Measure TX power & RX sensitivity

[dav01]

Hi, I know this topic has been up in several posts before, usually ends with, buy a LitePoint or Anritsu.

Assuming you want to do a poor mans measurement setup, just relying on a nrf52 being the measurement device,
what exact limitations would we get trying to measure tx power and rx sensitivity compared to a commercial test-system!?

let's say we calibrate (normalize) the text-jig once using a spectrum analyser, the nrf52840 can do -40dBm to +8dBm
something like a combination of a ~1mbps packet-rate measuring packet-loss on the other side (dropping tx power, test ad low/mid/high freq)

[run_ar]

Hi,

Is this for wireless or not? Either way you could probably use an attenuater together with the calibrated device lower the signal level. For TX ouputpower I think you can get ok results using rssi.

for RX sensitivity you need to lower the tx output of the tester to be close to the sensitivity limit. If you lower it so 8dBm out is close to 90 dBm in, you can probably do 1 dBm steps, use DTM or radio test firmware. Send a given number of packets and measure where it starts dropping packets. i.e. on the dut you could alter between tx/rx to indicate to the peer if it received packets or not.

What requirements do you have for test time and accuracy?

[dav01]

test-time can be "long", ok if the sequence takes 10-15 seconds

an attenuator around 50dB should be enough as I see it. needs to be tested in practice though and will be more sensitive in a 'non shield-box' environment.

[run_ar]

a 50 dB attenuator might be enough, just be aware that the accuracy at -40 dBm might be lower than for the output power setting from 0-8 dB. So maybe it's better to use more attenuation so you can measure sensitivity using the 2-8 dBm range. So use 2 or 0 for limit, then you can increase to up to 8 (maybe this gives to little range? Need to be tested).

I think bench measurement is likely to be fine, assuming this is for functional verification. I did some bench measurements using variable attenuation some time ago, and if I remember correctly this gave decent results for sensitivity measurements/verification. Packet loss is ramping up quite steeply, but if there are "busy" wifi blockers at the channel you are testing it will likely affect the result. Guess you could do a "sanity check" before testing by simply measuring the RSSI on the channel you want to test for a few seconds.