How to production test output power of 24L01+ module

Question

Does anybody have a suggestion for a relatively low cost way to test the output power of a nRF24L01+ module? We make the modules (which are identical to the demo module) in batches of approx 5000 and we have a tester which uses a second identical module and some firmware to both transmit and receive using the module under test. If all packets are sent and received okay, you get a green light. 
 We are seeing a small number of radios that pass the tests okay but actually have a very poor range when used as transmitters, around 30cm. Visually they look fine so we end up using them in production and we are seeing a few returns here and there for this reason. 
 We need to expand the tests we run to include output power in tx mode. I am experimenting with RPD and am having difficulties getting anything meaningful from it, so I'm wondering whether I should consider some other method but don't want to have to spend a huge amount on a spectrum analyser. Does anybody have any suggestions on a test that is suitable for a medium scale manufacturing environment? 
 Many thanks 
 Dave

Øyvind Karlsen · Accepted Answer

Hi Dave, 
 While I don't have any experience with large scale production I'll try to answer what can be done on single system level. 
 Typically when we see systems that have poor range there are two key things that went wrong: 
 
 Antenna is not matched to 50ohm, this means that all of the power we send into the antenna is not getting transmitted, and all the effect we receive through the antenna is not getting received by the rest of the circuit. 
 
 The way we test this here is to: 
 
 Measure the antenna impedance with a network analyzer and then make adjustments with a pi-network. 
 Once we have gotten the antenna to 50ohm we will remove the antenna, and make measurements on the signal coming from the radio. Here we check that the power is what we expect it to be using a spectrum analyzer, we also check that the noise on the harmonics is acceptable. 
 
 
 
 The layout is not ideal, poor grounding, wire wound inductors etc. 
 
 There's a lot that can be wrong here, I recommend sending in your layout for review, either here on DevZone or on the private support portal mypage . 
 
 Since you're in production I assume you've gone through these hoops already, for final testing we have the following application notes(be aware that they assume access to a spectrum analyzer): 
 
 RF Performance Test Guidelines 
 RF Test Instructions nRF24L01+ 
 
 Even though they are based on spectrum measurements, they give an idea of what needs to be checked. 
 Without a spectrum analyzer you will have to get creative, it might be possible to do a qualitative test using a dongle and verifying that they can actually achieve the range that you expect. 
 Best regards, 
 Øyvind