and connectors so as long as I use good quality 50Ohm antennas with low SWR then no other matching components are needed on pin 61?

Yes, that's what I meant. In theory, at least, and in most cases you shouldn't be too far off. However, there is always an advantage to be able to tune your antenna if it should be needed, as this may depend on casing etc. I don't know if you've had the chance to look at the Antenna and RF Interface Guidelines yet?

I'm sorry, I'm sure this is on me. But could you please clarify what you mean by internal TVS?

Best regards,
Mathias